Special Issue S1 (93) 2017
Vol. 24
ADDRESS OF PUBLISHER
& EDITORS OFFICE:
GDASK UNIVERSITY
OF TECHNOLOGY
Faculty of Ocean Engineering
& Ship Technology
G. Narutowicza 11/12
80-233 Gdask, POLAND
tel.: +48 58 347 13 66
fax: +48 58 341 13 66
EDITORIAL STAFF:
Jacek Rudnicki
| Editor in Chief
Rafa Szapczyski
| Associate Editor
Jerzy wirydczuk
| Associate Editor
Przemysaw Wierzchowski
| Associate Editor
Aleksander Kniat
| Editor for international
relations
Price:
single issue: 25 z
Prices for abroad
single ussue:
- in Europe EURO 15
- overseas USD 20
WEB:
pg.edu.pl/pmr
e-mail : pmr@post.pl
ISSN 1233-2585
Coastal, Offshore
and Ocean Engineering
4 Lech Baachowski
Physical modelling of geotechnical structures inports and
offshore
10 Katarzyna Bobkowska, Adam Inglot,
Miroslava Mikusova , Pawe Tysic
Implementation of spatial information for monitoring and
analysis of the area aroundthe port using laser scanning
techniques
16 Marcin Budzyski, Dawid Ry, Wojciech Kustra
Selected problems of transport in port towns Tri-city as an
example
25 Jacek Chrcielewski, Mikoaj Mikiewicz,
ukasz Pyrzowski, Krzysztof Wilde
Composite GFRP U-shaped footbridge
32 Marian Cichy, Zbigniew Kneba, Jacek Kropiwnicki
Causality in models of thermal processes in ship engine rooms
with the use of Bond Graph (BG) method
38 Mariusz Deja, Mieczysaw Siemitkowski, Sender P.
COMPARATIVE STUDY OF MACHINING TECHNOLOGY SELECTION TO
MANUFACTURE LARGE-SIZE COMPONENTS OF OFFSHORE CONSTRUCTIONS
46 Mariusz Deja, Micha Dobrzyski, Mieczysaw
S.Siemitkowski, Aleksandra Winiewska
SIMULATION STUDIES INTO QUAYSIDE TRANSPORT AND STORAGE YARD
OPERATIONS IN CONTAINER TERMINALS
53 Czesaw Dymarski, Pawe Dymarski, Aleksander Kniat
Searching for critical conditions during lifeboat
launching simulations
59 Czesaw Dymarski, Pawe Dymarski, Jdrzej ywicki
Technology concept of TLP platform towing and installation
in waters with depth of 60 m
67 Pawe Geesz, Artur Karczewski, Janusz Kozak,
Wojciech Litwin, ukasz Pitek
Design methodology for small passenger ships on the example
of the ferryboat MOTAWA 2 driven by hybrid propulsion system
ADDRESS OF PUBLISHER
& EDITORS OFFICE:
GDASK UNIVERSITY
OF TECHNOLOGY
Faculty of Ocean Engineering
& Ship Technology
G. Narutowicza 11/12
80-233 Gdask, POLAND
tel.: +48 58 347 13 66
fax: +48 58 341 13 66
74 M. Hossein Ghaemi, Henryk Olszewski
Total Ship Operability Review, Concept and Criteria
82 Radosaw Gleba, Sawomir Grulkowski, Jerzy Zariczny
EFFECT OF DIFFERENT STRUCTURE TYPE TRAFFIC ON RAILWAY LINE
CAPACITY
89 Sambor Guze, Tomasz Neumann, Przemysaw Wilczyski
MULTI-CRITERIA OPTIMISATION OF LIQUID CARGO TRANSPORT
ACCORDING TO LINGUISTIC APPROACH TO THE ROUTE SELECTION TASK
97 Rafa Hein
A discrete-continuous method of mechanical system modelling
108 Artur Janowski, Jacek Rapinski
THE ANALYZES OF PDOP FACTORS FOR A ZIGBEE GROUND BASED
AUGMENTATION SYSTEMS
115 Krzysztof J. Kaliski, Marek A. Galewski,
Micha Mazur, Marek Chodnicki
MODELLING AND SIMULATION OF A NEW VARIABLE STIFFNESS HOLDER
FOR MILLING OF FLEXIBLE DETAILS
125 Maciej Kashin
Numerical modelling of structures with uncertainties
133 Dariusz Mikielewicz, Jan Wajs
Possibilities of heat transfer augmentation in heat exchangers
with minichannels for marine applications
141 Dariusz Mikielewicz, Blanka Jakubowska
Calculation method for flow boiling and flow condensation of
R134a and R1234yf in conventional and small diameter channels
149 Mikoaj Mikiewicz, ukasz Pyrzowski,
Krzysztof Wilde, Oskar Mitrosz
Technical monitoring system for a new part of Gdask Deepwater
Container Terminal
156 Karol Niklas
Strength analysis of a large-size supporting structure for an
offshore wind turbine
166 Rafa Ossowski
Environmental aspects of coastal earth structures made of
soil-ash composites
174 Zygmunt Paszotta, Malgorzata Szumilo, Jakub Szulwic
Internet photogrammetry for inspection of seaports
182 ukasz Pyrzowski, Mikoaj Mikiewicz, Jacek Chrcielewski
THE EFFECT OF FISHING BASIN CONSTRUCTION ON THE BEHAVIOUR OF
A FOOTBRDGE OVER THE PORT CHANNEL
188 Grzegorz Rogalski, Dariusz Fydrych, Jerzy abanowski
UNDERWATER WET REPAIR WELDING OF API 5L X65M PIPELINE STEEL
195 Lech Rowiski
Living and working beneath the sea next approach
203 Jacek Rudnicki, Ryszard Zadrg
Technical state assessment of charge exchange system of self-
ignition engine, based on the exhaust gas composition testing
213 Pawel Sliwinski
The influence of water and mineral oil on volumetric losses
in a hydraulic motor
224 Piotr Zima
Investigations on water circulation in animal sea-water
basins on the example of seals breeding pools
230 Jdrzej ywicki, Pawe Dymarski, Ewelina Ciba, Czesaw Dymarski
Design of structure of Tension Leg Platform for 6 MW
offshore wind turbine based on FEM analysis
 Editorial
POLISH MARITIME RESEARCH is a scientic journal of worldwide circulation. The journal appears
as a quarterly four times a year. The rst issue of it was published in September 1994. Its main aim is to
present original, innovative scientic ideas and Research & Development achievements in the eld of :

Engineering, Computing & Technology, Mechanical Engineering,

which could nd applications in the broad domain of maritime economy. Hence there are published papers
which concern methods of the designing, manufacturing and operating processes of such technical objects
and devices as : ships, port equipment, ocean engineering units, underwater vehicles and equipment as
well as harbour facilities, with accounting for marine environment protection.
The Editors of POLISH MARITIME RESEARCH make also efforts to present problems dealing with
education of engineers and scientic and teaching personnel. As a rule, the basic papers are supplemented by
information on conferences , important scientic events as well as cooperation in carrying out international
scientic research projects.
Scientic Board
Chairman : Prof. JERZY GIRTLER - Gdask University of Technology, Poland
Vice-chairman : Prof. MIROSAW L. WYSZYSKI - University of Birmingham, United Kingdom

Dr POUL ANDERSEN Prof. STANISAW GUCMA Prof. EUGEN NEGRUS

Technical University of Denmark Maritime University of Szczecin University of Bucharest Romania
Denmark Poland
Prof. YASUHIKO OHTA
Dr MEHMET ATLAR Prof. ANTONI ISKRA Nagoya Institute of Technology
University of Newcastle United Pozna University of Technology Japan
Kingdom Poland
Dr YOSHIO SATO
Prof. GRAN BARK Prof. JAN KICISKI National Trafc Safety and
Chalmers University of Technology Institute of Fluid-Flow Machinery Environment Laboratory
Sweden of PASci Japan
Poland
Prof. SERGEY BARSUKOV Prof. KLAUS SCHIER
Army Institute of Odessa Ukraine Prof. ZBIGNIEW KORCZEWSKI University of Applied Sciences
Gdask University of Technology Germany
Prof. MUSTAFA BAYHAN Poland
Sleyman Demirel University Prof. FREDERICK STERN
Turkey Prof. JANUSZ KOZAK University of Iowa, IA, USA
Gdask University of Technology
Prof. VINCENZO CRUPI Poland Prof. ILCEV DIMOV STOJCE
University of Messina, Italy Durban University of Technology
Prof. JAN KULCZYK South Africa
Prof. MAREK DZIDA Wrocaw University of Technology
Gdask University of Technology Poland Prof. JZEF SZALA
Poland Bydgoszcz University
Prof. NICOS LADOMMATOS of Technology and Agriculture
Prof. ODD M. FALTINSEN University College London United Poland
Norwegian University of Science Kingdom
and Technology Prof. WITALIJ SZCZAGIN
Norway Prof. JZEF LISOWSKI State Technical University of
Gdynia Maritime University Poland Kaliningrad, Russia
Prof. PATRICK V. FARRELL
University of Wisconsin Madison, Prof. JERZY MATUSIAK Prof. BORIS TIKHOMIROV
WI Helsinki University of Technology State Marine University of St.
USA Finland Petersburg, Russia

Prof. HASSAN GHASSEMI Prof. JERZY MERKISZ Prof. DRACOS VASSALOS

Amirkabir University of Technology Pozna University of Technology University of Glasgow and
Iran Poland Strathclyde United Kingdom
 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 4-9
10.1515/pomr-2017-0014

PHYSICAL MODELLING OF GEOTECHNICAL STRUCTURES

INPORTS AND OFFSHORE

Lech Baachowski, Prof.

Faculty of Civil and Environmental Engineering, Gdask University of Technology, Poland

ABSTRACT

The physical modelling of subsoil behaviour and soil-structure interaction is essential for the proper design of offshore
structures and port infrastructure. A brief introduction to such modelling of geoengineering problems is presented and
some methods and experimental devices are described. The relationships between modelling scales are given. Some
examples of penetration testing results in centrifuge and calibration chamber are presented. Prospects for physical
modelling in geotechnics are also described.

Keywords: centrifuge, calibration chamber, CPT

in a series of tests in well defined and controlled boundary

INTRODUCTION
Interest in physical modelling of soil behaviour at high
stress levels has considerably increased in recent decades.
This is largely due to developments in offshore engineering,
laboratory testing and in situ soil investigations. The main
advantage of physical modelling is that it permits the
parametric studies to be performed including the model
geometry, different types or combinations of loading and soil
conditions [1,2,3]. It is also possible to study the behaviour
of the structure at different stages of construction process
[4]. The reduced-scale models can be subjected to complex
testing exposures including cyclic, dynamic or seismic forces
[5, 6]. Offshore engineering applications include: oil and gas
platform foundations, wind farms, pile foundations, anchors
and pipelines. In-situ soil investigations in marine civil
engineering should be performed after preliminary soundings
[7]. The calibration of new tools for in-situ investigations have
become a major problem in geotechnics [8, 9]. It is a question
of great interest in the behaviour of non-standard materials
from mineralogical point of view, cemented, containing
high amount of fines or shells [10, 11]. It is also important
when anthropogenic soils are used to improve the structure
response or to reduce the liquefaction risk [12]. It can be solved
conditions, using a repeatable and homogeneous soil mass
reconstituted in laboratory conditions, characterized by
its void ratio, stress level and stress and strain history.
Such samples can be also used in model tests thanks to
the development of the modern experimental stations for
physical modelling in geotechnics, like centrifuge facilities
or calibration chamber devices. The soil models and soil-
structure interaction under seismic loading can be also
studied in shaking table model test [13] where the inertia
forces can be taken into consideration.
Soil behaviour is highly non-linear concerning both stress
and strain levels. It implies that the same stress and strain level
is maintained in the reduced scale model and the prototype
construction. This level of stress is varying considerably for
different structures. It is about several kPa in case of cable
or pipeline on the seabed, but it can reach a dozen of MPa
under the platform foundation leg. In the first case a high
dilatancy will be observed in shearing, while grain crushing
and high soil contractancy will take place under high stress
level. One should also remember that the soil deformability
is also a function of stress level. Moreover, the degradation of
the shear modulus with shear strain should be also taken into
account when soil behaviour and soil-structure interaction
is considered. It is thus important to maintain, in physical

4 POLISH MARITIME RESEARCH, No S1/2017

modelling in geotechnics , the same stress and strain level The stress level can be increased either in centrifuge tests
for the model and the prototype construction. The soil is or calibration chamber. In centrifuge model tests, the stress
agranular material with pores partially or completely filled level increase is realized with the use of the soil container
with water. During cyclic, dynamic or seismic loading flying in a horizontal plane at constant radial velocity. In
some water pressure is generated in the pores, which leads case of calibration chamber the stress is applied by the
to effective stress reduction and in some cases may cause membranes around the soil specimen. The scale factors for the
the soil liquefaction and failure of nearby structures. The interpretation of centrifuge tests are summarized in Tab.1,
question is important in harbour design [12] and it needs where time scale factors depends on the phenomena to be
careful determination of environmental wave and wind studied. One should notice that in the small-scale model which
loading on offshore structures [14]. is geometrically identical to the prototype, the homologous
points are subjected to the same stresses and strains. Thus
it is possible to include stress and strain dependence of soil
SIMILITUDE CONDITIONS IN PHYSICAL characteristics and to properly describe the soil-structure
MODELLING interaction. More detailed information concerning the scale
factors in centrifuge modelling can be found in [15, 16].
Let us consider the similitude conditions between a model In calibration chamber tests the response of real size probes for
and its prototype. Let us assume a model of the dimension B in-situ testing like CPTU, DMT, pressure meter, cone pressure
and its prototype of the dimension b. The scale factor between meter and others , can be studied in well-defined boundary
*
the two variables can be defined as the geometrical scale x : conditions [10]. In this case no scale factor applies and the
results can be transfer directly to the prototype behaviour
after the examination of the influence of boundary effects.
B 1
x* (1) When the tests on so-called small models at 1g acceleration
b n are performed the strict similitude conditions are not fulfilled
where and some additional detailed analyses of the influence of
n modelling scale. size and scale effects should be done for the applied models.
A possible distortion in the similitude conditions arises if
The other modelling scales can be defined as: deformations appear in the soil mass or a shear band forms
VM in the soil-structure interface. When the size of the model is
stress scale V

getting closer to the mean grain size an eventual grain size
VP effect should be verified. The correct interpretation of such
gM
gravity scale g tests needs some correction factors to be applied [17].
gP UM Important advantage of physical modelling is that the tests
U
volumetric density scale UP are performed in reconstituted homogeneous soil mass with
well-known parameters established in extensive laboratory
where indexes M and P denote the model and the prototype, tests. The soil can be prepared to meet a certain range of
respectively. parameters including the stress history of soil deposit. It is
thus possible to calibrate the response measured in model
The equilibrium condition of continuum mechanics for tests to soil characteristics and to establish some correlations
static loading can be written as: for design.
Tab. 1 Scale factors in centrifuge tests
wV ij

j 1. 3 wx j
 U gi 0 (2) Parameter Prototype Model
length 1 n
Dimensional analysis of Eq. 5 ? needs the following surface 1 n2
equation to be satisfied: volume 1 n3
mass 1 n3
V
x g U

(3)
time (inertia) 1 n
acceleration 1 1/n
If the same material is used for the model and the prototype
( U 1), the same stress level is applied (V 1) to maintain stress 1 1
the soil rheology, and the model n-times smaller than the strain 1 1
prototype is used, the gravity should be n-times increased displacement 1 n
to respect the similitude condition (Eq. 3). In this case the force 1 n2
gravity scale should be n-times higher for the model:

1
g n (4)
x

POLISH MARITIME RESEARCH, No S1/2017 5

 GEOTECHNICAL CENTRIFUGE
Numerous experimental devices were constructed all over
the world to study physical modelling in geotechnics. The
first geotechnical centrifuge was constructed in the former
URSS, mainly for military purposes [18]. A view of the recent
centrifuge used in Ifsttar laboratories near Nantes is shown
in Fig.1. It is a beam centrifuge where the container with the
model of the mass up to 2000 kg is placed in the swinging
basket of the machine. Different stages of construction works
or combinations of loads can be modelled in-flight during
the experiment. Offshore engineering is one of the principal
domains for this modelling [20, 21, 22]. The effect of stress
level on the cone resistance mobilized during CPT penetration
test in centrifuge is given in Fig. 2. Here, amini-cone model of
12 mm in diameter is penetrating dense quartz sand medium
at the acceleration level of 60g at 100g. In this case the cone
resistance with vertical effective stress in the subsoil up to
400 kPa can be studied, which corresponds roughly to the
penetration depth of 40 m below seabed. New probes for
in-situ testing in offshore engineering like ball-cone, t-bar,
mini-pressure meter were extensively calibrated in centrifuge
tests.
Fig. 1. Geotechnical centrifuge at Ifsttar [19].
q [MPa]
c
0 4 8 12 16
0
50
100
150
V'v [kPa]
200
250
60g
300
350
60g - b= 720 mm
100g - b= 1200 mm
400
450
Fig. 2 Set of CPT results for Hostun dense quartz sand medium in centrifuge [23].
6 POLISH MARITIME RESEARCH, No S1/2017
CALIBRATION CHAMBER
Calibration chamber is another example of large
experimental device for physical modelling [10]. Numerous
chambers has been constructed in Italy, USA, Great Britain,
China and Japan [10, 23]. Such device was also designed and
built at the Geotechnical Laboratory of GUT ( = Gdask
University of Technology ? ) , which can house soil sample
of 53 cm in diameter and 100 cm high [24]. With self-reacting
design, the chamber is placed within the rigid steel frame
(Fig. 3). Hydraulic insertion cylinder with load cell and sand
storage bin for stationary sand spreader are mounted on the
frame. It is double wall chamber with independent pressure
control in internal and external chambers, which enables
complex boundary conditions to be applied with pneumatic
control system (Fig. 4). Top and bottom membranes and the
lateral cells are filled with water. Four air-water columns
and pressure transducers are fixed on the control panel.
Volumetric changes in the top and bottom membranes and
the internal chamber are measured with the position of the
floating element in air-water columns equipped with BTL2
micropulse transducers. Four applied pressures are controlled
with pneumatic system. The central hole in the bottom plate
permits to make tests with model piles which transverse the
chamber entirely.
20 24 28
Hostun medium
100g
Fig. 3 Calibration chamber at GUT [24]
 q [MPa]
c
0 4 8 12 16 20 24
0

10

20

cone penetration [cm]

30 BC1

40 ID=0.8

50
300 kPa
60
V'v = 50 kPa
70 150 kPa 250 kPa
70 kPa 100 kPa 400 kPa
80

Fig. 4 Scheme of the calibration chamber set-up with control system [24] Fig. 5. Cone resistance mobilization in Lubiatowo dense sand
at BC1 condition [25]

The reconstituted soil mass is not infinite, so the influence

of boundary conditions around the specimen in calibration
chamber should be evaluated. Five types of boundary PROSPECTS AND CONCLUSIONS
conditions expressed in stresses and strains can be applied
(Tab. 2). Special BC5 condition is intended to simulate infinite The experimental devices for physical modelling in
soil mass with a lateral stiffness corresponding to that of geotechnics offer the possibility of wide range parametric
surrounding soil. studies of complex foundations in offshore and port structures.
Tabl. 2 Boundary conditions to be applied Such study is substantially cheaper, effective and faster than
full scale experiment, often impossible to perform. The latter,
Boundary conditions Horizontal Vertical rare and expensive tests are however used to validate the
BC1 h=const v=const obtained results and to verify chosen assumption. In physical
BC2 h=0 v=0 modelling one can thus consider more options concerning soil
BC3 h=0 v=const and foundation model geometry and to check more loading
BC4 h=const v=0 conditions schemes. The most advanced laboratories are even
able to combine the advantages of two experimental devices,
lateral stiffness
BC5 k = const v=const i.e. the calibration chamber can be placed in the swinging
basket of centrifuge to study the effect of boundary conditions
The main application of calibration chamber tests is or the shaking table can be used to apply seismic exposure
averification of in-situ test response in well-defined soil to the model in centrifuge test.
conditions, applied stress level and boundary conditions. New advances in physical modelling in geotechnics
Some local correlations concerning the interpretation are related to the development of in-f light robotics,
of in-situ tests are developed including the effect of soil instrumentation and recording the results. One should,
mineralogy, compressibility, density and stress level [10]. The however, consider high complexity of problems to be studied.
mobilization of cone resistance in Lubiatowo dense sand at If the behaviour of the superstructure can be relatively easily
different vertical stress level applied to the chamber is given modelled the main difficulties are related to the soil response.
in Fig.5. Local correlation for the interpretation of CPTU Stress and strain level effects on the soil characteristics together
tests in quartz sands on the Baltic coast was proposed on with mobilization and accumulation of water pressures in
the basis of a series of calibration chamber tests and in-situ agranular material subjected to cyclic or dynamic loading
investigations in sands deposit on the Baltic coast [25]. Other should be analyzed [26]. The problem is how to consider
applications of calibration chamber are related to analyses of together the physical modelling of the structure, subsoil and
pile behaviour [1], installation effects and plate loading tests. their interaction. If a shallow foundation is considered [27]
The device can also operate as a large triaxial apparatus when the problem is less complex than in case of deep foundation
large grain material is used. embedded in the subsoil where comprehensive nonlinear
soil-structure interaction should be taken into account [27].
Then a simplified physical modelling at 1g acceleration could
be only the first attempt. Parallel physical modelling and

POLISH MARITIME RESEARCH, No S1/2017 7

numerical analysis could be very useful and promising to
fill the gap. They should be supported by detailed laboratory
analyses of soil behaviour and soil-structure interaction under
dynamic or cyclic loading to identify and better describe the
governing mechanisms.
REFERENCES
1. Foray, P., Baachowski, L., Colliat , J L.? : Bearing
capacity of model piles driven into dense over-consolidated
sand. Canadian Geotechnical Journal, 1998, 35(2), pp.
374-385, 10.1139/t97-082.
2. Tsuha, C. D. H. C., Aoki, N., Rault, G., Thorel, L., &
Garnier, J. Evaluation of the efficiencies of helical anchor
plates in sand by centrifuge model tests.Canadian Geo-
technical Journal,2012, 49(9), pp. 1102-1114.
3. Fioravante, V. : Load transfer from a raft to a pile with an
interposed layer. Gotechnique, 2011, 61(2), pp.121-132.
4. Le Kouby, A., Dupla, J. C., Canou, J., & Francis, R.:
The effects of installation order on the response of a pile
group in silica sand.Soils and Foundations, 2016,56(2),
pp.174-188.
5. Falborski T., Jankowski R.: Polymeric Bearings A New
Base Isolation System to Reduce Structural Damage during
Earthquakes. 10th International Conference on Damage
Assessment of Structures (DAMAS), 2013. http://www.
scientific.net/KEM.569-570.143
6. Jankowski R.: Experimental study on earthquake-induced
pounding between structural elements made of different
building materials. Earthquake Engineering & Struc-
tural Dynamics, 2010, Vol. 39, Issue 3, pp. 343-354, DOI:
10.1002/eqe.941
7. Kozaczka E., Grelowska G., Kozaczka S.: Images of the
Seabed of the Gulf of Gdansk Obtained by Means of the
Parametric Sonar. Acta Physica Polonica A., 2010,Vol.
118, No.1, pp. 91-94
8. Grelowska G., Kozaczka E.: Sounding of Layered Marine
Bottom - Model Investigations. Acta Physica Polonica A.,
2010,Vol. 118, No.1, pp. 66-70.
9. Grelowska G., Kozaczka E., Nowicki A., Kozaczka S.:
Investigation of Transmit and Receive Characteristics of
Laboratory Model for the Parametric Echosounder. Acta
Physica Polonica A., 2013,Vol. 123, No.6 , pp. 1094-1097.
10. Ghionna, V.N., Jamiolkowski, M. : A critical appraisal of
calibration chamber testing in sands. Proceedings of the
1st International Conference on Calibration Chamber
Testing, Clarkson Univ., Potsdam, USA, 1991. Edited by
A.-B. Huang, pp. 13-37.
8 POLISH MARITIME RESEARCH, No S1/2017
11. Dijkstra, J., Gaudin, C., & White, D. J. : Comparison
of failure modes below footings on carbonate and silica
sands.International Journal of Physical Modelling in
Geotechnics , 2013, 13 (1).
12. Hazarika, H., Kohama, E. and Sugano, T.: Underwater
shake table on waterfront structures protected with tire
chips cushion. J. Geotechnical and Geoenvironmental
Engineering, (ASCE), 2008, Vol. 134, No.12.
13. Mahmoud S., Abd-Elhameed A., Mahmoud S., Jankowski
R.: Earthquake-induced pounding between equal height
multi-storey buildings considering soil-structure interac-
tion. Bulletin of Earthquake Engineering, 2013, Vol. 11,
Issue 4 , pp.1021-1048. http://link.springer.com/10.1007/
s10518-012-9411-6
14. Dymarski C., Dymarski P.: Developing Methodology for
Model Tests of Floating Platforms in Low -Depth Towing
Tank. Archives of Civil and Mechanical Engineering,
2016, Vol. 16, Issue 1, pp. 159-167
15. Garnier, J., Gaudin, C., Springman, S. M., Culligan, P.
J., Goodings, D., Konig, D., ... & Thorel, L. : Catalogue
of scaling laws and similitude questions in geotechnical
centrifuge modelling.International Journal of Physical
Modelling in Geotechnics,2007, 7(3), 1.
16. Konkol, J. : Derivation of the scaling laws used in geo-
technical centrifuge modelling-application of dimensional
analysis and Buckingham theorem. Technical Sciences/
University of Warmia and Mazury, Olsztyn , 2014.
17. Foray, P., Baachowski, L., & Rault, G. : Scale effect in
shaft friction due to the localization of deformations.
InProceedings of the international conference on cen-
trifuge,Tokyo, 1998, Vol. 98, pp. 211-216.
18. Madabhushi, G. : Centrifuge Modelling for Civil Engineers.
CRC Press.
19, 2015, http://eurofuge2016.ifsttar.fr
20. Bayton, S.M., Black, J.A. : The effect of soil density on
offshore wind turbine monopile foundation performance.
Proc. of 3rd European Conference on Physical Modelling
in Geotechnics, Eurofuge 2016, Nantes, pp. 239-244.
21. Loli, M., Kassas, K. & Anastasopoulos, I. : Centrifuge
testing of a hybrid foundation for offshore wind turbines:
experimental technique and preliminary results. Proc.
of 3rd European Conference on Physical Modelling in
Geotechnics, Eurofuge 2016, Nantes, pp. 263-268.
22. Blanc, M., Thorel, L., Isorna, R., Dano, C., Kotronis,
P., Philippe, M. : Centrifuge investigation of axial cyclic
behaviour of a single pile used for the foundation of a jacket
 type offshore wind turbine. Proc. Frontiers in offshore
geotechnics, ISFOG 2015, Oslo, Norway, CRC Press, pp.
521-526.

23. Baachowski, L. : Size effect in centrifuge cone penetration

tests. Archives of Hydro-Engineering and Environmental
Mechanics, 2007, 54(3), pp. 161-181.

24. Baachowski, L. Physical modelling in sands in a wide

range of stress level: Application to the calibration of CPTU
and DMT tests. Wydawnictwo Politechniki Gdaskiej,
Gdask 2008.

25. Baachowski, L. : Penetration resistance of Lubiatowo

sand in calibration chamber tests.Archives of Hydro-
Engineering and Environmental Mechanics, 2006, 53(4),
pp. 311-329.

26. Mortara, G., Mangiola, A., & Ghionna, V. N. : Cyclic shear

stress degradation and post-cyclic behaviour from sand-
steel interface direct shear tests. Canadian Geotechnical
Journal,2007,44(7), pp. 739-752.

27. Dymarski C., Dymarski P., ywicki J.: Design and strength
calculations of the tripod support structure for offshore
power plant. Polish Maritime Research , 2015,Vol. 22,
No.1(85), pp. 36-46.

CONTACT WITH THE AUTHOR

Lech Baachowski

Faculty of Civil and Environmental Engineering,

Gdask University of Technology
Poland

POLISH MARITIME RESEARCH, No S1/2017 9

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 10-15
10.1515/pomr-2017-0015

IMPLEMENTATION OF SPATIAL INFORMATION

FOR MONITORING AND ANALYSIS OF THE AREA AROUND
THE PORT USING LASER SCANNING TECHNIQUES

Katarzyna Bobkowska1, Adam Inglot1, Miroslava Mikusova 2 , Pawe Tysic *1

1
Gdansk University of Technology, Poland
2
Univeristy of Zilina, Slovakia
* (corresponding author)

ABSTRACT

Nowadays, maritime infrastructure is heavily exploited, which requires monitoring. The article presents the
implementation of spatial information which are point clouds for monitoring and analysis of the area around the
port (buildings and wharves). For this study, point clouds coming from terrestrial (TLS) and airborne laser scanning
(ALS), each of them having different accuracy, were used. An important part of the analysis was the integration ofthe
two data sources. Through integration, the acquisition of information on areas not covered by the measurement
inthepresented case, one of the methods was possible for use (e.g. the roofs in case of TLS, or the lack of some of the
walls of buildings in case of ALS). Another aspect was to use this data. Measurement of the shape and geometry
ofobjects was executed. Additionally, the planeness analysis of individual elements of port infrastructure has been
carried out. An interesting analysis was to determine the water level, based on relation to specific characteristics
ofthe light reflectance.

Keywords: spatial information, monitoring of the area around the port, terrestrial and airborne laser scanning

of such a construction in the area of the coast of the Motlawa

INTRODUCTION
The Motlawa is a Polish river, which runs in the Pomerania
province. This river is a tributary of the Dead Vistula, which
has its estuary to Gdansk Bay. Motlawa in Gdansk has a very
rich port history. Since the middle ages it was a very important
point for the logistics of the Baltic shipping routes. A storage
district was located in the vicinity. Wood, grain, iron, copper,
lead, wax, leather, fur, eastern roots, cloth, herrings and salt
were trans-shipped in the Motlawa Port. Today, this part
of Gdansk is usually associated with a tourist attraction of
the city. The most popular of them is the Long Bridge. The
Motlawa Port functioned until 1939. Currently, Matlawa is
a harbor for the cruise company egluga Gdaska ships.
Figure 1 shows the map section of the old Motlawa Port area.
Figure 2 shows the same field section of orthophoto map,
which was developed in 2014. Due to historical aspects, it
is important to archive and inventory all historic buildings.
Here, laser and optical scanning is applicable. An example
is a wooden crane, which was first used in 1363. In the past,
it was the strongest crane in northern Europe1. Besides the
advantage of scanning - the ability to archive practically the
majority of information on the shape of any object - many
other possibilities for the use of scanning port infrastructure
can be indicated. These possibilities are discussed further in
this article. From todays point of view, they are very important
aspects of port building safety. Based on laser scanning,
port security systems can be developed, which are just as
important to the security systems of ships in the harbor2. Due
to the floods occurring in the history of Gdansk (in 1540 the
Granary Island was completely flooded), monitoring the water
level is quite important. Besides monitoring the water level in
the river, monitoring the technical condition of the buildings
is also necessary. All kinds of cracks, deformation and tilt,
should be detected as soon as possible in order to secure
and eliminate the total destruction of objects. It is possible
to monitor these values by using innovative measurement
technologies like laser scanning and create an application

10 POLISH MARITIME RESEARCH, No S1/2017

which could inform the user when the deformations reach
critical levels. Scientific research on the border of the water
and the ground are not widely implemented and the results
are not commonly in use.

Fig. 3 The results of TLS data acquisition.

Fig. 1. The Gdansk map section originating from 1940.

Area: The Motawa around Granary Island.

Fig. 4 Visualization of ALS data.

The accuracy of the spatial data, which has been used in

the current project is on the level of 5 mm in reference to TLS
in the case of scan positions alignment. The results of data
Fig. 2. The Gdansk orthophoto map section originating from 2014. alignment of the TLS are shown in Fig. 5.7
Area: The Motawa around Granary Island.

LASER SCANNING AS THE SOURCE

OFSPATIAL DATA
Laser scanning is an efficient tool for gathering spatial
information as well as it supports geodetic data in GIS
applications and provides support for city modelling3. In
addition to ranging, Light Detection and Ranging (LiDAR)
systems can provide additional information about the target
and information about the transmission path4,5. To gather the
spatial data, two of the laser scanning systems (terrestrial and
airborne laser scanning) have been used. Furthermore, based
on laser scanning measurements it is an obligation to regard
the influence of the heat caused by insolation and vibrations Fig. 5 The results of TLS data alignment.
interference. 6 On the base of collected data, a model of the
harbor was created. It was used for monitoring the area of
interest by analyzing the utility of proposed applications, The accuracy of the ALS data is up to 50 cm based on
which are described below. the information, provided by Head Office of Geodesy and
To create a point cloud, terrestrial and airborne laser Cartography. ALS point cloud basic operations of integration
scanning has been used. The results of used point clouds of the data which has been used come from the ISOK project.
have been shown in Fig. 3 and Fig. 4. The Polish ISOK abbreviation translates into IT System
of the Countrys Protection against hazards. It is aimed at

POLISH MARITIME RESEARCH, No S1/2017 11

providing a system that improves the protection of Economy,
Environment and Society against extraordinary threats,
especially against flooding8.
The point cloud is obtained from TLS measurements which
were made on the basis of measurements of 20 positions.
In order to connect the following stations geodetic prisms,
spheres reference or details of terrain, clearly identifiable
on the scan were used. In the case of measurements it was
important to give georeferencing because of ALS data
integration. For this purpose, measured geodetic Global
Navigation Satellite Systems (GNSS) receivers were used.
Control points were determined by the 3D error which was
no greater than 0.05 m. In the process of measurements of the
coastal zone, places where the horizon is too much obscured
should be considered. It has a direct impact on the accuracy
of the designated points of geodetic control network using
the GNSS receiver9,10.
The prepared laser data were subjected to further processing
that was foremost based on a proper integration of TLS and
ALS data. There are two main methods of integration of the
data and both of them rely on the users supervision. The first
one is to create tie points, which could be used to reference the
data. Whereas the latter is to align the data using temporary
tie points and manually create tie planes which are utilized
for automatic search for corresponding planes exploiting
iterative closest points algorithm (ICP). This algorithm is
based on detecting the closest planes in relative data sets and
matching them. To create automatic planes the Plane Patch
Filter has been used, implemented in RiScan Pro, courtesy
of Riegl GmbH. Plane Patch Filters basis is to divide the
3D space into equally sized cubes. For each cube the plane
is estimated using least squares method, when the standard
deviation of the plane is higher than the assumed value, the
cube is further divided into further eight smaller cubes, and
the process is repeated. The smallest size of a cube is defined
by the user. If the standard deviation is less than expected,
the algorithm saves the plane and takes it into consideration
during the final alignment process. Based on these algorithms,
the ALS and the TLS point clouds have been aligned. The
results are shown in Fig. 6.
Fig. 6 The results of TLS and ALS alignment.
12 POLISH MARITIME RESEARCH, No S1/2017
Based on the conclusive data set, the authors moved on to
the analysis in terms of port infrastructure with monument
elements. Besides laser scanning, an interesting form to obtain
data of objects is optical scanning11. It is being used increasingly
in the analysis of building shapes and in the analysis of shape
changes over time. Optical scanning is a method involving
the acquisition of data on the object based on the captured
images12 . Using the methods of photogrammetry, we can
pinpoint the coordinates (in local or global system global
requires input data with the coordinates of the markers)
characteristic points of the object by two images of the same
object. By automating the image processing we are able to
quickly obtain very dense and accurate point cloud objects.
Regarding to optical or laser scanning measurements the
appropriate kind of the platform should be considered. In the
project, the authors used a tripod for TLS, but it is possible to
acquire data by using a platform on a moving vehicle which
could be called as mobile13,14,15.
POSSIBLE APPLICATIONS OF HARBOR
INFRASTRUCTURE ANALYSIS
USINGINTEGRATED POINT CLOUD
VISUALIZATION + GEOMETRY
In the case of visualization, that could be supported with
geometry analysis, like the one shown in Fig. 7 (laser scanning
method provides metric system with accuracy at the level
of a few mm for TLS , whilst for ALS the accuracy reaches
approx. 20 cm) the application using integrated point cloud
could help establish the boundaries of the safe investments
zone or retreat of infrastructure facing real threats such as the
degradation of the objects of interests, by creating a warning
system from collected data16 . TLS gives the possibility to
analyze the geometry of the various objects (e.g. silos17, shell
structures, coastal cliffs18).
Fig. 7 Plan of the Granary Island, Gdansk with the metric grid, where side
length of the square is 5m.
 ESTIMATE THE WATER LEVEL USING
TERRESTRIAL LASER SCANNING
The estimation of the water level based on terrestrial laser
scanning could be executed with specific characteristics of
the light reflectance. Reflectance, sometimes called the 4th
coordinate beside X,Y,Z, is the ratio of the reflected beam
to the power of the incident light fraction, described on the
boundary of two media with different refractive coefficients,
expressed in decibels with a negative sign19. In the case of the
described application, the value of the amplitude has been
used. The amplitude is known as the intensity of reflection.
Relying on the fact that water absorbs the infrared part of
the electromagnetic spectrum(the amplitude equals 0) the
operator could with high precision point out the place of
water level by using adjacent harbor infrastructure. There
was a distinctive boundary, a cut off where there were no
valid points (reflections) on vertical walls (i.e. Larssen sheet
piling). Also, one could accurately estimate the height of
the water level according to the measurements conducted in
relation to the national vertical reference system. The chart
presenting reflection from different type of targets is shown
in Fig. 8. There is a line, marked with no. 7, corresponding
to remote sensing windows for water as an optical medium.

Fig. 9 Water level estimation, based on amplitude characteristic.

HARBOR INFRASTRUCTURE
MONITORING

Fig. 8 Reflectance of different types of targets [3].
To precisely point a place where a water level appears,
the user has to find a row of points where the amplitude
of these hits is approximately 1 dB. What is more, to avoid
misinterpretation, the operator should check for the close
neighbourhood of the marginal air-water laser scanning
reflexes. Those should have an amplitude value of no more
than a few dB, in this case: 6-7 dB. The example of water level
estimation is shown in fig. 9. The acquired data can also help
to gather information on the level of the seabed, in the case
of shallow areas. For greater depth, other methods should
be used.20,21 for example: acoustic waves are widely used 22.
Laser scanning datasets as a source for infrastructure
monitoring could be very useful in the evaluation of stability
and degradation level of the harbor. This method is based
on creating a 3D triangulated model. With this method
the operator could easily estimate the wall inclination or
deviation of any other type of element. This applies as much
to planar structures as for linear objects. In fig. 10 the authors
have shown an example of wharf deviation from assumed
plane. The different colors indicate local displacement from
a horizontal position.

Fig. 10 Inclination from horizontal position of the wharf.

POLISH MARITIME RESEARCH, No S1/2017 13

 Due to laser scanning limitations such as the inability to
penetrate the structure with a light beam, the degradation
monitoring can only be a guide to true objects condition.
Dense point clouds are subject to search for major defects:
missing, buckled or dented elements, cracks (also tiny ones)
and abrasions. The crack detection was further presented
in the paper23,24 . The operator can only rely on geometrical
characteristics of the monitored object, though it is still
very precise and effective information on general structural
integrity. The result of such monitoring can be easily compared
with some reference data (CAD drawings, original inventory
surveys). Hand in hand, with the addition of more complex
GIS analysis of cracks or displacements, the operator could
be provided with an advanced failure prediction.
CONCLUSIONS AND DISCUSSION
The article briefly presents the spatial analysis of former
Polish Motlawa Port, which today is an important tourist
attraction of the city of Gdansk. Due to historical buildings
located in in this area, archiving residues of these objects
requires a specific approach. The solution to this problem
can be achieved using the methods of photogrammetry and
remote sensing25, and specifically speaking - laser scanning.
The presented feasibility of collecting data on these objects
and their visualization, have great potential today. Collected
data on the Granary Island and surrounding areas may serve
as an interesting informant in the future. Besides archiving for
the purpose of data visualization, special attention was paid to
the safety aspect. Areas around the port, due to contact with
water, are more susceptible to the phenomenon of destruction.
The method that allows the analysis of vertical and planeness
waterfront has been presented. A deviation map presented in
this article was created based on data from laser scanning.
This map indicates the location where there are deviations
from the plane of the waterfront wall. The presented analysis
can be carried out for the needs of other facilities around
ports in Poland and in the world.
ENDNOTES
1. http://www.nordstern-gdansk.cba.pl/Port%20Gdanski/
gdansk%20port%20nad%20motlawa.html
2. Szlapczynski, R., Szlapczynska, J.: On evolutionary com-
puting in multi-ship trajectory planning, Appl Intell 37:
155, 2012, DOI:10.1007/s10489-011-0319-7
3. Tse R.O.C., Gold C., Kidner D.: 3D City Modelling from
LIDAR Data. In P. Van Oosterom et al., eds. Advances in
3D Geoinformation Systems. Springer Berlin Heidelberg,
pp. 161175, 2008
4. Introduction to LIDAR, NPS Lidar Workshop, May 24,
2007, Joe Liadsky, Optech Incorporated.
14 POLISH MARITIME RESEARCH, No S1/2017
5. Wehr A., Lohr U.,. Airborne laser scanning--an introduc-
tion and overview. ISPRS Journal of Photogrammetry
and Remote Sensing, Vol. 54, Iss. 2-3, pp.6882, 1999
6. Kaliski K. J., Galewski M. A. Kaliski K. J., Galewski
M.A.: Chatter vibration surveillance by the optimal-
linear spindle speed control. Mechanical Systems and
Signal Processing, Vol. 25, pp. 383-399, 2011, DOI:
10.1016/j.ymssp.2014.06.016
7. Kersten T, et al: Geometric accuracy investigations of the
latest terrestrial laser scanning systems. FIG Working
Week, (June), pp.116, 2008
8. Pawuszek K., Ziaja M., Borkowski A.: Accuracy Assess-
ment of the Height Component of the Airborne Laser
Scanning Data Collected in the ISOK System for the
Widawa River Valley. Acta Scientiarum Polonorum.
Geodesia et Descriptio Terrarum, Vol. 13, pp.2737, 2014
9. Tysiac, P., Wojtowicz A., Szulwic J.: Coastal Cliffs Moni-
toring and Prediction of Displacements Using Terrestial
Laser Scanning. Baltic Geodetic Congress (Geomatics),
IEEE, ISBN: 978-1-5090-2421-6, pp. 61-66, 2016, DOI:
10.1109/BGC.Geomatics.2016.20
10. Grejner-Brzezinska D.A., et al.: A robust solution to high-
accuracy geolocation: Quadruple integration of GPS,
IMU, pseudolite, and terrestrial laser scanning. IEEE
Transactions on Instrumentation and Measurement,
Vol. 60, Iss.11, pp.36943708, 2011
11. Paszotta, Z., Szulwic, J., Szumilo, M.: Internet photogram-
metry as a tool for e-learning. 8th International Confer-
ence of Education, Research and Innovation, ICERI2015,
ISBN: 978-84-608-2657-6, pp. 4565-4573, 2015
12. Bobkowska, K., Janowski, A., Przyborski, M. and Szul-
wic, J.: Analysis of High Resolution Clouds of Points as
aSource of Biometric Data, 2016 Baltic Geodetic Congress
(Geomatics), Gdansk, 2016, ISBN: 978-1-5090-2421-6,
pp.15-21, DOI: 10.1109/BGC.Geomatics.2016.12
13. Kalinski K., Buchholz C.: Mechatronic design o strongly
nonlinear systems on a basis of three wheeled mobile plat-
form. Mechanical Systems and Signal Processing, Vol.
52-53, pp. 700-721, 2015, DOI: 10.1016/j.ymssp.2014.06.016
14. Mikrut, S., Kohut P. et al.: Mobile Laser Scanning Systems
for Measuring the Clearance Gauge of Railways: State
of Play, Testing and Outlook. Sensors, Vol. 16, Iss. 5, pp.
683, 2016, DOI: 10.3390/s16050683
15. Pastucha, E.: Catenary System Detection, Localization
and Classification Using Mobile Scanning Data. Remote
Sens. Vol. 8, Iss. 801, 2016, DOI: 10.3390/rs8100801.
16. Burdziakowski P., Janowski A., Kholodkov A. Matysik CONTACT WITH THE AUTHOR
K., Matysik M., Przyborski M., Szulwic J., Tysiac P.: Mari-
time laser scanning as the source for spatial data. Polish Katarzyna Bobkowska
Marit. Res., Vol. 22, Iss. 4, pp. 9-14, 2015, DOI: 10.1515/ e-mail: katarzyna.bobkowska@pg.gda.pl
pomr-2015-0064 Gdansk University of Technology
Faculty of Civil and Environmental Engineering,
17. Gorski J., Mikulski T., Ozieblo M., Winkelmann K.: Effect Department of Geodesy
of geometric imperfections on aluminium silo capacities. ul. Narutowicza 11/12, 80-233 Gdansk
Stahlbau. Vol. 84, Iss. 1, 2015, pp. 52-57, DOI: 10.1002/ Poland
stab.201510224
Adam Inglot
18. Rohmer J. and Dewez T.: Analysing the spatial patterns e-mail: adam.inglot@pg.gda.pl
of erosion scars using point process theory at the coastal Gdansk University of Technology
chalk cliff of Mesnil-Val, Normandy, northern France. Faculty of Civil and Environmental Engineering,
Nat. Hazards Earth Syst. Sci., Vol. 15, pp. 349362, 2015, Department of Geodesy
DOI: 10.5194/nhess-15-349-2015 ul. Narutowicza 11/12, 80-233 Gdansk
Poland
19. Hfle B., Pfeifer N.: Correction of laser scanning intensity
data: Data and model-driven approaches. ISPRS Journal Miroslava Mikusova
of Photogrammetry and Remote Sensing, Vol. 62, Iss. 6, e-mail: mikusova@fpedas.uniza.sk
pp.415433, 2007 Univeristy of Zilina, Faculty of Operation and
Economics of Transport and Communications,
20. Kozaczka E., Grelowska G., Kozaczka S.: Images of the Department of Road and Urban Transport
seabed of the gulf of gdansk obtained by means of the Slovakia
parametric sonar. Acta Physica Polonica A. Vol. 118, no.1
s.91-94, 2010. Pawe Tysic
e-mail: pawel.tysiac@pg.gda.pl
21. Grelowska G., Kozaczka E.: Underwater Acoustic Imag- Gdansk University of Technology
ing of the Sea. Archives of Acoustics. Vol. 39, no. 4, pp. Faculty of Civil and Environmental Engineering,
439-452, 2014, DOI: 10.2478/aoa-2014-0048 Department of Geodesy
ul. Narutowicza 11/12, 80-233 Gdansk
22. Rucka M., Wilde K.: Experimental study on ultrasonic Poland
monitoring of splitting failure in reinforced concrete.
Journal of Nondestructive Evaluation. Vol. 32, Iss. 4,
pp.372-383, 2013, DOI: 10.1007/s10921-013-0191-y

23. Nagrodzka-Godycka K., Szulwic J., Ziolkowski P.: The

method of analysis of damage reinforced concrete beams
using terrestrial laser scanning. 14th International
Multidisciplinary Scientific GeoConference, Interna-
tional Multidisciplinary Scientific GeoConference &
EXPO SGEM, ISBN: 978-619-7105-12-4 / ISSN: 1314-
2704, Book2, Vol. 3, pp. 335-342, 2014, DOI:10.5593/
SGEM2014/B23/S10.042

24. Janowski A., Nagrodzka-Godycka K., Szulwic J.,

Ziolkowski P.: Remote sensing and photogrammetry
techniques in diagnostics of concrete structures. Comput-
ers and Concrete, Vol. 18, Iss. 3, pp. 405-420, 2016, DOI:
10.12989/cac.2016.18.3.405

25. Tong X., et al.: Integration of UAV-based photogrammetry

and terrestrial laser scanning for the three-dimensional
mapping and monitoring of open-pit mine areas. Remote
Sensing, Vol. 7, Iss. 6, pp.66356662, 2015

POLISH MARITIME RESEARCH, No S1/2017 15

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 16-24
10.1515/pomr-2017-0016

SELECTED PROBLEMS OF TRANSPORT IN PORT TOWNS TRI-CITY

AS AN EXAMPLE

Marcin Budzyski
Dawid Ry
Wojciech Kustra
Gdansk University of Technology, Poland

ABSTRACT

Port towns are strategic places from the point of view of transport systems. They form integration junctions for
various transport branches , apart from the traditional - road and railway ones , also for water( sea) transport which
is active there. Moreover, air transport comes also into consideration , whose efficient functioning must be connected
with good accessibility, that concerns sea transport as well. Efficient and safe servicing the ports is crucial for their
functioning. Problems associated with the overloading of lorries, which leads to degradation of road surface structure
, observed in Gdynia, are discussed as an example in this paper. Problems of road traffic safety (RTS) are presented in
this paper on the example of Gdask. The two issues : the road traffic safety and road surface degradation constitute
only some transport problems of port towns , but they are very important, from the point of view of their specificity,
for integration junctions of all the transport branches for people and goods. However, in discussing selected aspects of
transport in port towns it is necessary to refer to the managing of integrated transport system with taking into account
its traffic safety aspects.

Keywords: boil-off gas, KC-1 membrane, LNG carrier, rigid polyurethane foam, ozone depletion potential

INTRODUCTION x mutual relations between hazards to safety and effectiveness

Transport safety is considered a determined state of
the system whose operation runs without accidents and
other undesired events. The differentiation of particular
transport branches ( road, railway , water and air one) as
regards their organization , transport routes and means
makes reaching a common managing plane difficult.
But , despite the differences , philosophy of safety and the
goal which is aimed at in these aspects , are invariable.
Research on the state of transport safety may be reduced to
the analysis of :
x risk of hazards occurring in the transport system and its
surrounding ,
x possible counteraction against occurrence of hazards and
their consequences , to be undertaken by the transport
system and external rescue systems,
of counteraction and mitigation of consequences of
occurred events.
Consciousness of people on hazards to health and life,
associated with transport or travels , is an indispensable
factor for successful actions to improve transport safety. It
may be exemplified by the Integrated Transport Safety System
ZEUS developed under supervision of Prof. R. Krystek in
2009 [14]. In the frames of the project a method for integration
of transport safety managing was prepared. From the point
of view of port towns which serve as integration junctions
for all transport branches, a common platform for managing
their safety is a crucial problem to be solved. For example, in
maritime transport where safety issues undergo permanent
improvements [15, 16, 17, 18], adequate procedures of action are
determined and particular accidents are analyzed in detail. In
case of road transport such procedures are lacking , except for

16 POLISH MARITIME RESEARCH, No S1/2017

legal proceedings (in case of occurrence of fatality or heavy-
injured persons ), and no detail analyses of causes of such
accidents are made and no appropriate recommendations are
introduced into practice.
Another problem , characteristic for port towns where an
extensive heavy traffic on some street routes occurs , is the
degradation of road surface structure by overloaded lorries.
The problem of overloading the lorries operating on Polish
roads is serious and , in some points of observation, deals with
25% of lorries traffic stream . The dealings greatly contribute to
shortening life of road surface structure and greatly influence
rate and intensity of road surface failures, for this reason
weight control of lorries, aimed at protection of road surface
structure , is important and justified. One of the methods for
such control is the preselection , i.e. weighing the lorries in
motion and in consequence stopping the lorries identified to
be overloaded.
PROBLEM OF OVERLOADED LORRIES
ACCOMPANYING CONDITIONS
Gdynia is a transport junction of a great importance in a
European scale. Across the town two Paneuropean transport
passageways go. Aircraft connections are possible through
Gdask airport at Rbiechowo (2,46 mln passengers serviced
in 2011), in the future a part of the number will be serviced
by a new airport at Gdynia - Kosakowo (1,5 mln passengers
per year expected till 2030).The port in Gdynia ensures sea
connections with ports in Europe and worldwide. Municipal
transport in Gdynia belongs to the most ecological ones in
Poland and the trend will be certainly continued at least for
the reason that in its neighbourhood a few research centres
and universities which develop new ecological technologies,
are located [4, 11, 12, 13]. In Gdynia there is a network of
modern trolleybuses , in a part of autobus fleet , driving engines
are successively adjusted to combusting pressurized natural
gas , moreover, efforts are done to make inhabitants aware of
correct transport behaviour and to change their negative habits.
Yearly expenditures on transport infrastructure reach about
25 % of the town budget (this is the largest part of it , next to
that for schools and education ). In the town new transport
investment projects are permanently developed, both in the
area of common transport and of individual, pedestrian and
bicycle traffic , as well as transport of goods.
Total length of all the roads in Gdynia amounts to 395,1 km
(in 2011), 5,79 km of which belongs to the national highways :
DK6 - leading from JRZo via Gdynia up to Polish-German
border in Kobaskowo, and DK20 - connecting Gdynia with
Stargard Szczeciski.
In last years were completed very important investment
projects in the area of road infrastructure , that significantly
improved traffic conditions in Gdynia, namely , completion of
The Kwiatkowski Thoroughfare which connects Gdynia port
with The Tri-city Bypass Road, the putting into operation of
the Rose Road , the extension of Winiewski street which
connects the town centre with port area , the extension of the
transport junction Gdynia St. Maximilian Hill. In 2010 the
number of motor cars registered in Gdynia reached 111 966
cars, that gives motorization index equal to 450 cars per 1000
inhabitants (in 2008 the index amounted to 440 cars per 1000
persons). Share of intermodal railway transport , i.e. transport
of road lorries on railway lorries, continuously increases. In
2002 it amounted to 1,1% of railway transport, increased up
to 1,6% in 2004, up to 2,37%, in 2011, and in 2012 it reached
3,48%. In 2010 in Gdynia , turnover of goods transported
by paid-work lorries ( in domestic and international traffic)
reached 715 thou t ( by about 15% less than in 2008).
METHODS FOR DETERMINING PLACES FOR
PRE-SELECTION
In Poland heavy traffic means are classified according to
three categories :
x lorries without trailer,
x articulated vehicles (lorries with trailers or semitrailers),
x autobuses.
The classification is applicable to general traffic
measurements , analysis of traffic capacity and traffic modelling
as well as in road surface designing. To conduct a more precise
analysis of heavy traffic structure and impact of vehicles on
road surface, an additional classification with respect to
number and configuration of vehicle axles, may be introduced.
Fig. 1 shows an example of such classification.
Fig. 1 Schematic diagram of classification of heavy vehicles with respect to axle
configuration used in vehicle
Heavy traffic may be characterized by using mean
daily traffic of heavy vehicles as well as by load (total mass)
distributions and load- per- axle distributions of vehicles. The
mean daily traffic of vehicles is determined individually for a
given road section on the basis of traffic rate measurements.
In case of determining the load distribution of vehicles to
weigh the whole stream of vehicles is necessary. Execution
of measurements of mass of vehicles and weights exerted on
their axles requires to install special permanent devices into
road surface, therefore additionally use is also made of load
POLISH MARITIME RESEARCH, No S1/2017 17
distributions taken from other roads where such measurements
could be performed.
An analysis of the measurements conducted by these
authors makes it possible to preliminarily estimate traffic
rates and routes of heavy vehicles in Gdynia. The preliminary
cartograms of heavy vehicle traffic, developed on this basis,
served for formulating the concept of a system of road surface
protection against destructive action of overloaded vehicles.
The measurements were conducted in spring and fall of 2012
for the purposes of building the TRISTAR system. Analyzing
the obtained data one may observed the highest traffic rate
of heavy vehicles along The Kwiatkowski Flyover (1723%),
where the number of heavy vehicles reaches even 8.600 units
per day. In the next order, an extensive traffic of heavy vehicles
was observed : along Winiewski street, the end fragment of
the Tri-city Bypass-road as well as the exit of Morska street
toward Szczecin. The traffic rates of heavy vehicles are there
much lower than on The Kwiatkowski Flyover and amount to
2.3003.200 units/day (5,810%). The subsequent fragments
of municipal streets located in Centrum , on which number of
heavy vehicles varies about value of 1.500 units/day , are the
following: Wadysaw IV street, Zwycistwa avenue and the
route along Wielkopolska Chwaszczyska streets , where
the traffic rate of heavy vehicles is on the level of 3-5%. The
remaining streets are not so intensively loaded with traffic of
heavy vehicles, though the rate of the traffic can reach values
exceeding 500 units/day. Fig. 2 and 3 show, respectively, daily
road traffic and structure of vehicle types, measured on The
Kwiatkowski Flyover.
Fig. 2 Daily distribution of vehicle traffic rate measured on The Kwiatkowski
Flyover , Gdynia
Fig. 3 Vehicles type structure observed on the Kwiatkowski Flyover, Gdynia
18 POLISH MARITIME RESEARCH, No S1/2017
On the basis of the performed measurements and data
obtained from in-motion-weighing (WIM) stations, the
number of overloaded vehicles crossing each measurement
point , was estimated. The data from the existing WIM
stations provide information on the mean percentage share
of overloaded vehicles for each vehicle type, separately. The
mean daily number of overloaded vehicles was calculated by
using the following formula :
where:
LP mean daily number of overloaded vehicles
UPi - mean share of overloaded vehicles of i-th type ,
calculated on the basis of the data from WIM stations,
SDRi mean daily traffic rate of vehicles of i-th type,
determined on the basis of measurements.
The traffic rate measurements were conducted between
09 h and 17 h , for this reason the results were appropriately
converted into daily traffic rates. Means share of overloaded
vehicles for particular types of vehicles was determined on
the basis of the data measured by the continuously weighing
station on DK46 road at Grodziec, on a large sample of vehicles,
which contain measurement results collected for two years.
The problem of overloading the vehicles which operate on Polish
roads is serious and leads to degradation of existing roads.
According to the analyses of the data taken from WIM stations,
the share of overloaded vehicles on Polish roads reaches even
25%. Fig. 4 shows the share of overloaded vehicles on some
domestic roads and a highway in Poland. The dealings is greatly
conducive to the shortening of life of surface structure and
significantly influences rate and intensity of surface structure
failures , therefore it is important and justified to control load
of vehicles and , consequently, to protect surface structures.
Fig. 4 Share of overloaded vehicles operating on selected roads in Poland
A legal act in force in Poland , which regulates permissible
mass of vehicles and load exerted onto their axles is The
Decree of Ministry of Infrastructure on Technical Conditions
of Vehicles and Scope of Their Necessary Equipment , dated
December 31, 2002. The decree has been harmonized with
the EU directive WD 96/53 WE. In the above mentioned
regulations values of permissible mass and load per axle for
various types of vehicles are described in detail. The regulations
concerning mass and load per axle of vehicles operating on
public roads are also given in Article 41 of the Act on Public
Roads. According to the Act , the vehicles of the permissible
load per axle of 8 t are allowed to move over most of the roads.
The Ministry of Transport is authorized to determine , by
means of a decree, the list of the roads over which traffic of
vehicles of load per axle of 10 t or 11,5 t is allowed. Any vehicle
whose mass exceeds the limit determined in the Decree or
whose load per axle for at least one component axle exceeds
the limit assigned to a given road on which it moves, will be
considered overloaded. In analyzing influence of overloaded
vehicles on surface structure, are not taken into consideration
vehicles which exceed permissible loads per axle determined
by their producers , resulting from technical parameters of a
given vehicle.
In respect to a way of overloading the vehicles are classified
as follows :
x vehicles which exceed their permissible total mass,
x vehicles which exceed their permissible load per component
axle,
x overloaded vehicles which exceed either their permissible
mass or permissible load per component axle.
It should be added that in an overloaded vehicle
its permissible mass and permissible load per
component axle is not always exceeded simultaneously.
The Tri-city agglomeration together with sea ports and many
transport junctions generates traffic of a significant number of
lorries. A part of them , in view of the above given comments ,
is potentially overloaded and causes an accelerated degradation
of surface structures. For this reason implementation of a
preselective weighing system for vehicles over Gdynia
municipal area, is justified.
A surface structure protection system is in a sense already
in operation because control of vehicles is carried out by Road
Transport Inspection and Police, however its effectiveness is
not sufficiently high , therefore it still requires improvement.
One of the assumptions for a new surface structure protection
system for Gdynia is to make use of the existing potential of
equipment, technology and organization.
One of the main elements of the surface structure protection
system have to be modern devices for heavy vehicles weight
control as well as for traffic monitoring. As assumed in the
project of the surface structure protection system for Gdynia
, the stations for continuous weighing vehicles in motion
(WIM - weight in motion) will be installed as well as location
points for control of vehicles by using legalized balances will
be chosen.
A crucial assumption for such system is to develop a logic
structure for making it possible to connect the new equipment
with the existing town traffic management system TRISTAR.
It is also important to ensure, for the system, continuous flow
of information on potentially overloaded vehicles, from the
side of vehicles controlling services.
The choice of location points for control of vehicles as to
their correct loading was conducted in three phases:
x approval of a preliminary concept,
x detail verification of the preliminary concept acceptance
of its four variants,
x selection of the target variant intended for implementation
in practice.
The selection of the weight control points for vehicles in
motion was based on a multi-criterion analysis in the frames
of which the following was considered :
x expected effectiveness of solutions to be introduced,
x mean daily traffic rate of heavy vehicles over streets of
Gdynia,
x heavy traffic structure,
x number of potentially overloaded vehicles,
x technical conditions on a road and state of its surface
structure,
x grade of a road and character of traffic on it ( local, transit).
Preliminary concept of layout of weight control points for
vehicles
In the preliminary concept it was assumed to limit the
traffic of overloaded vehicles in Gdynia as much as possible.
The choice of the points in question was made on the basis of :
x traffic measurements,
x the model of heavy traffic in Gdynia , presented in Fig. 5 ,
x detail measurements concerning number of overloaded
vehicles,
x assessment of state of town streets.
Based on the model, the cartogram of heavy traffic rate,
shown in Fig. 5 , was produced. As can be seen in the figure,
the biggest heavy traffic rate was observed on the route
leading through The Tri-city Bypass Road , The Kwiatkowski
Thoroughfare , The Kwiatkowski Flyover up to the container
handling base ( at Kontenerowa street). A highly loaded entry
to Gdynia is also Morska street at Cisowa (1800 units/day)
and Hutnicza street (1200 units/day). Winiewski street and
Polska street which lead to port, are also highly loaded by
traffic. Less loaded entries to the town are Chwarznieska
street , Zwycistwa avenue and Wiejska street (the entry from
the side of Pogrze).
In the scope of the preliminary concept of the layout of
weighing points the following locations have been chosen :
Morska street at the boundary of Rumia entry to Gdynia,
Hutnicza street at the boundary of Rumia entry to Gdynia,
Kontenerowa street exit from the container handling base,
The Kwiatkowski Flyover, The Kwiatkowski Thoroughfare
entry to Gdynia from the side of The Tri-city Bypass Road,
Winiewski street its section between The Ofiar Grudnia
70s Junction and Energetykw street, Chwaszczyska street
entry to Gdynia from the side of Chwaszczyno, Chwarznieska
street entry to Gdynia , precisely its section between The
Tri-city Bypass Road and Witomino, Zwycistwa avenue
entry to Gdynia from the side of Sopot , precisely its section
between Kolibki and the supermarket Klif, Chwaszczyska
street, precisely its section between Nowowiczliska street and
Gryfa Pomorskiego street, Morska street , precisely its section
between Obwodowa street and The Kwiatkowski Flyover, The
POLISH MARITIME RESEARCH, No S1/2017 19
Tri-city Bypass Road , precisely its section before its crossing
with Morska street, Wiejska street (Pogrze) , precisely its
section before the traffic circle, The Tri-city Bypass Road ,
precisely its section between The Osowa Junction and The
Chwaszczyska Junction.
Fig. 5 Cartogram of heavy vehicles traffic in Gdynia
Verification of the preliminarily accepted concept of layout
of stations for weighing vehicles in motion
The verification has been aimed at the rejecting of such
points of location in which equipment of WIM stations
would not operate correctly, or their expected effectiveness
in lowering the number of overloaded vehicles would be low.
For verification of the preliminarily accepted points of location
of WIM stations the following was taken into account:
x making use of the existing infrastructure,
x effectiveness of solutions to be implemented,
 percentage of overloaded vehicles operating on the
considered roads,
 drop of the number of overloaded vehicles after putting
into operation the heavy vehicles control system,
 economic profits from the implemented solutions,
 administrative factors,
x technical factors :
 state of surface structures (load carrying capacity,
evenness),
 geometric layout of a road,
 access to electricity and possibility to transfer data.
As a result, were rejected the location points for WIM
stations where the assessment has revealed the following
drawbacks :
x insufficient state of surface structure ( too large
measurement error ),
x sections of too large longitudinal slopes or having to high
bendiness,
x sections on which frequent acceleration and braking
operations of vehicles may happen , i.e. those where too
many crossings equipped with traffic light signalling occur,
20 POLISH MARITIME RESEARCH, No S1/2017
x sections along bridge objects,
x sections not subjected to municipal administration of Gdynia.
To the second phase of planning the WIM stations system,
were taken into account those points of location which
could bring the greatest profits. The basis for the further
verification of the concept in question was a traffic analysis
of overloaded vehicles within the scope of which the
following was performed :
x studies of heavy traffic structure and that of overloaded
vehicles on the basis of data available from continuous
weighing the vehicles in motion,
x measurements of traffic structure , conducted in Gdynia,
x a detail analysis of heavy traffic structure and determination
of mean daily number of overloaded vehicles operating on
roads of Gdynia ,
x preparation of a traffic model of overloaded vehicles for
Gdynia,
x determination of product of daily traffic of overloaded
vehicles and length of their path covered along roads of
Gdynia.
Concept of the best solutions
Out of the preliminarily accepted fourteen location
points of WIM stations, the two were selected : at Morska
street and Winiewskiego street , shown in Fig. 6.
In these locations the possible installation of weighing stations
operating in two working modes was considered , namely :
x WIM-HS stations in which measurement of vehicle mass
and axle load is carried out at usual traffic speed of vehicles
such measurement is loaded by a larger error than in
the case of WIM-LS balances. The WIM-HS balances are
installed first of all for preselection purposes;
x WIM-LS stations in which measurement of vehicle mass
and axle load is carried out at a low traffic speed (210
km/h). This makes it possible to make measurements in
a more exact way than in the case of WIM-HS balances ,
however this is connected with the necessity of directing
the vehicles to control points.
It is technically possible to reach a high exactness of
WIM-HS balances sufficient for unambiguous indication of
vehicle overloading and automatic recognition of a overloaded
vehicle and assignment of a overloading fine to it. Such solutions
are more expensive than the equipment for weight preselection.
Moreover there is a problem of formulating legal regulations
for putting into force a system of automatic execution of fines
on the basis of information from WIM-HS stations only ;
for this reason the reaching of full effectiveness of WIM-HS
systems based on legalized measurements , may take a much
longer time.
In the case of the WIM-LS balances, the carrying out of
legalized control is possible if the stations are operated by
appropriate services. Implementation of a legalized automatic
control system into WIM-LS stations would be easier than in
the case of WIM-HS stations. A drawback of WIM-LS balances
is the necessity of reducing the traffic speed of vehicles.
 seriously and 90 casualties happened. Every year about
700 persons are embroiled in traffic accidents in the town.
The number of accidents and victims has showed an increasing
trend in the interval from 2009 to 2012. Attention should be
paid especially to increasing number of heavy injured persons
by 27% in 2012, compared with that in 2011. However a positive
trend is the decreasing number of casualties since 2011 as well
as the decreasing number of accidents and victims since 2012
(Fig. 7). It shows that the direction of undertaken actions aimed
at improvement of traffic safety on the roads of Gdask , may
be deemed right.

Fig. 6 Recommended location points for WIM stations

A system of preselective weighing the vehicles in motion

is composed of the following elements :
x stations for preselective weighing the vehicles in motion ,
for both traffic directions,
x video-recording zone, for both traffic directions,
x surface structure produced with the use of asphalt
technology together with substructure for traffic load of
KR6 category, placed along at least 50m before and 25m
behind the place of installation of gauges, over the full
breath of traffic lane as well as full breath of roadway within
the range of wearing course satisfying requirements for the
measurement exactness B+(7) according to the specification
COST 323: Weigh in Motion of Road Vehicles, Final
Report ,Appendix 1 European WIM Specification Version
Fig. 7 Number of accidents and their victims in Gdask
3.0, dated August 1999 ;
x internet connection allowing for real-time access to the
data from preselection station by means of an internet In the base year 2013 the road traffic safety indices
application accessible from an arbitrary place through an concerning casualties in the area of the town of Gdask
internet browser , as well as allowing for transmission of amounted, respectively, to:
data packets from a preselection station to a server; x 3,1 casualties per 100 thousand of inhabitants,
x protection of gantry lattice columns against impact, by x 0,6 casualties per 10 thousand of vehicles,
means of guard rails; x 3,1 casualties per 100 accidents,
x protection of gantry lattice columns against possible
climbing their structure by third party; Moreover , in 2013 it was 157,4 injured persons per 100
x electric supply for all the systems elements; thousand of inhabitants.
x design and building a road bay equipped with a weighing In the years 2009 2013 in the area of the town the following
stand. accidents were most frequent :
Additionally, implementation of GPS systems into heavy x runs into pedestrians 954 accidents happened, i.e. 5,1%
vehicles for monitoring their routes is also possible. In case of all accidents resulted in 61 casualties and 949 injured
of occurrence of vehicle overloading it will be easy to identify persons;
location of such vehicle [8]. x side collisions between vehicles 857 accidents happened,
i.e. 41,54 % of all accidents resulted in 10 casualties and
1135 injured persons;
SELECTED PROBLEMS OF ROAD TRAFFIC x back collisions between vehicles 447 accidents happened,
SAFETY i.e. 26,33 % of all accidents resulted in 4 casualties and
666 injured persons;
Crucial problems concerning the safety of road traffic x front collisions between vehicles 274 accidents happened,
participants and directions of actions necessary to improve i.e. 4,59 % of all accidents resulted in 13 casualties and386
it , have been presented on an example of the town of Gdask. injured persons.
For five last years in the area of Gdask , 2861 accidents
in which 3653 persons were injured including 245

POLISH MARITIME RESEARCH, No S1/2017 21

 Pedestrians belong to a group of a very high risk. They
constitute more than a half number of casualties and over
25% injured.
The analysis of safety level conducted according to streets
showed that the greatest number of accidents was recorded
on the following streets of Gdask :
x Grunwaldzka street 213 accidents; i.e. 8,9 % of accidents
in the whole town ;
x Kartuska street 103 accidents; i.e. 4,3 % of accidents in
the whole town ;
x The Z. Wojciech Thoroughfare 86 accidents; i.e. 3,6 %
of accidents in the whole town ;
x Zwycistwa avenue 83 accidents; i.e. 3,5 % of accidents
in the whole town ; Fig. 8 Strategic target of RTS Program for Gdask
x Haller street 70 accidents; i.e. 2,9 % of accidents in the
whole town ; The following directions of undertakings necessary to be
x Sowacki street 69 accidents; i.e. 2,9 % of accidents in implemented , split into pillars and priorities , were prepared
the whole town. to reach the strategic target (Fig. 9).

Based on the identified problems and assessment of

road traffic safety (RTS) in the last years , the target for the
program of improving the RTS in the town has been defined.
The determining of strategic target consists in searching for
such values of the indices which are a resultant of expectations
and possibility of their implementation. In the phase of
developing The Provincial Program of RTS GAMBIT
Pomorski, quantitative targets for all administrative districts
as well as cities were assigned , and the year of 2010 was chosen
to be base one ( in accordance with the assumptions for The
National RTS Program ).
The strategic target of the RTS Program GAMBIT
Gdaski is in line both with The National RTS Program and
The Provincial RTS Program till 2020 GAMBIT Pomorski.
The strategic target in question consists in decreasing the
mean yearly number of casualties by 50% and seriously injured
in road traffic accidents by 40% up to 2020. The assumptions
can be transformed into reaching the following values (Fig. 8):
x number of casualties in road traffic accidents not greater
than 11 persons up to 2020,
x lowering the number of seriously injured persons in road
traffic accidents to less than 28 in 2020.

Fig. 9 Strategy of actions in the RTS Program

One of the conditions for reaching the assumed target

is continuous improving the investigations in the area of

22 POLISH MARITIME RESEARCH, No S1/2017

RTS with taking into account especially control actions and
developing models of influence of selected factors on RTS
level, at simultaneous monitoring implemented solutions and
assessing their effectiveness [1, 10, 5, 6, 20].
Road network is basic element of transport system
which directly impacts daily life of millions of inhabitants
worldwide and economic development of countries of the
world. The system contributes in social, economic and cultural
development. Its great influence on life of people makes that
it is considered a kind of line of living resulting in that
the systems users expect high standards of this element of
transport system.
Risk management is a process of searching for an acceptable
risk level which is a balance between an ideal safety level and
demands of : the system, process or service itself as well as
such factors as profits for users, economic effectiveness or
standards accepted by social groups. It forces , in some way,
risk managing persons to define continuously a tolerated
risk level, especially when social, economy and technological
changes may have an influence on modification of the level [19].
In road engineering the risk management method is a
repeatable and formalized way of proceeding , which connects
risk assessment phase and phase of response to risk. Its task is
to support managing persons in making the decisions aimed
at improvement of road traffic safety as well as minimization
of losses incurred by users of the transport system [9].
One of the directions of actions is safety
state control over the existing network of streets.
Such control consists in a periodical identification of features
and faults found on an existing road , which may endanger
safety of road traffic participants [2].
In the control actions, the development of tools making it
possible to automate the process and improve its effectiveness,
is crucial; this is possible in case of adapting solutions already
implemented in other areas [3, 7].
SUMMARY
The above presented problems constitute only a few
elements selected out of many others dealing with the transport
system in port towns. Development of research methods and
tools for assessing state of the system form necessary directions
of further work. The indicated problems of surface structure
wearing by overloaded vehicles as well as safety of road traffic
participants belong to a very important part of hazards which
occur in port towns. The selection of the aspects discussed in
the above presented analysis was told for the reason that they
have been relatively best recognized in this phase of hazards
identification. The identification and counteraction against
such hazards are especially important from the point of view
of the correct functioning of transport system. The approved
program of improvement of road traffic safety in Gdask
should be under monitoring permanently and its effects -
assessed continuously. It is also necessary to especially identify
hazardous spots and eliminate them by introducing the safety
standards for road infrastructure.
BIBLIOGRAPHY
1. Bergel-Hayat R., Zukowska J. : Road Safety Trends at
National Level in Europe: A Review of Time-series Analysis
Performed during the Period 200012. , Transport Reviews,
Vol. 35, 2015 , No. 5 , pp. 650-671
2. Budzynski M., Jamroz K., Kustra W., Michalski L.:Tools for
road infrastructure safety management - Polish experiences,
Transp. Res. Procedia , 2014 , 3 , pp. 730739.
3. Burdziakowski P., Janowski A., Kholodkov A., Matysik K.,
Matysik M., Przyborski M., Szulwic J., Tysic P., Wojtowicz
A.: Maritime laser scanning as the source for spatial data.
Polish Maritime Research,Vol. 22, No. 4(88), (2015), pp. 9-14.
http://dx.doi.org/10.1515/pomr-2015-0064
4. Cichy M, Kropiwnicki J., Kneba Z: A model of thermal
energy storage according to the convention of bond graphs
(BG) and state equations (SE). Polish Maritime Research,
Vol. 22, No.4 (88) (2015), pp. 41-47.
5. Elvik R., Vaa T.: The Handbook of Road Safety Measures,
Elsevier, 2004.
6. Gaca S.: Speed tests of vehicles and its impact on road traffic
safety ( in Polish ). Politechnika Krakowska ( Cracow
University of Technology), 2002.
7. Janowski A., Jurkowska A., Lewczuk D., Szulwic J., Zaradny
A.: Assessment of cliff stability after the demolition of the
engineering facilities. 14th International Multidisciplinary
Scientific Geo-conference (SGEM) : Geo-conference on
science and technologies in geology, exploration and
mining, SGEM 2014, Vol. II, ISBN 978-619-7105-08-7 /
ISSN 1314-2704, 2014, DOI: 10.5593/SGEM2014/B12/S2.016
8. Janowski A., Nowak A., Przyborski M., Szulwic J.: Mobile
indicators in GIS and GPS positioning accuracy in cities.
2nd International Conference on Rough Sets and Emerging
Intelligent Systems Paradigms (RSEISP) held as part of Joint
Rough Set Symposium (JRS), 2014, Springer International
9. Jamroz K.: Risk management method in road engineering
( in Polish). Politechnika Gdaska, (Gdask University of
Technology) , 2011.
10. Jamroz K., Smolarek L.: Driver fatigue and road safety
on Polands national roads. International Journal of
Occupational Safety and Ergonomics, 19 , (2013) pp.
297309.
11. Kneba Z., Kropiwnicki J.: The initial selection of parameters
for dynamic brake for car engine test dynamometer. Key
Engineering Materials, Vol. 597 (2014), pp. 179-184.
POLISH MARITIME RESEARCH, No S1/2017 23
12. Kropiwnicki J., Kneba Z.: Carbon dioxide potential CONTACT WITH THE AUTHOR
reduction using Start-Stop system in a car. Key Engineering
Materials, Vol. 597 (2014), pp. 185-192. Marcin Budzyski

13. Kropiwnicki J., Kneba Z., Zikowski M.: Test for assessing e-mail: mbudz@pg.gda.pl
the energy efficiency of vehicles with internal combustion
engines. International Journal of Automotive Technology, Dawid Ry
Vol. 14, No. 3 (2013),pp. 479-487. Wojciech Kustra

14. Krystek R.: The integrated road transport safety system Gdask University of Technology
ZEUS ( in Polish), WK, 2009. 11/12 Narutowicza St.
80 - 233 Gdask
15. Szapczyski R.: Evolutionary Planning of Safe Ship Tracks Poland
in Restricted Visibility. Journal of Navigation ,Vol. 68, No.
01 (2015), pp.39-51

16. Szapczyski R., Szlapczyska J.: A Target Information

Display for Visualizing Collision Avoidance Manoeuvres
in Various Visibility Conditions. Journal of Navigation,Vol.
68,No. 06 (2015), pp.1041-1055

17. Szapczyski R., Szapczyska J.: Customized

crossover in evolutionary sets of safe ship trajectories.
International Journal of Applied Mathematics and
Computer Science. -Vol. 22, No.4 (2012),pp.999-1009 ,
ISBN:978-3-319-08729-0; 978-3-319-08728-3, DOI:
10.1007/978-3-319-08729-0_31. http://link.springer.
com/10.1007/978-3-319-08729-0_31

18. Szapczyski R., Szapczyska J.: On evolutionary computing

in multi-ship trajectory planning. Applied Intelligence ,2011,
pp.1-20

19. Technical Committee 18 : Study on Risk Management for

Roads, PIARC, 2004.

20. Wachnicka J., Jamroz K., Smolarek L.: Hierarchical random

models in road transport safety. Safety and Reliability:
Methodology and Applications, 2014, pp. 355360.

24 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 25-31
10.1515/pomr-2017-0017

COMPOSITE GFRP U-SHAPED FOOTBRIDGE

Jacek Chrcielewski
Mikoaj Mikiewicz
ukasz Pyrzowski
Krzysztof Wilde
Gdansk University of Technology, Poland

ABSTRACT

The paper presents proposals for the use of glass fiber reinforced polymer composites for the construction of engineering
objects, known and commonly used in the shipbuilding industry. An example of a pedestrian footbridge was used in
this case, which, despite the considerable thickness of the structural material, was made using infusion technology in
one production cycle. The designed and produced footbridge span is durable, dynamically resistant, incombustible, easy
to install and maintain, resistant to weather conditions and also aesthetically interesting. For footbridge production
environmentally friendly PET foam core may be used. It may come from recycling of used plastic packages and which
is produced with less energy consumption process and much less CO2 emission. The load bearing part of the structure
(skin) is made of polymer laminate reinforced with glass fabrics (GFRP).

Keywords: polymer composites, sandwich structures, footbridge, fobridge

INTRODUCTION they have been used from the very onset of their invention.
Composites now find use in the production of small craft as
Polymer composites are structures consisting of two or well as ships or parts thereof [22]. One example is the modern
more components. The primary element is reinforcement, ferry, Vision of The Fjords (see Fig. 1), launched in 2016.
which affects the strength and rigidity of the composite
element. In the composite constructions of the shipbuilding
industry and in the land transport infrastructure, the
reinforcement is usually made of glass, carbon, or aramid
fibers. The second essential component of the composite
is the matrix, which provides the shape, the appropriate
distribution of reinforcement, and the distribution of load
along the composite volume. The shell material may be, for
example, a polyester, vinyl ester or epoxy resin. Composite
structures are often combined into so-called sandwich
structures, which additionally employ a lightweight filler Fig. 1. The passenger ferry Vision of The Fjords [www.braa.no]
core to maintain the distance of the outer bearing skins. The
most important advantages of this material are: low volume Currently, several technologies of composite construction
weight, relatively high rigidity and strength, high resistance are in use. The most popular are hand layup, pultrusion,
to atmospheric [14] and chemical factors, and great flexibility and vacuum infusion. In shipbuilding applications, due
in geometrical shaping. to dominating production of surface elements, infusion is
All of the above factors cause the share of polymer primarily used. This technology is based on manufacture of
composites in various industries to grow year by year. This composite structures by placing dry reinforcement in mold,
is particularly evident in the shipbuilding industry, where

POLISH MARITIME RESEARCH, No S1/2017 25

on a pre-applied gel coat. Next, a delamination fabric and
a mesh to improve the flow of the resin are put into place,
and the whole assembly is sealed in a vacuum-sealed bag.
The final step is to create a vacuum to distribute the resin,
which must uniformly infuse the reinforcement. Infusion
technology has many advantages: it provides great elasticity
in forming, reduces the volatilization of chemicals in the
production process, the components are characterized by a
high degree of reinforcement and can be used for sandwich
structures. An example production photo is shown in Fig. 2.

Fig. 3. Composite span of the FOBRIDGE walking and cycling bridge

The characteristics of the composite material, especially its

low weight, offer designers much freedom. This is especially
apparent in drawbridge construction. The use of lightweight
Fig. 2. Preparation to infusion of a large-size object - footbridge span of platforms offers the possibility of designing more slender
Fobridge grant girders and mechanisms for lifting or rotating the spans
[18], [25]. One example of polymer composites used in a
The advantages of polymer composites make them likely drawbridge is the Dragons Bridge in Rhyl, Wales, put to
to be used not only in shipbuilding or aviation, but also in use in 2013 (Fig. 4).
land transport infrastructure. The specific application is
bridge-building [6], [7], [10], [26]. In recent years, a growing
number of engineering designs have been developed that
utilize this material in whole or in part. An example of a
composite structure is the Ubridge pedestrian-cyclist bridge,
which was created within the framework of the FOBRIDGE
project, realized by the consortium of the Gdask University
of Technology, the Military Academy of Technology in
Warsaw, and Roma Ltd. in 2015 [6],[20]. One of the goals of
the project was to make a footbridge competitive in terms
of price with traditional steel, concrete, or wood bridges. To
achieve this, both the material (glass reinforced composite
sandwich-construction based on vinyl ester resin) and the
production technology (infusion) used in other applications,
such as shipbuilding, were applied. The FOBRIDGE project Fig. 4. The drawbridge Dragons Bridge in the port of Rhyl in Wales [www.
assumes that the bridge span will be a freely supported and gurit.com]
constructed as U-shaped cross-section girder (Fig. 3) made
of a three-ply sandwich shell with laminated, glass fiber
reinforced skins, separated by a thick PET foam core. This
example of the use of composite materials in construction ARCHITECTURAL CONCEPTS
will be described in detail in this paper. AND DESIGN DESCRIPTION
Various forms of construction, with the assumed theoretical
single span in the range of 12 to 18 m were considered in the
architectural conceptual work. Selected architectural variants

26 POLISH MARITIME RESEARCH, No S1/2017

 Fig. 5. Architectural concepts of composite footbridge: color, light and texture [6], [7]

of the footbridge, using color, light and texture, are presented height of only a dozen or so centimeters. The side walls of
in Fig. 5. the U-shape are also handrails, with a height of 1.3 m. The
The footbridge is designed to facilitate pedestrian and axial geometry of the support construction is described by
bicycle traffic over obstacles as well as to incidentally allow an acircular arc, and the maximum longitudinal drop does not
ambulance or service vehicle to pass. The resulting structural exceed 8% (Fig.6).
design of the span, its lightness, and hence its small support
requirements make it perfect for use over any obstacle (road,
railway, waterway, off-shore platforms, etc.). Due to these
features and the short production time, the proposed spans
can be particularly useful in disaster areas, where immediate
reconstruction of destroyed communication routes with
minimum technical requirements is vitally important. In
addition, using the U-shape structural design, sandwich
components were incorporated into the structural work of
the system, hence the object has a very small construction

Fig. 6. Side view and cross-section of the span

POLISH MARITIME RESEARCH, No S1/2017 27

 The span was produced entirely with infusion technology,
in one technological cycle, and the produced full-scale
footbridge corresponds to the architectural vision (Fig.7).

Fig. 8. Selected material identification studies: laminate and foam and

validation of the sandwich structure calculation model [8], [10]

Fig. 7. FOBRIDGE span: top - architectural vision, bottom - finished product

IDENTIFICATION AND VALIDATION

STUDIES
The process of designing and manufacturing an innovative
composite bridge required solving various problems. The
most important were: absence of standard guidelines for
designing composite elements in construction applications,
no standardized material data, no guidelines for calculating
and evaluating material strain, and no guidelines for bulk
infusions and resulting thick spacers. Finding the answers
required a number of experimental, validation, and numerical
studies and many technological trials related to the selection
of material parameters and control of the infusion process. Fig. 9. Three-meter long validation footbridge segment
with target cross-section
Experimental studies performed included:
identification of mechanical parameters of materials
analogous to [1], [2]; For numerical investigations the usual solutions of finite
static and dynamic testing of samples in the technical element method, highly advanced calculation techniques
scale of the finished sandwich [23]; developed by these authors and commercial FEM tools were
structural and local tests with acoustic emission used (e.g. [5], [11], [13], [15], [16], [17], [24]. These programs
measurements [12], of a 3-meter segment with target have been tested for their suitability for modeling composite
cross section [8],[19] (Fig. 9); spacer structures. The ROBOT, SOFiSTiK, and more advanced
load-carrying test of full-size object [9]. ABAQUS and MSC Marc, NX NASTRAN (Fig. 10) programs
were acquired. As a result of these studies, an algorithm and
In this latter study group, long-term technical monitoring computational model was developed for solving engineering
of the bridge span [21] was performed. Among other things, problems with laminates, where high compatibility between
the purpose of experimental research was to obtain data for simulations and real experiments was achieved. The developed
the formulation and validation of analytical and numerical computational model was used to design a full bridge span.
computing models of laminate, foam, sandwich, and structural
work of the entire system.

28 POLISH MARITIME RESEARCH, No S1/2017


Fig. 10. Visualization of the selected target bridge span model (top)
and its 1st form of natural vibration
In the subsequent stages of the study, multiple selection
criteria were selected and finally a full-size test object
was produced using a vinyl ester resin as the matrix and
orthogonally stitched fabrics, quasi-balanced, made of
typeE fiberglass, as reinforcement of the shell laminate, and
perforated PET foam with a density of 100 kg/m3, forming
a100-mm core of a three-layer sandwich.
INSTALLATION AND ACCEPTANCE TESTS
The research object of the FOBRIDGE project, in the form
of a full bridge span of a total length of 14.5 m, was transported
from Grabowiec near Toru to the Gdansk University of
Technology by means of standard wheel transport (Fig. 11)
and assembled using a small crane (Fig. 12). The total mass
of the span was approximately 3.3 t.
Fig. 11. Transporting the footbridge span
In order to facilitate further research, the span was situated
on reinforced concrete road-slabs, simulating low-rise
bridgeheads. Ultimately, a special bearing design was used
to secure the bridge from breaking off, which can be caused
by winds in the case of such lightweight constructions. The
span is only 14 m long, hence the bridge is classified as a small
object, as bridges go. Nevertheless, for research purposes,
it was decided to monitor a wide range of its parameters.
For this reason, a total of 216 different types of measuring
points were assumed in the main static and dynamic surveys,
including the typical bridge approval load tests. In the
basic static test, concrete road slabs were laid in different
configurations, with a total weight of 141 kN (Fig.12) [9].
In addition, a live crowd load of over 100 people was used.
Laser scanning was performed on one side of the span [3],
[4] and measurement points included: strains (electrical
resistance strain gauges, extensometric and fiber optic gauges),
displacements (induction sensors, geodetic measurements),
bearing distortion, support settlement, and construction
temperature measurements. Dynamic tests (Fig.12) included
measurements of strains (electrical resistance strain gauges
and fiber optic gauges), displacement (induction sensors),
accelerations, and angular velocities.
Fig. 12. Example tests: static loading with concrete road slabs (top); dynamic,
synchronous jumps by a group of pedestrians (bottom)
SUMMARY
As a result of the study, a lightweight bridge-type object
has been developed and manufactured with many desirable
properties: durable, dynamic, flame retardant, easy to
assemble and maintain, resistant to atmospheric factors, and,
moreover, aesthetic. The sandwich core material is made of
environmentally friendly PET foam, which can be produced
using significantly less energy-consuming processes than
POLISH MARITIME RESEARCH, No S1/2017 29
those used for producing more traditional materials, with
reduced CO2 emissions, and utilizing materials from recycled
plastic bottles. A modern construction material, glass fiber
reinforced polymer laminate, was used to make the footbridge
structure, and the entire girder is one compact piece - without
foreign parts that could weaken the object and corrode
faster. It is assumed that upon termination of the footbridge
usability period, the fiberglass and the PET foam core will
be recovered and will continue to be used in subsequent
constructions. Another advantage is that the process of
manufacturing takes only a few days utilizing the standard
infusion technology. This process allows us to obtain a high-
quality product every time with a significantly reduced price
from other manufacturing methods. The U-bridges final price
will be comparable to traditional solutions (steel, concrete,
wood) already at the time of purchase and, taking into
account the 50-year period of usability with zero necessity of
maintenance, will ultimately be significantly lower. Due to the
low weight of the construction, only lightweight equipment
will be required for assembly, and the supports will be small
in even the most unfavorable geotechnical conditions. Other
properties of note include the low constructional height
(0.14m) and excellent dynamic parameters of the object
(7.8Hz first frequency of vibration). It is assumed that the
target footbridge, even at the prefabrication stage, due to the
use of fiberglass as reinforcement, will be able to be equipped
with a technical monitoring system based on fiber Bragg
grating (FBG) sensors.
BIBLIOGRAPHY
1. Ambroziak A., Kosowski P.: Mechanical properties of
polyvinyl chloride-coated fabric under cyclic tests. Journal
Of Reinforced Plastics And Composites. Vol. 33, No. 3
(2014), pp. 225-234
2. Ambroziak A., Kosowski P.: Mechanical testing of
technical woven fabrics. Journal Of Reinforced Plastics
And Composites. Vol. 32, No. 10 (2013), pp. 726-739.
3. Bobkowska K., Janowski A., Przyborski M., Szulwic J.:
Analysis of High Resolution Clouds of Points as a Source
of Biometric Data. 2016 Baltic Geodetic Congress (BGC
Geomatics)/ : IEEE, 2016, pp.15-21
4. Burdziakowski P., Janowski A., Kholodkov A., Matysik K.,
Matysik M., Przyborski M., Szulwic J., Tysic P., Wojtowicz
A.: Maritime laser scanning as the source for spatial data.
Polish Maritime Research, Vol. 22, No. 4(88) (2015), pp.
9-14.
5. Burzyski S., Chrcielewski J., Witkowski W.:
Geometrically nonlinear FEM analysis of 6-parameter
resultant shell theory based on 2-D Cosserat constitutive
model. ZAMM- Zeitschrift fuer Angewandte Mathematik
und Mechanik, Vol. 96, No. 2 (2015), pp.191-204
30 POLISH MARITIME RESEARCH, No S1/2017
6. Chrcielewski J., Klasztorny M., Mikiewicz M.,
Romanowski R., Wilde K.: Innovative design of GFRP
sandwich footbridge, Proc. 5th International Conference
on Footbridges: Past, Present & Future, Footbridge 2014,
London, England, 1618 July 2014, Paper #1250, pp. 18
7. Chrcielewski J., Klasztorny M., Wilde K., Mikiewicz M.,
Romanowski R.: Composite sandwich bridge for pedestrians
and bicyclists ( in Polish), Materiay Budowlane 7/2014
(No. 503), pp. 1-2.
8. Chrcielewski J., Mikiewicz M., Pyrzowski ., RuckaM,
Ferenc T.: Experimental tests of a validation segment of
composite foot - bicycle bridge (in Polish). Materiay
Budowlane 4/2015 (No. 512), pp. 72-73.
9. Chrcielewski J., Mikiewicz M., Pyrzowski ., Wilde K.:
Tests of a composite footbridge ( in Polish). Mosty, 1/2016,
pp. 44-49.
10. Chrcielewski J., Mikiewicz M., Wilde K. A composite
foot - bicycle bridge. ( in Polish). II Konferencja Naukowo-
Techniczna Wspczesne Materiay, Techniki i Technologie
we Wspczesnym Budownictwie ( 2nd Scientifical Technical
Conference on Contemporary Materials, Techniques and
Technologies in Today Building Industry), Cracow, 2123
-10-2015, pp. 67-74
11. Chrcielewski J., Witkowski W.: Four-node semi-EAS
element in six-field nonlinear theory of shells. International
Journal for Numerical Methods In Engineering 68(11),
2006, pp. 1137-1179.
12. Grelowska G., Kozaczka E., Nowicki A., Kozaczka S.:
Investigation of Transmit and Receive Characteristics of
Laboratory Model for the Parametric Echosounder. Acta
Physica Polonica A., Vol. 123, No. 6 (2013), pp.1094-1097
13. Iwicki P., Tejchman A., Chrcielewski J.: Dynamic FE
simulations of buckling process in thin-walled cylindrical
metal silos. Thin-Walled Structures. Vol. 84 (2014),
pp.344-359.
14. Jakubowski M.: Influence of pitting corrosion on fatigue
and corrosion fatigue of ship and offshore structures. Part
II: Load - pit crack interaction. Polish Maritime Research.
-Vol. 22, No. 3 (2015), pp. 57-66.
15. Klasztorny M., Bondyra, A., Szurgott, P., Nycz, D.:
Numerical modelling of GFRP laminates with MSC.
Marc system and experimental validation. Computational
Materials Science, vol. 64, 2012, pp. 151-156. DOI: 10.1016/j.
commatsci.2012.05.024
16. Kaliski K.: The finite element method application to linear
closed loop steady system vibration analysis. International
Journal of Mechanical Sciences 39 (3), 1997, pp. 315330.
17. uczak M., Manzato S., Peeters B., Branner K., BerringP.,
Kahsin M.: Updating Finite Element Model of a Wind CONTACT WITH THE AUTHOR
Turbine Blade Section Using Experimental Modal Analysis
Results. Shock And Vibration, No. 1 (2014), pp. 71-82. Jacek Chrcielewski

18. Mikielewicz J., Mikielewicz D. : A simple model of circular Mikoaj Mikiewicz

hydraulic pump. International Journal of Heat and Mass
Transfer, Vol. 52(1), (2008) pp.17-21,. ukasz Pyrzowski

19. Mikiewicz M, Daszkiewicz K., Ferenc T. Witkowski W Krzysztof Wilde

and Chrcielewski J.: Experimental tests and numerical
simulations of full scale composite sandwich segment of a Gdask University of Technology
foot-and-cycle bridge. Advances in Mechanics: Theoretical, Faculty of Civil Engineering and Environment
Computational and Interdisciplinary Issues Kleiber et al. 11/12 Narutowicza St.
(Eds), Taylor & Francis Group, London, 2016, pp. 401-404. 80 - 233 Gdask
Poland
20. Mikiewicz M., Okraszewska R., Pyrzowski . : Composite
footbridge synergy effect in cooperation between universities
and industry. ICERI2014: 7th International Conference of
Education, Research and Innovation, ICERI Proceedings,
(2014), pp. 2897-2903.

21. Mikiewicz M., Pyrzowski ., Chrcielewski J., Wilde K.:

Structural Health Monitoring of Composite Shell Footbridge
for Its Design Validation. Proceedings 2016 Baltic Geodetic
Congress (Geomatics)/ ed. Juan E. Guerrero Los Alamitos:
IEEE Computer Society Order Number E5972, 2016, pp.
228-233.

22. Niklas K., Kozak J.: Experimental investigation of Steel

ConcretePolymer composite barrier for the ship internal
tank construction. Ocean Engineering. Vol. 111, (2016),
pp. 449-460.

23. Pyrzowski ., Sobczyk B., Witkowski W., Chrcielewski J.:

Three-point bending test of sandwich beams supporting
the GFRP footbridge design process validation. 3rd Polish
Congress of Mechanics (PCM) / 21st International
Conference on Computer Methods in Mechanics (CMM),
2016, Taylor & Francis Group, London, pp. 489-492.

24. Sabik A., Kreja I.: Large thermo-elastic displacement and

stability FEM analysis of multilayered plates and shells.
Thin-Walled Structures 71, (2013), pp.119-133.

25. Siemitkowski M., Przybylski W. : Simulation studies of

process flow with in-line part inspection in machining cells.
Journal of Materials Processing Technology Volume 171,
No. 1, January 10, 2006, pp. 27-34.

26. Siwowski T., Kulpa M., Poneta P.: Research on application

of FRP composites to bridge building in Poland ( in Polish).
Materiay Budowlane 11/2014

POLISH MARITIME RESEARCH, No S1/2017 31

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 32-37
10.1515/pomr-2017-0018

CAUSALITY IN MODELS OF THERMAL PROCESSES

IN SHIP ENGINE ROOMS
WITH THE USE OF BOND GRAPH (BG) METHOD

Marian Cichy
Zbigniew Kneba
Jacek Kropiwnicki
Gdansk University of Technology, Poland

ABSTRACT

With a single approach to modeling elements of different physical nature, the method of Bond Graph (BG) is
particularly well suited for modeling energy systems consisting of mechanical, thermal, electrical and hydraulic
elements that operate in the power system engine room. The paper refers to the earlier presented [2] new concept of
thermal process modeling using the BG method. The authors own suggestions for determining causality in models
ofthermal processes created by the said concept were given. The analysis of causality makes it possible to demonstrate
the model conflicts that prevent the placement of state equations which allows for the direct conduct of simulation
experiments. Attention has been drawn to the link between the energy systems models of thermal processes with
models of elements of different physical nature. Two examples of determining causality in models of complex energy
systems of thermal elements have been presented. The firs relates to the electrical system associated with the process
of heat exchange. The second is a model of the mechanical system associated with the thermodynamic process.

Keywords: Bond Graph (BG) method, causality, ship engine room, energy system modeling

conditions. Marine vessels are characterized by the autonomy

INTRODUCTION
The social growth of expectations in terms of increasing
the efficiency of machinery and equipment as well as reducing
greenhouse gas emissions and air pollutants forced in recent
years the intensification of research and development to
improve the efficiency of heat exchange processes [4, 18],
the structural improvement of working machines and power
equipment [7, 14, 16, 24], improving the quality of diagnostics
of these devices [11, 12] and the improvement of efficiency of
the manufacturing process of machine elements [6, 9]. One
of the more complex energy systems, which is covered by the
above-mentioned development work is a ships engine room.
A ship engine room is an energy system which includes
machinery and equipment whose task is to convert chemical
Energy into mechanical Energy, electrical and thermal
quantities and specifications to ensure the correct operation
of receivers necessary for the fulfillment of the assumed unit
functions, in all states of operation and the variable external
of energy, which means that all the energy needed must be
met by using fuel consuming elements like main engines,
auxiliary boilers, etc. . The proper design, as well as the
operation of such a system of energy requires a comprehensive
examination of thermal, mechanical and electric processes.
The method of Bond Graphs (BG) which is the core of this
study was initiated by Paynter [20] and has been developed in
a number of monographs [1, 3, 10, 26, 27, 28] and publications
[5, 8, 13, 15, 17, 19, 21, 22, 23, 25]. Thanks to a singular approach
to modeling elements of different physical nature, this method
is especially suited for modeling energy systems consisting of
mechanical, thermal, electrical and hydraulic elements which
operate in the power system of an engine room.
The flow of Energy between the individual components
of the Energy system is marked with graphs (half-arrow)
automatically introducing information about the direction
of the flow. The energy flow is always described by two
parameters of energy: metastatic potential (e) and generalized
flow (f ). In the case of the internal combustion engine,

32 POLISH MARITIME RESEARCH, No S1/2017

the parameters characterizing the energy receiver to be
transmitted are respectively Mo (e) and (f ). The product of
Energy parameters (e and f ) must always give the dimension
of power.
Models of Bond Graphs composed of external elements:
(fig. 1): S (source of energy), L (kinetic energy storage), C
(potential energy storage), R (dissipative element) and the
internal components: energy converters, node 0 of the same
generalized potential and nodes 1 of the same generalized
flow. The energy sources present in the modeled object,
according to the formalism of BG may exist in two forms:
as a source of potential: SE (systemic independent potential
is a function of time), or as a source of flow: SF (systemic
movement is an independent function of time).
The model itself in the form of BG reflects clearly the
dynamic structure of the object and can be easily modified,
but does not allow for the direct conduct simulation
experiments [3]. The BG model allows for a manual or
automatic generation of equations of state, which means
creating a mathematical model of cause and effect.
Using the causality discussed in this paper, one can
describe the relationship between metastatic potential (e)
and generalized flow (f). By convention, the designation of BG
causality edge is represented by short dash perpendicular line
at one end of the Edge (fig. 4). The position of this line indicates
which value e or f is a dependent and which independent value
(e.g. fig. 4 a). Such value is T(t)). The analysis of causality
makes it possible to demonstrate conflicts that prevent the
placement of state equations. These conflicts can be of two
types:
a conflict related to the use of improper or inadequate
energy source for the energy source used, as of the internal
structure of the model,
a conflict caused by the introduction of interdependent
power accumulators into the model.
Fig. 1 Diagram and designation of BG model Energy system to determine the
efficiency of: egeneralized potential, f generalized flow, p generalized
momentum, q generalized displacement [3]
The basic theory of BG concept of causality and use of it
to analyze the BG models i clearly shown [1, 3, 10, 26, 27]
in the construction of models of mechanical, electrical and
hydraulic processes. Doubts arise when creating BG models
of thermal processes, especially regarding the new elements
introduced into these models as a dual battery [1, 27] (2-port)
or the modeling of thermal processes using so-called pseudo
Bond Graphs [1, 3, 8, 10, 26, 27]. In [2, 3] a new method of
modeling thermal processes has been presented using only
elements introduced in the basic BG theory form modeling
of mechanical, hydraulic and electrical processes. In these
works the analysis of causality was deliberately not used
since according to the theory proposed in separate models
of thermal processes, there was no possibility of a conflict
of causality [3].
In [2] it was found that the Energy storing models of
thermal processes, as in the basic theory, should be in the form
of a graph (bond) with a single node (1-port). The concept of
an energy transducer has been introduced [2, 3] (fig. 2). In
the basic theory of BG the transducer exists in the form of a
transformer or gyrator. In general theory of thermodynamics
and heat exchange there are physical concepts that could
justify the introduction of such a distribution of energy
converters. Introduced in [2] energy converters must comply
only with the basic theory of the BG relationship (1) of the
net amount of energy flow in any multiple node (multiports),
in which the n graphs (bonds).
Fig. 2. Power converter in BG model [2]
n
e f i i =0 (1)
i =1
When modeling the extracted thermal processes, the
available literature does not provide a specified example
where a conflict of causality would be shown. Numerous
examples of creating equations of state based on the BG
model in [2, 3] showed that the determination of causality
can only serve to illustrate the flow of energy. However, the
need for analysis of causality exists only when the systems
are modeled, in which the nest thermal processes taking
place relate to processes of different physical nature such as
the mechanical or electrical one. In real power systems such
links always take place. Simulation studies using BG model
can be carried out when the graphic model allows to generate
equations of state in the form of:
 = f (X,U )
X (2)
where: X vector of state variables,
U vector of controls.
The conflict of causality prevents the arrangement of the
equation in the form of (2).
POLISH MARITIME RESEARCH, No S1/2017 33
 CAUSALITY IN BG ELEMENTS IN
THERMAL PROCESSES MODEL
The basic elements of BG, which should clearly define
the causality are the heat energy accumulators and energy
sources. As shown in [2] the thermal processes, there is only
one type of energy storage accumulator wherein the state
variable characterizing is the temperature of the kinetic
energy of the molecules. There is no physical evidence to
assign two kinds of causality to heat accumulators just as it
has been, for example, in the mechanical processes, namely
integral and differential causality. According to this reason,
causality of heat accumulator is marked clearly as shown in
Fig. 3. According to the general theory of BG it is causality of
the generalized potential of the kinetic energy accumulator. Fig. 4. Sources of Energy and determination of causality in the thermal
processes BG model

CAUSALITY IN COMPLEX BG MODELS

Fig. 3. Determination of causality of thermal energy accumulator
in the BG model
The general form of the dependence defining the flow
of the accumulated heat is presented by the relation:
d(Cv mT )
TS A = E A =
dt
For many cases, for example, heat conduction, it can be
assumed that the product is Cvm=const, then . environment F.
In work [2] various types of heat sources have been shown,
which are graphically shown in Fig. 4. In the general theory
of BG [1, 3, 10, 27] the causality of energy sources is uniquely
determined for a source of potential or sources of movement.
Case b on Fig. 4 shows an energy source in the form of
mass flow of agent of enthalpy i. There is here the special
case where both generalized parameters can be independent
time functions. Fig. 4 indicates, as example, a generalized
flow causality. Determination of the energy gates, which are
regarded arbitrarily as a source of Energy, depends on the
purpose served by the built model. For example, in the electric
power network as an energy source may be indicated the
energy flux in the fuel supplied to the boiler steam turbine,
or otherwise as an energy source one can treat the electrical
energy received from the generator current. In the first case,
causality will be designated as the source of the flow, and in
the second one, as source of potential.
As a first example of a complex system in which the thermal
processes are associated with processes of different physical
nature an electric circuit with a resistor will be considered
(fig. 5) in which the heating and heat exchange with the
environment takes place. The BG model of these processes
is shown in Fig. 6 and consists of the electrical circuit model
(3) and the model of the heating system. The source of power
is the voltage generator V(t) of the internal resistance R s.
The external resistor R is the thermal energy accumulator of
mass m, specific heat Cv and surface of heat exchange with
R
i
m, F, Cv ,
Q
RS
V(t)
SE
Fig. 5. The electrical circuit in conjunction with a thermal process
The output equation of the state follow directly from the
BG model shown in Fig. 6:

(4)

In this equation, the flow of heat exchange with the

environment is determined by the relationship:

34 POLISH MARITIME RESEARCH, No S1/2017

 (5) The output equation of state, as before, follows directly
form the BG model shown in Fig. 9:

where: heat transfer coefficient of the resistor surface area (6)

F and temperature T to ambient temperature TO.

Fig. 6. BG model of related systems: electrical and thermal

As already mentioned, the determination of causality of

the BG model of the electrical system in Fig. 6 has been made
according to the general principles of the BG theory [3, 10, 27]. Fig. 8. Diagram of the related systems: mechanical and thermodynamic
According to these rules, the causality scattering element R
can be arbitrary. Assuming the causality of heat accumulator
C according to the above accepted rule, the causality of other
thermal model graphs can be marked in different ways as
shown in Fig. 6 and 7. In any of these cases, there is no
conflict of causality. Assuming the rules introduced above,
the causality of the PE energy transducer results from the
adopted arbitrary causality of dissipation elements and can
be gyrator causality (fig. 6) or of the transformer (fig. 7). Fig. 9. The BG model of the related systems: mechanical and thermodynamic

V(t) VR T T
a) SE
i 1 i PE S SR
R
Using the basic physical dependence in both of the above
VS i T SA examples, the output equations of state (4) and (6) should be
presented in the form of (7) [2] enabling simulation research:
R C (7)

b) V(t) VR T T
SE 1 PE R
i i S SR
CONCLUDING REMARKS
VS i T SA
In previous papers [2, 3] a new method of modeling thermal
R C processes using the BG method has been presented. It has
Fig. 7. Two additional versions of the possible designation of BG causality been shown that the thermal process models can be created
model from Fig. 6 without entering new elements into the BG theory such as
Pseudo Bong Graphs and dual elements of capitalization and
The second example concerns the relationship of the distraction. There is also no need to introduce the gyrator [1]
mechanical system in which the change in volume V into the BG models as in the theory of thermodynamics and
by changing the predetermined piston speed v(t), of the heat exchange there is no equivalent of this concept, as is the
thermodynamic system, wherein the solid mass of gas m is case of mechanical and electrical processes. By creating a BG
in variable volume V(t) (fig. 8). The existence of friction force Ft model covering only thermal processes, there is no possibility
was assumed in a piston cylinder and the introduction of the of a conflict of causality, mainly due to the existence of only
heat flux Q(t) to the gas medium. The BG model of the related one type of heat energy accumulator. In complex energy
system indicating causality is shown in Fig. 9. In this system, systems, the heat processes are associated with the processes
there are two energy converters. The first one is atransformer of different physical nature, and in this case, the analysis of
(TR) with unambiguous causality, the second a transduce in causality BG model is justified and necessary.
a thermal system (PEVS), whose causality can be arbitrary.

POLISH MARITIME RESEARCH, No S1/2017 35

BIBLIOGRAPHY
1. Borutzki W.: Bond Graphs a Metodology for Modelling
Multidisciplinary Dynamic Systems. Springer, (2010).
2. Cichy M., Kropiwnicki J., Kneba Z.: A Model of Thermal
Energy Storage According to the Convention of Bond
Graphs (BG) and State Equations (SE). Polish Maritime
Research, Vol. 22, nr 4 (88) (2015), pp. 41-47.
3. Cichy M.: Modelowanie systemw energetycznych (Mod-
eling of energetic systems). Wydawnictwo Politechniki
Gdaskiej. Gdask, (2001), (in Polish).
4. Cieliski J. T., Mosdorf R.: Gas bubble dynamics -
experiment and fractal analysis. International Journal
of Heat and Mass Transfer Volume 48, Issue 9, (2005),
pp. 1808-1818.
5. Creyx M., et al.: Dynamic modelling of the expansion
cylinder of an open Joule cycle Ericsson engine: A bond
graph approach. Energy 102 (2016), pp. 31-43.
6. Deja M., Siemitkowski M. S.: Feature-based generation
of machining process plans for optimised parts manufac-
ture. Journal of Intelligent Manufacturing (2013), Volume
24, Issue 4, pp. 831-846.
7. Domachowski Z., Dzida M.: Inlet Air Fogging of Marine
Gas Turbine in Power Output Loss Compensation. Polish
Maritime Research 4 (88) (2015), Vol. 22, pp. 53-58.
8. Hubbard M., Brewer J. W.: Pseudo Bond Graphs of cir-
culating fluids with Application to Solar Heating Design.
Journal of the Franklin Institute Vol. 311, No 6, (1981),
pp. 339-354.
9. Kaliski K. J., Galewski M. A.: Chatter vibration sur-
veillance by the optimal-linear spindle speed control.
Mechanical Systems and Signal Processing Volume 25,
Issue 1, (2011), pp. 383-399.
10. Karnopp D. C., Margolis D. L., Rosenberg R. C.: System
dynamics: a unified approach. Wiley, New York, (1990).
11. Korczewski Z., Zacharewicz M.: Alternative diagnostic
method applied on marine diesel engines having limited
monitoring susceptibility. Transactions of the Institute
of Measurement and Control. Vol. 34, No. 8 (2012),
pp.937-946.
12. Korczewski Z.: Exhaust Gas Temperature Measurements
in Diagnostics of Turbocharged Marine Internal Com-
bustion Engines. Part II Dynamic Measurements. Polish
Maritime Research 1 (89) (2016) Vol. 23, pp. 68-76.
36 POLISH MARITIME RESEARCH, No S1/2017
13. Kortas P., Kropiwnicki J.: Analysis of accumulation pos-
sibility of energy dissipated in the braking process of
train driven by hybrid locomotive. Combustion Engines,
(2015), pp. 631-638.
14. Kowalczyk T., Guch J., Zikowski P.: Analysis of Possible
Application of High-Temperature Nuclear Reactors to
Contemporary Large-Output Steam Power Plants on
Ships. Polish Maritime Research 2 (90) (2016), Vol. 23,
pp. 32-41.
15. Kropiwnicki J.: Comparison of energy efficiency of
vehicles powered by different fuels. Combustion Engines,
nr 3, (2012), pp .34-43.
16. Litwin W., Olszewski A.: Water-Lubricated Sintered
Bronze Journal Bearings - Theoretical and Experimental
Research. Tribology Transactions. Vol. 57, No. 1 (2014),
pp.114-122.
17. M.S. Jha, et al.: Particle filter based hybrid prognostics
of proton exchange membrane fuel cell in bond graph
framework. Computers and Chemical Engineering 95
(2016), pp. 216-230.
18. Mikielewicz D., Mikielewicz J., Tesmar J.: Improved semi-
empirical method for determination of heat transfer coef-
ficient in flow boiling in conventional and small diameter
tubes. International Journal of Heat and Mass Transfer
Volume 50, Issues 19-20, (2007), pp. 3949-3956.
19. Mishra C., Samantaray A.K., Chakraborty G.: Bond graph
modeling and experimental verification of a novel scheme
for fault diagnosis of rolling element bearings in special
operating conditions. Journal of Sound and Vibration
377 (2016), pp. 302-330.
20. Paynter H.M.: Analysis and Design of Engineering Sys-
tems. The MIT Press Cambridge, Massachusetts (1961).
21. Sagawa J.K., Nagano M.S., Neto M.S.: A closed-loop
model of a multi-station and multi-product manufac-
turing system using bond graphs and hybrid controllers.
European Journal of Operational Research (2016), pp.
1-15.
22. Shoureshi R., McLaughlin K. M.: Analytical and Experi-
mental Investigation of Flow-Reversibile Heat Exchangers
Using Temperature-Entropy Bond Graphs. Journal of
Dynamic Systems, Measurement, and Control, Vol. 106
(2), (1984), pp. 170-175.
23. Silva L.I., et al.: Coupling Bond Graph and Energetic
Macroscopic Representation for Electric Vehicle Simula-
tion: Mechatronics 24 (2014), pp. 906-913.
24. Sliwinski P.: The basics of design and experimental tests CONTACT WITH THE AUTHOR
of the commutation unit of a hydraulic satellite motor.
Archives of Civil and Mechanical Engineering, vol. 16, Jacek Kropiwnicki
iss. 4 (2016), pp. 634-644.
Gdask University of Technology
25. Sosnovsky E., Forget B.: Bond graph representation of Faculty of Mechanical Engineering
nuclear reactor point kinetics and nearly incompressible 11/12 Narutowicza St.
thermal hydraulics. Annals of Nuclear Energy 68 (2014), 80 - 233 Gdask
pp. 15-29. Poland

26. Thoma J. U., Boumama B. O: Modelling and Simulation

in Thermal and Chemical Engineering - a Bond Graph
Approach. Springer, (2000).

27. Thoma J. U.: Simulation by Bondgraphs. Springer, Berlin,

(1990).

28. Wellstead P.E.: Introduction to System Modeling. Aca-

demic Press, London (1979).

POLISH MARITIME RESEARCH, No S1/2017 37

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 3845
10.1515/pomr-2017-0019

COMPARATIVE STUDY OF MACHINING TECHNOLOGY SELECTION

TO MANUFACTURE LARGE-SIZE COMPONENTS OF OFFSHORE
CONSTRUCTIONS

MariuszDeja 1
Mieczysaw Siemitkowski 1
Sender P.2
1
Gdansk University of Technology, Mechanical Engineering Faculty, Dept. of Manufacturing Eng.
and Automation, Gdansk, Poland
2
Elektromonta-Pnoc Gdynia S.A. (EPG), Poland

ABSTRACT

The focus of this paper is on process planning for large parts manufacture in systems of definite process capabilities,
involving the use of multi-axis machining centres. The analysis of machining heavy mechanical components used in
off-shore constructions has been carried out. Setup concepts applied and operation sequences determined in related
process plans underwent studies. The paper presents in particular a reasoning approach to setup sequencing and machine
assignment in manufacturing large-size components of offshore constructions. The relevant reasoning mechanism
within a decision making scheme on generated process plan is shown based on a case study derived from the offshore
sector. Recommendations for manufacture of selected exemplary and typical parts were formulated.

Keywords: off-shore construction, heavy mechanical component, multi-axis machining centre, process planning

INTRODUCTION
Large-size and heavy mechanical components have
numerous applications in up-to-date industry. Consequently,
there is an increased demand on them, especially in regards
to fast-growing sectors, among others: renewable energy
power stations (e.g. wind turbines), maritime industry,
including offshore platforms as well as shipping equipment.
Components of such kind are often machined by specially
developed machine tools: workshop or portable machines
for in-situ manufacture and inspection [9, 11, 13, 16].
Significant aspects related to manufacturing and possibly
maintenance and repairing of large-size and accurate parts
are indicated in Fig.1. Related equipment of large-size
machines involving laser trackers and rotary tables help in
part handling while locating and centring the part in order
to minimise the number of setups and shortening the overall
machining time Fig. 1b. The specific shape and weight of
big components, especially those of prismatic configuration
with protruding features, require special considerations on
design of supporting elements needed for safety reasons and
to avoid possible deformation of the part on the machine tool
Fig. 1 c. Portable machines are used due to their mobility in
those branches of the industry where the size of components
and their availability is considered to be a problem in the
production technology, the assembly area and repair [6, 17].
They form an alternative to stationary machining devices and
eliminate the need for their disassembly and transportation
Fig. 1d. Regeneration works performed on maritime internal

38 POLISH MARITIME RESEARCH, No S1/2017

combustion engines, water power plants or steam turbines
may also require a narrow technical specialisation of the
staff operating the mobile machines to eliminate or reduce
the costs of transport and preparation of a repair service of
large-parts. The continuity of the production process can
be restored much quicker than after parts reassembling.
As reported in [12], turbine housings of more than 9000
mm in diameter can be machined with the use of a mobile
system based on a mobile parallel kinematic machine tools
for 5-axis machining and equipped with a highly accurate
angle measuring system inside the rotary table and tactile
measuring of reference geometries on the work part.
Constructions associated with application of large-size
machine tools have not been searched sufficiently by research
institution. The reason for that might be the relatively high
cost involved in necessary experimentation. There are also
relatively few reports derived from industrial area as well.
Selected issues of full scale parts manufacture are discussed
in following chapters of the paper work with illustrative case
study derived from industrial practice of a middle-sized
production plant operating in an offshore engineering area.
PROCESS REQUIREMENTS VERSUS
MACHINE CAPABILITY OF LARGE-
SIZE COMPONENTS IN OFFSHORE
CONSTRUCTIONS
Kinematical possibilities of modern CNC centres with
driven tools allow for processing machining features
characteristic for both: prismatic and rotational parts.
Fig. 1. Manufacturing aspects of large-size parts for offshore structures: a) a winch dedicated to shipping and offshore industry [21], b) a winch drum with 800 tons
pulling force mounted for machining on a machine tool in EPG (Energomonta-Pnoc Gdynia S.A. plant, c) supporting a large part on a machine tool in EPG
plant, d) a mobile precise lathe for flanges of diameter 1143 6000 mm [22]
Components with prismatic and rotational features are
defined as prisronal parts in[18]. The main tendency in the
design of large- size machine tools is to ensure possibility of
machining axial-symmetric parts on CNC milling centres
and prismatic parts on CNC lathe with a controlled rotation
of a rotary table for obtaining required geometry and tight
tolerances [5, 13]. Technological problems which can be easily
solved on conventional size machine tools represent a huge
challenge for big and accurate dimensions, mostly within a
tolerance grade of IT7, and obtained on large-size machine
tools. Cylindrical shapes, like in cylindrical bars, can be
made by turning with fixed cutting tools or with driven tools.
Conventional turning is possible if the machined diameter
and available rotations (sometimes limited to 25 rpm or
even lower for non-symmetric parts) ensure obtaining the
required cutting speed.
The main kinematics configurations of milling machines
for large parts are with movable column (Fig. 2a) or gantry
type. Controlling 5 axes of the machine is possible by the
installation of automatic swivelling head on a multi-tasking
milling machine tool [7, 9]. The machining features of a part
clamped in a fixed position are easily accessible by the tools
mounted in a spindle of movable column machines. A vertical
lathe is dedicated to large rotational parts for turning the
diameters ranging from 1.5 m to more than 6 m with the
possibility of complete machining, by conducting milling,
drilling, boring and even gear cutting in a single set-up using
additional spindles and equipment, e.g. a swivelling head
Fig. 2b, changed manually or automatically.
Static deformation of large parts due to their own weight
must be considered in detail while choosing fixtures to locate,
clamp and support a workpiece during machining. Fixtures
POLISH MARITIME RESEARCH, No S1/2017 39
should be designed so that the workpiece is not deformed
under its own weight due to plastic deformation. Finite
element analysis was carried out with the use of Autodesk
Inventor software for a sample component made of high-
strength low-alloy steel (Youngs modulus E=200GPa,
Poissons ratio =0.29) to predict its deformation (Fig. 3b).
The related deformation can be significantly minimised
by additional technological fixturing features (Fig. 3 c, d),
intelligent fixtures, CNC part program compensation [2, 8,
10] and relying on the experience of the operator [17].
Fig. 2. Configurations of large workshop machines: a) 5-axis CNC horizontal
milling centre, b) 5-axis CNC vertical turning centre (a multi-tasking vertical
lathe); all systems equipped with a 2-axis swiveling head
Fig. 3. Placement and orientation of a prismatic part on the horizontal milling centre (a), along with the FEA analysis of a part deformation for a chosen support
(b) and with welded technological fixturing features for additional clamping (c, d)
40 POLISH MARITIME RESEARCH, No S1/2017
MANUFACTURING PROCESS PLANNING
FRAMEWORK FOR FULL-SCALE PART
COMPONENTS
Fabrication of part machining features in proper setups and
sequence is particularly significant as far as the manufacture
of large- size and heavy components is concerned. This is
because of extremely long times involved in such preparatory
operations as laying out and checking the material allowance
available, performed within intermediate storage sites, and
preceding also time consuming part positioning in working
space of a definite machine tool. As a result, process planning
involving the selection of adequate processing technologies
under consideration of available machine capabilities is the
matter of great importance in here.
INTEGRATED GENERIC REASONING SCHEME FOR
SETUP PLANNING AND MACHINE ASSIGNMENT
In this research, the task of machining process sequencing
alternatives for parts types under consideration encompasses
sequencing features provided in the form of a list. Within this
context, the related task can be accomplished by grouping
features of the attributed list into setups first, along with
their assignment to machines available in a facility, and
then possible sequencing the features in each setup based
on definite reasoning technological rules. The former stage
of the two-stage decision framework as orientated to full-
scale parts manufacture is discussed most extensively in
this research. The proposed decision making scheme for
generation of process alternatives assumes input data model
including the description of both features themselves and
also concerning precedence relationships for the features,
while considering constraints related to setup planning in
parts manufacture. Therein, it seems to be reasonable to apply
specific heuristics to process sequencing that are valid in the
area of large-size parts manufacture and often observed in
industrial practice. The decision making process entails the
need for further consideration on datum reference selection
and part placement on individual machines that is assigned
to the stage of feature grouping for consecutive setups created
that is discussed closer underneath.
OPERATION SEQUENCING STRATEGY FOR LARGE-
SIZE PARTS MANUFACTURE
The approach applied to solving process sequencing
problem utilises heuristic premises coming from accumulated
technological knowledge and reported industrial experience.
Each of machining technologies like milling, drilling, turning
or grinding might have its own specified tool approaching
direction (TAD) [19] which can be assigned to each feature
[3, 20] for automatic process planning realised with computer
support. If features have more than one TAD, definite TAD
should be assign to every feature according to some priorities
of tight tolerance relation between features, the number of
features and good machining practices [20]. Providing more
TADs by a machine tool can reduce the number of setups [3].
To avoid relocation of large parts, features in one TAD should
be machined in a single setup although features machined
in one setup are not necessary in one TAD [20]. Most parts
have surfaces to be machined parallel to the natural axes
and planes of the workpiece linear coordinate system but
positioning of large parts on the machine could be a very
long process taking even a few days.
As noted, most of the research works on machining feature
sequencing focused either on prismatic or rotational parts.
For effective setup planning, machining features need to
be properly oriented with regard to the spindle axis of the
machine tool [1, 3, 4] which is typically a Z-axis. Prismatic
features made on four- or five-axis lathes must be oriented
with regard to the spindle axis of the driven milling-drilling
tools.
The demanded orientation of features is defined by the
feature working direction which is the same as TAD vector
along the spindle axis of the machine tool for turning features
made on lathes or prismatic features made on milling
machines. Prismatic features made on turn-mill centres are
oriented along the spindle axis of a driven tool. Like in [1] the
feature working direction is described by the direction cosines
of a main TAD vector along a spindle axis with respect to an
assigned workpiece linear coordinate system.
General description of the working direction of a feature
Fi, which should be considered together with the feature type,
is given by an ordered triple:
(1)
where:
-
cosines of the angles between the TAD of the feature Fi and
three workpiece coordinate axes.
As a consequence, the implied approach to setup planning
and operation sequencing is based on unified model involving
the specifications of feature working directions with respect
to an assigned work part linear coordinate system as well
as the direction cosines used to describe its orientation
with regard to the machine linear coordinate system.
Assigning features to appropriate setups proceeds after the
determination of a primary locating direction and selection
of datum reference feature. The decision making procedure
involves considerations on configuration and kinematics
characteristics of machines available in the definite machining
facility, as outlined in Fig.6.
AN INDUSTRIAL CASE STUDY OF
PART MANUFACTURE FOR OFFSHORE
APPLICATIONS
In order to illustrate the developed reasoning framework
with broader reference to diverse aspects of large-size parts
manufacture, an industrial case study derived from EPG
S.A. - a production plant located in Gdynia (Poland) is
presented. It is intended primary to show the machining
process alternatives together with adequate part fixturing
strategies on available CNC multi-axis machine tools of
different configuration and processing capabilities.
The drum of a winch shown in Fig. 1b, has been selected
as the exemplary large-size component applied to diverse
offshore constructions for testing the proposed decision
making scheme. Sets of representative machining features
for the component, that principally falls into the category of
mill-turn parts, were determined on the basis of technological
knowledge and practical applications Fig. 4 and Table1.
Most of the features with TAD lying along the axis of apart
rotation has the same cos{TADi, xwp}, cos{TADi, ywp} Fig 4.
The difference in cos{TAD, zwp} indicates working direction
change and the need for a new subset or changing the part
position around 180. One of the features distinguished (side
hole) has the working direction [0, -1, 0] that is perpendicular
to the main datum axis of the analyzed mill-turn work part.
Possible orientations of an exemplary part, with the
overwhelming number of rotational surfaces, on typical
and available machine tools are presented in Fig. 5. In
chosen machine configurations there is a problem with
the accessibility of features from both working directions
[0,0, -1] and [0, 0, 1] in a single setup, therefore two setups
are necessary for complete machining. This, in turn, gives
rise to additional difficulties in the reorientation of large
and heavy parts and its proper positioning for continued
machining. Moreover, the diameter of an outer cylindrical
surface is too small for obtaining required cutting speed.
Part orientation on a vertical turning machine equipped
with rotary table, shown in Fig. 1 b, allows for changing the
part position around 180, although the symmetry axis of a
POLISH MARITIME RESEARCH, No S1/2017 41
Fig. 4. Standard placement of a sample part (3787 mm in length and outer diameter of 3340 mm) in working space of the horizontal type machine and definition
of working directions of machining features with respect to workpiece coordinate system
part is located horizontally and perpendicularly to the axis
of a rotary table. It might be noted that machining of outer
cylindrical surface Tab. 1, is enabled owing to the available
equipment for turning or milling solely in the horizontal part
orientation on a vertical lathe or a horizontal milling machine.
a)
b)
Fig. 5. Possible placement and orientation of the exemplary part type on a
rotary table of a vertical lathe using independent chucks for centering (a) and
a horizontal milling machine using V-block locating (b)
42 POLISH MARITIME RESEARCH, No S1/2017
Alternative process capabilities have been considered in
this research assuming an extended set of machine resources
and related equipment. Three machine alternatives, however,
are included into further comparative analysis, and with
reference to the specified machining features of a winch
drum as the exemplary part design, and envisaged as those
performed within a primary part setup (Table 1). Following
the suggestions reported in [14, 15], the operation alternatives
are depicted in this Table by definite schemata, associated with
machining the specified set of features of the primary part
setup. Thus, in accordance with the reports mentioned, an
individual schema is due to be understood as a technologically
meaningful combination of a tool type (accordingly to a
machining method) and a set of relative motions between
a work part (WP) and the tool (T) of a specific geometry.
Each of the schemata defined involves the designation of
operation kinematics with the primary motion expressed
by the cutting speed vc , and the feed speed vf attributed
to a definite system component, a.e. a work part or the tool
applied, respectively. With reference to the working step of
face milling for instance: the cutting speed vc=240m/min, the
feed per tooth fz=0.3 mm and working engagement ae - equal
to 75% of the head diameter were applied for machining using
the working head equipped with octagon milling inserts.
Such a tool type is extensively used in manufacturing offshore
components while the removal of high volumes of excess
stock material and high process productivity are demanded.
The aforementioned rules are generally valid as far as the
consecutive part setups are concerned, and throughout
the decision steps of the developed reasoning algorithm,
summarised in Fig. 6.
 Table. 1. Designation of process alternatives applied to machining features for a winch drum as the exemplary large-size component
Fig. 6. Reasoning approach to setup sequencing and
machine assignment in manufacturing large-size components
of offshore constructions ( a simplified outline)
As it can be seen, the developed algorithm allows the
interchangeability of generated process alternatives, resulting
from the utilisation of both single-point cutting tools and
the tools with multiple edges. It is primarily for enabling a
proper selection of the most suitable machining strategies
related to consecutive working steps of formulated machining
operations.
CONCLUSIONS
This research work outlines the developed framework
for the reasoning approach to effective feature-based
machining process sequencing and machine assignment in
manufacturing so called turn-mill part components of large
dimensions that are typical for offshore applications. It enables
the generation of interchangeable process alternatives that
involve the utilisation of processing capabilities of available
multi-axis milling and turning CNC centres, equipped
with relevant fixtures, and tools, as far as the technological
specification was concerned. This approach has been validated
through a selected case study derived from an industrial
practice of the offshore sector. The results demonstrate
the assumed feasibility of the feature-based approach
for generic and adaptive machining sequence generation
and its usability across the numerical study performed.
Consequently, the generated process plan solutions could
be identified, considering actual production conditions
with regard to the constraints of a fixturing strategy, and
the availability of a specific set of machine tools of definite
capabilities (machine configuration and kinematics, incl.
NC-axes), potential part positioning (standard placements),
setup realisation conditions (the equipment available for
relevant part locating and clamping) and the specification
of machining features based on defined working directions.
At the operation planning stage, as noted, the machining
sequence along with cutting parameters and tool paths
could be optimised specifically for the selected machine tool
based on best practices and the equipment - related data in
considerably confined search space and with less complexity.
ACKNOWLEDGMENTS
The authors would like to thank the management board
of the manufacturing plant Energomonta-Pnoc Gdynia
S.A. (EPG),http://www.epgsa.com, for providing relevant
data used in the research. Computations related to testing of
the proposed procedure were accomplished with the use of
Matlab software, available in Academic Computer Centre
in GdanskTASK (http://www.task.gda.pl).
POLISH MARITIME RESEARCH, No S1/2017 43
 Fig. 6. Reasoning approach to setup sequencing and machine assignment in manufacturing large-size components of offshore constructions ( a simplified outline)
BIBLIOGRAPHY
1. Celano G., Costa A., Fichera S. and Santangelo B.: Pallet
configuration for approaching mapping requirements
on devices, Chapters 6 in: Design of flexible production
systems: methodologies and tools, Tolio T (Ed.), Springer
-Verlag 2009.
2. Choi J.P., Min B.K. and Lee S.J.: Reduction of machining
errors of a three-axis machine tool by on-machine
measurement and error compensation system. Journal of
Materials Processing Technology, Vol. 155-156 (2004), pp.
2056-2064.
3. Deja M. and Siemiatkowski, M.S.: Feature-based generation
of machining process plans for optimised parts manufacture.
Journal of Intelligent Manufacturing, Vol. 24 (2013), pp.
831-846.
44 POLISH MARITIME RESEARCH, No S1/2017
4. Gologlu, C.: Machine capability and fixturing constraints-
imposed automatic machining set-ups generation, Journal of
Materials Processing Technology, Vol. 148 (2004), pp. 83-92.
5. Groover M.P.: Fundamentals of modern manufacturing,
Materials, processes and systems, J. Wiley & Sons Inc., 2010.
6. Haddag B., Nouari M. and Mouf ki A.: Some cases
of machining large-scale parts: characterization and
modelling of heavy turning, deep drilling and broaching,
AIP Conference Proc. 1716, AIP Publishing 2016, pp. 1-9.
7. Jang S.H, Choi Y.H., Kim S.T., An H.S., Choi H.B. and
Hong J.S.: Development of core technologies of multi-tasking
machine tools for machining highly precision large parts.
Journal of the Korean Soc. for Precision Engineering, Vol.
29, No.2 (2012), pp. 129-138.
8. Kaliski K.J. and Galewski M.: Chatter vibration
 surveillance by the optimal-linear spindle speed control.
Mechanical Systems and Signal processing, Vol. 25 (2011),
pp. 383-399.
9. Lizarralde R., Azkarate A. and Zelaieta O.: New
developments in lathes and turning centres, Chapter 7 in:
Machine tools for high performance machining, Lopez de
Lacalle L.N and Lamikiz A. (Eds), Springer-Verlag 2009.
10. Mares M., Horejs O. and Hornych J.: Advanced thermal
error compensation of a flor type machining centre allowing
for the influence of interchangeable spindle heads, Journal
of Machine Engineering, Vol. 15, No. 3 (2015), pp. 19-32.
(2007), pp. 205-217.
20. Zhang, H-C. and Lin, E.: A hybrid-graph approach foe
automated setup planning in CAPP. Robotics and Computer-
Integrated Manufacturing, Vol. 15 (1999), pp. 89-100.
21. http://www.gospodarkamorska.pl/wydarzenia/wciagarka-
z-energomontazu-polnoc-gdynia-dla-norweskiego-
przemyslu-offshore.html, accessed on February 16th 2017.
22. http://en.nglmachining.com/park-maszynowy.html,
accessed on February 16th 2017.

11. Matuszewski A. Musia J. and Styp-Rekowski M.: Design CONTACT WITH THE AUTHOR
and technological issues related to large parts of the
modular structure (in Polish), Technologia i Automatyzacja Mariusz Deja
Monau, No. 2 (2012), pp. 28-30.
Gdask University of Technology
12. Neugebauer R., Wabner M., Rentzsch H. and Ihlenfeldt S.: 11/12 Narutowicza St.
Structure principles of energy efficient machine tools. CIRP 80 - 233 Gdask
Journal of Manufacturing Science and Technology, Vol. 4 Poland
(2011), pp. 136-147.

13. Onozuka H., Utsumi K., Kato T., Takahashi H. and

Obikawa T.: Optimal design of a damped arbor for heavy
duty machining of giant parts. Journal of Advanced
Mechanical Design, Systems, and Manufacturing, Vol. 7,
No. 2 (2013), pp. 171-186.

14. Ratchev T.M.: Concurrent process and facility prototyping

for formation of virtual manufacturing cells. Integrated
Manufacturing Systems, Vol. 12, No.4 (2010), pp. 306-315.

15. Siemitkowski M. and Przybylski W.: Modelling and

simulation analysis of process alternatives in the cellular
manufacturing of axially symmetric parts. Intl. Journal of
Advanced Manufacturing Technology, Vol. 32, Nos.5-6
(2007), pp. 516-530.

16. Simpson B. and Dicken P.J.: Integration of machining and

inspection in aerospace manufacturing, IOP Conf. Series:
Materials Science and Engineering, Vol. 26 (2011), pp. 1-4.

17. Uriarte L., Zatarain M., Axinte D., Yague-Fabra J., Ihlenfeldt
S., Eguia J. and Olarra A.: Machine tools for large parts.
CIRP Annals-Manufacturing Technology, Vol. 62 (2013),
pp. 731-750.

18. Waiyagan K. and Bohez E.L.J.: Intelligent feature based

process planning for five-axis multi-turn parts. Computers
in Industry, Vol. 60 (2009), pp. 296-316.

19. Yao S., Han X., Yang Y., Rong Y., Huang S., Yen D. and
Zhang G.: Computer aided manufacturing planning for
mass customization: part 2, automated setup planning. Intl.
Journal of Advanced Manufacturing Technology, Vol. 32

POLISH MARITIME RESEARCH, No S1/2017 45

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 46-52
10.1515/pomr-2017-0020

SIMULATION STUDIES INTO QUAYSIDE TRANSPORT AND

STORAGE YARD OPERATIONS IN CONTAINER TERMINALS

Mariusz Deja
Micha Dobrzyski
Mieczysaw S.Siemitkowski
Aleksandra Winiewska
Gdansk University of Technology,
Mechanical Engineering Faculty, Dept. of Manufacturing Engineering and Automation, Gdansk, Poland

ABSTRACT

The focus of the paper is in quayside transport and storage yard operations in container terminals. Relevant algorithms
have been applied and a simulation model adopted. Evaluative criteria chosen for that model were: the total time of
ship unloading and the truck utilization level. Recommendations for unloading in berth and yard areas were analysed
in three different case studies. Results of simulations and deterministic model based analyses are included.

Keywords: container terminals, mechanical handling, simulation modelling, quantitative analysis.

INTRODUCTION a traffic zone or a set of transfers for trucks and transporters,

The container terminal enables efficient container flow
between ships and inland transport means. Containers
transported from ships to trucks or trains are called import
containers, while those taking the reverse route bear the
name of export containers [3]. Transshipment containers are
transported between ships. In all cases, the containers are
delivered to the terminal yard and stored for few days before
being transported either to the berth for ship loading (export,
transshipment), or by inland transport means (import) [10].
The yard planning involves allocation of storage spaces for
import, export and transshipment containers [13, 17]. The
yard is divided into several blocks, and each block consists
of several bays. Each bay in the block has the same number
of stacks. One of most common problems in the terminal
transportation system is the container relocation problem
(CRP) which has to be solved when containers, piled up
in stacks, need to be transported to a ship or to trucks in
a predefined sequence. Forster and Bortfeldt propose an
effective tree search procedure for the CRP [3].
As assumed in [13], container terminals perform various
handling operations by utilizing such resources as quay cranes
(QCs), storage yards (SYs), yard cranes (YCs), travelling areas
(TAs), and transporters (TRs). The travelling area represents
including internal moving vehicles (IMVs) or automated
guided vehicles (AGV). Xin et al. [14] propose a methodology
to generate collision-free trajectories of free-ranging AGVs
in automated container terminals, while minimizing the
makespan of the whole container handling system. Two most
important performance measures in the container terminal
are the turnaround time of ships, trucks, and trains in the
terminal. Proper yard planning with adequate CRP solving
are crucial for minimizing the turnaround time. This paper
addresses mostly terminal operation activities related to its
landside, such as outbound container transportation with
IMVs and, consequently, their stowage into the assigned
storage yards. Vessels are unloaded by one or more quay
cranes (QCs) according to the unloading plan. Containers
are then relayed to transfer vehicles (IMVs or AGVs),
which transport them to the storage yard where they are
temporarily stored by yard cranes (YCs). Depending on their
destination, containers may be transshipped to another vessel,
or dispatched, via terminal gates, for transport by trucks or
trains after being inspected. The objective of this paper is
minimizing the turnaround time of the vessel unloading at
a definite berth location with the utilisation of the required
number of IMVs. An efficient operational plan in container
terminals requires a large number of factors to be considered

46 POLISH MARITIME RESEARCH, No S1/2017

for decision-making [13]. The simulation of vessel unloading
with available IMVs to a chosen storage yards can help to
make the proper decision concerning the storage yard choice,
to minimize the turnaround time of vessel unloading and
to utilise the minimum number of IMVs required to unload
the vessel without breaks [15, 16].
LITERATURE OVERVIEW
Problems with seaport container terminal planning and
operation have attracted a great deal of research attention,
and there is abundant literature available on this subject.
However, despite a significant progress in this area, it is
particularly the issues associated with load transportation and
storage yard operation which still remain to be fully solved.
Much of current effort is aimed at the container relocating
problem (CRP) which occurs in container terminals of
seaports [3]. Efficacious integration of individual subtasks
into the planning and operation task of terminals still remains
a problem in engineering practice of container terminals,
and thus, it constantly attracts research attention. In general
terms, container terminals can be described as open systems
of material flow with two external interfaces [9]. These
interfaces are: (1) the quayside with loading and unloading
of ships, and (2) the landside where containers are loaded and
unloaded on/off trucks and trains. Loading and unloading of
vehicles for horizontal transport is done by cranes, either quay
cranes or gantry cranes [5, 9]. The frameworks provided for
integrating various planning activities in container terminals
consider two basic planning problems:
planning problem in the quayside: berth allocation, ship
stowage planning, quay crane scheduling;
planning problem in the landside: yard storage and stacking
planning, yard crane scheduling, vehicle scheduling.
From a logistic point of view, terminals only consist of
two components: stocks and transport vehicles. The yard
stacks, ships, trains, and trucks belong to the category stock
[9]. Problems with yard storage and stacking planning are
considered with maximum utilization of the storage space,
minimum transport distance from quay to stack and vice
versa, and minimum number of unproductive moves [11]. The
main constraints in yard storage and stacking planning are
maximum stack tiers depending on the stacking equipment
and container attributes, such as: size, weight, reefer,
dangerous, damaged.
A stowage (or loading) instruction is necessary to define
the rules how and where containers have to be stored [1,
2]. Problems with yard crane scheduling refer to minimum
waiting time of the truck at the yard and to minimum
travelling time of the yard crane in the yard zone. The
main constrains in yard crane scheduling are: the process
of physical handling of yard cranes and the accessibility of
ayard crane in the yard. Problems with vehicle scheduling are
also related to the minimisation of the lateness of container
deliveries, empty travel distances, and loaded travel times,
like in production systems [4].
Quayside problems analysed in the literature are related to
the trade-off optimization between the vessels turnaround
time and the number of the used berths and quay cranes.
The utilization of berths and quay cranes taking into account
random vessel arrivals and some fluctuations in the analysed
system is also optimised. Different logistic concepts, decision
rules and optimization algorithms have to be compared by
simulation before they are implemented into real systems
[68].
This paper addresses mostly the terminal operation
activities related to its landside, such as outbound containers
transportation with IMVs and consequently their stowage
into the assigned storage yards. The objective is aimed
at minimizing the turnaround time of vessel unloading
at a definite berth location. In this research, the issues of
operative planning of overall logistic processes performed in
seaport terminals underwent a comprehensive analysis and
resultant evaluation. Following the preliminary assumptions
based on the concept of deterministic models, a discrete
event simulation approach was applied to system analysis
and relevant quantitative evaluation of its operational
characteristics was obtained. Three alternative scenarios
were neatly considered throughout the presented quantitative
study.
PRELIMINARIES AND PROBLEM
OVERVIEW
The founded model assumes that the transport of
containers takes place by/within ship yards storage area.
The ship waiting at quayside for discharge is divided into 2
to 4 sections (berths), depending on the ships length, Fig.1.
Each section is supported by one Quay Crane moving along
the quay and the ship. Transport is done by using IMV trucks.
Containers are taken by the crane from the ship and loaded on
trucks, which carry the container to a designated location on
the yards storage space using a determined path. There, the
container is being received by the yards crane and is placed
in the right/dedicated place on the yards storage area bay,
Fig.2. One gantry crane supports two yards storage areas.
Fig. 1. Loading/unloading operations general view
In the analysed case the yards storage area consists of
two parts, Fig.3. The first part has 12 storage fields for both
POLISH MARITIME RESEARCH, No S1/2017 47
conventional and refrigerated containers, while the second
area has two fields used for of hazardous load and undersized
containers. The storage area is a cube with dimensions of
7x10x5 (numbers of containers in width x length x height).
For logistic reasons, the statistical fulfilment of a cube of
containers is 80%. This value has resulted from the need
to perform manoeuvres of containers. When the container
tobe loaded on a ship is in one of the lower or middle layers,
to access it the gantry needs to make so called relocation
operation, i.e. to download and put in free space containers
stored on the sought container. The founded model assumes
that the use of the fields (operational) is at 60% due to low
load orders number of storing containers.
IMV trucks move along designated routes between the
quay crane and the yard crane (gantry of RTG type). Most
often they do it at a speed of up to approx. 30km/h, and the
assumed model adopted the average truck speed of 20 km/h.
a)
b)
Fig. 2. Yard crane and containers stacked in tiers (a)
and in individual block (b)
Containers are the object of loading and unloading
operations. They can be of different types and sizes:
conventional containers of small size (length 20 ), conventional
containers of large size (length 40), refrigerated containers
(requiring storage oriented to the direction of the location
of power connections), containers carrying hazardous loads
(requiring usually dedicated fields and sometimes even the
space separating them from other containers), and containers
with protruding elements, so called POG. It is conventional
and refrigerated containers which prevail in logistical
practice, while the containers with hazardous loads and
projecting parts are generally up to 10% of the total number
of containers transported.
48 POLISH MARITIME RESEARCH, No S1/2017
SIMULATION CASE STUDY OF LOAD
TRANSPORTATION AND STORAGE SPACE
ALLOCATION
In the simulation project, a model with quayside unloading
operations performed in regard to a medium-sized vessel of
medium tonnage has been adopted as the illustrative case
study. However, it involved carrying containers of different
types, including those of 20 and 40 in length, as well as
reefer or hazardous goods containers, or oversized ones.
The containers within the developed model are available in
three categories: A, B and C, correspondingly to their type,
i.e. large and small containers (without distinction between
conventional or reefer ones), and hazardous load or non-
standard-sized containers. Containers A and B are deposited
within the designated storage yard area and divided into
anumber of blocks (colour blue, block numbers from B1 to
B12), while the C-type containers are deposited on a dedicated
special blocks (red, B13 and B14), Fig. 3.
Fig. 3. Layout of the seaport terminal under investigation
In order to find the best design solution, different alternative
scenarios were neatly considered throughout the presented
quantitative study. Basic attributes for terminal equipment
and their operating characteristics used for the simulations
are given in Table 1. A simplified flowchart of the container
terminal operation of the analysed quayside is presented in
Fig. 4.
Table1. Basic attributes for terminal equipment and their operating
characteristics a)
50
Type of equipment / C B A

Number of containers
Resource characteristic 40
parameters
30

Number of quay cranes 2

20

(QC) 10

0
Cycle time of quay crane Uniform (2.0,3.0) 1
3
5
(QC) [min] 7
9
11
13
15
17
Number of rubber-tyred 7
19
21
23
25
gantry cranes (RTG) 27
29
31 A
Inflow order of containers 33
35
37 B
Cycle time of rubber-tyred Uniform (2.5,3.5)
39
41
43 C
gantry crane (RTG) [min] 45
47
49

Number of IMV 14; 7; 10*; 11**

multipurpose vehicles
b)
IMV speed [m/min] 420 (unloaded)
80
C B A
IMV speed [m/min] 333 (loaded) 70

Number of containers
60
50
* for scenario 1; ** for scenario 2 40
30
20
10
0

SHIP arrival 1
3
5
7
9
11
13
15
17
19
21
23
25
27
29 A
31
Inflow order of containers 33
35
37 B
39
Unloading order formulation 41
43
45
C
47
49

Fig. 5. Input profiles for intermittent inflow of containers by unloading plan for
QC1 (a) and QC2 (b), respectively.
Berths reservation
i including the
number of QC cranes
The adopted model was analysed in a series of experiments,
taking into account the following criteria of quantitative
system analysis:
Unloading plan the total time of unloading a ship with relocation of
containers from the ship to definite blocks in the storage
yard space,
the level of utilisation of individual vehicles assumed as
IMV calling
lli v the target function.
v=1n
B1 B7 B1 B7

Containerr allo
allocation to B2 B8 B2 B8

Yard blocks;
B3 B9 B3 B9
RTG crane calling
B4 B10 B4 B10

B5 B11 B14 B5 B11 B14

N
v t n* B6 B12 B13 B6 B12 B13

Y
B1 B7
STOP
B2 B8

Fig. 4. Simplified flowchart of container terminal operation of the analysed quayside

B3 B9

B4 B10
The detailed unloading plan for a vessel, prepared for
QC1 and QC2, is characterised by intermittent inflow of B5 B11 B14

containers of different types: A, B, and C, Fig. 5. The sequences

B6 B12 B13
of containers stowage in the storage area for two analysed
scenarios are illustrated by directed graph models with shaded Fig. 6. Directed graph models showing the sequence of containers stowage
nodes denoting starting block locations, Fig. 6. inthe storage area for Scenario 1 (a) and Scenario 2 (b); shaded nodes denote
starting block locations.

POLISH MARITIME RESEARCH, No S1/2017 49

 The following assumptions formed the basis for analytical % Idle % Busy
considerations used to obtain starting estimates of system Scenario 1 Scenario 2 Scenario 1 Scenario 2
performance, and further on at the stage of simulation
modelling: RTG2 71,43 70,70 28,57 29,30
number of IMV vehicles equal to the number of fields of RTG3 70,54 71,99 29,46 28,01
storage yard area, RTG4 71,04 69,63 28,96 30,37
number of IMV vehicles equal to the number of RTG
RTG5 70,57 71,43 29,43 28,57
cranes serving the storage yard area,
optimum number of vehicles at the shortest ship unloading, RTG6 70,52 72,46 29,48 27,54
and the assumption of continuity of the QC crane working RTG7 86,64 86,64 13,36 13,36
cycle.
QC1 0,32 99,68
In the wake of the above, three adequate cases were
consequently adopted within the framework of the simulation QC2 0,18 99,81
analysis, correspondingly with the ascertained number of
transport resources operating in the storage yard area. Following the established assumptions, the calculated total
The transport path in the frame of the definite mesh vessel unloading time (the turnaround time) amounted to
routing for IMV_n (Vehicle_n) passes through the elements 3,358 min, and particularly, for the case of 7 IMVs working
shown in Fig. 3 and requires some identifications: in the determined storage area the calculated time was 3,371
Parking lot the route (T_n) QuayCrane_n min. Thus, the observed time difference for the instances
identification of the type of container (A, B, C) truck_n studied was less than 0.5%.
Yard Storage Space (TP_n) unloading RTG_n path As shown in Table 3 in turn, there are no significant
transport T_n parking lot. differences concerning the use of QC and RTG cranes for
The simulation investigations of process alternatives have both scenarios and the instance cases using 7 IMVs. As it
been carried out in the environment of Witness [12], the can be also seen, the limited number of transport resources
interactive software package, with the use of programmable and strategies adopted for the distribution of containers in
models to map real operating characteristics of the seaport the storage yard has no effect on the discussed process output
container terminal. characteristics. The lower utilisation level of RTG cranes
is associated with a relatively small number of containers
allocated to definite blocks and supported by these container
SIMULATION RESULTS handling resources.
AND RELATED DISCUSSION Table 3. Utilisation of RTGs and QCs for the number of IMVs = 7

Various data have been derived from experiments, and Scenario 1 Scenario 2
among other those related to resource utilisation during % Idle % Busy % Idle % Busy
stowage of containers within the assigned storage area and QC1 0,16 99,84 0,34 99,66
stacking the unit loads in tiers within destination blocks.
The utilisation characteristics of individual QC and RTG QC2 0,2 99,8 0,18 99,82
cranes are shown in Table 2, provided that the demanded RTG1 71,61 28,39 69,43 30,57
operation continuity of the related equipment is met RTG2 71,55 28,45 70,82 29,18
(regarding comparable Cases 1 and 2). Obviously, this is
consistent with the objective function aimed at minimising RTG3 70,66 29,34 72,11 27,89
the turnaround time (vessel servicing time at the berth). RTG4 71,15 28,85 69,75 30,25
Busy state percentage includes two components, which are RTG5 70,69 29,31 71,55 28,45
the operation time and the waiting time. In the case of QCs,
RTG6 70,63 29,37 72,58 27,42
the operation time component encompasses picking up
acontainer from the ship and transporting it to the IMV RTG7 86,69 13,31 86,69 13,31
loading position, while the waiting time component is the
time needed for the access of the vehicle to its loading position. Table 4. Utilisation statistics of IMVs for analysed cases and Scenario 1
For the RTG crane, the operation time includes picking up % Idle % Demand % Transfer % Loaded
acontainer from the IMV and its transportation to a specific
storage location in a block of storage yard. Min 78,24 3,02 2,09 14,72

Table 2. Utilisation levels of QC and RTG cranes, following the condition of Max 79,43 3,16 3,26 15,81
continual supply of container units at the process input Case 1
Mean 79,00 3,08 2,63 15,29
% Idle % Busy Median 79,07 3,08 2,64 15,31
Scenario 1 Scenario 2 Scenario 1 Scenario 2

RTG1 71,50 69,30 28,50 30,70

50 POLISH MARITIME RESEARCH, No S1/2017

 % Idle % Demand % Transfer % Loaded % Idle % Demand % Transfer % Loaded
Min 57,87 5,94 4,85 29,67 Min 72,35 4,21 1,21 20,77
Max 58,70 6,42 5,67 30,81 Max 73,47 4,40 1,97 21,55
Case 2 Case 3
Mean 58,36 6,14 5,15 30,34 Mean 72,96 4,31 1,63 21,09
Median 58,48 6,18 4,99 30,37 Median 73,01 4,31 1,68 21,06
Min 72,54 3,66 2,76 18,97
Considerable differences in the lengths of paths travelled by
Max 73,97 4,05 4,29 20,25 the IMVs, as well as in the numbers of loading cycles executed
Case 3
Mean 73,27 3,92 3,35 19,46 by QCs can be observed in the performed simulation studies,
Table 6. As noted further, in each case under consideration
Median 73,14 3,95 3,24 19,49
the distance travelled by the transport resources turned out
With the commitment of 14 IMVs, about 15% of the total to be directly dependent on the number of IMVs involved
vehicle working time is spent in the loaded state, regardless into the overall transportation task and indirectly related to
of the selected scenario (see Table 4 and Table 5). For Case 3, the applied allocation strategy.
characterised by the optimum number of IMVs assigned to Table 6. Total distances travelled by IMVs and the number of related loads
the transportation task, the time when the available vehicles
remain in the loaded state increases by about 5%. Moreover, Scenario 1 Scenario 2
Distance
with 7 IMVs this quoted level is nearly twice as high. Loads Distance Loads Distance
Due to the adopted strategies for vehicle allocation to Min 371305 149 372795 148
appropriate blocks, the container loading and unloading
operation times at the dissipated structure of possible Max 388730 152 385751 152
Case 1
allocations are almost twice as long, compared to Scenario2. Mean 381073 150 378033 150
This is because the individual RTG cranes operating in Median 380901 150 376831 150
definite storage blocks have to handle the container units
dispatched from both QCs. Such a mode of system operation Min 750682 296 750540 295
results in the formation of queues in front of storage blocks, Max 770988 306 759642 306
Case 2
which consequently leads to the increase of the unloading Mean 762131 300 756051 300
time at the assigned destination of the storage yard.
Median 762790 299 758181 299
As mentioned earlier and in order to diminish this adverse
effect, the farthest and the nearest load allocation rules to Min 473248 188 521377 206
storage blocks have been applied, respectively, for QC1 and Max 495485 193 539075 214
QC2. Case 3
Mean 484999 191 529241 210
The shortest queues of vehicles waiting to be unloaded
which are observed in instances of Scenario 2 are the result Median 487438 191 529316 211
of the assumed allocation strategy that allows the separation
of the stream of IMVs into two storage areas, Fig.6b. It is
also worth mentioning that the demand percentage is at a CONCLUSIONS
comparable level for both scenarios, as it can be noted in
Table 4 and Table 5. Container terminals represent highly dynamic logistics
Table 5. Utilisation statistics of IMVs for analysed cases and Scenario 2 systems of stochastic nature, where the use of deterministic
analysis is insufficient for their proper analysis and
% Idle % Demand % Transfer % Loaded quantitative evaluation. Hence a comprehensive simulation
Min 80,28 3,03 0,97 14,66 study seems to be adequate in this research. The numerical
Max 81,23 3,19 1,63 15,25
results of the performed simulation runs give insight into their
Case 1 diverse performance characteristics of relevant integrated
Mean 80,69 3,08 1,16 15,07 transportation issues and storage loads in yard blocks (yard
Median 80,68 3,06 1,08 15,10 storage space) under study (strategies for dispatching /
Min 61,45 6,07 1,80 29,55
assignment of IMV vehicles), along with their interactions
with the handling equipment. As a result, the objective of
Max 62,18 6,20 2,45 30,19 determining the optimised storage strategy and the utilisation
Case 2
Mean 61,82 6,14 2,16 29,89 of container transport and handling resources has been
Median 61,85 6,14 2,15 29,91 provided to a great extent, at the least according to the humble
opinion of the authors team.
Further research will be aimed at the container relocation
problem (CRP) with the utilisation of both deterministic and
simulation modelling.

POLISH MARITIME RESEARCH, No S1/2017 51

 Acknowledgments: Authors would like to thank the
management board of DCT Gdansk SA (http://dctgdansk.pl)
for valuable tips and advice on the operation of container
terminals
BIBLIOGRAPHY
1. Carlo H.J., Vis I.F.A. and Roodbergen K.J.: Storage yard
operations in container terminals: Literature overview,
trends, and research directions, European Journal of
Operational Research, Vol. 235 (2014), pp. 412-430.
2. Carlo H.J., Vis I.F.A. and Roodbergen K.J.: Transport
operations in container terminals: Literature overview,
trends, research directions and classification scheme. Euro-
pean Journal of Operational Research, Vol. 236 (2014),
pp. 1-13.
3. Forster F. and Bortfeldt A.: A tree search procedure for the
container relocation problem. Computers & Operations
Research, Vol. 39 (2012), pp. 299-309.
4. Groover M.P.: Automation, production systems, and
computer-integrated manufacturing, Chapter 10, Pearson
Prentice Hall, New Jersey 2008.
5. Gnter H.O. and Kim K.H.: Container terminals and
terminal operations, OR Spectrum, Vol. 28 (2006), pp.
437-445.
6. Kia M., Shayan E. and Ghotb F.: Investigation of port
capacity under a new approach by computer simulation,
Computers & Industrial Engineering, Vol. 42 (2002), pp.
533-540.
7. Murty K.G., Liu J., Wan Y. and Linn R.: A decision support
system for operations in container terminal, Decision
Support Systems, Vol. 39 (2005), pp. 309-332.
8. Rashidi H. and Tsang E.P.K.: Novel constraints satisfaction
models for optimization problems in container terminals.
Applied Mathematical Modelling, Vol. 37 (2013), pp.
3601-3634.
9. Steenken D., Vo S. and Stahlbock R.: Container terminal
operation and operations research a classification and
literature review, OR Spectrum, Vol. 26 (2004), pp. 3-49.
10. Urbanyi-Popiolek I. and Klopott M.: Container terminals
and port city interface a study of Gdynia and Gdansk
ports, Transportation Research Procedia, Vol. 16 (2016),
pp. 517-526.
11. Vis I.F.A.: A comparative analysis of storage and retrieval
equipment at container terminal, Intl. Journal of Produc-
tion Economics, Vol. 103 (2006), pp. 680-693.
52 POLISH MARITIME RESEARCH, No S1/2017
12. Witnessver.13. Visual Interactive Simulation Software,
User manual, Lanner Group Ltd, Redditch-Worcs, UK
2010-2013.
13. Won S.H. and Kim K.H.: An integrated framework for
various operation plans in container terminals. Polish
Maritime Research, No. 2 (2009), pp. 51-61.
14. Xin J., Negenborn R.R., Corman F. and Lodewijks G.:
Control of interacting machines in automated container
terminals using a sequential planning approach for col-
lision avoidance. Transportation Research, Part C, Vol.
60 (2015), pp. 377-396.
15. Yun W.Y. and Choi Y.S.: A simulation model for container
terminal operation analysis using an object-oriented
approach, Intl. Journal of Production Economics, Vol.
59 (1999), pp. 221-230.
16. Zhang X., Zeng Q. and Yang Z.: Modeling the mixed
storage strategy for quay crane double cycling in container
terminals. Transportation Research, Part E, Vol. 94 (2016),
pp. 171-187.
17. Zhen L., Xu Z., Wang K. and Ding Y.: Multi-period yard
template planning in container terminals. Transportation
Research, Part B, Vol. 93 (2016), pp. 700-719.
CONTACT WITH THE AUTHOR
Mariusz Deja
Micha Dobrzyski
Mieczysaw S.Siemitkowski
Aleksandra Winiewska
Gdansk University of Technology,
Mechanical Engineering Faculty, Dept. of Manufacturing
Engineering and Automation
Gdansk
Poland
 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 53-58
10.1515/pomr-2017-0021

SEARCHING FOR CRITICAL CONDITIONS DURING LIFEBOAT

LAUNCHING SIMULATIONS

Czesaw Dymarski
Pawe Dymarski
Aleksander Kniat
Gdansk University of Technology, Poland

ABSTRACT

The article describes numerical simulations of the process of lifeboat launching at the ships side. The research aimed
at finding the values of ship motion parameters which appear to be most dangerous for people in the lowered lifeboat
due to the generated accelerations. The simplified model of ship hull motion adopted at this research stage bases on
a superposition of harmonic motions with given amplitudes and periods in six degrees of freedom. The range of the
amplitude of motion for each degree of freedom corresponds approximately to that of possible motion of the PANAMAX
type ship on the Baltic sea. In total, 120 960 cases of ship hull motion were examined.

Keywords: lifeboat launching, numerical simulations, kinematics

INTRODUCTION
The aim to ensure the best possible safety of navigation for
the crew and passengers on seagoing vessels has always been
an important aspect for ship designers and owners. However,
until the Titanic disaster, no consolidated requirements
concerning lifesaving appliances and devices had been
formulated for ship designers. Only after this tragedy the
need for such requirements was recognised, which resulted in
working out the SOLAS Convention (part III) and, based on
this Convention, certain regulations adopted by classification
societies. With time, as the number of vessels in operation
increased and their design solutions and material technology
developed, the level of requirements for both the lifesaving
equipment, and the devices used for launching it on and lifting
from water has gradually increased, which was reflected in
introducing subsequent amendments, concerning, among
other aspects, increased safety of evacuation in rough sea
conditions. For instance, Ref. [13] describes a modern
procedure of material property testing, while Refs. [5] and
[8] publish results of calculations performed using the Finite
Element Method (FEM).
The safety requirements which are currently in force
are laid down in the International Life Saving Appliance
Code (LSA Code). This code has been brought into force in
accordance with the amendments to the SOLAS Convention
adopted by the Maritime Safety Committee (MSC) at its
66-th session in June 1996: - Resolution MSC.47(66) and
Resolution MSC.48(66).
An essential requirement here is ensuring safe motion of
the lifesaving device, a life raft or lifeboat for instance, over
the ships side and launching it on water, under unfavourable
conditions of trim of up to 10 and list of up to 20, either way,
using the gravity force or the collected energy, irrespective
of the energy sources on the ship:
a) when the boat/raft is loaded with the full number of people
provided for it, and
b) when only crew in on its deck.
Only for the life raft, moving it over the ships side with the
aid of the manual drive is allowed. The principles of design
and use of these devices are described in [15].
The present article discusses the kinematics of the lifeboat
during its launching. Like in [10], substituting expensive field
tests by numerical simulations the authors aim to identify
most dangerous conditions for the motion of the lifeboat
with people at the ships side. The results of these simulations
will be used to eliminate dangerous scenarios of lifeboat
launching and finding most favourable conditions for this

POLISH MARITIME RESEARCH, No S1/2017 53

action. Calculating the trajectory of motion, complemented
by accelerations and speed of the lifeboat lowered on ropes,
allows to learn better the effect of individual components
of the designed device on parameters of its motion. This,
in turn, will lead to the improvement of the designs of
lifesaving devices, with respect to both their structure and
applied materials, and subsequent softening of the effects of
possible boat strokes against the ships side. The results of
the simulations performed for a large number of different
lifesaving device types and launching scenarios will make the
basis not only for finding the most favourable design solution
and geometrical parameters of the device and its location,
but also for selecting the optimal start time of launching in
relation to the instantaneous ship position.
PROCESS OF LIFEBOAT LAUNCHING
FROMSHIP
The increasing number of ships in operation and the
abovementioned requirements concerning the lifesaving
and rescue devices have stimulated intensive development
of both design solutions, and drive and control systems of
these devices.
At present, various types of davits are in common use.
With respect to the way in which the boat or raft is moved
over the ships side, their designs can be divided into:
level-luffing sample davit of this type is shown in Fig. 1.
luffing,
travelling-luffing,
travelling-telescope,
round-bar,
Fig. 1. Photo and kinematic scheme of the level-luffing davit in three
characteristic positions: - cruising, - at ships side when the people are directed
from the evacuation deck to the boat, - during boat launching.
The system of davits with lifeboats is sometimes
complemented by the Marine Evacuation System (MES),
which usually consists of large rafts and, optionally, platforms.
This solution is mainly used on new passenger ships taking
large numbers of people. In the MES system, the people slide
down, using special chutes, from the ship deck onto the
platform or directly to the raft. Due to highly unpredictable
behaviour of passengers, attempts are made to simulate the
evacuation process, see [9] for instance, to be able to predict
possible difficulties and problems.
54 POLISH MARITIME RESEARCH, No S1/2017
A common feature of all devices in which the lifeboat is
lowered on ropes is the similar procedure of evacuating people
from the ship. This procedure includes the following stages:
removing protections mounted for the time of voyage and
moving the davit with the boat over the ships side,
lowering the boat on the ropes to the evacuation deck level
and pulling the boat to the ships side,
preparing passages, directing people from the deck to the
boat, locking safety belts, and closing the boat door,
detaching the ropes which pressed the boat to the ships
side and, once the boat hangs free, releasing the rope winch
brake to allow the boat to lower gravitationally at the speed
stabilised by the centrifugal brake of the winch,
when the boat nears the water, lifting the hook release lever
to allow the hydrostatic release to act when the boat is on
water. Its action will automatically release both hooks and
disconnect the boat from the ropes, thus allowing it to sail
away from the ship.
CALCULATION MODEL USED IN
SIMULATION
The calculation model elaborated to simulate the process of
boat launching consists of the following components, shown
in a simplified form in Fig. 2:
a) ship hull subjected to harmonic motions in all six degrees
of freedom with respect to its centre of gravity,
b) davit arms fixed to the permanent upper deck structure
of the ship hull,
c) ropes connecting the davit with the boat,
d) mechanism for running out the rope (rope winch
with centrifugal brake and locking brake, and passive
compensator),
e) lifeboat with fenders.
Fig.2. Simplified view of the model used in the simulation.
Relevant kinematic constraints were created for particular
model components. The ship hull experiences harmonic
motions in all degrees of freedom, thus defining the excitations
in the remaining model components. The linear motions of
the boat are described by Equations (2.1), while its angular
motions are given by Equations (2.2). Both equation sets were
assumed according to the literature [7].
 Table 1. Ship hull motion parameter values assumed in the simulation.

(2.1) Min Max Step Number

value value of cases
amplitude with respect

linear hull motions

0m 6.0 m 0.3 m 21
to Z-axis

amplitude with respect 0m 0.5 m 0.5 m 2

to Y-axis

amplitude with respect 0m 0.5 m 0.5 m 2

(2.2) to X-axis

amplitude with respect 0o 4o 2o 3

to Z-axis

angular hull motions

amplitude with respect 0o 8o 2o 5
to Y-axis

amplitude with respect 2o 16o 2o 8

where:
xc, yc, zc coordinates of the boats mass centre
xo, yo, zo directions of principal axes of inertia of the
boat
instantaneous angular velocities of the boat
rotating about its principal axes of inertia
principal moments of inertia of the boat
force vectors along principal axes of inertia
, , vector moments of inertia in relation to
principal axes of inertia
SELECTING RANGES OF HULL
MOTIONPARAMETERS
An important limitation for the performed calculations is
the time of computation required to obtain final results. The
above described calculation model was implemented in the
form of a computer code, written in C++ by dr. Pawe Dymarski
and described in [2] and [3]. The code allows to simulate the
entire process of boat lowering, from the time of boat hook
release to boat setting on water and disconnecting from the
ropes on which it was lowered. The time of computations for
a single set of parameters defining the simulation conditions
is equal to about 1.5 minute on amid-class PC. That means
that the computations for 121000 cases would require 181500
minutes, i.e. 3000 hours (or 126 days = 4.2 months). Such
along computing time is inacceptable in practice.
That is why a decision was made to perform computation
at a professional computer centre. For this purpose, the
code was modified in such a way as to enable running it on
acomputing cluster available at the Academic Computer
Centre (CI TASK) in Gdansk. The cluster comprises 32000
processors and allows to perform parallel computations.
However, due to a large number of users, the real access to
the cluster is limited. In the present case, about a thousand
of parallel computations could be performed simultaneously.
The available data on sea waves in the Baltic sea water
region have made the basis for selecting parameters for
calculations. The selected parameter values are given in
Table1. The periods for particular motions were assumed
based on the wave data published in [4] and [11].
to X-axis
phase with respect to
X-axis 0o 330o 30o 12
Total number of cases: 120960
The maximal wave height recorded on the Baltic sea is
about 15m, but this extreme value occurs very rarely. The
research performed at the PAS Institute of Hydroengineering,
Gdansk, Poland, and published in [11] indicates that even
during heavy storms the wave height rarely exceeds 6m.
This assessment is also confirmed by global wave statistics
presented in [4]. Moreover, the design practice recommends
to assume the so-called significant wave height, which is
even smaller and acc. to [1] does not exceed 4,3m on the
Baltic sea. Having taken all this into consideration, it was
assumed that the ship hull heaving, damped with respect to
sea waves, does not exceed 6m.
The remaining parameters of motion were selected based
on classification regulations. Ref. [16] in Chapter 10 Item 1.1
clearly states that each device used for launching and lifting
a lifeboat should be designed in such a way as to ensure safe
operation under conditions of ship trim of up to 10 and list
of up to 20, either way. The regulations do not say whether
these limits are static or dynamic in nature. In the engineering
practice they are considered static, but in the present case
they were assumed as dynamic conditions, therefore their
ranges were slightly decreased: list to 16o, and trim to 8o.
Due to insufficient data on phenomena taking place during
lifeboat launching and the fact that the present research was
mainly focused on this process, a relatively simple model of
waves and effects of their activity was used, compared to that
described in [6], for instance.
The selected values of ship hull motion amplitudes in
particular degrees of freedom, complemented by relevant
phase shifts in the angular motion about the X-axis, resulted
in a total number of 120 960 examined cases. The adopted
parameters were considered most representative and best
describing possible and permissible variants of ship hull
motion in this water region. For each case, a simulation was
performed using the lifeboat of 8.8 m in length and having
the mass of 15550 kg.

POLISH MARITIME RESEARCH, No S1/2017 55

 Table 3. Ship hull motion parameters for which the highest value of linear
RESULTS OF SIMULATION motion acceleration acting on the launched boat was recorded.

amplitude period
Parallel computations performed at CI TASK allowed to

Linear hull
Z-axis 5.7 m 8.4 s
collect results for all 120960 cases after about two hours.

motions
For each case, a check was made whether or not the lowered Y-axis 0.5 m 12.0 s
boat struck against the ships side and what the maximum X-axis 0.0 m -
accelerations acting on the lowered boat were. Z-axis 4.0o 12.0 s

Angular hull
All cases in which there was no boat stroke against the

motions
Y-axis 0.0o -
ships side were considered safe. The total number of such
X-axis 12.0o 7.3 s
cases amounted to 4 470. The maximal accelerations recorded
phase with respect to X-axis 30o
in these cases were much lower than when the stroke took
place. The recorded maximal acceleration took place when the
Table 2 collects rates of incidence of particular acceleration lowered boat struck against the ships side. Figures 4 and
values obtained from the calculations. Figure 3 presents 5 show, respectively, the time-histories of changes of the
the same data in graphical form. They reveal that in 108210 lifeboats centre of gravity when the boat was lowered under
cases, which is more than 89% of all examined cases, the the above critical conditions, and linear acceleration changes
accelerations did not exceed 3,5 g. for the same case.
Table 2. Rates of incidence of particular acceleration values.
Acceleration range

10,7g
2,5g

4,5g

6,5g

8,5g
3,5g

5,5g
1,5g

9,5g
7,5g

10g
6g
4g
2g

8g

9g
3g

5g
1g

7g
Number of cases

26894
32820

29915

15312
3085

7237

1320
3135

250
184

552

133

60

27

12

12

Fig. 4. Time-history of centre of gravity position changes for the lifeboat

launched under critical conditions.
Fig. 3. Incidence rate diagram for particular acceleration values.

The maximal acceleration, amounting to: 103.55 m/s2

(components: X=-10,65; Y=-102.06; Z=13.92) was recorded
after 17.8 seconds from the beginning of boat lowering, for
ship parameters given in Table 3.

56 POLISH MARITIME RESEARCH, No S1/2017


Fig. 5. Time-history of linear accelerations acting on the boat launched under
critical conditions.
A distinctive peak in the green curve in Fig. 4, between
15 and 20 second, illustrates the lifeboats stroke against the
ships side. A negative peak, observed for the same time in
the green line representing the Y-component of acceleration
(perpendicular to the ships side) in Fig. 5, is even more
prominent.
CONCLUSIONS
The case selected from among all 120 960 cases in the
previous Chapter is the extreme case. The recorded linear
acceleration amounting to 103.55 m/s2, which was recorded
during lifeboats stroke against the ships side, is unacceptable
for people in the boat and is likely to destroy the boat itself.
It turns out that for the same ship hull motion amplitudes,
but different phase shifts, the recorded accelerations can be
lower, as seen in Table 4.
Table 4. Maximal linear accelerations for the same amplitudes of ship hull
motion and different phase shifts with respect to the X-axis.
Linear motions [m] Angular motions [o]
Item
amp. Z amp. Y amp. X amp. Z amp. Y amp. X
That is why of great importance is the selection of the start
time for lifeboat launching. Minimising the boat rocking in
the beginning phase of lowering to protect the boat against
kinetic energy gaining is very important, as this kinetic energy
is extremely dangerous when the boast strikes against the
ship hull. An attempt to evaluate possible effects of such
strokes is made in [12].
The lifeboat launching process is affected by numerous
factors, including flexibility of particular structural elements
of davits, characteristics of their mechanisms, as well as the
ship position and instantaneous motion parameters at the
time when the boat launching is started. For instance, such
simple elements as fenders can have a significant impact on
the values of the accelerations recorded when the boat strikes
against the ship hull. The boat fender which was taken into
account in the present calculations had a linear characteristic,
i.e. its deflection was proportional to the applied force. Further
research oriented on developing a new characteristic of the
fender would undoubtedly lead to changes in the course of
motion of the lifeboat bouncing off the ships side and could
considerably reduce the level of accelerations recorded in this
motion. However, this issue is very wide and worth analysing
in a separate article.
The calculation model presented in this article allows
to perform a detailed analysis of a single case of lifeboat
launching from ship at given parameters. Transferring the
calculations to the cluster and performing them in parallel
have provided opportunities for effective analysis of a large
number of cases. Such a two-stage analysis allowed first to
identify most dangerous cases based on the assumed limits,
the critical acceleration value for instance, and then to
analyse in detail possible causes of their appearance. This
identification and analysis will make it possible to introduce
structural and functional changes to the lifeboat launching
system, or at least to introduce a procedure which will force
a more favourable course of events during boat lowering.
However, a detailed study of possible solutions intended to
improve the lifeboat launching goes beyond the scope of this
article and will be discussed in further publications.
Linear acceleration [m/s2]
phase
X
X Y Z Resultant

1. 0 -3.81 -19.33 4.26 20.19

2. 30 -10.65 -102.06 13.92 103.55
3. 60 -1.75 1.88 19.36 19.53
4. 90 1.74 4.32 23.65 24.11
5. 120 4.15 8.59 20.31 22.44
6. 150 -5.20 -4.54 22.08 23.13
0.00
0.50

12.0
5.70

4.0

0.0

7. 180 6.81 6.02 25.53 27.10

8. 210 -3.18 -29.49 -9.10 31.02
9. 240 -0.56 -19.22 15.13 24.47
10. 270 8.39 -37.93 10.94 40.36
11. 300 8.73 -36.85 1.90 37.92
12. 330 0.99 -18.83 -4.60 19.41

POLISH MARITIME RESEARCH, No S1/2017 57

BIBLIOGRAPHY
1. Dudziak J., The theory of the ship (in Polish).
Wydawnictwo Morskie, Gdask 1988.
2. Dymarski P. Dymarski Cz., COMPUTATIONAL
MODEL FOR SIMULATION OF LIFEBOAT MOTIONS
DURING ITS LAUNCHING FROM SHIP IN ROUGH
SEAS, Polish Maritime Research Vol. 19 No. 3(75), 2012.
3. Dymarski P., Dymarski Cz., Safecrafts GUT_WP2_D2.3
Report, 2006-04-07 Gdask.
4. Hogben, N., Da Cunha, N. M. C. and Oliver, G. F.,
GLOBAL WAVE STATISTICS. Compiled and edited
by British Maritime Technology, Unwin Brothers, UK,
1986.
5. Iwicki P., Tejchman A., Chrcielewski J., DYNAMIC FE
SIMULATIONS OF BUCKLING PROCESS IN THIN-
WALLED CYLINDRICAL METAL SILOS, Thin-walled
structures. Vol. 84 (2014), pp.344-359.
6. Kozicki J., Niedostatkiewicz M., Tejchman A., Muhlhaus
H., DISCRETE MODELLING RESULTS OF A DIRECT
SHEAR TEST FOR GRANULAR MATERIALS VERSUS
FE RESULTS. Granular Matter, Vol. 15, iss. 5 (2013),
pp.607-627.
7. Veic D., Kraskowski M., Bugalski T., BOTTOM FIXED
SUBSTRUCTURE ANALYSIS, MODEL TESTING
AND DESIGN FOR HARSH ENVIRONMENT. Osta-
chowicz W. et al. (eds.), MARE-WINT, Chapter 13, DOI
10.1007/978-3-319-39095-6_13.
8. Kruszewski J., Ostachowicz W., Wittbrodt E., General
Mechanics (in Polish), Gdansk University of Technology,
Gdansk 1993.
9. Kujawa M., Szymczak C., NUMERICAL AND EXPERI-
MENTAL INVESTIGATION OF ROTATIONAL
STIFFNESS OF ZED-PURLINS CONNECTION WITH
SANDWICH PANELS, Thin-walled structures, Vol. 75,
(2014), pp.43-52.
10. Magluta C., Roitman N., Batista R.C. DYNAMIC BEHAV-
IOUR ANALYSIS OF A LIFEBOAT SYSTEM UNDER
SIMULATED ACCIDENTS, Mechanical Systems and
Signal Processing, Volume 10, Issue 6, November 1996,
Pages 763-774.
11. Morch H. J., Peric M., Schreck E., Moctar O., Zorn T.,
SIMULATION OF FLOW AND MOTION OF LIFE-
BOATS, ASME 2009 28th International Conference on
Ocean, Offshore and Arctic Engineering, Volume5: Polar
and Arctic Sciences and Technology; CFD and VIV, Paper
No. OMAE2009-79608, pp. 595-605; 11 pages, doi:10.1115/
58 POLISH MARITIME RESEARCH, No S1/2017
OMAE2009-79608, Honolulu, Hawaii, USA, May 31
June5, 2009.
12. Papliska-Swerpel B., Paprota M., Przewcki J., Sulisz
W., ANALYSES OF WAVE RECORDS FROM THE
SOUTHERN BALTIC SEA WITH THE EMPHASES
ON LARGE WAVE EVENTS. Dynamics of coastal waters
and their modelling: Proceedings of the 9th German
Polish Seminar. Ed. J. Sndermann, Hamburg: IFM,
GKSS Printing Office, 2005, Berichte aus dem ZMK, B
49, s. 5-16.
13. Raman-Nair W., Chin S.N., ESTIMATION OF IMPACT
FORCES BETWEEN SMALL BODIES IN WAVES,
Ocean Engineering, Volume 46, June 2012, Pages 46-51.
14. Rucka M., Wilde K., EXPERIMENTAL STUDY ON
ULTRASONIC MONITORING OF SPLITTING FAIL-
URE IN REINFORCED CONCRETE. Journal of nonde-
structive evaluation. Vol. 32, Issue 4 (2013), pp.372-383.
15. Sding H., FLOW COMPUTATIONS FOR SHIP SAFETY
PROBLEMS, Ocean Engineering, Volume 29, Issue 7, June
2002, Pages 721-738.
16. Taylor D.A., INTRODUCTION TO MARINE ENGI-
NEERING (Second Edition), Chapter 10 - Deck machin-
ery and hull equipment, 1996, Pages 180-199.
17. Non-classification regulations concerning the equip-
ment of sea-going vessels (in Polish), Polish Ship Register,
Gdask, 1990.
CONTACT WITH THE AUTHORS
Czesaw Dymarski
Pawe Dymarski
Aleksander Kniat
Gdansk University of Technology
Faculty of Ocean Engineering and Ship Technology
11/12 Narutowicza St.
80 - 233 Gdask
Poland
 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 59-66
10.1515/pomr-2017-0022

TECHNOLOGY CONCEPT OF TLP PLATFORM TOWING AND

INSTALLATION IN WATERS WITH DEPTH OF 60 M

Czesaw Dymarski
Pawe Dymarski
Jdrzej ywicki
Gdansk University of Technology, Poland

ABSTRACT

The article is part of the design and research work conducted at the Gdansk University of Technology, Faculty of Ocean
Engineering and Ship Technology, in cooperation with a number of other research centres, which concerns offshore
wind farms planned to be built in the Polish zone of the Baltic sea in the next years. One of most difficult tasks in this
project is building suitable foundations for each power unit consisting of a tower and a wind turbine mounted on
its top. Since the water regions selected for building those wind farms have different depths, there was need to study
different possible technical variants of this task, with the reference to both the foundation structures themselves, and
the technology of their transport and setting, or anchoring. The article presents the technology of towing, from the
shipyard to the setting place, and installation of the foundation having the form of a floating platform of TLP (Tension
Leg Platform) type, anchored by tight chains to suction piles in the waters with depth of 60 m.

Keywords: platform anchoring, towing, TLP platform installation

INTRODUCTION source is not equally reach in different areas and, what is

The permanently increasing demand for energy, with
simultaneous dwindling of the worlds fossil energy sources,
leads to growing interest in acquiring energy from renewable
sources and from various unconventional sources, such as
waste materials for instance [1]. Along with that, higher and
higher attention is paid to rational energy use, which makes
combined energy generation systems and hybrid drives
become very attractive and worth studying [2]. Here, special
attention is given to the efficiency of the energy production
methods and their impact on the environment [3,4].
A renewable energy source which is ecological and available
in all countries is the wind. That is why the development of
windpower engineering is very dynamic, although this energy
more, reveals high variability of parameters [5]. Here, the
areas of interest refer to selecting most favourable types and
dimensions of wind turbines and their control systems, and
finding places with relatively high and stable wind speeds,
preferably situated far from residential areas. Such favourable
conditions are observed at sea, and therefore a large number
of maritime countries have made decisions to build offshore
farms of wind power stations [6,7]. However, these projects
are very difficult and require solving numerous complex
problems, with the resultant high costs involved, particularly
the investment cost. The most difficult problems refer to
wind turbine foundations, bearing the name of supporting
structures, on which the towers with wind turbines are

POLISH MARITIME RESEARCH, No S1/2017 59

mounted. Types of these structures, as well as the level
of difficulties and costs of their design, necessary tests,
production, transport, and installation, mainly depend on
the depth of the water region in which they are to be installed
[8,9]. In general, the supporting structures used in the waters
with depth not exceeding 40 m have the form of founded
supports, such as, for instance, mono-piles, and jacked or
triode type platforms [10], while in deeper waters, standard
floating platforms are in common use [6].
The article presents the technology of towing and anchoring
(installation) of a TLP-type floating platform intended to work
as the supporting structure on which the tower with wind
turbine will be installed. Selecting the type of platform and
the method of its anchoring was preceded by the analysis of
the research results published in [11,12,13,14]. It also based
on personal experience gained by the authors in their past
research, and on the results obtained by them.
The platform, shown in Fig. 1 as ready for towing, is
planned to be installed on the Baltic sea, in waters with depth
of 60 m. The basic structural elements of this platform are the
hull and the column, on which the tower with the 6MW wind
turbine will be mounted. The turbine axis is assumed to be at
the height of 90 m above sea level. The 10-meter high tripod
hull of the platform, shown in Fig. 1, will be submerged in
the water in such a way that its bottom surface will be 40 m
above the seabed, while the upper flange of the column will
protrude 15 m above the sea level. Each arm of the hull will be
connected, via tight anchor chains, to one of the three suction
piles set in the seabed. Further in the article, successive steps
of platform installation, including its preparing, towing, and
anchoring at the destination place, are discussed.
PREPARING THE PLATFORM FOR TOWING
It is assumed that the platform will be built in one of
shipyards situated in Gdansk or Gdynia. A more detailed
description of the structure, with drawings and strength
analyses, will be provided in a separate article prepared for
publication. The platform is planned to be equipped with
the basic ballast system adapted for cooperation with an
external pump unit.
The platform will be able to be installed and anchored
using a number of different systems. For each system, a
different structural solution was designed in the end parts
of the arms in the underwater hull section. The system of
platform anchoring which is described in this article makes
use of six vertical tight anchor chains, mounted in pairs at
each end of the three platform hull arms, as shown in Fig. 1.
These chains are planned to be installed on the platform at the
shipyard before towing. The upper chain parts will be fixed to
the platform arms using a ratchet stopper and a tension wedge,
which will facilitate the assembly and provide opportunities
for better distribution of chain forces into the fastening
elements and the arm structure. Before towing, special lugs
or chain stoppers and rope pulleys will be installed on the
upper surfaces of the platform arms, near the column, while a
60 POLISH MARITIME RESEARCH, No S1/2017
removable protection cap with cylindrical part covered with
polyamide, or a set of two wide chain rollers for directing the
anchor chain during its lowering, will be mounted at each arm
end edge. In the latter case, two smaller rope pulleys, intended
to be used for auxiliary works, will be installed between the
two chain rollers. Parts of arm surfaces on which the chains
are expected to be laid should be covered with wooden
boards, or other type of facing to protect the hull plating
from direct contact with the chain. Once this preparatory
work is done, lower parts of the chains will be lifted, one by
one, by a crane and, after embracing front arm surfaces with
protecting facing or pulleys, will be arranged along the arms,
on the upper arm surfaces. The end link of each chain will be
fastened with a pin to the lug or chain stopper situated close
to the platform column. The method of chain attachment
to the platform column is schematically illustrated in Figs.
1 and 2. It is expected that in this attachment method, part
of the chain of about 8 m in length will hang down under
the platform bottom to the approximate depth of 4 m. It has
been assessed in preliminary calculations that the platform
draught during towing will be h0 = ~3m, while the height of
its centre of gravity will be hsc =~8 m, all this ensuring good
platform stability during the towing operation.
Vertical fragments of small-diameter pipes (not shown
in Fig. 1) will be fastened to the side walls of the arm ends.
These pipe fragments will be used for guiding lines for remote
releasing or locking of the mechanisms connecting the chains
with the anchor piles shown in Fig. 5.
Fig. 1. Simplified isometric drawing of the platform prepared for towing. The
enlarged section shows the method of fastening of upper anchor chain ends to
the platform arm.
PREPARING THE SEABED FOR PLATFORM
ANCHORING
Preparing the seabed for anchoring of a TLP-type platform
requires numerous evaluative and research studies in order
to assess possible extreme sea conditions (wind, waves, sea
currents, temperature changes), as well as to identify the
structure and strength of the seabed [15,16], the effect of
platform foundation installation on the environment, and
the economic and technical conditions. In the present case,
the analysis was limited to technical and, to some extent,
economic issues which allowed to select most profitable
devices and technologies.
Anchoring of the TLP-type platform is characterised by
the appearance of extremely high loads, mainly in vertical
direction, acting on the elements anchored in the seabed [17,
18, 19, 20]. Constructional elements which best carry this
type of load are piles, driven into the seabed or introduced
using the suction method. The type of the applied pile and
its geometrical parameters depend on numerous factors, in
particular on the expected loads, the seabed structure [21],
the water region depth, as well as on the available specialised
units and the cost of their service. In the present case,
taking into account the abovenamed factors, a preliminary
decision was made to use suction piles. Decisive arguments
for this choice included relatively short time of piling, and
favourable shapes and proportions of pile dimensions, large
diameter in particular, thanks to which two anchor chains
will be able to be fastened to each pile. Precise evaluation
of geometrical and strength parameters of the piles, and
selection of a specialised unit with the equipment needed for
driving them into the seabed will be discussed in a separate
article. Based on preliminary calculations performed for
the assumed type of soil (fine sand), depth of 60 m, and the
required vertical component of the holding force, the pile
diameter was assessed as equal to D = 5m and the length equal
to L=18 m. The pile positions in the seabed should correspond
as precisely as possible to the platform position, for the anchor
chain pipe axes in the platform hull arms to coincide with
the axes of the fastening mechanisms, mounted on the pile to
which the anchor chains will be fastened (their displacement
should not exceed ~0,10 m). To facilitate the operation of
fastening the lower chain end to the lug on the pile, these
mechanisms will have in their upper part simple leading
constructions, for the end link of the lowered chain to settle
automatically in a relevant seating of the mechanism and to
take the correct position for introducing the pin connecting
it with the lug (see Fig. 5)
It is planned to install a control lever on the mechanisms
connecting the anchor chain with the pile. This lever will be
able to be operated directly by a diver, or remotely via a thin
control cord, the upper end of which will be temporarily fixed
to a small buoy floating on the sea surface. Upon request of
the owners, a remote cordless system for switching on the
above mechanism can also be provided.
TOWING
An assumption has been made that the towing operations
can be performed when the sea state does not exceed 4 on the
Beaufort scale and the wind speed does not exceed W= 8m/s.
The towrope will be fastened at the height of 4 m above the
platform bottom, i.e. about 1 m above the sea surface. For
the towing speed of V = 2 m/s with respect to water and the
opposite wind direction, the required towrope force is equal
to Fh = 380 kN. The highest overturning moment acting on
the towed platform Mw = 970 kNm will occur for the above
maximal towing and wind speeds, but for the same wind
direction as the towing direction. The platform should be
towed by at least two tug units, one towing and one dragged
unit, both connected with the platform via ropes in the way
accepted by classification societies. The calculated stability
of the towed platform in the above defined conditions is
sufficiently high to ensure its safe delivery to the anchoring
place.
PREPARING THE PLATFORM FOR
ANCHORING LOWERING ANCHOR
CHAINS
When the platform reaches the anchoring place, but before
keeping it precisely in the final position, subsequent anchor
chains can be lowered.
This operation can be executed in two ways:
with the assistance of three tugs,
with the assistance of a specialist vessel.
Further in the article, only the latter method is described
in detail.
Using a specialist vessel for chain lowering is considered
justified, as this vessel is also planned to be used in other,
frequently more difficult, dangerous and expensive operations,
such as platform ballasting and anchoring.
In authors opinion, this vessel should have a properly
shaped hull in the stern section and two azimuth propellers
for precise position keeping. It should be equipped with a
slewing deck crane and a set of six hydraulic servomotors for
connecting the vessel with the platform column, as shown
in a simplified way in Figs. 2, 3 and 4. It should also have a
properly trained crew and other specialist equipment for
performing technical operations on the vessel or platform,
or certain underwater actions. This vessel is also planned to
be used in further operations of platform installation, which
will be discussed in the next Chapters. For the purpose of
chain lowering, at least two technical servicemen should be
brought onto the platform hull.
The vessel comes up to the platform and moors with its
stern between the platform arms in the way shown in Fig.
2, for the crane jib to be able to reach over the entire surface
of the platform hull arm. The crane operator sets the jib in
proper positon and lowers the rope with the hook to the place
of chain fastening to the lug (or stopper) on the platform hull.
People on the platform hull connect the hook lowered with
the rope to the end of the laying chain, at the lug (or stopper)
to which it was fixed. The crane operator slowly heaves the
rope with the hook to tighten it slightly, then the servicemen
on the platform pull out the pin connecting the end chain
link with the lug (or release the stopper) on the hull. At the
same time the crane operator, manoeuvring with the jib and
POLISH MARITIME RESEARCH, No S1/2017 61
Fig. 2. Simplified top and side view of the specialist vessel position with respect to the platform in the beginning and final phase of chain lowering with the aid of the
deck crane. The maximal required crane radius is R=~23 m.
the rope winch, slowly lowers the chain, until it hangs with
its end fastened in the chain pipe, thus unloading the rope,
as shown in Fig. 2. In this situation, further hook lowering
will cause its automatic loosening and disconnecting from the
chain. Then the crane operator takes the rope with the hook
from water and moves it towards the lug with the other chain
on the same arm to lower it in the same way. The procedure
of chain lowering from other arms is identical, but is to be
preceded by vessel sailing to the next mooring position with
respect to the arm.
PREPARING FOR BALLASTING,
BALLASTING, AND ANCHORING OF THE
PLATFORM
After lowering all anchor chains, their fastening in the
chain pipes are to be inspected visually, then the hull deck is
to be cleared from plates and other redundant elements, and
the control cords of mechanisms connecting anchor chains
with piles, which were earlier fixed to the floating buoys, are
to be now fixed to the cords on the platform, the other ends
of which are in the control room in the upper part of the
column. Further operations refer to direct preparation for
platform ballasting. The type of these operations depends
on the available devices and the adopted procedure. Below is
presented the technology of platform ballasting and anchoring
with the aid of the specialist vessel. In authors opinion, this
technology provides opportunities for maximal shortening
of the operating time. It is assumed that this operation will
62 POLISH MARITIME RESEARCH, No S1/2017
be able to be conducted at sea state below 3, which means the
wind force 3B and wave height of up to 0,5 m.
In the first stage after lowering the anchor chains, the
platform servicemen are to be relocated via the crane to a
bridge situated in the upper part of the platform column.
This is the place where terminals and valves of the platforms
ballast system are situated. Here, after passing elastic hoses
from the ship and connecting the platform ballast system
to that of the ship, the ballasting process is started, causing
slow submergence of the platform hull to the depth of 9,5
m. Further platform submergence should be carried out at
simultaneous stability control, as the platform loses stability
when the hull is fully submerged. Then cameras and reflectors
are installed underwater on the suction piles, along with
control cords for the mechanisms fastening and releasing
anchor chains. Before starting the final platform installation
operation, the specialist vessel is to be connected with the
platform column in such a way as to ensure platform stability
after full submergence of its hull. To make it possible, the stern
section of the vessel with a U-shaped symmetrical channel
situated above the water line is equipped with three special
devices symmetrically distributed with respect to the vessels
symmetry axis, in 120 degree steps, as shown in Figs. 3 and
4. Each device consists of the stand, situated on the upper
deck of the stern section, with an arm leaned out over the
channel, and a set of two main hydraulic servomotors and
one auxiliary servomotor. The two main servomotors are
connected together by a chain or cable rod, which allows the
auxiliary servomotor to lean them out, as shown in Fig. 4.
Special lugs with rollers having elastic raceways and equipped
with slide bearings are installed at the ends of the main
servomotor piston rods. These servomotors are mounted in
the vertical channel walls, and the lugs of each pair of them
are connected together by a chain, which then goes through
the chain wheel of the stand situated over the channel. On
the other side of the stand it goes through the stopper and
the auxiliary servomotor lug roller, to be finally fixed to the
stand structure. For the time of voyage, the piston rod of
the auxiliary servomotor is pulled inside the cylinder and
the chain is locked in the stopper, which corresponds to the
servomotor position: lifted up.
Before this operation starts, elastic fenders are mounted
on those vessel channel sides which are situated close to the
platform structure. The main servomotors are lifted up, to
increase the available channel width when the vessel nears
the platform column. The vessel slowly moves back, keeping
symmetrical position with respect to the platform hull arms
in such a way that the symmetry plane of the stern section
Fig. 3. Stern section of the specialist vessel during TLP platform installation
channel continuously coincides with the column axis. The
mounted fenders protect against contacts between the vessel
and the platform and predetermine their relative positions.
When the vessel stops in this preliminary position, which is
close to the final correct position with respect to the platform
column, the stoppers on the stands are released and the piston
rods of the auxiliary servomotors are pulled out, this way
setting the servomotors installed on the channel walls in the
operating position. After checking the correctness of their
positions, the chain stoppers on the stands are blocked and oil
supply for the main servomotors is initiated, to ensure even
and automatically controlled motion of their piston rods with
rollers out of the cylinders. What is noteworthy, the option of
individual manual control of their motion is also provided.
At this time, proper pressure contact of all rollers with the
column and the accuracy of the column axis and platform
position with respect to the elements on the seabed are to
be re-checked and, if necessary, the platform position is to
POLISH MARITIME RESEARCH, No S1/2017 63
 Fig. 4. Positions of servomotors positioning the platform with respect to the vessel during three basic platform installation stages: 1 stage of bringing vessels
stern near the platform column leading servomotor rollers lifted up to increase the save space around the column; 2 stage of symmetrical setting of the vessel
stern channel with respect to the platform column leading servomotor rollers lowered to the operating position and pressed to the column; 3 stage of platform
ballasting and platform hull submergence servomotors ensure controlled movement of the rollers and their pressure contact with the column; the diameter of the
column in the planes of contact with servomotors decreases with the increasing submergence depth.

be corrected with the aid of the vessel propulsion system.
Then the process of platform ballasting and submerging
begins. During this process, continuous pressure contact
of the servomotor rollers with the platform column is to be
ensured, and the position of the column is to be continuously
supervised, and accompanied with additional observation
of motion of the last links of the anchor chains and their
settlement in the seats of their fastening mechanisms, as
shown in Figs. 5 and 6.
When slight clearance between two last chain links is
observed, along with their deflection, the ballasting process
is stopped. Releasing the control cords moves the pins in
the mechanisms connecting the chain with the lugs on the
suction piles.
After checking the blockages of all anchor chains, the
platform should be slowly unballasted until all anchor chains
become properly tight. The next step may be the start of
removing the auxiliary equipment from the platform and
transporting servicemen from the bridge on the column with
the aid of the slewing deck crane. After completing these
operations, the hydraulic supply system is switched on, to
pull the main servomotor piston rods with rollers inside the
cylinders. Once this process is completed, the chain stoppers
on the stands over the channel are released and the hydraulic
supply of auxiliary servomotors is switched on, to pull their
piston rods inside the cylinders, thus setting the servomotors
on the side walls of the channel to the voyage position, after

Fig. 5. Axial section through the mechanism connecting the anchor chain with the suction pile, just before and after chain blocking: 1 piles body, 2 chain link,
3 blocking pin, 4 lever for chain blocking and release, 5 connector between the lever arm and the pin moved by it, 6 pins for optional lever position blocking

64 POLISH MARITIME RESEARCH, No S1/2017

 Fig. 6. Axial section through the mechanism fastening the anchor chain to the suction pile after blocking and tightening of anchor chains: half section (left) and
bottom view (right)
which the chains are blocked in the stoppers. Then the vessel
can start moving out of the area occupied by the platform.
CONCLUSIONS
In authors opinion, the presented technology concept
of towing and installation of an offshore wind turbine
supporting structure having the form of TPL-type platform
is simpler and cheaper than other presently used technologies,
at the same time fulfilling the safety requirements to the
same extent. This assessment takes into account the number
of objects to be installed in the Polish zone of the Baltic sea,
in waters with depth exceeding 50 m, as well as the distance
of the installed farms from the shipyard and sea conditions
observed in these water regions. The use of well-known
technologies making use of large self-lifting vessels with
large cranes was assessed in this case as too expensive and
not bringing substantial benefits with respect to both time
consumption and safety of the operations to be performed.
The authors believe that this article will find interest among
institutions and companies involved in the here analysed
problem matters, which will create space for meetings and
discussions on substantive issues concerning operations
of this type, all this resulting in wider and more objective
evaluation of the proposed technology.
ACKNOWLEDGMENT
This research was supported by The Polish National Centre
for Research and Development (NCBR) under the project
WIND-TU-PLA ERA-NET MARTEC II (Agreement No.
MARTECII/1/2014)
BIBLIOGRAPHY
1. Mikielewicz D., Wajs J., Zikowski P., Mikielewicz J.:
Utilisation of waste heat from the power plant by use of
the ORC aided with bleed steam and extra source of heat,
Energy, 97, 11-19, 2016.
2. Kropiwnicki J., Kneba Z., Zikowski M.: Test for assessing
the energy efficiency of vehicles with internal combustion
engines. International Journal of Automotive Technology,
Vol. 14, nr 3 (2013), s. 479-487.
3. Hirt ., Lampart P.: Complex multidisciplinary
optimization of turbine blading systems// ARCHIVES
OF MECHANICS. -Vol. 64, nr. 2 (2012), s.153-175.
4. Barthelmie R., Pryor S., Frandsen S., Hansen K., Schepers
J., K. Rados K., Schlez W., Neubert A., Jensen L. and
Neckelmann S.: Quantifying the Impact of Wind Turbine
Wakes on Power Output at Offshore Wind Farms. Journal
of Atmospheric and Oceanic Technology Vol. 27, 2010,
5. Ackermann T., Sder L.: Wind energy technology and
current status: a review, Renewable and Sustainable Energy
Reviews, 4 (2000), pp. 315374
6. Markard J., Petersen R.: The offshore trend: structural
changes in the wind power sector. Energy Policy, 37 (2009),
pp. 35453556.
7. Musial W., Butterfield S., Ram B.: Energy from offshore
wind. Proc. offshore wind Conference Houston (2006),
pp. 18881898.
POLISH MARITIME RESEARCH, No S1/2017 65
8. Denis Matha: Model Development and Loads Analysis
of an Offshore Wind Turbine on a Tension Leg Platform,
with a Comparison to Other Floating Turbine Concepts.
University of Colorado - Boulder Subcontract Report
NREL/SR-500-45891, 2010.
9. Wandji W., Natarajan A., Dimitrov N.: Development and
design of a semi-floater substructure for multi-megawatt
wind turbines at 50+ m water depths. Ocean Engineering,
Volume 125, 1 October 2016, Pages 226-237.
10. Dymarski C., Dymarski P., ywicki J.: Design and strength
calculations of the tripod support structure for offshore
wind power plant. Polish Maritime Research 01/2015
20. Masciola M., Robertson A., Jonkman J., Driscoll F.:
Investigation of a FAST-OrcaFlex coupling module for
integrating turbine and mooring dynamics of offshore
f loating wind turbines, Proc. of the International
Conference on Offshore Wind Energy and Ocean Energy
(ICOWEOE), 2011.
21. Youhu Zhang Y., Husmann K., Andersen K., Tedesco G.:
Ultimate bearing capacity of laterally loaded piles in clay
Some practical considerations. Marine Structures, Volume
50, November 2016, Pages 260-275.

11. Roddier D., Cermelli C., Aubault A., Alla Weinstein

A.: WindFloat: a floating foundation for offshore wind
turbines. JRenew Sustain Energy, 2 (2010), p. 033104 CONTACT WITH THE AUTHOR

12. Adam F., Myland T., Schuldt B., Gromann J., Dahlhaus Czesaw Dymarski
F.: Evaluation of internal force superposition on a TLP for Pawe Dymarski
wind turbines. Renewable Energy, Volume 71, November Jdrzej ywicki
2014, Pages 271-275
e-mail: jedzywic@pg.gda.pl
13. Butterfield S., Musial W., Jonkman J., Sclavounos P.:
Engineering challenges for floating offshore wind turbines. Gdask University of Technology
Proc. offshore wind conference Copenhagen (2005). Faculty of Ocean Engineering and Ship Technology
11/12 Narutowicza St.
14. Bachynski E., Torgeir Moan T.: Design considerations for 80 - 233 Gdask
tension leg platform wind turbines. Marine Structures, Poland
Volume 29, Issue 1, December 2012, Pages 89-114.

15. Grelowska G., Kozaczka E., Kozaczka S., Szymczak W.:

Sea bottom structure investigation by means of acoustic
methods// POLISH JOURNAL OF ENVIRONMENTAL
STUDIES. -Vol. 19, nr. No. 4A (2010), s.35-38.

16. Kozaczka E., Grelowska G., Szymczak W., Kozaczka S.:

The Examination of the Upper Layers of the Seabed by
the Means of the Acoustic Methods// ACTA PHYSICA
POLONICA A. -Vol. 119, nr. No. 6A (2011), s.1091-1094.

17. Benassai G., Campanile A., PiscopoV., Scamardella A.:

Ultimate and accidental limit state design for mooring
systems of floating offshore wind turbines. Ocean
Engineering, Volume 92, 1 December 2014, Pages 64-74.

18. Johanning L., Smith G.: Station keeping study for WEC
devices including compliant chain, compliant hybrid
and taut arrangement, Proc. of the 27th International
Conference on Offshore Mechanics and Arctic Engineering
(OMAE), 2008

19. Fitzgerald J., Bergdahl L.: Considering mooring cables for

offshore wave energy converters, Proc. of the 7th European
Wave and Tidal Energy Conference (EWTEC), 2007.

66 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 67-73
10.1515/pomr-2017-0023

DESIGN METHODOLOGY FOR SMALL PASSENGER SHIPS ON THE

EXAMPLE OF THE FERRYBOAT MOTAWA 2 DRIVEN BY HYBRID
PROPULSION SYSTEM

Pawe Geesz1
Artur Karczewski
Janusz Kozak
Wojciech Litwin
ukasz Pitek 3
1
Faculty of Architecture and Design, Academy of Fine Arts in Gdask
2
Gdask Univerity of Technology, Poland
3
Warsaw University of Technology, Poland
ABSTRACT

Dynamic development in practically all fields of science and engineering has not passed over shipbuilding . In last years ,
engineers got to their use computer software which makes it possible to perform strength and hydrodynamic calculations
as well as to visualize design projects in 3 D space [1-4]. At their disposal they have full spectrum of modern solutions
associated with the use of advanced materials and technologies [5-7]. More and more attention is also paid to impact
onto the natural environment [8,9]. Every new object must influence the environment as low as possible, beginning from
building phase through its service life up to final utilization such approach is called green-shipping. However , not
only practical reasons are important . Clients , i.e. ship owners and passengers of ships paid more and more attention
to image of floating units. During decision taking on that from whom a transport service has to be ordered , the most
modern ships of an attractive image matching with place and time, are often taken into consideration.
Such situation has become a basis for an idea of working out a concept of a new ferryboat for National Maritime
Museum. As the ferry has to navigate in the heart of the town, then , apart from strictly marine and engineering
aspects , an important factor of its designing is its expected image - a set of significant meanings and emotions written
in architecture language. The new ferryboat , like its historical predecessors , will never leave urban water routes.

Keywords: small passenger ships, urban ship, design methodology, hybrid propulsion system

INTRODUCTION

Urban ship, like autobus or tramway, is a permanent

element of town space. For many inhabitants it is an object
of adoration. On the one hand, it is also a symptom of multi-
age tradition, attachment to maritime culture, on the other
hand - a symptom of modernity of the town, an effective
communication mean, and sometimes even a catalogue
visit card. Such role is served by Venetian vaporettas, Fig. 1-1. The excursion ship Hibiko approaching the quay at Odaiba island,
Parisian bateau mouche, rondvaartboten in Amsterdam, Tokio,photo by P.Gelesz
or Hotaluna i Himiko in Tokyo [Fig.1-1], which, apart
from strictly utilitarian functions, are one of very important This is exemplified by a ferry which operate on Motawa
components of cultural identity of the town. river between north cape of Granary Island (Wyspa
Spichrzw ), Oowianka quay and uraw (Crane)- just in the

POLISH MARITIME RESEARCH, No S1/2017 67

heart of Gdask. Though the currently used ship Motawa
maintains its full serviceability, its 20-year service resulted
in its technical and esthetic obsolescence. Its propulsion
system and functionality are far from today standards, and
its image especially in the context of the marine located
in the direct neighbourhood of National Maritime Museum,
and yachts moored in it, as well as Maritime Culture Centre
may be taken rather not adequate.
Such situation has become a basis for an idea of working out a
concept of a new ferryboat for National Maritime Museum. As
the ferry has to navigate in the heart of the town, then, apart
from strictly marine and engineering aspects, an important
factor of its designing is its expected image - a set of significant
meanings and emotions written in architecture language. The
new ferryboat, like its historical predecessors, will never leave
urban water routes.
The design project has assumed that an entirely new concept
of the ferry has to be worked out possibly without copying
the solutions used in the existing ferryboat Motawa under
operation for a few dozen years. Because, as said Prof.A.Rylke,
one of the fathers of post-war shipbuilding in Poland :
Contemporary ship, both in its structure and technical or
economic functioning, is a very complex object . It must satisfy
a large number of various requirements which are contradictory
to each other, already due to their assumptions. To compromize
the requirements and next to indicate a way toward their most
proper realization constitutes the subject of one of the most
crucial disciplines of shipbuilding taken in a modern way. Such
discipline is theory of ship design... Its task is to create an
ordered world from chaos of issues and various requirements,
whose rational, harmonious and mutual connection results in
this what constitutes a ship of today. [10]
The so formulated assumption of design methodology has
made it necessary to assume a non-standard methods of the
searching for and working out of ship hull parameters , based
on the packet of initial parameters external limitations,
hard design assumptions (rules and regulations) and soft
design postulates ( whishes and visions of designers and
to-be-operator ).
Within the group of external limitations, the dominating
requirements resulted from the character of sailing waters
area and a mode of the ferryboat operation. Size of mooring
places and a way of positioning the ship against water lane axis
during mooring operation, forced to limit ships hull length and
breadth to L=12,00 m and B=5,00 m, respectively. Similarly,
data on water depth distribution over the assumed service route
of the ferry made it possible, in this design stage, to assume
its full draught Tmax=1,30m.
The precision improving of expected ferryboats parameters
made it possible to start working on the underwater hull
form design. It was necessary to find compromise between
contradictory parameters.
The ship should have a required displacement at fulfilled
requirements for transverse and longitudinal stability,
unsinkability, and, in view of its propulsion it should have
as far fine form of underwater hull as to show possibly low
resistance to motion. For searching for the best solution
68 POLISH MARITIME RESEARCH, No S1/2017
fulfilling the given criteria, some number of hull underwater
form models was preliminarily selected and multi-criteria
poly-optimization was performed to find a solution close to
the nearest optimum.
ISSUES OF MULTI-CRITERIA
OPTIMIZATION
Optimization models make it possible to organize complex
systems of mathematical equations and inequalities which
represent target functions and constraints, in order to make
an optimum decision selected out of a large number of
possible ones. Optimum decision is that which satisfies the
target given by decision maker, however within the field of
various limiting conditions.
Optimization may be divided into:
one-criterion (one-dimensional) optimization, when
choice is made on the basis of one representative
assessment criterion,
multi-criteria optimization (multi-dimensional, poly-
optimization), when choice is made on the basis of more
than one assessment criterion.
One-criterion issue is characteristic of a strict structure,
i.e. it is solved by using a model in which both target function
and constraints are determined. The issue is solved by means
of optimization tools, and the itself process of searching for
an optimum solution consists in solving the set of appropriate
equations.
Multi-criteria issue may have, but not necessarily, a strict
structure. In the case of a multi-criteria programming model,
the multi-dimensional issue has strict structure, however in
case of specific ones (choice, classification, ranking) there is
a lack of a strict structure. In case of strict structure this is
multi-criteria optimization, otherwise the issue of decision
aiding.
In case of application of a large number of assessment
criteria for the best solution, contradiction between the
criteria often takes place. It means that the searched solution
does not extremize all the criteria considered separately, but
constitutes a kind of compromise between them. The multi-
criteria optimization issue first of all consists in defining the
compromise. In many cases it is possible to formulate, based
on heuristic knowledge on the process under optimization,
another, substitute criterion in respect of which a compromise
solution is searched for.
The multi-criteria optimization issue may be formulated
as follows [11].
Let :
X = {xl}, l = 1, 2,...., N be a vector of decision variables
taken as independent ones,
F = {fi}, i = 1, 2,....,M be a set of criteria (functions) in
respect of which solutions are assessed during searching
for a compromise.
Moreover, let be given the constraints put upon solution
values:
inequalities:
 (1)
- equations:
(2)
Multi-criteria optimization is aimed at finding the solution
for which the following condition will be satisfied :
(3)
If maximization of a certain function f* is required then
an auxiliary criterion may be introduced according to the
relation :
(4)
Historically, attempts to find an optimum solution were
made by Leibniz and Euler; solutions belonging to I. Newton,
J. Bernoulli and D. Bernoulli should be also mentioned.
Mathematical background for optimization was formulated
by Lagrange and Hamilton, and finally, Pareto, economist
of Italian French origin, formulated the principle of multi-
criteria optimization for economic issues, which was later
called the optimization in the sense of Pareto, if to find a
better solution against at least one criterion is not possible
without its worsening in respect of the remaining ones.
The principle is graphically presented in Fig. 2-1.
Fig. 2-1. Definition of optimum in the sense of Pareto
The solutions situated within the area indicated with red
line, belong to optimum solutions in the sense of Pareto. Each
of them is characteristic of that any attempt to bettering one
of the criteria results in worsening the other.
Among multi-criteria optimization methods can be
numbered a. o. the weighing criteria method, hierarchy
optimization method, and also method of evolutionary
algorithms [12-15].
The method of weighing criteria consists in the reducing
of multi-criteria optimization to single-criterion one by
introducing a substitute criterion based on a weighing sum
of criteria:
(5)
where:
and
Its solution can be graphically presented as a point of
intersection of area of acceptable solutions and the hyper
straight line P dependent on value of the criteria significance
coefficient wq (Fig. 2-2).
Fig. 2-2. Graphical presentation of optimum according to the method of
weighing criteria
In order to balance impact of particular criteria, their
normalization may be introduced. A problem in the method
is a-priori choice of values of weights of criteria, that may
obviously lead to different solutions.
In the analyzed issue the set of criteria (functions) F ={fi},
i = 1, 2,....,M against which a compromise was searched
for between a few contradictory parameters, resulted from
the requirements for transverse and longitudinal stability,
unsinkability and simultaneous need to minimize resistance
criterion.
DESIGN PROCESS
When preliminary design assumptions are made, a CAD
software which allows to make spatial visualizations and
parametric form of the design, was used. The applied software
made it possible to carry out graphical programming without
any need of knowledge of programming languages, due to
having projects in a parametric form, and to obtain new free
shapes by building connections between various geometrical
objects, their transformations and functions.
During initial phase of the design process 8 different
hull geometry forms were generated (Fig. 3-1). All of them
were longitudinally and transversely symmetrical and free
of any appendages at this design stage. They were based on
traditional hull forms including : flat-bottomed, spheroidal,
keel - fitted, and cruiser stern- ended. During searching
POLISH MARITIME RESEARCH, No S1/2017 69
 Fig. 3-1. Considered variants of underwater part of designed ferryboat
for an optimum solution, stability criteria were tested by
means of CAD software. Resistance qualities calculated by
using the parametric method, were analyzed with the use of
a spreadsheet program.
Results obtained in this stage made it possible to select the
hull form somewhere between cubical and spheroidal one,
intended for further optimization. In the final phase of the
design process a genetic algorithm contained in a software for
searching for optimum in the sense of Pareto, was used [16].
The hull No. 07 which showed optimum features for the
assumed criteria, was selected for further actions. Since the
moment, the design work was running along many paths:
design of hull,
design of propulsion system,
design of ship body, deck arrangement, wheelhouse
position and system.
PROPULSION AND SUPPLY OF THE
FERRYBOAT
Progress in electric drive technology and supply systems
more and more influences solutions used in shipbuilding.
Fig. 4-1. Schematic diagram of the propulsion and supply system of the ferryboat Motawa 2
70 POLISH MARITIME RESEARCH, No S1/2017
Dynamically developing hydraulic drive is also widely
applied, though less often, to main propulsion systems [17-19].
The hybrid parallel drive based on two motors high-
pressure engine and electric motor as well as the hybrid
serial drive based on one electric motor, are more and more
often installed on inland floating units . Main reasons
for growing popularity of such solution are : propulsion
flexibility, easiness and speed of performing manoeuvres
and negative environmental impact reduced to a minimum
[20-22]. It results from the fact that fully electrical propulsion
does not emit exhaust gases and is practically noiseless. The
largest pollution burden takes place during production and
utilization of motor, especially electric batteries containing
lithium.
In view of specificity of water area on which the ferryboat
is intended to operate,especially its small size and high
traffic rate, it was decided that the ferry should be of high
maneouvrability qualities. The more so, because the ferry
which operates in a shuttle mode, does not navigate straight
ahead between quays but every time it makes slight turn by
900. To satisfy owners expectations it was decided that the
ferry in question will be fitted with electric propulsion based
on pod propellers installed at bow and stern. The selected
 Fig. 5-1. Visualization of the designed ferryboat
propellers were azimuthal ones driven by electric vertical-
shaft motors which drive in turn, through reduction gear,
the dedicated propellers. Rotation of the propellers around
their axis is executed electrically. Paying attention to possible
simplification of construction, in preliminary design stage
one selected the propellers with the air-cooled engine of 15kW
output. As the supply system was based on lithium batteries
charged from electric network ( grid), and an independent
block of photovoltaic panels, the propulsion system was
qualified to be hybrid serial one. Based on experience
gained from other designs it was stated that especially good
maneouvrability qualities are provided by towing propulsion.
Therefore, in the preliminary design stage it was concluded
that in an ideal case, during bringing up to speed, only
one of the propellers always that towing will operate.
The other will be used during stopping braking just before
mooring the ferry. During move in opposite direction the
roles of propellers are reversed. In case when a situation forces
helmsman to make sudden manoeuvres, the two propellers
are always instantly ready to work in order to execute various
maneouvres, e.g. side drifting, which is non-available for
ships propelled conventionally.
Having in mind that a high degree of navigation safety is
expected from the side of passenger ship, one decided to split
the supply system into two autonomous circuits. Schematic
diagram of the propulsion and supply system is presented
below (Fig 4-1).
A problem unsolved in preliminary design stage was the
preparation of a reliable energy balance. As results from
measurements taken on the currently used floating unit, it
usually moves with the speed below 5 km/h and the whole
trip lasts only about 2 minutes.
In order to verify experimentally value of power demand,
a hull model was built in 1:10 scale to perform tests aimed at
power demand identification and maneouvrability verification
of the ferryboat.
The executed model tests made it possible to confirm the
suppositions that two azimuthal propellers of 15 kW output
each can be used to propel the ferryboat in question. The
lithium batteries installed on the ferryboat are of sufficient
capacity to supply the ships propulsion system during
twenty-four-hour intensive operation. Owing to that, the
necessary charging process can be carried out only once a
day in the night. This is of a decisive impact on the batteries
life, which is an important issue for ship owner as it influences
profitability of the proposed solution.
It may be added that in view of the specific area of
navigation over Motawa and Dead Vistula (Martwa Wisa)
river, and operation not farther than 100 m from river side,
the classification society did not required to install a reserve
electric generating set. In case of the necessity to make a
longer trip, e.g. for repair, such emergency electric generating
set may be temporarily placed on the deck. One decided to
install battery charger under the deck. It was possible to do
because, in order to obtain possibly long life of the batteries,
only a slow (10 h - lasting) small - current charging process
was assumed. Therefore, the chargers have small dimensions
and their connection to a power network is made through
typical three-phase contactors. Moreover, one more practical
advantage results from such solution, namely, the low power
absorption from the network does not lead to the necessity
of building a high -power terminal.
The ship owner was proposed to cover the roof slope of
one of the Museums building by photovoltaic panels with
total output of over 20 kW. It could reduce cost of power
supply to the ship.
POLISH MARITIME RESEARCH, No S1/2017 71
 CONCLUSIONS
Design process of small, untypical, non-serial floating
unit requires to apply a modern approach based on
optimization algorithms.
Parametric design of small ship hull with application
of optimization algorithms seems to be a promising
approach.
Advantages resulting from possibility of projects
modification in real time make it possible to carry out
multi-path work by a few design teams simultaneously.
ACKNOLEDGEMENTS
The design project has been worked out in the frame of
the grant : Project of ecological ferryboat for operation
over water area of the town of Gdask, financed- in the
interval from June 2015 to August 2016 - from the resources
of the Provincial Fund for Environmental Protection and
Water Economy in Gdask (Wojewdzki Fundusz Ochrony
rodowiska i Gospodarki Wodnej w Gdasku ) (http://www.
wfosigw-gda.pl).
BIBLIOGRAPHY
1. Sabik A., Kreja I.: Large thermo-elastic displacement and
stability FEM analysis of multilayered plates and shells.
Thin-Walled Structures. Vol. 71, (2013), pp.119-133.
2. Iwicki P., Tejchman A., Chrcielewski J.: Dynamic FE
simulations of buckling process in thin-walled cylindrical
metal silos. Thin-Walled Structures, Vol. 84 (2014), pp.
344-359.
3. Kozak J., Tareko W.: Case study of masts damage of the sail
training vessel POGORIA. Engineering Failure Analysis,
Vol. 18, No. 3 (2011),pp. 819-827
4. Kaliski K.: The finite element method application to linear
closed loop steady system vibration analysis. International
Journal of Mechanical Sciences 39, 3 (1997), pp. 315330.
5. M. Siemitkowski, W. Przybylski: Simulation studies of
process flow with in-line part inspection in machining cells.
Journal of Materials Processing Technology Vol.171, No.
1, ( 2006), pp. 27-34.
6. A. Zieliski, H. Smoleska, W. Serbiski, W. Koczewicz,
A. Klimpel: Characterization of the Co-base layers obtained
by laser cladding technique. Journal of Materials Processing
Technology,Vol. 164165, ( 2005), pp. 958963.
7. M. Deja, M. S. Siemitkowski: Feature-based generation of
machining process plans for optimised parts manufacture.
Journal of Intelligent Manufacturing Vol. 24, (2013),No.
4, pp. 831846.
72 POLISH MARITIME RESEARCH, No S1/2017
8. Kropiwnicki J., Kneba Z.: Carbon dioxide potential
reduction using Start-Stop system in a car. Key Engineering
Materials, Vol. 597 (2014), pp. 185-192.
9. Kropiwnicki J., Kneba Z., Zikowski M.: Test for assessing
the energy efficiency of vehicles with internal combustion
engines. International Journal of Automotive Technology,
Vol. 14, (2013), No. 3, pp. 479-487.
10. Rylke A.: Theory of ship design ( in Polish), L.M.Nogid ? .
Wyd. Morskie Gdynia 1962.
11. Eschenauer H., Koski J., Osyczka A.: Multicriteria Design
Optimization. Springer-Verlag, Berlin 1990.
12. Osiski Z., Wrbel J.: Theory of structures ( in Polish).
PWN, Warsaw 1995.
13. Peschel M., Riedel C.: Poly-optimization: methods of
compromise decision making in engineering problems (in
Polish). WNT, Warsaw, 1979.
14. Seidler J., Badach A., Molisz W.: Methods for solving
optimization tasks ( in Polish) . WNT, Warsaw 1980.
15. Stadnicki J.: Theory and practice of solving optimization
tasks with examples of technical applications ( in Polish).
WNT, Warszawa 2006.
16. Karczewski A., Pitek .: Reducing the Environmental
Impact of the Public Water Transportation Systems by
Parametric Design and Optimization of Vessels Hulls.
Study of the Gdasks Electric Passenger Ferry (2015-2016),
Architecture for the Society of Knowledge, Vol. 1, Faculty
of Architecture, Warsaw University of Technology, Warsaw
2016, pp. 314-321.
17. liwinski P. : Flow of liquid in flat gaps of the satellite motor
working mechanism. Polish Maritime Research, Vol. 21,
No. 2 (82) (2014), pp. 50-57.
18. liwinski P.: The basics of design and experimental tests of
the commutation unit of a hydraulic satellite motor. Archives
of Civil and Mechanical Engineering, vol. 16, (2016), No.
4, pp. 634-644.
19. Mikielewicz J., Mikielewicz D. : A simple model of circular
hydraulic pump. International Journal of Heat and Mass
Transfer, Vol. 52(1), (2008) pp.17-21,
20. B. Zahedi, L.E. Norum, K.B. Ludvigsen: Optimized
efficiency of all-electric ships by dc hybrid power systems.
J. Power Sources. 255 (2014) pp. 341354.
21. V. Lady, N.- German, L.N. Gas, V. Lady, ? Fuel cells
Norwegian hybrid ship aims to cut CO2-emissions by 50
percent . 02/13/2011, (2011) pp. 24.
22. M. Lapuerta, O. Armas, J. Rodrguez-Fernndez: Effect CONTACT WITH THE AUTHOR
of biodiesel fuels on diesel engine emissions. Prog. Energy
Combust. Sci. 34 (2008) pp. 198223. doi:10.1016/j. Pawe Geesz1
pecs.2007.07.001 Artur Karczewski
Janusz Kozak
Wojciech Litwin
ukasz Pitek 3

1
Gdask Academy of Fine Arts
6 Targ Wglowy St.
80-836 Gdask

2
Gdask University of Technology, Faculty of Ocean
Engineering and Shipbuilding
11/12 Narutowicza St.
80 - 233 Gdask

3
Warsaw University of Technology, Faculty of
Architecture
Pl. Politechniki 1
00-661 Warsaw
Poland

POLISH MARITIME RESEARCH, No S1/2017 73

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 74-81
10.1515/pomr-2017-0024

TOTAL SHIP OPERABILITY REVIEW, CONCEPT AND CRITERIA

M. Hossein Ghaemi1
Henryk Olszewski2
1
Gdansk University of Technology, Poland
2
The State University of Applied Sciences in Elblg, Poland

ABSTRACT

The ship operability can be understood and analysed in different manners, and its quality is described using different
indexes by different authors. Practically, there is no general and widely accepted description of total ship operability
in the literature, nor in the rules of classification societies, which would include both seakeeping and manoeuvrability
characteristics of a ship, and simultaneously take into account all ship subsystems and, what is most important,
comfort and safety of people on board. The aim of this paper is to propose a general definition of total ship operability
and name, adjust, describe and justify criteria which should be considered in the ship operability analysis, as well
as to provide a relevant algorithm paving the road for further investigation on total ship operability determination.

Keywords: ship operability, operability limitation criteria, safe operation level

INTRODUCTION The term operability is used in computer science,

At the beginning of the 21st century we observe intensive
technology development in many areas, including ocean
engineering (for example see Burdziakowski 2015, and
Janowski 2014). It has been a usual practice now to perform
complicated numerical analyses using different FEM or CFD
techniques (for example see Sabik 2013). Advanced materials
with high performance level are now used in practice. Safety
plays a fundamental role in design and decision making
processes. All of these advantages make it possible to solve the
ship operability problem not only for a particular subsystem
or ship hull, but also for the entire ship system. This refers
to total ship operability.
medicine, and engineering to describe the ability to keep
a whole system, installation, equipment or a set of facilities
in a safe and reliable functioning condition, according to
pre-defined operational requirements. Therefore, in the case
of a ship as a whole, with all subsystems and equipment, its
operability should take into account four basic elements:
1. The ship type, purpose and structure (hull and ship
body), its subsystems, equipment, installations, devices,
extra modules, people on board, and cargo, as well as all
interactions between them for a considered type of vessel,
2. Ship environment, external conditions, and ship-
environment interaction for a specific route,

74 POLISH MARITIME RESEARCH, No S1/2017

3. Requirements for safe and reliable functioning of all items
named under point 1 which will respect all regulations,
rules and owner demands,
4. A set of criteria which should be satisfied to respect all
constraints specified under point 3.
Till now, there is no general and widely accepted
description of total ship operability in the literature, nor in
the rules of classification societies, which would include both
seakeeping and manoeuvrability characteristics of the ship,
and simultaneously take into account all ship subsystems
and, what is most important, comfort and safety of people
on board.
The aim of the paper is to deliver a general definition of
total ship operability and name, adjust, describe and justify
criteria which should be considered in the ship operability
analysis, as well as to provide a relevant algorithm.
STATE OF THE ART
Various models and a wide range of criteria for analysing
the ship operability can be found in the literature. However,
there is no method which would simultaneously include
all elements mentioned in Introduction and, what is more
important, refer to the ship as a whole. There is no general
and widely accepted description of total ship operability in
the literature, nor in the rules of classification societies. An
overview of the state of the art in this area is given below.
The conventional method to determine ship operability
bases on a model of ship responses to a variety of sea states
having the form of wave statistics tables (Naito et al., 2006).
The fundamental operability method, which utilizes the
scatter diagram, has been proposed by Nordenstrom (1973).
This method is often referred to as the scatter analysis and
determines thresholds in the wave scatter diagram.
To determine the thresholds, ship responses are to be
calculated and then prepared in the form of most probable
maximum value. This method has been proposed by Ochi
(1981). The scatter analysis neglects chronology and duration
of the accounting operations, and the influence of weather
changes. To include these factors, Dallinga (2004) and Grin
and Van De Voorde (2004) have proposed the scenario
simulation technique.
De Wilde (2009) has proposed the persistency analysis
method based on a large environment statistics dataset. This
method takes into account the influence of changing weather.
Feikens (2011) has proposed to adapt the persistency analysis
to an offshore operation with multiple operational phases.
This method makes it possible to determine the influence of
weather variation on the assessed operations.
Vidal et al. (2015) have developed a parametric model
to assess the operability of Offshore Support Vessels (OSV)
via configuration-based design, considering the availability
and capacity of the ship and its installations. The method
generates a ranking of different designs with respect to
the operability level and capital cost. The developed model
consists of a series of parametric equations, solved via a
configuration-based approach for specific modules including
cranes, extra accommodation, and larger propulsion. The
analysis takes into account different operating areas and
different scenarios. In this study the operability is determined
based on the percentage of time during which the ship and
its installations are available for operation, so that 100%
operability means the capability of 365 days of ship sailing
with faultless operation of its equipment, subsystems and
installations, see Fig. 1. That study rationally addresses the
following options: extra accommodation, extra bollard pull,
extra power, extra crane, dynamic positioning (DP1, DP2
or DP3), ice class; x-bow; extra moon pool, extra ROV, and
helipad, which should be additionally considered for an OSV
and are directly related to the operability of the vessel.
Figure 1. Relations between the ship and its operability, mission and route or
place (Vidal, 2015).
Rusu and Bernardino (2009) have applied an assessment
method to determine the operability index of a container ship
in the Black Sea. In that study the operability is considered
as the percentage of time during which the considered vessel
is operational, depending on: (1) wave characteristics of the
sailing area, (2) dynamic ship response to the waves, and (3)
the ship mission. The relation between the ship operability
and its mission was determined based on the seakeeping
criteria. The operability index is understood in that study as
degradation of ship ability to carry out its mission in calm
water conditions.
Beside wave characteristics, Mudronja et al. (2014) have
added some ship manoeuvrability characteristics to the
operability analysis and presented the numerical results for
a 9200 TEU container ship in extreme seas. They showed
the influence of manoeuvring on wave loads, which then
affects the ship operability. They suggested improving the
approach to manoeuvring calculations by connecting them
to the ship operability and wave load calculations in future
studies. The operability was defined by them as the sustainable
level of course and speed of navigation for each sea state
shown by polar plots, or appropriate manoeuvres for bow
heading navigation in different sea states illustrated by the
so-called operability diagram, which splits the manoeuvring
conditions into 4 groups of commands: no change, speed
reduction, route change, and no solution.
POLISH MARITIME RESEARCH, No S1/2017 75

Figure 2. Sample percentages of limiting values for two different sea states (top) and the
operability diagram (bottom), Mudronja et al. (2014).
The sample operability diagram is shown in Fig. 2.
The results of the study by Pipchenko and Zhukov (2011),
which concerned ship control optimization in heavy weather
conditions, show how the criteria for general ship operability
limits influence the ship control strategy. The study takes
into account risk and seaworthiness assessments and then
applies the fuzzy logic multi-criteria assessment system to
evaluate the decision by calculating the seakeeping efficiency.
It is noteworthy that the problem of ship control should be
understood as total ship control, taking into account collision
avoidance manoeuvres and collision risk (Szapczyski 2015).
Tezdogan et al. (2014) have assessed the operability of
high speed passenger ships based on human comfort criteria.
Understanding the operability in a similar way, Samson and
Parsons (2002) have proposed its definition based on general
motion limits related to biomechanical and physiological
effects, which, when exceeded, lead to the degradation of
human comfort.
Hutchison (1981) has proposed to analyse the ship
operability in two manners: first as a multiple-criteria
problem leading to the generation of the operability diagram
in which the operability domain is described in terms of
significant wave height and mean period (Hs and Tp in Fig.
3, respectively). Here, the operability is determined based
on the so-called downtime during which the ship is out
of operation, with the seakeeping characteristics being the
main decision criteria. The second manner concerns projects
in which one critical activity occurs repeatedly in time, and
the results include the so-called delayed time of each ship
system function resulting from the inoperability of that
system or the ship as a whole.
Figure 3. Operability domain and illustration of downtime regions, (Hutchison
76 POLISH MARITIME RESEARCH, No S1/2017
Zeraatgar et al (2012) have presented a study in
which a naval ship was considered as a complete
system. They proposed a method to establish which
subsystem of the ship fails to perform its function
in harsh sea conditions. They showed how a ship
designer can predict which subsystem is more
sensitive to wave height and can become inoperable
when encountering the sea waves.
The method applied by Mudronja et al (2015)
concludes the above presented methods, delivering,
as a result, operability guidelines for product tankers.
It makes use of the seakeeping characteristics in
relation to the following operability criteria: propeller
emergence, deck wetness, and bow acceleration of a
product tanker, to determine the sustainable speed.
The authors pay attention to the influence of ship
operability on ship manoeuvring characteristics in
heavy seas, for instance on necessary route and/or
speed changes.
An issue which is worth addressing here refers
to the database for wave and wind characteristics
in a given water region. Traditionally, the data are
taken from wave atlases, which are mainly based
on observations rather than measurements. Wing
and Johnson (2011) have worked out a method to
determine wave characteristics from the long term hourly
historical wave buoy data, which seem to be more accurate and
useful for further embedding into the operability calculation
algorithms.
As far as ship operability calculations are concerned,
an advanced tool has been delivered by the Department of
Marine Technology, Norwegian University of Science and
Technology. The basic idea behind ShipX is to make a platform
that integrates all kinds of hydrodynamic analysis into an
integrated design tool in which the ship operability analysis
is available as part of the software.
CONCEPT AND CRITERIA
The operability limits are usually introduced based on one
or a number of the following criteria or standards:
1981)
 i. Motion Induced Interruptions (MII), Baitis (1995), 9. Propeller emergence
ii. Motion Sickness Incidence (MSI), OHanlon and Mc
Cauly (1974) Sample limiting criteria are given in Table 1. There are also
iii. The North Atlantic Treaty Organization Standard some experts recommendations for each limiting criterion.
Agreement 4154 (NATO STANAG 4154), For instance, Pipchenko (2011) reports that according to the
iv. U.S. Navy and U.S. Coast Guard Cutter Certification inquiry of 100 captains and chief mates trained in TCCS
Plan, ONMA, the limiting criteria may be formulated in the way
v. NORDFORSK (1987) presented in Table 2. Their illustration is shown in Fig. 4.
vi. ISO 2361/3-1985 (in relation to vertical acceleration) The total ship operability analysis should deliver the level for
each abovementioned index or limitation criterion and each
The ship operability limitation criteria usually refer to 9 state of the sea. Then, the multi-criteria decision algorithm
items: is to be used to establish the probability of operability loss
1. Slam in the context of these limitations and indexes regarding the
2. Deck wetness ship mission for possible modes of operation. Additionally,
3. Accelerations (vertical and lateral) the interactions between the people on board, different ship
4. Roll subsystems, the vessel, the facilities and equipment, the cargo,
5. Pitch and external loads should be clearly defined. Depending on
6. Heave the type of connections between different subsystems, the
7. Displacement (particularly, vertical displacement) resultant operability indexes may be the maximum values of
8. Local relative motion, which is obtained by subtracting indexes (for parallel configuration) or the minimum values
the wave depression from the absolute motion and is (for serial configuration). The subsystems, equipment and
positive for the increasing immersion installations should be divided into two main categories, the
Table 1. General operability limiting criteria for ships (Stevens, 2002, Pipchenko, 2011)

NORDFORSK 1987
Ref. Cruikshank &
NATO STANAG U.S. Coast Tasaki et al.
4154 Guard Cutter Landsberg
Certification Plan (Japan)
Naval Fast small (USA)
Criterion Merchant ships
vessels craft
0.275g (L100 m)
or 0.275g 0.65g
0.05g (L330m)
Vertical acceleration at 0.20g for light manual work
0.2g RMS 0.4g SSA 0.80g @ P=0.001 0.25g
forward perpendicular 0.15g for heavy manual work
0.10g for intellectual work
0.05g for transit passengers
0.02g for cruise liner
Vertical acceleration at 0.1g RMS 0.2g SSA 0.15g 0.2g 0.275g 0.20g
bridge
0.12g 0.1g 0.1g
0.10g for light manual work
Lateral acceleration at 0.10g RMS 0.2g SSA 0.60g @ P=0.001
0.07g for heavy manual work
bridge 0.05g for intellectual work
0.04g for transit passengers
0.03g for cruise liner
Motion Sickness 20% of crew in 4 5% in a 30 minute
Incidence (MSI) hours exposure
Motion Induced 1 tip per minute 2.1 tip per minute
Interruption (MII)
6.0 O 4.0 O 4.0 O
6.0 o for light manual work

Roll amplitude 4.0 o for heavy manual work

4.0O RMS 8.0o SSA 25.0o @ P=0.001 3.0 o for intellectual work
15.0o
2.5o for transit passengers
2.0 o for cruise liner
Pitch amplitude 1.5o RMS 3.0O SSA
0.03 (L100 m)
Slamming (probability) 0.01 or 0.03 0.03 0.06
0.01 (L300 m)
Deck wetness
0.01 0.05 0.05 0.05 0.07
(probability)
Propeller Emergence
0.1 0.25
(probability)
* RMS: Root Mean Square, SSA: Significant Single Amplitude (2 times of RMS)

POLISH MARITIME RESEARCH, No S1/2017 77

first of which is related to those subsystems or equipment,
whose lack of operability means the lack of operability
of the entire ship, while the second group composes the
units whose lack of operability does not result in the loss
of operability of the whole ship system, although they can
provoke inoperability of some other subsystems or equipment.
The total operability assessment should provide the
following results:
1. The percentage of time during which the ship and its
installations are available for operation, or inversely the
time of degradation of ship ability to carry out its mission
in calm water conditions.
2. The interaction between the characteristics related to ship
manoeuvrability and the operability indexes or limitation
criteria, including the sustainable level of the course and
speed of navigation for each sea state.
3. Mutual impact of human comfort criteria and the
operability, based on general motion limits in relation
to biomechanical and physiological effects, which, when
exceeded, lead to the degradation of human comfort.
4. Indication which ship subsystem, equipment or
installation, and in which order of priority, may fail to
perform its function in harsh sea conditions.
a. Deck wetness (DW)
The deck wetness, DW, takes place when the relative motion
of any part of the hull exceeds the height of the freeboard and
may cause, for instance, damage of the equipment or cargo
loss. The probability of DW is defined as:
(1)
where FBi denotes the freeboard at the i-th section and m0
is the zero moment of th relative motion spectrum.
b. Vertical Acceleration (VA)
Like DW, the vertical acceleration, VA, may be dangerous
for the crew or people on board, or lead to the damage of
equipment, for instance in the case of Offshore Supply
Vessels equipped with a crane or gangway at bow. Here, the
Root Mean Square (RMS) of the mean acceleration value is
calculated and then multiplied by a constant depending on
the ship type.
c. Propeller Emergence (PE)
The propeller emergence, PE, occurs when the relative
motion of the ship aft part is faster than the height between
the propeller tip and the water line. The probability of PE is
defined as follows:

(2)

where Hi denotes the distance between the water line and

the propeller tip.

d. Response Amplitude Operators (RAOs)

Figure. 4. Safe operation level based on experts recommendations (Pipchenko,
2011)
For some subsystems or installations, the risk assessment can
be considered as a parallel procedure which delivers valuable
input data for determining the probability of operability loss.
However, the majority of indexes can be calculated directly.
Some examples are given below.
Table 2. Limiting criteria recommended by experts (Pipchenko, 2011).
The response amplitude operator, RAO, is often used
to denote the ratio between the response amplitude of any
analysed variable to the wave amplitude. Its advantage is
simple application, while the main disadvantage is the fact
that it is a transfer function for linear or linearized systems.
In general, the RAOs are the function of wave number (k),
ship length (L), ship width (B), Froude number (Fn), and
block coefficient (Cb). A semi-analytical method to calculate
the RAOs has been developed and described by Jensen et al.
(2004) using the closed-form expression.

Deviation
Roll motion Slamming Deck wetness Speed
from course
amplitude [Iph] [Iph] reduction [%]
[deg.]

Small (up to) 7 5 5 13 20

Not dangerous (up to) 14 11 10 24 38
Substantial (up to) 23 19 20 46 40
Dangerous (greater than) 26 23 23 58

IpH: Intensity per hour

78 POLISH MARITIME RESEARCH, No S1/2017

 e. Motion Sickness Incidence (MSI) It should be emphasised that the majority of the existing
The motion sickness incidence, MSI, can be calculated based operability assessment methods do not include interactions
on the McCauley and OHanlon equation that makes it possible between the ship and ship subsystems. A good example here
to determine the impact of ship motions on the percentage of is the influence of pendulum motion of the cargo suspended
people that would suffer from sea sickness: by a crane on the ship motion. Table 3 shows essential ship
subsystems for which these interactions may have place
(3) regarding different operability criteria.
To illustrate the problem of total ship operability, the
following example can be considered. The ship selected for
where the error function (erf) is: the analysis can be a Crane/Workboat Barge or similarly
an AHTS, which is equipped with large telescopic boom
(4) crane at bow for offshore operations. The crane and the ship
are treated as a single rigid body, while the crane barge is
and av is the mean value of vertical acceleration at a selected not a semisubmersible vessel. It can be supposed that the
point. stabilization of vessel oscillations is controlled only by using
The structure of the total operability assessment algorithm passive healing control tanks, and no additional method or
is illustrated in Fig. 5. stabilizer is taken into account. The rules which are applicable
here are those given in LRS Code Lifting Appliances Rules
(Jan. 2013), particularly for offshore cranes - Chapter 3,
Sections 2 and 3. The vessel is supposed to operate for offshore
purposes, mainly loading/unloading processes. The external
loads are those imposed by the sea state 3 and the vessel is
to be positioned dynamically. In this case there are at least
4 subsystems to be taken into account, which are: the load
(acontainer, for instance) and its pendulum motion, the crane,
the propulsion system, and the dynamic positioning system.
Additionally, the ship motion and the crew safety and comfort
should also be taken into account, see Fig. 6. Consequently, 7
items interact with each other and these interactions should
be clearly defined and modelled. The calculation results may
include: positions and motions of the load, the crane hook,
and the vessel (see Fig. 7), as well as forces and moments acting
on each. Then, the abovementioned 4 items of operability
Figure 5. Structure of the total operability assessment algorithm.
results can be determined. Such a study has been done by the
authors and the results will be presented in the next article.

Table 3 Interactions between the ship and ship subsystems regarding different criteria (mainly based on the ShipX User Guide).

Criteria with regard to

Lateral acceleration

Deck wetness
Acceleration

emergence
Propeller
Relative
Vertical

Vertical
velocity

Motion
Heave

Ship subsystem
Pitch
Slam

MSI
Roll

MII

Ship hull X X X
Propulsion system and X X
machinery
Ship equipment X X X X
Cargo X X X X X
Personnel effectiveness X X X X X X X
Passenger comfort X X X X X X X
Helicopter X X X X
Sonar X
Li f t i ng a nd load i ng / X X X X X X X
unloading operations
Anchoring and Mooring X X X X X
system
Dynamic positioning X X X X X X X X

POLISH MARITIME RESEARCH, No S1/2017 79

 Figure 6. Subsystems and installations considered for operability assessment.
Figure 7. The coordinate system and the load, crane and vessel motions.
REFERENCES
1. Baitis, A. E., Holcombe, F. D., Conwell, S. L., Crossland, P.,
Colwell, J., and Pattison, J. H.: Motion Induced Interruptions
(MII) and Motion Induced Fatigue (MIF) Experiments
at the Naval Biodynamics Laboratory, Technical Report
CRDKNSWC-HD-1423-01, Bethesda, MD: Naval Surface
Warfare Center, Carderock Division, 1995.
2. Burdziakowski P., Janowski A., Kholodkov A., Matysik K.,
Matysik M., Przyborski M., Szulwic J., Tysic P., Wojtowicz
A.: Maritime Laser Scanning as The Source for Spatial
Data. Polish Maritime Research, Vol. 22, No. 4(88) (2015),
pp.9-14., 2015.
80 POLISH MARITIME RESEARCH, No S1/2017
3. Dallinga R., Daalen E. V., Grin R., Willemstein T.: Scenario
Simulations in Design for Service. Proc. of 9th Symposium
of Practical Design of Ships and Other Floating Structures
(PRADS2004), Volume 2, pp 604611, 2004.
4. Faltinsen O. M., Sea Loads on Ships and Offshore Structures.
Cambridge University Press, ISBN 0521458706, 1998.
5. Fathi D., Hoff J. R., MARINTEK A/S, ShipX User Manual,
2004.
6. Grin R. and Van De Voorde E.: Weather-Related Economics
of Natural Gas Transport for Two Propulsion Plant
Configurations. RINA 2004 Conference: Design and
Operation of Gas Carriers. CD-ROM, 2004.
7. Hutchison B. L.: Risk and Operability Analysis in the
Marine Environment. SNAME Transactions, Vol. 89, pp.
127-154, 1981.
8. Janowski A., Nowak A., Przyborski M., Szulwic J.: Mobile
indicators in GIS and GPS positioning accuracy in cities.
2nd International Conference on Rough Sets and Emerging
Intelligent Systems Paradigms (RSEISP) held as part of Joint
Rough Set Symposium (JRS), 2014, Springer International
Publishing, ISBN:978-3-319-08729-0; 978-3-319-08728-3,
DOI: 10.1007/978-3-319-08729-0_31, 2014.
9. Jensen J., Mansour A. E., Olsen A. S.: Estimation of
Ship Motions Using Closed-form Expressions. Ocean
Engineering, 2004, 31, pp. 6185, 2004.
10. Wilde J. J. de, Van Dijk A. W., Van Den Berg J., Dekker
J., et al.: Direct Time Domain Downtime Assessment for
Long Operations Using Computer Cluster, The Nineteenth
 International Offshore and Polar Engineering Conference, Display for Visualising Collision Avoidance Manoeuvres
International Society of Offshore and Polar Engineers, in Various Visibility Conditions. Journal of Navigation,
2009. Vol. 68, No. 06 (2015), pp.1041-1055, 2015.

11. Feikens H., Verwey R., Serraris J. J., Huijsmans R.: Wet
handshake: Workability Study of an Offshore Thruster
Exchange Operation. ASME 30th International Conference
on Ocean, Offshore and Arctic Engineering, pp 609616,
American Society of Mechanical Engineers, 2011.
23. Tezdogan T., Incecik A.: Operability assessment of
high speed passenger ships based on human comfort
criteria. Ocean Engineering, Volume 89, 1 October
2014, Pages 3252, ISSN 0029-8018, DOI: 10.1016/j.
oceaneng.2014.07.009, 2014.

12. Mudronja L.: Numerical Evaluation of Operability
Envelope for Ultra Large Container Ship in Extreme Seas
and Influence of Maneuvering on Wave Loads, Presented in
the 21st Symposium on Theory and Practice of Shipbuilding
- SORTA 2014 (2 to 4 October 2014 in Baka, Island of Krk,
Croatia), 2014.
13. Mudronja L., Katalini M., Bonjak R., Vidan P.: Operability
Guidelines for Product Tanker In Heavy Weather in The
Adriatic Sea. Annual of Navigation 21 /2014, DOI: 10.1515/
aon-2015-0008, 2014.
14. Naito S., Minoura M., Hamanaka S., Yamamoto T.: Long-
Term Prediction Method Based on Ship Operation Criteria.
International Shipbuilding Progress, 53 (3):229252, 2006.
24. Vidal, H.L., Gaspar, H.M., Weihmann, L. et al.: Mar Syst
Ocean Technol (2015) 10: 47. doi:10.1007/s40868-015-0001-
8, 2015.
25. Wing D. A., Johnson M. C.: Ship Operability Predicted
from Long Term Directional Wave Records, Transactions
RINA, Vol 153, Part A2, International Journal Maritime
Engineering, DOI No: 10.3940/rina.ijme.2011.a2.wf5, Apr-
Jun 2011.
26. Zeraatgar, H.; Rostami, A. B.: An Investigation on Ship
Operability Versus Equipment Operability in Irregular
Waves, Brodogradnja, Brodarski Institut doo, pp 30-34,
2012.

15. Nordforsk, Assessment of Ship Performance in a Seaway:

The Nordic Co-operative Project: Seakeeping Performance
of Ships, ISBN 8798263714, 9788798263715, 1987. CONTACT WITH THE AUTHOR

16. Ochi M. K.: Principles of Extreme Value Statistics and Their M. Hossein Ghaemi
Application. In: Extreme Loads Response Symposium,
SNAME, Arlington, VA, pp 1530, 1981. Gdask University of Technology
Faculty of Ocean Engineering and Ship Technology
17. Pipchenko A. D., Zhukov D. S.: Ship Control Optimization 11/12 Narutowicza St.
in Heavy Weather Conditions. International Association 80 - 233 Gdask
of Maritime Universities AGA11, 2011. Poland

18. Rusu L., Bernardino M.: Estimation of The Operability

Index of A Containership Operating In The Black Sea. The
Annuals of University Dunrea De Jos of Galati Fascicle
VIII, 2009 (XV), ISSN 1221-4590 TRIBOLOGY, 2009.

19. Sabik A., Kreja I.: Large Thermo-elastic Displacement and

Stability FEM Analysis of Multilayered Plates and Shells.
THIN-WALLED STRUCTURES. -Vol. 71, (2013), pp.119-
133, 2013.

20. Stevens S. C., Parsons M. G.: Effects of Motion at Sea on

Crew Performance: A Survey. Marine Technology, Vol.
39, No. 1, pp. 2947, January 2002.

21. Szapczyski R.: Evolutionary Planning of Safe Ship Tracks

in Restricted Visibility. Journal of Navigation, Vol. 68, No.
01 (2015), pp 39-51, 2015.

22. Szapczyski R., Szlapczynska J.: A Target Information

POLISH MARITIME RESEARCH, No S1/2017 81

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 82-88
10.1515/pomr-2017-0025

EFFECT OF DIFFERENT STRUCTURE TYPE TRAFFIC ON RAILWAY

LINE CAPACITY

Radosaw Gleba
Sawomir Grulkowski
Jerzy Zariczny
Gdansk University of Technology, Poland

ABSTRACT

The article points to methods of analyzing railway traffic conditions based on two parameters: capacity and delay
of trains. The impact of the differentiated railway type structure on the capacity of the railway line was presented.
Particular attention has been paid to the assessment of commonly used simplifications in analyzes.

Keywords: railway line capacity, railway traffic conditions, train delays

INTRODUCTION the traffic flow required by the demand. This phenomenon is

Modernization activities at railway stations and railways
bring tangible benefits in terms of improving the quality and
competitiveness of the railways, which results in a significant
increase in traffic on major rail lines. Investments in railway
infrastructure should be of adequate durability (longevity) not
only for the construction of railway pavements, earthworks or
engineering facilities, but also to ensure adequate conditions
for long-distance rail traffic. There are many methods for
evaluating rail traffic conditions, the usefulness of which
depends on the purpose, the input data, the appropriate
computing tools, and the expectations of the accuracy of
the mapping of reality. Analytical, optimization, simulation,
and hybrid models, i.e. methods that combine both analytical
and simulation elements, are distinguished. The simulation
method, which takes into account the greatest number of
variables and assumptions, is the simulation method done
by the creation of computerized microsimulation model of
the analyzed railway [3, 4].
Most often, however, due to the complex analysis process,
the simplest analytical models are used, which often lead to
incorrect conclusions at the pre-design stage. In practice, in
many cases, this results in a mismatch of the railway line to
most noticeable in the large agglomerations, where demand
for passenger transport is increasing in demand for freight.
An example is the Tri-City agglomeration, where the
growing demand for railways has a growing phenomenon
of suburbanization and the increase in trans-shipments in
the Tri-City seaports, which are one of the largest railway
freight generators in Poland [2, 10, 14].
An additional factor that often misinterprets the results
obtained with analytical models is the large variation in
freight technology and the speed of freight trains [7, 8].
PARAMETERS AND FACTORS
DETERMINING RAIL TRAFFIC
CONDITIONS
One of the main characteristics of a railway line is its
capacity, the largest number of trains or pairs of trains that
can flow smoothly over a given line within a specified time
unit (typically day or peak time) [18]. However, the capacity
itself does not say much about the conditions and fluidity of
rail traffic, as it is not a synthetic indicator. Railway lines with
similar capacity values can in practice vary radically in terms

82 POLISH MARITIME RESEARCH, No S1/2017

of traffic flow. This is due to the large number of different
factors affecting the capacity value. Comparative analysis of
two railway lines or even two sections of the same railway line
is possible only after comparing the capacity value with the
current or predicted traffic volume. The synthesized indicator
thus obtained is referred to as the capacity utilization rate
and allows for an assessment of traffic conditions, including,
among others, the infrastructure features of the reserve in
case of traffic disruptions, the impact of different types of
traffic [9, 16, 20, 21, 23, 24, 29].
Capacity is defined primarily for routes, i.e. part of a
railway line between two adjacent watch posts - stations or
branch outlets. The value of the parameter is determined by
the critical route or the critical section (in the case of routes
equipped with self-locking line locks). This is a trail (stretch),
which is characterized by the longest time of attachment of
infrastructure by a passing train. A given time is referred to
as the traverse time interval of trains and depends on:
number and layout of tracks;
arrangement and configuration of track circuits;
the technical parameters of the line (in particular the
permissible driving speed for each type of train);
line geometry (values of horizontal radius and
longitudinal radius);
type of rail traffic control devices;
type structure of movement (movement diversity);
train speed spectrum (variation of permissible train
speed);
parameters of dynamic traction vehicles (traction
characteristics);
traffic organization (including train hierarchies and train
initialization: eg under green light or at green light);
driving techniques;
location of passenger stops on the trails [1, 5, 17, 18, 30].
LINE CAPACITY
The basic requirement for a complex analysis of traffic
conditions on a railway line is to determine the capacity of
individual routes. Under ideal conditions, where the structure
of movement on the analyzed line segment is homogeneous,
and trains run at the minimum tramline intervals, the line
capacity is calculated according to (1) [12, 18]. The given
physical situation is shown in Figure 1.
Fig. 1. Homogeneous movement of trains running at minimum intervals
(source: own study)
The motion organization shown in Figure 1 provides the
opportunity to launch the largest possible number of trains on
the trail, but this is an idealized situation because, due to the
lack of time between successive trains, it is highly susceptible
to the possible formation and spread of primary traffic
disruption, causing a high risk of secondary interference.
In addition, most of the railway lines have a variation in
the type structure of the trains running, so the load time of
the route is variable. This situation is most likely to occur
near large transport generators (e.g. harbor agglomerations).
Variety of traffic influences the increase in speed range with
which individual types of trains travel, and their movement
is not evenly distributed in successive time intervals (so
called windows). In fact, there are additional time reserves
between successive trains to enable neutralization of possible
interference (primary or secondary), thereby increasing
system reliability [22, 23, 26, 27, 28, 31]. Therefore, the value
of the track successive time interval is individual for each train
included in the timetable and consists of the time taken by
the particular train and the preceding reserve of time (2). The
heterogeneous structure of train traffic is shown in Figure 2.
(2)
where:
In,i track time sequence corresponding to train and
non-homogeneous motion [min];
tob,i unit time of track occupancy [min];
tb,i unit time reserve [min].

(1)

where:
Nmax maximum theoretical capacity [train/day or pair
of trains/day];
T analyzed time window (essentially 24h: T=1440 min)
[min];
In track time interval for trains [min].

Fig. 2. Non-homogeneous structure of train traffic (source: own elaboration)

POLISH MARITIME RESEARCH, No S1/2017 83

 The variety of train traffic combined with a large range of
speed affects the difficulty in determining the capacity of the
routes [18]. In the case of application of methods of analysis
of railway traffic conditions in which some parameters
describing railway transport system are standardized (e.g.
parameters related to track geometry - arc rays, values of
longitudinal gradients) and the relation between subsequent
trains is not taken into account (skipping of the timetable),
it is possible to calculate the theoretical capacity according
to the relation (3).
(3)
where:
Nt Theoretical capacity [train/day or pair of trains/day];
Fluidity coefficient of motion (for calculation is
assumed =0,20,3);
T Analyzed time window (essentially 24h: T=1440
min) [min];
tob,max Maximum load time of the trail in the analyzed
time window [min].
The approach assumes the possibility of the randomness of
train reports, characterized by the greatest load on the trail.
In the case, where trains with the highest infrastructure,
take time in succession, the value of tob,max is actually the
minimum train line interval. Where a track is being loaded by
faster trains, the difference between the peak load time of the
trail and the load of the trail over the faster train is treated as
a time reserve. In addition, the so-called motion fluctuation
factor , which reduces the duration of the analyzed time
window, provides a reserve for diagnostic and maintenance
activities and possible neutralization of interference.
In many situations, calculations based on dependence
(3) are unrepresentative. This particularly applies to routes
where there is a significant difference in the minimum and
maximum time of seizure of infrastructure. The method
does not take into account the share of the different types of
trains during the total time of the route during the 24-hour
period, which can lead to the situation where one type of train,
which has fixed time intervals, is a train that determines the
capacity values for the whole analysis period (essentially 24
hours). Hence, using a given way of analyzing railway traffic
conditions, it is important to be aware of the large number of
simplifications that can affect the reliability of the research
carried out.
These simplifications resulting from the application
of dependence (3) can be depicted on the example of
Kocierzyna- Gdynia railway section of line No 201 Nowa
Wie Wielka - Gdynia Port. Figure 3 shows the values of the
maximum unit time of seizure of particular routes in selected
time intervals. The value of this parameter is variable over
time, and the difference between maximum and minimum
can be significant. It is particularly noticeable in the case
of an even track of Gdask Osowa (Os) - Gdynia Gwna
(GO) line section. The difference in the value of tob,max at
84 POLISH MARITIME RESEARCH, No S1/2017
night in relevance to the remaining time periods, is due to
the fact that free freight trains (2 trains / d) are essentially
launched in the given time interval. Due to the intensity and
priority of passenger trains, there is no possibility of trains to
travel at other time intervals. Therefore, for this case it is not
appropriate to determine the capacity for the whole analysis
period based on the value of tob,max determined by trains
whose movement is not reliable for the whole section. In
addition, the type of train is not representative of the section
under consideration because of its low participation in the
quantitative structure of trains - passenger trains: 48 trains/d;
Freight trains: 2 trains / d [6].
Fig. 3. The maximum busy time of individual pathways analyzed section of
the railway line (markings: Kc - Kocierzyna, GK - Goubie Kaszubskie, So -
Somonino, GL - Glincz, ZW - ukowo Wschodnie, Os - Gdask Osowa, GWK
- Gdynia Wielki Kack, GO - Gdynia Gwna) (source: own elaboration)
As already mentioned, traffic on most railway lines is
characterized by significant heterogeneity and high variability
of train speeds, which translates into significant differences
between minimum and maximum occupation times. This
results in the need to take into account the appropriate
simplifications, using the mean values (4) [5, 18]. As a result,
the calculation methodology presented below is representative
of most railway lines in Poland (including the railway line
201 Nowa Wie Wielka - Gdynia Port in the Kocierzyna
- Gdynia section) as based on the average values taken
into consideration is the structure of generic trains, their
participation in the total time of route occupation during
the day and time intervals.
(4)
where:
Npr Practical capacity [train/day or pair of trains/day];
Coefficient of motion fluidity (for calculations it was
assumed =0,2);
T Analyzed time window (essentially 24h: T=1440 min)
[min];
tob,sr Average unit load time of the trail [min];
tb,opt,sr Acceptable average unit time interval between
train reports [min].
 The average unit load time of the trail in the analyzed time
window is calculated as the quotient of the total time of the
route load and the number of trains running:
(5)
where:
tob,sr Average unit load time of the trail [min];
tob,i Unit load time of the trail [min];
n Number of trains on the trail [-].
Determining the allowable average unit interval between
subsequent train reports requires analyzing the time reserves
in timelines. The first step is to isolate all of those intervals,
which are smaller than tob,sr (6). It is assumed that these are
intervals that cannot be used to trample subsequent trains,
and therefore add to the average unit load time of the route,
thus increasing the stability of the traffic graph and thus the
reliability of the rail transport system on the analyzed section.
(6)
where:
tb,i Interval before the train i [min];
tob,sr Average unit load time of the trail [min].
The values of time intervals satisfying the inequality (6)
are then summed and divided by the number of trains on the
path that meet the inequality (6). The value thus obtained is
defined as the permissible average unit time interval between
train reports:
(7)
where:
tb,opt,sr Acceptable average unit time interval between
train reports [min];
tb,i Interval before the train i smaller than tob,sr [min];
li The total number of trains on the trail that are
characterized by value [-].
The calculation methodology is well illustrated in Figure 4,
where the time reserve between the SKM96349 and SKM96351
trains cannot be used to tracer another train because it is
less than the average unit time of the track. Consequently,
the value of the interval time is taken into account when
determining the acceptable average unit interval time
between subsequent train reports and improves the stability
of the timetable. The intervals between the SKM96341 and
SKM96349 and SKM96351 and SKM96357 are higher than
the average unit time of the trains, so they can be used to
trample trains.
Fig. 4. Reserves of time between successive trains on track no. 1 of Os - GWK
route
(source: own elaboration)
Practical throughput capacity based on the above presented
methodology is the most representative basis for further
analysis for lines characterized by high train speed variability
and uneven traffic. Calculations based on the average load
time of the trail allow to include the traffic structure in the
calculation, because the greatest impact on the capacity
value is the trains that have the greatest total infrastructure
occupancy per day. The reliability of calculations is also
enhanced by taking reserve of the time that affects the stability
of the timetable. This is particularly important for punctuality
of trains. In 2014, of all freight trains operating in Poland, as
many as 58.6% were delayed, of which over 40% had delays
of more than 16 minutes. In the same year punctuality of
passenger trains was 91.1% [19].
TRAIN DELAYS
One of the most important indicators that best illustrates
the reliability of rail operations is the spread of delays. Train
delays can be divided into primary ones (caused by, for
example, rolling stock or infrastructure failure) and secondary
ones that have arisen as a result of primary delays. The risk
of secondary delays is all the greater, the smaller the value of
time intervals between subsequent trains. As a consequence,
the greatest risk of secondary delays occurs when trains run at
minimum track intervals (Figure 1). Determining the amount
of secondary delays, depending on the value of the original
delay of one of the trains in heterogeneous motion, requires
a detailed analysis in which it is necessary to determine what
kind of train follows its preceding, what is the time distance
between them, and where observation is done on the line
[11, 13]. The issue is highly complex, which in practice uses
simplifications, which, as in the case of capacity calculation,
are based on average values. The basis for the analysis is the
time reserve between train reports based on the average value:
(8)
where:
POLISH MARITIME RESEARCH, No S1/2017 85
 tb,sr Average unit time reserve between train reports
[min];
tb,i Unit time reserve [min];
n Number of trains on the trail [-].
Determining the average unit value of the time reserve
allows to specify the number of trains that will be subject to
a secondary delay resulting from the original delay of one
of the trains [13]:
(9)
where:
j Number of trains covered by the secondary delay
[-];
td,1,i Value of original delay time of one of the trains
[min];
tb,sr Average unit time interval between train reports
[min].
The next step is the sum of the delays [6]:
(10)
where:
td Sum of delays [min];
j Number of trains covered by the secondary delay
[-];
td,1,i Value of original delay time of one of the trains
[min];
tb,sr Average unit time interval between train reports
[min].
The final result of the analysis is a graph showing the
process of increasing the sum of delays depending on the
original delay value of one of the trains (Figure 5).
Fig. 5. The increase of the sum of delays depending on the original delay
(source: own elaboration)
On the basis of the analysis, it was found that the
calculations of the increase of delays made on the basis of
the mean unit time interval between the reports of subsequent
trains are characterized by considerable simplifications. This
86 POLISH MARITIME RESEARCH, No S1/2017
is due to significant unevenness in traffic and a large variation
between actual reserves of time. Therefore, it is advisable
to investigate the process of acceleration of delays only for
specific traffic situations on a critical route, where it is possible
to refer to actual rather than averaged time intervals.
SUMMARY
Two of the main parameters mentioned above: line capacity
and trains delay are just a starting point for analysis with more
parameters such as practical capacity, capacity utilization,
timetable stability and organizational reliability. Only a
comprehensive analytical approach, based on the individual
characterization of each train, allows for a fair assessment of
traffic conditions on lines of varying traffic pattern.
From the point of view of the management of the transport
process, where intramodality is a major challenge and a
contemporary duty, optimizing system reliability in terms
of technological and technical aspects becomes absolutely
desirable and above all else - possible [15, 19, 24, 25].
BIBLIOGRAPHY
1. Abril M., Barber F., Salido M.A., Tormos P., Lova A.: An
assessment of railway capacity, Transportation Research
Part E: Logistics and Transportation Review, Volume 44,
Issue 5, September 2008.
2. Burdziakowski P., Janowski A., Kholodkov A., Matysik K.,
Matysik M., Przyborski M., Szulwic J., Tysic P., Wojtowicz
A.: MARITIME LASER SCANNING AS THE SOURCE
FOR SPATIAL DATA// Polish Maritime Research. -Vol.
22, Iss. 4(88) (2015), pp.9-14.
3. Dymarski C., Dymarski P.: Computational simulation of
motion of a rescue module during its launching from ship
at rough sea// Polish Maritime Research. -Vol. 21, nr. 3(83)
(2014), s.54-60.
4. Dymarski P., Dymarski C.: A numerical model to simulate
the motion of a lifesaving module during its launching
from the ships stern ramp// Polish Maritime Research.
-Vol. 21, nr. 2(82) (2014), s.34-40.
5. Gapark J., Zitrick V.: Aspects of railway capacity and
occupation time estimation, Journal of Civil Engineering
and Architecture, Volume 8, No. 3, March 2014.
6. Gleba R., Grulkowski S., Szczepiski D., Zariczny J.:
Znaczenie analiz ruchowych w procesie optymalizowania
zakresu infrastrukturalnych inwestycji kolejowych,
Materiay Konferencyjne, Nowoczesne technologie i
systemy zarzdzania w transporcie szynowym. Cz II:
Transport szynowy, sterowanie ruchem kolejowym, SITK,
Krakw 2016, s. 63-82.
7. Gleba R., Grulkowski S., Zariczny J.: Problem zdolnoci
przepustowej linii kolejowych w obszarze aglomeracji
trjmiejskiej, Materiay Konferencyjne, Nowoczesne
technologie i systemy zarzdzania w transporcie szynowym,
SITK, Krakw 2014, s. 145-166.
8. Gleba R., Grulkowski S., Zariczny J.: Wielowariantowa
analiza wpywu modernizacji linii kolejowej nr 201
na dostpno kolejow gdyskiego portu, Materiay
Konferencyjne, Nowoczesne technologie i systemy
zarzdzania w transporcie szynowym. Cz , SITK,
Krakw 2014, s. 167-182.
9. Grzeczka A., Kaczyski M., Wittbrodt E., Uhl T., Kohut
P.: 2015. Motion analysis of a kitesurfer employing a vision-
based measurement system// Journal of Vibroengineering.
-Vol. 18, iss. 3 (2016), s.1884-1892
10. Janowski A., Nowak A., Przyborski M., Szulwic J.: Mobile
indicators in GIS and GPS positioning accuracy in cities.
2nd International Conference on Rough Sets and Emerging
Intelligent Systems Paradigms (RSEISP) held as part of Joint
Rough Set Symposium (JRS), 2014, Springer International
Publishing.
11. Krlicka A.: State equations in the mathematical model
of dynamic behaviour of multihull floating unit// Polish
Maritime Research. -Vol. 17, nr. iss. 1 (2010), s.33-38.
12. Lai Y.C., Barkan C.P.L.: An enhanced parametric railway
capacity evaluation tool (RCET), Proceedings of 88th
Transportation Research Board, Washington, DC, 2009.
13. Landex A.: Methods to estimate railway capacity and
passenger delays PhD thesis. Technical University of
Denmark, Department of Transport, November 2008.
14. Lindfeldt A.: Railway capacity analysis methods for
simulation and evaluation of timetables, delays and
infrastructure. Doctoral Thesis in Infrastructure. KTH
Royal Institute of Technology, School of Architecture and
the Built Environment, Department of Transport Science,
Stockholm 2015.
15. Lipiski K., Docquier N., Samin J., Fisette P.: mechanical
engineering education via projects in multibody dynamics//
COMPUTER APPLICATIONS IN ENGINEERING
EDUCATION. -Vol. 20, nr. iss. 3 (2012), s.529-539.
16. Mioduszewski P., : Inhomogeneity of low-noise wearing
courses evaluated by tire/road noise measurements using
the close-proximity method// APPLIED ACOUSTICS.
-Vol. 111, (2016), s.58-66.
17. Mioduszewski P., Ejsmont J., Grabowski J., Karpiski D.:
noise map validation by continuous noise monitoring//
APPLIED ACOUSTICS. -Vol. 72, nr. iss. 8 (2011), s.582-589
18. Nowosielski L.: Organizacja przewozw kolejowych. KOW,
Warszawa 1999.
19. Ocena funkcjonowania rynku transportu kolejowego i
stanu bezpieczestwa ruchu kolejowego w 2014 roku, Urzd
Transportu Kolejowego, Warszawa 22 wrzenia 2015.
20. Sandberg U., Mioduszewski P.: The importance of the
bottom layer in double-layer porous asphalt for noise
reduction// Journal of the Acoustical Society of America.
-Vol. 131, iss. 4 (2012), s.1-6.
21. Szapczyski R., Szlapczynska J.: A Target Information
Display for Visualising Collision Avoidance Manoeuvres
in Various Visibility Conditions// JOURNAL OF
NAVIGATION. -Vol. 68, iss. 06 (2015), s.1041-1055.
22. Szapczyski R., Szapczyska J.: Customized crossover in
evolutionary sets of safe ship trajectories// International
Journal of Applied Mathematics and Computer Science.
-Vol. 22, nr. No.4 (2012), s.999-1009.
23. Szapczyski R., Szapczyska J.: On evolutionary
computing in multi-ship trajectory planning, Applied
Intelligence// APPLIED INTELLIGENCE. -, (2011), s.1-20.
24. Szapczyski R.: Evolutionary approach to ships trajectory
planning within Traffic Separation Schemes// Polish
Maritime Research. -Vol. 19, nr. iss. 1(72) (2012), s.11-20.
25. Szapczyski R.: Evolutionary Planning of Safe Ship Tracks
in Restricted Visibility// JOURNAL OF NAVIGATION.
-Vol. 68, iss. 01 (2015), s.39-51.
26. Szapczyski R.: Evolutionary Sets of Safe Ship Trajectories
Within Traffic Separation Schemes// JOURNAL OF
NAVIGATION. -Vol. 66, nr. iss. 1 (2012), s.65-81.
27. Szapczyski R.: Evolutionary sets of safe ship trajectories
with speed reduction manoeuvres within traffic separation
schemes// Polish Maritime Research. -Vol. 21, iss. 1 (2014),
s.20-27.
28. Szapczyski R.: Evolutionary Sets Of Safe Ship Trajectories:
A New Approach To Collision Avoidance// JOURNAL OF
NAVIGATION. -Vol. 64, nr. no. 1 (2011), s.169-181.
29. Szymkiewicz A.: Modelaling water flow in unsaturated
porous media: Accounting for nonlinear permeability and
material heterogeneity. Heidelberg: Springer, 2013.
30. Weryk M., Kozaczka E., Grelowska G.: Study of Noise
Propagation for Small Vessels// Archives of Acoustics.
-Vol. 40, nr. 2 (2015), s.267-272.
POLISH MARITIME RESEARCH, No S1/2017 87
31. Woch J.: Narzdzia analizy efektywnoci i optymalizacji
sieci kolejowej. Wydawnictwo Politechniki OVNLHM,
Gliwice 2001.

CONTACT WITH THE AUTHOR

Sawomir Grulkowski

e-mail: slawi@pg.gda.pl

Gdask University of Technology

11/12 Narutowicza St.
80 - 233 Gdask
Poland

88 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 89-96
10.1515/pomr-2017-0026

MULTI-CRITERIA OPTIMISATION OF LIQUID CARGO TRANSPORT

ACCORDING TO LINGUISTIC APPROACH TO THE ROUTE
SELECTION TASK

Sambor Guze
Tomasz Neumann
Przemysaw Wilczyski
Gdynia Maritime University, Poland

ABSTRACT

The main aim of the paper is to present the possibility of use of the multi-criteria optimization method Analytical
Hierarchy Process (AHP) to liquid cargo transportation by sea. Finding the optimal solution is not simple. There are
many factors influencing the shipping process. In the case of liquid cargo, the most important thing is the safety of the
crew, ship, and environment. Therefore, the Mathematical Theory of Evidence is introduced and used to determine
the optimal path in terms of time and safety of transport. Moreover, the details of liquid cargo transport process are
described with particular attention to ship to ship operations. Besides, the basic concept of the AHP method, steps
of the algorithm are introduced. Finally, the multicriteria optimization of the transport of the liquid cargo from the
Persian Gulf to Port of Gdansk is done. It is based on the experts opinions.

Keywords: liquid cargo transport, Dempster-Shaffer Theory, STS, AHP, multi-criteria optimization

INTRODUCTION detailed construction solution. This area concerns research

Maritime transport, like air, is exposed to several
unfavorable factors from the environment. These includes,
among other, time varying hydro-meteorological conditions.
Thus, it is crucial to neutralize or to resist these negative factors.
Considering the liquid cargo like oil, gas, chemicals, etc., the
problem of transport via sea is more complicated. Due to the
deep draft of tankers, not every destination can be reached.
Thus, the current trends in the industry of carriage demonstrate
its strong connection to the decision and optimization theory.
The first step of the process takes place already during
the designing and ship building. One of the important thing
in these processes are the decision about materials, ship
equipment, i.e., navigational, electrical or handling [14], and
not only on a new composite structure for a ship hull what
should increase structural safety during a collision [18],
[21]. Also, it concerns the processing technologies of various
parts of the ship to reduce the impact of adverse factors [7],
[9], [21]. Moreover, the significant field of interest is also the
effectiveness and fuel consumption of ship engines [13]. The
international regulations on emission become more restrictive,
so it is important to seek such solutions of marine engines that
the vessels comply with these strict standards.
The next step after designing and building of the ship is its
exploitation process. A distinctive feature of the transport of
liquid cargo is possible to do ship-to-ship operations (STS)
[24]. Performing these operations is possible after meeting

POLISH MARITIME RESEARCH, No S1/2017 89

very strict rules related to the place, hydro-meteorological
conditions, handling technology and qualifications of the crew
[6], [20], [24]. Transportation of liquid cargo by sea demands
ensuring an appropriate monitoring level to prevent an accident
or find the source of the environment pollution. Nowadays,
such control is possible using mobile units, including flying or
floating drones. The research on usage of mobile laser scanning
is introduced in [3].
In the research on transportation systems, a lot of papers
are devoted to the mathematical method of analysis, modeling,
and optimization. Many research teams conduct their studies
of the analysis, modeling, and optimization of the safety and
reliability of transportation systems [4], also taking into
account the operation process and semi-Markov approach
[4], [10] [12]. The different way of transportation systems
safety description is used in the time series analysis [1].
The second group of mathematical tools is those related to
improving and optimization of the objectives for operation
transportation systems. A lot of attention is paid to the
optimization methods and tools on algorithms for determining
safe ship trajectories [15], [16], [17], [23], and the analysis and
optimizing of the flow of people or cargo transportation system
with accordance to graph theory methods [5].
As it was mentioned above, the problem of liquid cargo is
complicated. Thus, one criterion optimization is not enough.
In the paper, the Analytic Hierarchy Process method of multi-
criteria optimization, [19], [22], is used to solve the problem
of liquid cargo transport per route selection task according
to costs. This approach is described in the Subsection 2.4. To
realize this goal, the Dempster-Shafer Theory combined with
fuzzy modeling as the tool to find the safe ships trajectory is
introduced. Furthermore, the description of the liquid cargo
transport is presented in Section 3. There is some information
about the fleet used in liquid cargo transportation and about
realization for the STS operations. Chapter 5 describes the
results on multi-criteria optimization with AHP (the Analytical
Hierarchy Process )method for the Persian Gulf Port of
Gdansk route.
METHODOLOGY
The three methods of research: Mathematical Theory of
Evidence, fuzzy modelling, and AHP method are discussed in
detail as the basic tools used in the paper. The first two of these
are used to determine the shortest in terms of time, and the
safest route of the ship. The third one is used to perform multi-
criteria optimization based on the opinions of the experts.
The input data is based on the exemplary record vessel
properties, motion report data, and digital climate prognosis
data. Furthermore, the case in the paper will join consumption
curves, velocity diminution curves, vessel class, ship wind
and weather sea borders, motion statement velocity, maximal
permitted speed, motion statement trace data to contain
waypoints, their latitude, and longitude. On top of data related
to the motion of the ship, it is indispensable to include the
specification of the surroundings. In particular, significant
90 POLISH MARITIME RESEARCH, No S1/2017
is the specification of the practicable routes between the first
point and the last one. This approach was also presented in
[15], [16].
The approach presented in this paper can be the conception
of accident risk on each curve of the transportation network by
taking account of the sensitivity of the curve in question and
of the cost created in the event of the accident on this curve
with respect to the various impacts considered. To model these,
the authors proposed Dempster-Shafer Theory.
DEMPSTER-SHAFER THEORY
The Dempster-Shafer Theory, also called Mathematical
Theory of Evidence, deals with function combining
information contained in two sets of assignments, subjective
expert ratings. This process may be interpreted as a knowledge
update. Combining sets results in forming of new subsets of
possible hypotheses with new values characterising probability
of specific options occurrence. The aforementioned process
may continue as long as provided with new propositions.
This function is known as Dempsters rule of combination. If
more than one factor appears on an edge, then it is possible
to cumulate them based on the following formula, where A is
the investigated set, B, C are elements of P(:).
This equation is proposed by Dempster:
(1)
where m(A), m(B) are probability mass assignments.
Combination rules specify how two mass functions,
presented as m1 and m2, are fused into one combined belief
measure. Many combination rules have been suggested
(several are presented in [2]).
For a published source apex (node) in the graph, the
algorithm discovers the way with smallest cost (i.e. the
shortest path) among that vertex and every other one. It could
also be used for discovering the smallest cost way from one
vertex to a goal vertex by the stoppage algorithm is intended
by the smallest way to the goal vertex. For instance, if the
apexes of the graph describe the cities and there are given costs
of flowing ways distances among pairs of points combined
immediately to the road, Dijkstras algorithm can be used
to discover the briefest route between one city and all other
cities. Consequently, the briefest path algorithm is highly
used in routing protocols in a web network, in particular the
ISIS and Open Shortest Path First [15].
For a given source vertex (node) in the graph, the algorithm
finds the path with lowest cost (i.e. the shortest path) between
that vertex and every other vertex. It can also be used for
finding the shortest cost path from one vertex to a destination
vertex by the stopping algorithm which is determined by
the shortest path to the destination node. For example, if
the vertices of the graph represent the waypoints and are
the costs of running paths edge distances between pairs of
waypoints connected directly, Dijkstras algorithm can be
used to find the shortest route between start and the exit
point in the maritime area (see Figure 1). This approach was
presented in [17].

Fig. 2. Typical membership function for the famous age example

Calculations in the fuzzy sets of values can be added to

the popular graph algorithms. One of them is presented in
Figure 3. We initialize the algorithm source with valued fuzzy
graph G(:,V,P), where V=G. The S is the sequence of vertices,
which is empty at the start algorithm. The Q = G.V is the set
of all unvisited vertices that are to be removed. In this set,
u represents the vertex with the smallest distance from the
Fig. 1. Scheme of maritime routes among islands
source. The G.Adj[u] describes the neighbour of the vertex u.

FUZZY MODELLING

The fuzziness shortest path finding problem from start

position - specific node of source to the end position - other
node is presented in many papers. In the field of maritime
transport systems, the suitable networks using fuzzy data of
the curves, are supposed to represent time of move from one to
another waypoint or economic costs as traffic flow, etc. These
data are supple and could be expressed by fuzzy numbers or
fuzzy sets [26].
In this approach, one implements the problem of defining
a fuzzy data retrieval model. In fuzzy model, queries are
Fig. 3. Pseudocode of single shortest path
defined as linguistic generalizations in the weighted model.
To this end, linguistic descriptors are introduced in the query
language to express the importance that a term must have in
the wished records and in the classification mechanism to AHP METHOD DESCRIPTION
label the restored records in relevance classes.
In the paper, the authors use one of the most useful multi-
DESCRIPTION OF FUZZY GRAPH criteria optimization methods, i.e. the Analytical Hierarchy
Process (AHP) method, which was introduced by Thomas
Let : be a complete set supposed to be equal to {1, 2, ., Saaty in the 70s. The general concept of this approach is as
n}. The framework G(:,V,P) can represent valued fuzzy graph follows [19]:
on : where: a. the goals hierarchisation;
 :[0,1] and means level of membership of any node, b. paired comparison of the objectives being on the same
level.
 :[0,1] and means level of membership of any curve On Figure 4 the general structure of AHP method is
presented.
According to the proposed approach, it is possible to use It should be noted that this is a helpful method when the
various functions. Most popular are trapezoidal or triangular expert opinions are collected during the research.
shaped membership functions. Figure 2 presents typical Let one assume, that there is a set of n-variants (options)
functions regarding to age of people. to consider. Every one of these components take the value for
k-criteria. Thus, there is the decision matrix
.

POLISH MARITIME RESEARCH, No S1/2017 91


Fig. 4. General diagram of the AHP method [19]
The steps of the AHP algorithm are as follows [19], [22]:
1. Hierarchization of the problem.
2. Paired comparison of the objectives being on the same
level matrixes of the paired comparisons.
3. Definition of the mutual weight of the criteria and
decision variants.
4. Choosing the best options.
In step 1, the detailed description of a problem, definition
of the primary goal and expectations of them is done. The
decomposition of the problem in the form of the principal
criteria and the main options considered, which generate a
certain degree of fulfillment of objectives of the function at
different levels of the hierarchical model is defined (see Fig. 4).
In step 2, the decision maker compares together in pairs
criteria in relation to the primary goal and the options to
the specific guidelines. A subjective determination indicates
which of the criteria and options, and to what extent are more
important than the other.
Relations between the elements is determined based on a
9-point scale [19], [22]:
a. 1 a same significance;
b. 3 - a small advantage;
c. 5 - a strong advantage;
d. 7 a very strong advantage;
e. 9 an absolute advantage;
f. 2, 4, 6, 8 an intermediate value.
Evaluation of the inverse relations is determined as a
reciprocal of integers.
This step completes the formation of a matrix B ,
level
level 2,3 , size k u k and n u n in case of the second and
third levels, respectively, which is made of k ( k  1) 2 and
n(n  1) 2 of these comparisons. The characteristic feature
of this matrix is a diagonal equal to 1, which consists of the
following property [19]:
(2)
where is element in i-th row and j-th column and is
element in j-th row and i-th column.
In step 3, the mutual weights for criteria and variants
(options) are calculated. The normalized rows of the matrix
92 POLISH MARITIME RESEARCH, No S1/2017
B level , level=2,3, are summed and the eigenvector of it is
level
found. Furthermore, the matrix B , level=2,3, satisfies [19]:
(3)
where
level
w - the eigenvector of matrix B ,
O - the eigenvalue of matrix B
level
.
The experts assessments are not always completely neutral,
so it is necessary to introduce the inconsistency coefficient
IF defined as follows [19]
(4)
where
CI - consequence ratio,
RI - random index.
It should be less than or equal to 0.2. In the case when
CI=0 , then the value of coefficient IF is calculated in respect
to the random index RI. It is the average CI for a large number
of randomly generated matrix of comparisons.
Moreover, the consequence ratio CI for matrix size n is
given by [19]:
(5)
where Omax is maximal eigenvalue of matrix B
level
,
level=2,3, calculated with equation (3).
It is believed that the data are consistent to the value of the
ratio CI, given in (5), which is less than 0.1 [19].
Finally, in the step 4, the decision-maker chooses the
optimal option for established criteria.
LIQUID CARGO TRANSPORT
Sea transport of crude oil & oil products generally covers
all sea routes between sources of crude oil in the world and
the places where refineries producing various type of fuels,
lubricating oils and others are located. There are many places
in the world where crude oil is produced on shore & also at
the off shore - oil field.
Crude oil is carried by large tankers (size ULCC or
VLCC) to minimize cost of the transport cargo between
continents to deliver cargo to the shore storage tanks or to the
fix installations like the pipelines, connecting oil terminals
with the refinery.
Ultra Large Crude Carriers (ULCCs), are tankers able to
transport very large volumes of oil, up to three million barrel
cargoes. A typical Double Hull Ship is of 410.000 Dwt. LOA
337 m Breadth 68 m Draft 23 m, Light ship 45.000 tons.
Very Large Crude Carriers (VLCCs), are tankers able to
transport large volumes of oil, including two million barrel
cargoes, over relatively long distances. Typical Double Hull
Ship is of 280.000 Dwt. LOA 335 m, Breadth 57 m Draft 21
m, Light ship 35.000 tons (see Figure 5).
Fig. 5. VLCC Typical Double Hull Tanker [25]
However, some destinations for tankers are restricted
by maneuverability and draft of the large ships. One of the
solution is to reduce the size of a tanker, but quantity of cargo
is automatically reduced too and the cost of the transport
is arises.
Maximum size for Baltic Sea are Suez Max tankers generally
identified as those capable of transporting one million barrel
cargoes. Typical Double Hull Ship is of 150.000 Dwt. LOA
274 m, Breadth 50 m, Draft 14,5 m, Light ship 20.000 tons.
Another solution to achieve the same volume of delivered
cargo to a destination and to reduce the cost to a reasonable
level is to organize the STS operation before the port of
destination.
Generally, STS operations take place at open sea or on
the road before restricted draft approaches to ports, rivers,
straits for large vessel. The main goal of an STS operation
is to reduce the draft of the large tankers up to accepted
value, which allows them to enter such areas. Part of the
liquid cargo from a large tanker is discharged to a smaller
size tanker to reduced draft. Discharging a quantity of liquid
cargo should always correspond to minimum numbers of
small size tankers. Liquid cargo that remains on onboard
should be distributed on board in such way to keep large
tanker always on even keel without heel and avoid forming
free surface in the cargo tank (see Figure 6).
Fig. 6. Example of lightering operation La Plata river [photo P. Wilczyski]
The size of small tankers designated to STS operation
should be adjusted to the quantity of cargo to be discharged
(exemplary see Figure 7). All ships parameters like
deadweight, draft, mooring arrangement, max. loading rate
etc. contain Q88 form exchanged between vessels involved
in STS operation. On the base of Q88 tankers operators
and masters take the decision which ship could be used for
planned STS.
Fig. 7. Typical Double Hull tankers of 60.000 Dwt. LOA 228,6 m, Breadth
32,2 m Draft 12,6 m, Light ship 12.000 tons [25]
Also distance between place where STS is planned to the
port of cargo destination is a very important factor to take
decision, how many small tankers should be involved in such
operation.
There are not so many places where STS operation could
be performed for VLCC proceeding to the Baltic Sea. Such
an area should provide enough space and depth for both
maneuvering vessels, additional support and service should
be available to fulfill all national regulation.
STS operations required proper hydro-meteorological
conditions to perform safe transfer of cargo between the
tankers. Sea current, wind and waves above the certain limits
do not allow to start the STS operation.
RESULTS
For consideration, the authors took into account the oil
transport from the Persian Gulf to the Port of Gdansk. After
discussion with experts, in transport of oil and oil products,
the following optimization criteria are fixed:
a. C1 - number of STS operation,
b. C2 - total load [mt],
c. C3 - time route [h],
d. C4 - total cost [USD].
As it was mentioned in the first step of AHP algorithm,
the first step and taking into account experts opinions gives
the hierarchization which is presented in Figure 8.
As it is presented above on the AHP method schema, six
options are under investigation. The decision matrix of the
considered problem is given in Table 1. Calculations find the
data for time route column based on Mathematical Theory of
Evidence and fuzzy modeling, which is described in Section
2. According to information mentioned in Section 3, the two
main routes are possible to realize the transport between
the Persian Gulf and Port of Gdansk. The shortest path goes
through the Suez Canal and its time length is 514 [h]. The
second route runs around Africa with length equal to 837 [h].
Both results are calculated numerically.
POLISH MARITIME RESEARCH, No S1/2017 93
 b) in case the criterion C2
Tab. 4. Matrix of comparisons of third level for criterion C2

C2 O1 O2 O3 O4 O5 O6
O1 1.0000 3.6667 3.6667 8.1111 9.0000 9.0000
O2 0.2727 1.0000 1.0000 5.4444 6.3333 6.3333
O3 0.2727 1.0000 1.0000 5.4444 6.3333 6.3333
O4 0.1233 0.1837 0.1837 1.0000 1.8889 1.8889
O5 0.1111 0.1579 0.1579 0.5294 1.0000 1.0000
Fig. 8. AHP schema for liquid cargo transport O6 0.1111 0.1579 0.1579 0.5294 1.0000 1.0000

Tab. 1. Decision matrix

c) in case the criterion C3
Number Tab. 5. Matrix of comparisons of third level for criterion C3
Total Load TIME TOTAL
of STS
[mt] ROUTE [h] COST [$]
operation C3 O1 O2 O3 O4 O5 O6
(C2) (C3) (C4)
(C1) O1 1.0000 2.0686 2.9592 8.9109 9.0000 8.3766
Option 1 0 140 000 514.00 2178333.33 O2 0.4834 1.0000 1.8905 7.8423 7.9314 7.3080
Option 2 1 200 000 586.00 2512500.00 O3 0.3379 0.5289 1.0000 6.9518 7.0408 6.4174
Option 3 2 200 000 646.00 2567500.00 O4 0.1122 0.1275 0.1438 1.0000 1.0891 0.6518
Option 4 3 300 000 1047.00 3825000.00 O5 0.1111 0.1261 0.1420 0.9182 1.0000 0.6160
Option 5 3 320 000 1053.00 3862500.00 O6 0.1194 0.1368 0.1558 1.5343 1.6234 1.0000
Option 6 4 320 000 1011.00 3875000.00
d) in case the criterion C4
The second step of AHP algorithm is to build the matrix Tab. 6. Matrix of comparison of third level for criterion C4
of criteria comparison according to the formula (2), i.e. the
2
matrix B . It is given in Table 2.
C4 O1 O2 O3 O4 O5 O6
Tab. 2. Matrix of comparisons for second level
O1 1.0000 2.5756 2.8350 8.7642 8.9411 9.0000
O2 0.3883 1.0000 1.2593 7.1886 7.3654 7.4244
O3 0.3527 0.7941 1.0000 6.9293 7.1061 7.1650
O4 0.1141 0.1391 0.1443 1.0000 1.1768 1.2358
O5 0.1118 0.1358 0.1407 0.8497 1.0000 1.0589
O6 0.1111 0.1347 0.1396 0.8092 0.9443 1.0000

In the third step of the algorithm, the normalized

3 2
Moreover, the four matrices B of third level are obtained eigenvector of matrix B is calculated via formula (3) and
for criteria C1, C2, C3 and C4 from experts and filled in presented in Table 7.
accordance to formula (2). The matrixes of comparisons are
Tab. 7. Normalized weighted vector of matrix B2
presented in Table 3 (for C1), Table 4 (for C2), Table 5 (for
C3), Table 6 (for C4). The results are as follows:
a) in case of the criterion C1 Criterion C1 C2 C3 C4

Tab. 3. Matrix of comparisons of third level for criterion C1 Weight 0.0507 0.0955 0.3080 0.5458

C1 O1 O2 O3 O4 O5 O6 The conclusion is that the highest weight has the C4

criterion - total cost (see Table 7).
O1 1.0000 3.0000 5.0000 7.0000 7.0000 9.0000
Finally, the vector of the rank is calculated and given in
O2 0.3333 1.0000 3.0000 5.0000 5.0000 7.0000 Table 8.
O3 0.2000 0.3333 1.0000 3.0000 3.0000 5.0000
O4 0.1429 0.2000 0.3333 1.0000 1.0000 3.0000 Tab. 8. Rank vector for choosing the option
O5 0.1429 0.2000 0.3333 1.0000 1.0000 3.0000
Opt. O1 O2 O3 O4 O5 O6
O6 0.1111 0.1429 0.2000 0.3333 0.3333 1.0000
Rank 0,4256 0,2497 0,2092 0,0408 0,0367 0,0380

94 POLISH MARITIME RESEARCH, No S1/2017

 The above rank vector indicates that the optimum option
for our problem is O1. Thus, we know, that:
a) to transport the liquid cargo, one should use the Suez
Max with total load 140000 [mt] without the need of
STS operation,
b) the route leads through the Suez Canal and the
Mediterranean Sea at total time equal to 514 [h],
c) the cost of transportation is equal to 2178333,33 [USD].
The alternative is the Suez Max with full tanks equal to
200000 [mt] (second in the ranking of options, see Table 8). In
this case, before the Danish Straits, it is necessary to unload
part of the cargo to a tanker with a capacity of 60000 [mt].
Thus, the cost of transport is higher - about 334166,67 [USD].
The final remark is that the experts opinions are consistent,
because IF = 0.04, with accordance to formulae (4) (5).
CONCLUSION
The article is devoted to the problem of multi-criteria
optimization of liquid cargo transport in relation to the Persian
Gulf - Port of Gdansk rout. This objective was made possible by
obtaining information from experts. The information has made
possible to determine the cost, route and organization of STS
operations. Furthermore, the time of passage has been fixed
by the linguistic approach to the route selection task taking
into account the uncertainties. The ship to ship operations
have been described. Moreover, the details of tankers have
been introduced.
Finally, the optimization problem has been solved using the
AHP method. The steps of the algorithm have been described
in details and used to optimize the transport of liquid cargo.
Further work in this area will be aimed to optimize the
return route.
ACKNOWLEDGMENTS
The paper is partially supported (Sambor
Guze) by financial sources of the EU-CIRCLE
project titled A pan European framework
for strengthening Critical Infrastructure
resilience to climate change that has received
funding from the European Unions Horizon 2020 research and
innovation program under the grant agreement No 653824.
http://www.eu-circle.eu/
BIBLIOGRAPHY
1. Bergel-Hayat R., Zukowska J.: Road Safety Trends at
National Level in Europe: A Review of Time-series Analysis
Performed during the Period 200012. Transport Reviews
Vol. 35, Iss. 5, 2015.
2. Bostrom H.: On evidential combination rules for ensemble
classifiers. Presented at the 11th International Conference
on Information Fusion, 2008.
3. Burdziakowski P., Janowski A., Kholodkov A., Matysik K.,
Matysik M., Przyborski M., Szulwic J., Tysic P., Wojtowicz
A.: Maritime laser scanning as the source for spatial data,
Polish Maritime Research. -Vol. 22, Iss. 4(88) (2015),
pp.9-14.
4. Girtler J., _ M.: Four-state stochastic model of changes
in the reliability states of a motor vehicle, Eksploatacja i
Niezawodno - Maintenance and Reliability. -Vol. 15, nr.
2 (2013), s.156-160
5. Guze S.: Graph Theory Approach to Transportation Systems
Design and Optimization. TransNav, the International
Journal on Marine Navigation and Safety of Sea
Transportation, Vol. 8, No. 4, pp. 571-578, 2014.
6. International Safety Guide for Oil Tankers and Terminals,
5th Edition, June 2006, London.
7. Jakubowski M.: Influence of pitting corrosion on fatigue
and corrosion fatigue of ship and offshore structures. Part
II: Load - pit crack interaction. Polish Maritime Research.
-Vol. 22, nr. 3 (2015), s.57-66..
8. Janowski A., Jurkowska A., Lewczuk D., Szulwic J., Zaradny
A.: Assessment of cliff stability after the demolition of the
engineering facilities. 14th International Multidisciplinary
Scientific Geoconference (SGEM): GEOCONFERENCE
ON SCIENCE AND TECHNOLOGIES IN GEOLOGY,
EXPLORATION AND MINING, SGEM 2014, VOL II,
ISBN 978-619-7105-08-7 / ISSN 1314-2704, 2014.
9. Kaliski K. J., Mazur M., Galewski M. A.: The Optimal
Spindle Speed Map for Reduction of Chatter Vibration
During Milling of Bow Thruster Blade. Solid State
Phenomena 198 (2013) 686-691.
10. Koowrocki K.: Reliability of Large and Complex Systems.
Amsterdam, Boston, Heidelberd, London, New York,
Oxford, Paris, San Diego, San Francisco, Singapore, Sidney,
Tokyo, Elsevier, 2014
11. Koowrocki K., Soszyska-Budny J.: Reliability and Safety
of Complex Technical Systems and Processes: Modeling
- Identification - Prediction Optimization. London,
Dordrecht, Heildeberg, New York, Springer, 2011.
12. Koowrocki K., Soszyska-Budny J.: Reliability prediction
and optmization of complex technical systems with
application in port transport. Journal of Polish Safety and
Reliability Association, Summer Safety and Reliability
Seminars, Vol. 3, No 1-2, pp. 263-279, 2013.
13. Kropiwnicki J., Kneba Z., Zikowski M.: Test for assessing
POLISH MARITIME RESEARCH, No S1/2017 95
 the energy efficiency of vehicles with internal combustion
engines. International Journal of Automotive Technology, CONTACT WITH AUTHORS
Vol. 14, nr 3 (2013), s. 479-487.
Sambor Guze
14. Mikielewicz J., Mikielewicz D.: A simple model of circular Department of Mathematics
hydraulic pump. International Journal of Heat and Mass Email: s.guze@wn.am.gdynia.pl
Transfer, Vol. 52(1), 17-21, 2008. Tel.: +585586179

15. Neumann T.: The Shortest Path Problem with Uncertain Tomasz Neumann
Information in Transport Networks. In Challenge of Department of Navigation
Transport Telematics J. Mikulski, Ed. Springer International Email: t.neumann@wn.am.gdynia.pl
Publishing, 2016. Tel.: 585586168

16. NeumannT.: Good choice of transit vessel route using Przemysaw Wilczyski
Dempster-Shafer Theory. Presented at the 2015 International Department of Ships Operation
Siberian Conference on Control and Communications Email: p.wilczynski@wn.am.gdynia.pl
(SIBCON), Omsk, 2015, pp. 14. Tel.: 585586168

17. Neumann T.: Vessels Route Planning Problem with Gdynia Maritime University
Uncertain Data. TransNav, the International Journal on Faculty of Navigation
Marine Navigation and Safety of Sea Transportation, vol. 3 Jana Pawla II Avenue
10, no. 3, pp. 459464, Sep. 2016. 81-345 Gdynia
poland
18. Niklas K., Kozak J.: Experimental investigation of Steel
ConcretePolymer composite barrier for the ship internal
tank construction. OCEAN ENGINEERING. -Vol. 111,
(2016), s.449-460;

19. Saaty T.L.: The Analytic Hierarchy Process. RWS

Publications, Pittsburgh, 1998.

20. Ship to Ship Transfer Guide for Petroleum, Chemicals and

Liquefied Gases, November 2013, London

21. Skarzynski L., Kozicki J., Tejchman A.: Application of DIC

technique to concrete - study on objectivity of measured
surface displacements. EXPERIMENTAL MECHANICS.
Vol. 53, iss. 9 (2013), pp.1545-1559.

22. Szapczyska J.: Review of the multi-criteria decision

problems solving methods application in navigational
processes. Prace Wydziau Nawigacyjnego, No 17, pp. 136
150, 2005 (in Polish).

23. Szapczyski R., Szapczyska J.: Customized crossover

in evolutionary sets of safe ship trajectories. International
Journal of Applied Mathematics and Computer Science.
-Vol. 22, nr. No.4 (2012), s.999-1009

24. Wilczyski P.: STS Plan for m/t Icarus III. Course materials
- PRS approved, Gdynia 20 May 2014.

25. Transas homepage, source: http:// www.transas.com.

26. Zadeh L.A.: The concept of a linguistic variable and its

application to approximate reasoning. Information Sciences,
vol. 8, pp. 199249, 1975.

96 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 97-107
10.1515/pomr-2017-0027

A DISCRETE-CONTINUOUS METHOD OF MECHANICAL

SYSTEM MODELLING

Rafa Hein
Gdansk University of Technology,Poland

ABSTRACT

The paper describes a discrete-continuous method of dynamic system modelling. The presented approach is hybrid in
its nature, as it combines the advantages of spatial discretization methods with those of continuous system modelling
methods. In the proposed method, a three-dimensional system is discretised in two directions only, with the third
direction remaining continuous. The thus obtained discrete-continuous model is described by a set of coupled partial
differential equations, derived using the rigid finite element method (RFEM). For this purpose, firstly the general
differential equations are written. Then these equations are converted into difference equations. The derived equations,
expressed in matrix form, allow to create a global matrix for the whole system. They are solved using the distributed
transfer function method. The proposed approach is illustrated with the examples of a simple beam fixed at both ends
and a simply supported plate.

Keywords: modelling, mechanical system, dynamic systems, vibrations, hybrid modelling methods.

systems. Such a model also requires considerable computer

INTRODUCTION
Continuous systems with spatially distributed parameters
are described by partial differential equations. However, exact
solutions to such equations exist only for a small class of
simple technical problems. For this reason, the continuous
structure is usually discretized and divided into small,
spatially lumped parameter elements. Such elements are
called finite elements and the method is well known as the
finite element method. A discrete model obtained in this way
is described by the set of ordinary differential equations. The
solutions to such equations pose no problem but the model
is approximate and its accuracy depends on the number of
finite elements. The greater their number, the more accurate
the model. However, there is an optimal division density,
above which the rounding errors start to seriously affect the
numerical calculation. Additionally, a very large number of
finite elements increases the order of the differential equation,
which is not convenient when designing, for instance, control
run time and huge memory capacity to obtain and store
numerical calculation results.
In the case of slender elements, such as: strings, bars and
beams, for which the analytical and exact solutions are well
known, it is better to use an accurate, continuous model
description.
Consequently, there is no universal approach which is both
accurate and applicable to a wide range of dynamic systems.
That is why the modelling methods of physical systems are
constantly being developed and improved.
The paper proposes a hybrid method of modelling that
combines the advantages of the spatial discretization methods
with those of continuous system modelling methods.
In the classical finite element method, the body is
discretized in all three spatial directions (Fig. 1a). In the
proposed method, the same body is discretized only in two
spatial directions (Fig. 1b), with one direction remaining
continuous. Such a division results in the appearance of
discrete-continuous elements, so-called prisms (Fig. 1b), with

POLISH MARITIME RESEARCH, No S1/2017 97

parameters distributed along one of the axes. Consequently, qr +1 = col ( qr +1,1 , qr +1, 2 , qr +1,3 , qr +1, 4 , qr +1,5 , qr +1, 6 ) , (3)
the obtained elements are one-dimensional distributed
elements, described by second-order partial differential
equations. However, these equations also include terms related q p = col ( q p ,1 , q p , 2 , q p , 3 , q p , 4 , q p ,5 , q p , 6 ) , (4)
to interactions between elements, as a result of which the
entire system is described by coupled second-order partial a) k
differential equations.

a) b)

r p

finite element prism

Fig. 1. Spatial discretization of 3D body:

a) conventional finite element method, b) proposed hybrid method

The obtained equations are solved by using the distributed

transfer function method [15, 16, 18]. This method enables to
obtain analytical or semi-analytical solutions for 1D and 3D b)
systems, respectively. A similar idea was considered in [12] z
and [15]. However, in some aspects there are fundamental
r-1
differences. In [12] and [18] the body is treated as deformable sr-1,i,1
in the directions in which the discretization was made, x i
while in the proposed method it is treated as a rigid and y
non-deformable solid in two discretization directions. This r k p
approach has many advantages. Firstly, it makes it possible sr,i,1
to obtain a simple and clear system of partial differential sr,j,1
equations. Then, by using the modal decomposition method, j x3
it is easy to determine the modal model and analyse it in the q6
time domain [1, 9, 10, 11]. In a simple way, the modal model r+1 q3 q5
can be reduced to a low order model, which is convenient in sr+1,j,1
q4
design and analysis of control systems [2, 3, 4]. The proposed
method has been applied in modelling 1D, 2D and 3D systems
q1 q2 x2
x1
[8, 10, 12, 17] and may be used to modelling and solving many
practical engineering problems [6, 7]. Fig. 2. General model of the analysed system:
a) discrete-continuous, b) discrete

GENERAL MODEL OF THE DISCRETE-
CONTINUOUS SYSTEM
In order to derive a general model of the discrete-continuous
system, let us consider two prisms, r and p, connected by
aspring-damping layer, k, continuous in the x direction, with
distributed parameters (Fig. 2a). The discretized model of the
analysed system is shown in Fig. 5. Each element has 6 degrees
of freedom expressed by means of general displacements. The
first three of them are translational displacements along the
x1, x2, and x3-axes, while the remaining represent rotational
displacements around these axes:
qr 1 = col ( qr 1,1 , qr 1, 2 , qr 1,3 , qr 1, 4 , qr 1,5 , qr 1, 6 ) ,
For the r-th rigid element of the discrete model presented
in Fig. 2b, a system of ordinary differential equations was
written and verified by using the rigid finite element method
[5, 13, 14]. The obtained equations have the following form:
f r ,1 = mqr ,1 + c i ,1 (q r ,1 q r 1,1 ) c j ,1 ( q r +1,1 q r ,1 ) +
+ c k ,1 (q r ,1 q p ,1 ) + c k ,1 ( s r , k ,3 q r ,5 s p , k ,3 q p ,5 ) +
c k ,1 ( s r , k , 2 q r , 6 s p , k , 2 q p , 6 ) + (5)
+ c i ,1 ( s r ,i ,3 q r ,5 s r 1,i ,3 q r 1,5 ) c i ,1 ( s r ,i , 2 q r , 6 s r 1,i , 2 q r 1, 6 ) +
+ c j ,1 ( s r , j ,3 q r ,5 s p , j ,3 q r +1,5 ) c j ,1 ( s r , j , 2 q r , 6 s p , j , 2 q r +1, 6 ),
f r , 2 = mqr , 2 + c i , 2 (q r , 2 q i 1, j , 2 ) c j , 2 (q r +1, 2 q r , 2 ) +
(1) + c k , 2 (q r , 2 q p , 2 ) c k , 2 ( s r ,k ,3 q r , 4 s p ,k ,3 q p , 4 ) +
+ c k , 2 ( s r ,k ,1 q r , 6 s p ,k ,1 q p , 6 ) + (6)

+ c i , 2 ( s r ,i ,1 q r , 6 s r 1,i ,1 q r 1, 6 ) + c j , 2 ( s r , j ,1 q r , 6 s p , j ,1 q r +1, 6 ) +
qr = col ( qr ,1 , qr , 2 , qr ,3 , qr , 4 , qr ,5 , qr , 6 ) , (2)
+ c j , 2 ( s p , j ,3 q r +1, 4 s r , j ,3 q r , 4 ) + c i , 2 ( s r 1,i ,3 q r 1, 4 s r ,i ,3 q r , 4 ),

98 POLISH MARITIME RESEARCH, No S1/2017


f r ,3 = mqr ,3 + c i ,3 (q r ,3 q r 1,3 ) c j ,3 ( q r +1,3 q r ,3 ) +
+ c k ,3 (q r ,3 q p ,3 ) + c k ,3 ( s r , k , 2 q r , 4 s p , k , 2 q p , 4 ) +
+ c p ,3 ( s p , k ,1 q p ,5 s r , k ,1 q r ,5 ) +
c i ,3 ( s r , i ,1 q r ,5 s r 1,i ,1 q r 1,5 ) c j ,3 ( s r , j ,1 q r ,5 s r +1, j ,1 q r +1,5 ) +
+ c i ,3 ( s r ,i , 2 q r , 4 s r 1,i , 2 q r 1, 4 ) + c j ,3 ( s r , j , 2 q r , 4 s r +1, j , 2 q r +1, 4 ),
f r , 4 = xI 0 x qr , 4 + c i , 4 (q r , 4 q r 1, 4 ) c j , 4 (q r +1, 4 q r , 4 ) +
+ c k , 4 ( q r , 4 q p , 4 ) + c k ,3 (s r , k , 2 q r ,3 s r , k , 2 q p ,3 ) +
+ c k ,3 ( s 2
r ,k , 2 q r , 4 s r ,k , 2 s p ,k , 2 q p , 4 ) +
f r , 6 = zI 0 z qr , 6 + ci , 6 (qr , 6 qr 1, 6 ) c j , 6 (qr +1, 6 qr , 6 ) +
+ c j , 2 sr , j ,1 (qr , 2 qr +1 ,2 ) ci , 2 sr , i ,1 (qr 1 ,2 qr , 2 ) +
+ ck , 6 (qr , 6 q p , 6 ) ck ,1sr , k , 2 (qr ,1 q p ,1 ) +
(7)
ck ,1 (sr , k , 2 sr , k ,3qr ,5 sr , k , 2 s p , k ,3q p ,5 )
+ ci , 2 (sr2, i ,1qr , 6 sr 1, i ,1sr , i ,1qr 1, 6 ) +
+ c j , 2 (sr2, j ,1qr , 6 sr , j ,1sr +1, j ,1qr +1, 6 ) +
+ ck ,1 (sr2, k , 2 qr , 6 sr , k , 2 s p , k , 2 q p , 6 ) +
+ ck , 2 sr , k ,1 (qr , 2 q p , 2 )
ck , 2 (sr , k ,1sr , k ,3qr , 4 sr , k ,1s p , k ,3q p , 4 ) + (10)

+ c k , 2 ( s r2, k ,3 q r , 4 s r ,k ,3 s p , k ,3 q p , 4 ) + c k , 2 ( s r , k ,3 q p , 2 s r , k ,3 q r , 2 ) + + ck , 2 (sr2, k ,1qr , 6 sr , k ,1s p , k ,1q p , 6 )

c k , 2 ( s r , k ,1 s r , k ,3 q r , 6 s r , k ,3 s p , k ,1 q p , 6 ) + ci ,1sr , i , 2 (qr ,1 qr 1 ,1 ) + c j ,1sr , j , 2 (qr +1,1 qr ,1 ) +
c k ,3 ( s r , k ,1 s r , k , 2 q r ,5 s r , k , 2 s p , k ,1 q p ,5 ) + + ci ,1 (sr 1, i ,3 sr , i , 2 qr 1,5 sr , i , 2 sr , i ,3qr ,5 ) +
c i , 2 ( s r ,i ,3 q r , 2 s r ,i ,3 q r 1, 2 ) c j , 2 ( s r , j ,3 q r , 2 s r , j ,3 q r +1, 2 ) + + c j ,1 (sr , j , 2 sr +1, j ,3qr +1,5 sr , j , 2 sr , j ,3qr ,5 ) +
+ c i ,3 ( s r ,i , 2 q r ,3 s r ,i , 2 q r 1,3 ) + c j ,3 ( s r , j , 2 q r ,3 s r , j , 2 q r +1,3 ) + + ci , 2 (sr 1, i , 3 sr , i ,1qr 1, 4 sr , i ,1sr , i ,3qr , 4 ) +
(8)
+ c i ,3 ( s r 1,i ,1 s r ,i , 2 q r 1,5 s r ,i ,1 s r ,i , 2 q r ,5 ) + + c j , 2 (sr , j ,1sr +1, j ,3qr +1, 4 sr , j ,1sr , j ,3qr , 4 ) +
+ c j ,3 ( s r , j , 2 s r +1, j ,1 q r +1,5 s r , j ,1 s r , j , 2 q r ,5 ) + + ci ,1 (sr2, i , 2 qr , 6 sr 1, i , 2 sr ,i , 2 qr 1, 6 ) +
+ c i ,3 ( s r2,i , 2 q r , 4 s r 1,i , 2 s r ,i , 2 q r 1, 4 ) + + c j ,1 (sr2, j , 2 qr , 6 sr , j , 2 sr +1, j , 2 qr +1, 6 ),
+ c i , 2 ( s r2,i ,3 q r , 4 s r 1,i ,3 s r ,i ,3 q r 1, 4 ) +
where: f r, is the general force applied to the r-th element
+ c j ,3 ( s r2, j , 2 q i , j , k , 4 s r , j , 2 s r +1, j , 2 q r 1, 4 ) + (excitation) in the x -axis direction (Fig. 2b); m is the mass of
+ c j , 2 ( s r2, j ,3 q r , 4 s r , j ,3 s r +1, j ,3 q r +1, 4 ) + the rigid finite element; is the density of the rigid element; I0
is the geometric, polar moment of inertia around the x -axis;
+ c i , 2 ( s r 1, i ,1 s r , i ,3 q r 1, 6 s r , i ,1 s r , i ,3 q r , 6 ) +
x, y, z are the elementary dimensions of the finite element
+ c j , 2 ( s r , j ,3 s r +1, j ,1 q r +1, 6 s r , j ,1 s r , j ,3 q r , 6 ), (Fig. 2b); qr, is the general displacement of the r-th element
in the x -axis direction; q p, is the general displacement of
f r ,5 = yI 0 y qr ,5 + ci ,5 (qr ,5 qr 1,5 ) c j ,5 (qr +1,5 qr ,5 ) + the p-th element in the x -axis direction; s,, is the distance
+ ck ,5 (qr ,5 q p ,5 ) ci ,3sr ,i ,1 (qr , 3 qr 1 ,3 ) ck ,3sr , k ,1 (qr ,3 q p , 3 ) between the body and the spring-damping element in
the x -axis direction, and c,is the stiffness coefficient of the
+ ci ,3 (sr2,i ,1qr ,5 sr 1,i ,1sr , i ,1qr 1,5 ) +
spring element in the x direction [5,13,14]:
+ c j ,3 (sr2, j ,1qr ,5 sr , j ,1sr +1, j ,1qr +1, 5 ) +
+ ck ,1sr , k ,3 ( qr ,1 q p ,1 ) + ck ,3 ( sr2, k ,1qr ,5 sr , k ,1s p , k ,1q p ,5 ) + Eyz Gxz
c i ,1 = c j ,1 = , c k ,1 = ,
x y
+ ck ,1 ( sr2, k ,3qr ,5 sr , k ,3s p , k ,3q p ,5 ) + c j ,3sr , j ,1 (qr +1,3 qr ,3 ) +
Gyz Exz
ck ,3 ( sr , k ,1sr , k , 2 qr , 4 sr , k ,1s p , k , 2 q p , 4 ) + ci , 2 = c j , 2 = , ck ,2 = ,
x y
ck ,1 ( sr , k , 2 sr , k ,3qr , 6 sr , k ,3 s p , k , 2 q p , 6 ) + (9) Gyz Gxz
ci ,3 = c j ,3 = , c k ,3 = ,
+ ci ,1sr , i ,3 (qr ,1 qr 1,1 ) + c j ,1sr , j ,3 ( qr ,1 qr +1,1 ) + x y
(11)
+ ci ,1 ( sr2,i ,3qr ,5 sr 1,i ,3 sr , i ,3qr 1,5 ) + GI 0 x EI xy
ci , 4 = c j , 4 = , ck , 4 = ,
+ c j ,1 ( sr2, j ,3qr ,5 sr , j , 3 sr +1, j ,3qr +1,5 ) + x y
EI yx GI
+ ci ,3 ( sr 1, i , 2 sr , i ,1qr 1, 4 sr , i , 2 sr , i ,1qr , 4 ) + ci , 5 = c j , 5 = , c k ,5 = 0 y ,
x y
+ c j ,3 ( sr , j ,1sr +1, j , 2 qr +1, 4 sr , j ,1sr , j , 2 qr , 4 ) +
EI zx EI
+ ci ,1 ( sr 1, i , 2 sr , i ,3qr 1, 6 sr , i , 2 sr , i ,3qr , 6 ) + ci , 6 = c j , 6 = , c k , 6 = zy ,
x y
+ c j ,1 ( sr +1, j , 2 sr , j ,3qr +1, 6 sr , j , 2 sr , j ,3qr , 6 ),
In Eq. (11): E is the Youngs modulus, G is the Kirchhoff
shear modulus, I, is the geometric moment of inertia of
the cross-section area perpendicular to the x-axis around
the x -axis, and is the numerical shape factor of the cross
section area.
The rigid finite element model described by Equations
(5)(10) is next transformed to a discrete-continuous model.
We assume that the discretization is not performed in the x
direction. By putting:

POLISH MARITIME RESEARCH, No S1/2017 99

 x x =0
s r 1,i ,1 = s r , j ,1 = , s r ,i ,1 = s r +1, j ,1 = , z 1
-
2 2 (12) f xr , 6 = xy (y + x 2 )qr , 6 +
2

s r ,i , 2 = s r ,i , 3 = s r , j , 2 = s r , j , 3 = s r +1, j , 2 = s r +1, j , 3 = s r ,k ,1 = s p ,k ,1 = 0 x 12
q qr , 6 q qr 1, 6
( r +1, 6 ) ( r ,6 )
and (11) into Eqs. (5)(10), and dividing these equations by EI zx x x +
'x we obtain the following formulas: x
Gyz x qr , 2 qr +1 ,2 Gyz x qr 1 ,2 qr , 2
+ ( )+ ( )+
qr +1,1 qr ,1 qr ,1 qr 1,1 x 2 x x 2 x

f xr ,1 = yzqr ,1 Eyz x x + 1
- =0

x E zx x 2
Gz Gz + 12 ( qr , 6 q p , 6 )
+ (qr ,1 q p ,1 ) + ( s r , k , 3 qr , 5 s p , k , 3 q p , 5 ) + (13) yx
y y
Gxz (18)
Gz sr , k , 2 (qr ,1 q p ,1 ) +
( sr , k , 2 qr , 6 s p , k , 2 q p , 6 ), yx
y
Gxz
q qr , 2 q qr 1, 2 (sr , k , 2 sr , k ,3qr ,5 sr , k , 2 s p , k ,3q p , 5 ) +
( r +1, 2 ) ( r ,2 ) yx
f xr , 2 = yzqr , 2 Gyz x x + 2
x Gyz x
+ (qr , 6 + qr 1, 6 ) +
Ez Ez xx 2
+ ( qr , 2 q p , 2 ) ( sr , k , 3 q r , 4 s p , k , 3 q p , 4 ) +
y y (14) Gyz x
2

+ (qr , 6 + qr +1, 6 ) +
Gyz qr +1, 6 qr , 6 Gyz qr , 6 qr 1, 6 xx 2
+ ( )+ ( ),
2 x 2 x Gxz 2
+ (sr , k , 2 qr , 6 sr , k , 2 s p , k , 2 q p , 6 ),
(qr +1,3 qr ,3 ) (qr ,3 qr 1,3 ) yx

f xr ,3 = yzqr ,3 Gyz x x + where fxr, is the distributed force along the x1-axis applied to
x
the r-th element (excitation) in the x -axis direction (Fig. 2).
Gz Gz
+ ( qr , 3 q p , 3 ) + ( s r , k , 2 qr , 4 s p , k , 2 q p , 4 ) + (15) By assuming 'x0, we can obtain the following partial
y y
differential equations:
Gyz qr +1,5 qr ,5 Gyz qr ,5 qr 1,5
( ) ( ),
2 x 2 x
f xr ,1 = Aqr ,1 EAqr,1 +
qr +1, 4 qr , 4 qr , 4 qr 1, 4
( )( ) Gz Gz
x x + (qr ,1 q p ,1 ) + ( s r ,k , 3 q r , 5 s p ,k , 3 q p , 5 ) + (19)
f xr , 4 = I 0 x qr , 4 GI 0 x + y y
x
Gz
EI xy Gz ( sr ,k , 2 qr , 6 s p ,k , 2 q p , 6 ),
+ (qr , 4 q p , 4 ) + (sr , k , 2 qr ,3 sr , k , 2 q p ,3 ) + (16) y
yx y
Gz 2 f xr , 2 = Aqr , 2 GAq r, 2 + GAq r , 6 +
+ ( sr , k , 2 qr , 4 sr , k , 2 s p , k , 2 q p , 4 ) +
y
Ez Ez (20)
Ez 2 Ez + (qr , 2 q p , 2 ) ( s r ,k ,3 q r , 4 s p ,k ,3 q p , 4 ),
+ ( sr , k ,3qr , 4 sr , k ,3s p , k ,3q p , 4 ) + ( sr , k ,3q p , 2 sr , k ,3qr , 2 ), y y
y y
- =0 f xr , 3 = Aqr ,3 GAq r,3 GAqr ,5
y 1
f xr ,5 = xz ( x 2 + z 2 ) qr ,5 + Gz Gz (21)
x 12 (q r ,3 q p ,3 ) + ( sr ,k , 2 q r , 4 s p ,k , 2 q p , 4 ),
qr +1,5 qr ,5 q qr 1,5 y y
( ) ( r ,5 )
EI yx x x +
x f xr , 4 = I 0 x qr , 4 GI 0 x q r, 4 +

1
- =0
Ez 3 Gz
G xz ( x 2 + z 2 ) + (qr , 4 q p , 4 ) + (s r ,k , 2 q r , 3 s r ,k , 2 q p ,3 ) +
12y y
+ 12 ( qr ,5 q p ,5 ) +
xy Gz 2
(17) + ( s r ,k , 2 q r , 4 s r ,k , 2 s p ,k , 2 q p , 4 ) +
Gy z x qr ,3 qr 1 ,3 Gy z x qr +1,3 qr ,3 y (22)
+ ( )+ ( )+
x 2 x x 2 x Ez 2
+ ( s r ,k , 3 q r , 4 s r ,k , 3 s p ,k , 3 q p , 4 ) +
Gy z x
2
Gy z x
2
y
+ (qr ,5 + qr 1,5 ) + (qr ,5 + qr +1,5 ) + Ez
xx 2 xx 2 + ( s r ,k , 3 q p , 2 s r ,k ,3 q r , 2 ),
Gxz Gxz 2 y
+ sr , k , 3 ( qr ,1 q p ,1 ) + ( s r , k , 3 q r , 5 sr , k , 3 s p , k , 3 q p , 5 ) +
yx yx
Gxz
( sr , k , 2 sr , k ,3qr , 6 sr , k ,3 s p , k , 2 q p , 6 ),
yx

100 POLISH MARITIME RESEARCH, No S1/2017

 f xr ,5 = I yx qr ,5 EI yx qr,5 + GAqr ,3 + GAqr ,5 The above equations are obtained for prism r (connected
to prism p). However, by exchanging subscripts r p, we
Gz 3 Gz
( qr ,5 q p ,5 ) + s r ,k ,3 ( qr ,1 q p ,1 ) + can easily obtain appropriate equations related to prism p.
12y y
The general equations for the whole system, consisting
Gz 2
+ ( s r , k , 3 q r , 5 s r ,k , 3 s p , k , 3 q p , 5 ) + (23) in this particular case of r and p elements, can be written in
y the matrix form
Gz
( s r ,k , 2 sr ,k ,3 qr , 6 sr ,k ,3 s p ,k , 2 q p , 6 ),
y A02q( x, t ) + A20q( x, t ) + A10q( x, t ) + A00q ( x, t ) = f ( x, t ) (32)
f xr , 6 = I zx qr , 6 EI zx q r, 6 GAq r , 2 + GAq r , 6 +
with the boundary conditions
Gz Gz
s r ,k , 2 (q r ,1 q p ,1 ) + (s r2,k , 2 q r , 6 s r ,k , 2 s p ,k , 2 q p , 6 ) + (24)
y y
M j q ( 0, t ) + N j q ( l , t ) = j ( t ) , j=1,2 (33)
Gz
(s r ,k , 2 s r ,k ,3 q r ,5 s r ,k , 2 s p ,k ,3 q p ,5 ),
y
where:
where: r p
1 1
A = yz , I yx = yz 3 , I zx = zy 3 , q = d dt q , q = d dx q .
12 12 0 0  0 0 0  0 0 0  0

0 0  0 0 0  0 0 0  0
In order to obtain the general form of Equations (19)(24),            
the following substitutions are applied:
0 0  0 0 0  0 0 0  0
Gz Ez Gz EI zy
0 0  0 A02 r 0  0 0 0  0 r
cxk ,1 = , cxk , 2 = , cxk ,3 = , c xk , 6 = lim = 0,
y y y x 0 y x 0 0  0 0 0  0 0 0  0
A02 = (34)
           
EI xy Ez 3
GI Gz 3 (25)
cxk , 4 = lim = , cxk ,5 = lim 0 y = . 0 0  0 0 0  0 0 0  0
x 0 yx 12y x 0 yx 12y
0 0  0 0 0  0 A02 p 0  0 p
Thus Equations (19)(24) can be presented in the final 0 0  0 0 0  0 0 0  0

form:            
0 0  0 0 0  0 0 0  0
f xr ,1 = Aqr ,1 EAq r,1 +

+ c xk ,1 ( q r ,1 q p ,1 ) + c xk ,1 ( s r ,k , 3 qi , j , 5 s p ,k , 3 q p , 5 ) + (26) or A02 = diag[0,0,,0, A02 r ,0,,0, A02 p ,0,,0] ,

c xk ,1 ( s r ,k , 2 q r , 6 s p ,k , 2 q p , 6 ),
where A02 r = diag [ A, A, A, I 0 x , I yx , I zx ] ,
f xr , 2 = Aqr , 2 GAqr, 2 + GAqr , 6 +
(27) A02p can be obtained from the matrix A02r by replacing the
+ c xk , 2 (qr , 2 q p , 2 ) c xk , 2 ( sr ,k ,3 qr , 4 s p ,k ,3 q p , 4 ),
parameters and dimensions of the r prism with appropriate
parameters of the p prism,
f xr , 4 = I 0 x qr , 4 GI 0 x q r, 4 +
r p
+ c xk , 4 (q r , 4 q p , 4 ) + c xk ,3 (s r ,k , 2 q r ,3 s r ,k , 2 q p ,3 ) +
+ c xk ,3 ( s r2,k , 2 q r , 4 s r ,k , 2 s p ,k , 2 q p , 4 ) + (28) 0 0  0 0 0  0 0 0  0

+ c xk , 2 ( s 2
q
r ,k , 3 r , 4 s r ,k ,3 s p ,k ,3 q p , 4 ) + c xk , 2 ( s r ,k ,3 q p , 2 s r ,k ,3 q r , 2 ), 0 0  0 0 0  0 0 0  0
           

f xr , 4 = I 0 x qr , 4 GI 0 x q r, 4 + 0 0  0 0 0  0 0 0  0
0 0  0 A20 r 0  0 0 0  0 r
+ c xk , 4 (q r , 4 q p , 4 ) + c xk ,3 (s r ,k , 2 q r ,3 s r ,k , 2 q p ,3 ) + (35)
0 0  0 0 0  0 0 0  0
+ c xk ,3 ( s 2
q r , 4 s r ,k , 2 s p ,k , 2 q p , 4 ) + (29) A20 =
r ,k , 2
            ,
+ c xk , 2 ( s 2
q
r ,k , 3 r , 4 s r ,k ,3 s p ,k ,3 q p , 4 ) + c xk , 2 ( s r ,k ,3 q p , 2 s r ,k ,3 q r , 2 ), 0 0  0 0 0  0 0 0  0

0 0  0 0 0  0 A20 p 0  0 p
f xr , 5 = I yx qr ,5 EI yx qr,5 + GAqr , 3 + GAqr ,5 + 0 0  0 0 0  0 0 0  0

+ c xk ,5 (qr ,5 q p ,5 ) + c xk ,1 sr ,k ,3 (qr ,1 q p ,1 ) +            
0  0
+ c xk ,1 ( sr2,k ,3 qr ,5 sr ,k , 3 s p ,k ,3 q p ,5 ) + (30) 0 0  0 0 0  0 0
c xk ,1 ( sr ,k , 2 sr ,k ,3 qr , 6 sr ,k ,3 s p ,k , 2 q p , 6 ),
or A20 = diag[0,0,,0, A20 r ,0,,0, A20 p ,0,,0] ,
f xr , 6 = I zx qr , 6 EI zx qr, 6 GAqr , 2 + GAqr , 6 +
c xk ,1 sr ,k , 2 (qr ,1 q p ,1 ) c xk ,1 (s r ,k , 2 sr ,k ,3 qr ,5 sr ,k , 2 s p ,k ,3 q p ,5 ) + where A20 r = diag [ EA,GA,GA,GI 0 x , EI yx , EI zx ] .
(31)
+ c xk ,1 (sr2,k , 2 qr , 6 s r ,k , 2 s p ,k , 2 q p , 6 ).

POLISH MARITIME RESEARCH, No S1/2017 101

A20p can be obtained from matrix A20r by replacing the and
parameters and dimensions of the r prism with appropriate
parameters of the p prism, a rrk11 0 0 0 a rrk15 arrk16

p arrk 22 0 arrk 24 0 0
r
a rrk 33 a rrk 34 0 0
Arrk = ,
0 0  0 0 0  0 0 0  0 arrk 44 0 0
a rrk 56
0 0  0 0 0  0 0 0  0 a rrk 55

            sym arrk 66

0 0  0 0 0  0 0 0  0
a rrk11 = c xk1 , a rrk 15 = c xk 1 s rk 3 , a rrk16 = c xk1 s rk 2 ,
0 0  0 A10 r 0  0 0 0  0 p
a rrk 22 = c xk 2 , a rrk 24 = c xk 2 s rk 3 ,
0 0  0 0 0  0 0 0  0
A10 = a rrk 33 = c xk 3 , a rrk 34 = c xk 3 s rk 2 ,
            (36)
0 a rrk 44 = c xk 4 + c xk 2 srk2 3 + c xk 3 s rk2 2 ,
0  0 0 0  0 0 0  0
r a rrk 55 = c xk 5 + c xk1 s rk2 3 , a rrk 56 = c xk 1 s rk 2 s rk 3 ,
0 0  0 0 0  0 A10 p 0  0
0 0  0 0 0  0 0 0  0 a rrk 66 = c xk 6 + c xk1 s rk2 2 ,

           
0 0  0 0 0  0 0 0  0 a rpk11 0 0 0 a rpk15 a rpk16


0 a rpk 22 0 arpk 24 0 0
or A10 = diag [0,0, ,0, A10 r ,0, ,0, A10 p ,0, ,0] , where 0 0 a rpk 33 a rpk 34 0 0
Arpk = ,
0 a rpk 42 arpk 43 arpk 44 0 0
0 0 0 0 0 0
a
0 0 0 a rpk 55 a rpk 56
0 0 0 0 0 GA rpk 51
a rpk 61 0 0 0 arpk 65 arpk 66
0 0 0 0 GA 0
A10 r = ,
0 0 0 0 0 0 arpk11 = cxk1 , arpk15 = c xk1 s pk 3 , a rpk 16 = c xk 1 s pk 2 ,
0 0 GA 0 0 0
arpk 22 = c xk 2 , arpk 24 = c xk 2 s pk 3 ,
0 GA 0 0 0 0 arpk 33 = c xk 3 , a rpk 34 = c xk 3 s pk 2 ,

A10p can be obtained from the matrix A10r by replacing the arpk 42 = c xk 2 srk 3 , a rpk 43 = c xk 3 srk 2 , arpk 44 = c xk 4 c xk 2 srk 3 s pk 3 c xk 3 srk 2 s pk 2 ,
parameters and dimensions of the r prism with the appropriate arpk 51 = c xk1s rk 3 , arpk 55 = cxk 5 cxk1srk 3 s pk 3 , arpk 56 = cxk 1srk 3 s pk 2 ,
parameters and dimensions of the p prism, a rpk 61 = c xk1 s rk 2 , a rpk 65 = c xk1 s rk 2 s pk 3 , a rpk 66 = c xk 6 c xk1 s rk 2 s pk 2 ,
A prk = Arpk
T
,
A00=Ak+As, (37)
Appk is obtained from the matrix Arrk by replacing indexes r
~
where Ak has a form analogous to the stiffness matrix K k [5] with p,

r p
r p
0 0  0 0 0  0 0 0  0 0 0  0 0 0  0 0 0  0

0 0  0 0 0  0 0 0  0
0 0  0 0 0  0 0 0  0
            
          

0 0  0 0 0  0 0 0  0
0 0  0 0 0  0 0 0  0
0 0  0 Arrk 0  0 Arpk 0  0 r 0 0  0
0  0 Asr 0  0 0 r

0 0  0 0 0  0 0 0  0 (38)
0 0  0 0 0  0 0 0  0
Ak = As =
              
(39)
        
0 0  0 0 0  0 0 0  0 0
0  0 0 0  0 0 0  0
0 0  0 A prk 0  0 A ppk 0  0 p p
0 0  0 0 0  0 Asp 0  0
0 0  0 0 0  0 0 0  0
0 0  0 0 0  0 0 0  0
           
           
0 0  0 0 0  0 0 0  0 0 0  0 0 0  0 0 0  0

102 POLISH MARITIME RESEARCH, No S1/2017

and by replacing the time-derivative operators w/wt and w2/wt2
with s and s2, respectively.
Equation (40) can be transformed into the set of first-order
0 0 0 0 0 0
equations (state equations):
0 0 0 0 0 0
0 0 0 0 0 0
Asr = . ( x , s ) = F ( s ) ( x , s ) + u ( x , s ) (42)
x
0 0 0 0 0 0
0 0 0 0 GA 0

0 0 0 0 0 GA M ( s ) ( 0, s ) + N ( s ) ( l , s ) = ( s ) (43)

Asp is obtained from the matrix Asr by replacing the parameters where:
and dimensions of the r prism with the appropriate parameters
and dimensions of the p prism. ( x , s ) q ( x, s )
The equations derived in the above way were then verified ( x, s ) = 1 = q ( x, s ) ,
2 ( x, s ) x
in three steps. In the first step, continuous parts of Equations
(26)(31) were compared to the appropriate components of the 0 I
F = 1 1 ,
well-known equations describing one-dimensional elements A
20 02 ( A s 2
+ A00 ) A A
20 10
with distributed parameters. As can be seen, the continuous u ( x, s ) 0 M1 M2
part (first line) of Equation (26) is the same as the equation u( x, s ) = 1 = 1 , M = ,
for a longitudinal bar element. In turn, the continuous parts u2 ( x, s) A20 f ( x, s) 0 0

of Equations (27) and (31) are coupled together and are the 0 0 1 ( s )
N = , ( s) = .
same as the Timoshenko beam model equations (vibrating N1 N2 2 ( s)
in the x1-x2 plane). Similarly, Equations (28) and (30) also
describe the Timoshenko beam model, but vibrating in the The solution to Equation (42) can be presented as [12, 18]
perpendicular plane x1-x3. Lastly, the continuous part of
Equation (29) expresses a torsion bar element. x2

In the second step, Equations (26)(31) were written in ( x, s ) = G ( x, , s ) u( , s ) d + H ( x, s ) ( s ) (44)

the matrix form and the obtained stiffness matrix Ak (38) was x1
~
compared to the corresponding matrix K k which describes
the discrete model in the rigid finite element method [5, 13, where:
14]. As it turns out, the discrete parts of Equations (26)(31)
G ( x, , s ) =
are analogous to those in the rigid finite element method [5].
In the last step, easy examples were used and presented in e F ( s ) x [ M ( s )e F ( s ) x1 + N ( s)e F ( s ) x2 ]1 M ( s )e F ( s )( x1 ) , x (45)
= F (s) x
this paper, some of which have analytical solutions and are e [ M ( s )e F ( s ) x1 + N ( s)e F ( s ) x2 ]1 N ( s)e F ( s )( x2 ) , x
thus easily verifiable.
A global model for the whole system is built in the same H ( x, s ) e F ( s ) x [ M ( s )e F ( s ) x1  N ( s )e F ( s ) x2 ]1 (46)
way as the finite element method (FEM) model. Global
matrices A02, A20, A10 include sub-matrices of each prism If we denote:
element, located on a main diagonal. Matrix A00 is the sum of
all the prism element stiffness matrices in the global system.
G ( x, , s ) G12 ( x, , s )
To solve the partial differential equation (32), the G ( x, , s ) = 11 , (47)
distributed transfer function method [15, 16, 18] is used. This G 21 ( x, , s ) G 22 ( x, , s )
method gives semi-analytical results for the tree-dimensional
H ( x, s) H12 ( x, s )
structure. To use it, Equation (32) is to be subjected to Laplace H ( x, s) = 11 , (48)
transformation, after which it takes the form H 21 ( x, s ) H 22 ( x, s )
the solution to Eq. (44) can be written as

( A02 s 2 + A20 + A10 + A00 )q ( x, s) = f ( x, s ) (40)
x 2 x
2 x2 2
with the boundary conditions i ( x, s ) = Gij ( x, , s ) u j ( , s ) d + H ij ( x, s ) j ( s ) , (49)
j =1 x1 j =1

M j q (0, s) + N j q (l , s) = j (s) ,j=1,2, (41)

i=1,2, or
where:
q (x, s ) , f (x, s ) denote the Laplace transformation of the x2 2
corresponding functions q(,t), f(,t), respectively, and M j , N j 1
i ( x, s ) = Gi 2 ( x, , s ) A20 f ( , s )d + H ij ( x, s ) j ( s ) , i=1,2 (50)
j =1
are the operator matrices formed from the matrices M j , N j x1

POLISH MARITIME RESEARCH, No S1/2017 103

where the expression The calculations were performed for two cases. In the
first case the beam was treated as a single prism, description
of which comes down to the coupled two equations of the
G ( x, , s) A20 1

K ( x, , s) = 12 1
(51) Timoshenko beam model, Eqs. (27) and (31), or (28) and (30)
G
22 ( x, , s ) A20 without discrete parts. In the second case the beam is divided
into four prisms (Fig. 4) and four distributed spring elements.
is the matrix of transfer function between the i-th output Each prism has three degrees of freedom: displacements along
(row) and the j-th input (column). In the case of a force the x1 and x2-axes, and angular displacement around the
concentrated at point x0, the relation (50) can be written as x3-axis.
1 2
The stiffness coefficients of the distributed spring layers
i ( x, s ) = Gi 2 ( x, , s) A20
1
f ( , s) ( x0 )d + H ij ( x, s ) j ( s ) , i=1,2 (52) are calculated using the formulas:
0 j =1

and after integrating Gz Gz 3

c x11 = c x13 = c x 41 = c x 43 = , cx15 = cx 45 = ,
2y 2 12 y
2
1
i ( x, s ) = Gi 2 ( x, x0 , s ) A f ( x0 , s ) + H ij ( x, s ) j ( s ) , i=1,2. (53) Gy Ey
20
j =1 c x 21 = c x 31 = c x 22 = c x 32 = , c x 25 = c x 35 = 0 , c x 23 = c x 33 =
2z 2z
Assuming that the boundary conditions (s)=0, from Eq.
(53) we obtain the equation and the fixing coordinates:
1
i ( x, s ) = Gi 2 ( x, x0 , s ) A20 f ( x0 , s ) . y
(54) s1,1, 2 = s3, 4, 2 = s1, 2, 2 = s3, 2, 2 = ,
4
The condition (s)=0 occurs in most cases. It means that y
s2,1, 2 = s 4, 4, 2 = s 2,3, 2 = s 4,3, 2 = ,
the external force does not act on beam supports [12]. 4
z
In the specific case, when the excitation point = 0.1 and s1,1,3 = s 2,1,3 = s1, 2,3 = s 2,3,3 = ,
the response observation point x=0.4 and (s)=0, the transfer 4
z
function of the distributed parameter system is calculated by s 3 , 4 , 3 = s 4 , 4 , 3 = s 3 , 2 , 3 = s 4 , 3, 3 = .
putting corresponding submatrices of the relation (45) into 4
Eq. (51). In this way we obtain:
a) for < x a) y

K < x ( s ) = e F ( s )0.4 [ M ( s )e F ( s )0 + N ( s )e F ( s )1 ]1 M ( s )e F ( s )0.1 A20

1
(55)
z
b) for > x

K < x ( s ) = e F ( s )0.4 [ M ( s )e F ( s )0 + N ( s )e F ( s )1 ]1 M ( s )e F ( s )0.9 A20

1
(56) b) y
2
After putting s=j into the transfer functions (55) and cx1 z
(56), the frequency characteristics can be calculated and the 1 2
2
eigenvalues of the system may be determined.
cx2 cx3

FIRST EXAMPLE OF METHOD 3 cx4 4

APPLICATION
c)
As the first example, let us consider a simple beam fixed
at both ends (Fig. 3) with the following data: E=21011Pa,
G=81011Pa, =8000kg/m3, y=0.25 m, z=0.25 m, l=1m,
v=0.25, =1.
x3 x=0.1
P=1 (force) 1 2

h=0.25 x3
x1
x2
x=0.4 displacement 3 4 x1
l=1 b=0.25

Fig. 3. Fixed beam Fig. 4. Discrete model of beam: a) primary division, b, c) secondary division

104 POLISH MARITIME RESEARCH, No S1/2017

 The matrices of boundary conditions for the system divided the 20 element FEM model. Although the first two frequencies
into four prisms have the form: of the Timoshenko beam model are less accurate, the next
three frequencies are more accurate. But both these models
I 01212 01212 01212 are not accurate enough.
M = 1212 , N = .
01212 01212 I1212 01212 When the equivalent diameter d of the beam is greater
than l/(2 S ), we should use a higher-density mesh or, for
For the discrete-continuous model prepared in the above example, the here proposed discrete-continuous method of
way, the frequency characteristics were calculated (Fig. 5) for modelling.
the harmonic force excitation with amplitude equal to one
and acting at prism No. 2, at point =0.1 m. The displacement
output signal was observed at prism No. 4, at the x=0.4 m SECOND EXAMPLE OF METHOD
point. The obtained characteristics were compared with those APPLICATION
developed by using the finite element method and the ANSYS
system (Fig. 5). The calculated natural frequencies are given As the second illustrative example, let us consider a simply
in Table 1. supported plate along its outline (Fig. 6a) with the following
-100 data: E=21011Pa, G=81010Pa, =81010kg/m3, v=0.25,
discrete-continuous method 1 prism (Timoshenko beam model)
-120
-140
=1, a=1 m, b=0.6 m, h=0.03 m.
discrete-continuous method 4 prisms
-160 Comparative calculations were performed after
magnitude [dB]

-180
-200 implementing the proposed discrete-continuous method in
-220 the MATHEMATICA system and using the finite element
-240
-260 method in the ANSYS system. The calculations were made
-280
-300
FEM - 2x2x5=20 finite elements for different numbers of prisms and finite elements. A sample
FEM - 20x20x80=32000 finite elements
-320
0 0.5 1 1.5 2 2.5 3 3.5 4
discrete model of plate divided into twelve prisms is shown
frequency [rad/s] x 104 in Fig. 6. All prisms are connected via spring layers with
Fig. 5. Frequency characteristics distributed parameters. The stiffness coefficients of these
layers were calculated from the following equations (25):

Table 1. Natural frequencies of the analysed beam [rad/s]

Ez Gz Ez 3 Gz 3
cx12 = , c x13 = , cx14 = , c x15 =
Proposed method FEM model Error with reference to FEM model y y 12y 12y
202080 elem. [%]
No.
225 202080
1 prism 4 prisms
(20 el.) (32000 el.)
1 prism 4 prisms 225 and are the same for spring layers from cx1 to cx11. For the
1 6211.15 6191.7 6164.68 6080.93 2.14 1.8 1.5 supports along the plate outline, the stiffness coefficients
2 13848.36 13568.31 13682.26 13381.3 3.5 1.4 2.2 were calculated using the following formulas: cx02=5 cx12,
3 22763.9 21184.5 22817.3 21901.29 4 3.2 4.1 cx03=5 cx13, cx04=0, cx05=0.
4 32186.47 29587.12 33288.92 30888.13 4.2 4.2 7.7 Each prism has three degrees of freedom, which are:
5 41982.17 38913.2 42763.4 40216.78 4.4 3.2 6.3
translational displacement along the x3-axis and angular
6 47172.88 41321.6 45807.18 43225.17 9.1 4.4 5.9
displacements around the x1 and x2-axes.
a)
x3
The discrete FEM model used as the reference consist of
32000 twenty-node hexahedron elements. x2

The characteristics in Fig. 5 show that the first two

frequencies of the proposed discrete-continuous model,
consisting of 4 prisms, and the reference FEM model with a

32000 elements are very similar and differ by about 1.8

and 1.4% respectively (Table 1). In turn, the corresponding
frequencies of the FEM model consisting of 20 finite elements x1 b
h

differ respectively by about 1.5 and 2.2%.

While the first frequency of the discrete-continuous b)
x3
model is slightly higher than the corresponding frequency
of the FEM model with 20 elements, the second frequency x2
is much lower and closer to the frequency of the reference
FEM model (32000 elements). As the frequency grows, the
discrete-continuous model appears to be more precise than
the 20- element FEM model, although it consists only of
four prisms. lines of
physical division
The model consisting of one prism, which is equivalent to
the Timoshenko beam model, turns out to be comparable with x1 distributed parameter spring element

POLISH MARITIME RESEARCH, No S1/2017 105

c) x3
d) -80
x2 -100 discrete-continuous method 10 prisms
-120

magnitude [dB]
-140
-160
-180
-200
-220
-240 finite element method 1x3x5=15 elements
-260
0 2000 4000 6000 8000 10000 12000 14000 16000 18000
x1
y y/2 z frequency [rad/s]

Fig. 7. Frequency characteristics

d) support stiffness coefficients
Table 2. Comparison of natural frequencies
cx0 cx1 cx2 cx11 cx0 Error in reference
Exact Proposed method FEM to FEM model
1 11 12 Mode frequency (3000elem.)
No. (57) 135 23050 13x5
s112 s212 4 prisms 7 prisms 10 prisms (15 el.) (3000 el.) 10 prisms (15 el.)
1 1692 2130.35 1812.36 1632.08 1847.63 1633.88 0.1 13.08
s112=y/4 s212= -y/2
2 3036 4003.84 3274.93 2916.00 3468.88 2907.732 0.28 19.29
Fig. 6. The analysed plate: a) general scheme, b) primary virtual division, 3 5276 6042.83 5359.35 4996.67 7598.88 5039.429 0.84 50.78
c)secondary physical division, d) equivalent scheme 4 5425 6314.78 5411.67 5121.91 9417.23 5219.819 1.87 80.41
5 6769 8727.93` 7117.69 6417.73 10717.23 6427.07 0.14 66.75
6 8412 9259.70 8263.06 7850.63 14031.61 7987.185 1.7 75.67
The frequency characteristics were calculated for the
7 9009 9305.27 9571.59 8487.07 19156.8 8459.681 0.32 126.44
harmonic force excitation having the amplitude equal to 8 11648 12852.87 10958.05 10754.84 23491.57 11017.57 2.38 113.21
one and acting at the point with coordinates: x1=0.4 m and
x2=0.2 m. The displacement output signal was observed at the The obtained results prove that the proposed method is
same point. The received characteristics were compared with effective and more accurate than the finite element method.
those obtained by using the finite element method and the Figure 7d and Table 2 show that for a smaller number of
ANSYS system, and also with the exact frequencies calculated elements (10 prisms) the presented method gives results
from the relation closer to the accurate values than the finite element method
with a larger number of elements (15 elements). The relative
i 2 j 2 E h3 error of the proposed method with respect to the reference
(i, j ) = 2 + . finite element method model with 3000 elements is smaller
a b 12(1 ) h
2 (57)
than that of the finite element model with a large number of
All frequency values are compared in Table 2. elements (15 elements), Table 2.
a) -80
-100 discrete-continuous method 4 prisms
-120
magnitude [dB]

-140 CONCLUSIONS
-160
-180
-200 The paper presents a hybrid discrete-continuous method
-220
-240 finite element method 2x30x50=3000 elements of modelling. For the proposed approach, general partial
-260
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 differential equations were derived. These equations were
frequency [rad/s] verified and then written in a formalized matrix form, which
is very convenient in application to numerical computations.
b) -80
discrete-continuous method 7 prisms
A beam fixed at both ends and a simply supported plate were
-100
-120 used to illustrate the general concept. The proposed method
magnitude [dB]

-140 is described in a general form and can be also easily applied

-160
-180 to other cases, not presented in this paper, including simply
-200
-220 supported or one side fixed beams, as well as rectangular
-240 finite element method 2x30x50=3000 elements plates with different boundary conditions.
-260
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 The performed numerical calculations and computer
frequency [rad/s] simulations showed that the proposed method, especially in
c) -80 the higher frequency range, gives more accurate results than
-100 discrete-continuous method 10 prisms
-120 the finite element method with the same or similar number
magnitude [dB]

-140 of elements. In addition, even at a small number of elements,

-160
-180 it gives a very good model accuracy and high convergence of
-200
-220
numerical calculation results. The proposed method has been
-240 finite element method 2x30x50=3000 elements proved to be efficient and applicable to discrete-continuous
-260
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 modelling of dynamic systems. It can be applied especially
frequency [rad/s] to the analysis, simulation and numerical calculations of

106 POLISH MARITIME RESEARCH, No S1/2017

nontypical elements or systems requiring high accuracy.
Semi-analytical solutions make the proposed method more
accurate than the finite element method, especially in
predicting high-frequency dynamics.
BIBLIOGRAPHY
1. Hein R., Orlikowski C.: Hybrid reduced model of rotor,
The Archive of Mechanical Engineering, Vol. LX, No 3,
pp. 319-333, 2013.
2. Hein R., Orlikowski C.: Optimum control of gyroscopic
systems, Solid State Phenomena, Vol. 164, pp. 121-126,
2013.
3. Kaliski K. J., Galewski M. A.: A modified method of
vibration surveillance by using the optimal control at
energy performance index. Mechanical Systems and
Signal Processing 58-59 (2015) 41-52.
4. Kaliski K. J., Galewski M. A.: Chatter vibration sur-
veillance by the optimal-linear spindle speed control.
Mechanical Systems and Signal Processing Volume 25,
Issue 1, January 2011, Pages 383399.
5. Kruszewski J., Gawroski W., Wittbrodt E., Najbar F.,
Grabowski S.: Metoda sztywnych elementw skoczonych
[Rigid finite element method], Arkady, Warszawa 1975.
6. Kujawa M., Szymczak C.: Numerical and experimen-
tal investigation of rotational stiffness of zed-purlins
connection with sandwich panels// THIN-WALLED
STRUCTURES. -Vol. 75, (2014), s.43-52.
7. Lipiski K.: Modeling and control of a redundantly actu-
ated variable mass 3RRR planar manipulator controlled
by a model-based feedforward and a model-based-pro-
portional-derivative feedforwardfeedback controller.
Mechatronics 37 (2016), 42-53.
8. Orlikowski C., Hein R.: A simplified model of 3-D pipe
system conveying flowing liquid, Solid State Phenomena,
Vol. 198, pp. 621-626, 2013.
9. Orlikowski C., Hein R.: Modelling of geared multi-rotor
system, Solid State Phenomena, Vol. 198, pp. 669-674,
2013.
10. Orlikowski C., Hein R.: Modelling and analysis of beam/
bar structure by application of bond graphs. Journal of
Theoretical and Applied Mechanics, Vol. 49, No. 4, 2011.
11. Orlikowski C., Hein R.: Modal reduction and analysis
of gyroscopic systems, Solid State Phenomena, Vol. 164,
pp. 189-194, 2010.
12. Park D.-H., Yang B.: Distributed transfer function analysis
of multi-body prismatic elastic solids, Int. J. of Structural
stability and dynamics, Vol. 1, No. 1, pp. 83-104, 2001.
13. Wittbrodt E., Adamiec-Wjcik I., Wojciech S.: Dynam-
ics of Flexible Multibody Systems. Rigid Finite Ele-
ment Method. Foundations of Engineering Mechanics.
Springer, Germany 2006. Pp. 225 (ISBN 3-540-32351-1,
SPIN 11593553).
14. Wittbrodt E., Szczotka M., Maczyski A., Wojciech S.:
Rigid finite element method in analysis of dynamics of
offshore structures. Ocean Engineering & Oceanography
1. Springer-Verlag Berlin Heidelberg, 2013. Pp. 252 (ISBN
978-3-642-29885-1, ISSN 2194-6396).
15. Yang B., Tan C.A.: Transfer functions of one-dimensional
distributed parameter systems, ASME Journal of Applied
Mechanics, Vol. 59, December, pp. 1009 1014 , 1992.
16. Yang B.: Distributed transfer function analysis of complex
distributed parameter systems, ASME Journal of Applied
Mechanics, Vol. 61, pp. 8492, 1994.
17. Yang B., Zhou J.: Semi-analytical solution of 2-D elastic-
ity problems by the strip distributed transfer function
method, Int. J. Solids Structures, Vol. 33, No. 27, pp.
3983-4005,1996.
18. Zhou J., Feng Z.: Transient response analysis of one-
dimensional distributed parameter systems, Int. J. of
Solids and structures, Vol. 36, pp. 2807-2824, 1999.
CONTACT WITH THE AUTHOR
Rafa Hein
e-mail: rahe@pg.gda.pl
Gdansk University of Technology
Narutowicza 11/12, 80-233 Gdansk
Poland
POLISH MARITIME RESEARCH, No S1/2017 107
 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 108-114
10.1515/pomr-2017-0028

THE ANALYZES OF PDOP FACTORS FOR A ZIGBEE GROUND

BASED AUGMENTATION SYSTEMS

Artur Janowski1
Jacek Rapinski 2
1
Institute of Geodesy, Faculty of Geodesy, Geospatial and Civil Engineering University of Warmia and Mazury in Olsztyn Poland
2
Institute of Geodesy, Faculty of Geodesy, Geospatial and Civil Engineering University of Warmia and Mazury in Olsztyn Poland
These authors contributed equally to this work.

ABSTRACT

This paper presents the analysis of PDOP factors for a ZigBee ground based augmentation systems. It presents the idea
of such a system followed by the results of assessment of application in Gdansk Marina. The results of the experiment
show that the application of ZigBee can significantly improve PDOP value in harsh measurement environment. The
analysis shows that it is possible to select an optimal location of ground-based transceiver on the basis of predicted
trajectory and obstructions measured with laser scanning.

Keywords: point cloud processing, GNSS, ZigBee, GBAS, PDOP, navigation in harbor

Harbor infrastructure is an important part of the economy.

INTRODUCTION
The need for localization based services in almost every
place is constantly growing. GNSS systems are widely used
to provide position, velocity and time solutions. The major
drawback of these systems is the lack of satellite signals when
the sky is obstructed [14, 20, 26]. For this reason, ground
based augmentation systems (GBAS) [5] are gaining more
and more attention in the modern world. GBAS systems
are meant to improve the GNSS system performance in the
terms of availability, accuracy and continuity. Dedicated
GBAS systems are widely used to support precision aircraft
approach operations. GBAS can be created in various ways
from differential corrections transmitted using very high
frequency data broadcast to pseudolite systems.
The most crucial part of ship navigation is in the harbor
or in a water channel. Limited space and high obstacles on
the shore are making the GNSS navigation less reliable then
in the open sea (where the requirements for the navigation
are smaller). In order to mitigate the limitations of GNSS in
such places a GBAS system can be introduced.
A seaport could operate freely, if the following criteria are
fulfilled:
access to other kinds of transport,
winds and currents protection,
sufficiently deep water channel,
storage space of goods (warehouses, storage yards, tanks,
siloes, with different strength, internal structure, height
and capacity with a huge influence to the use of harbor
space) [1,2,4,7,9,7,14,16-17]
In this paper we would like to present the impact of a low-
cost, ZigBee based GBAS on the accuracy and availability
of GNSS in a harbor. ZigBee is one of the IEEE 802.15.5
low-rate wireless personal area networks (LR-WPAN) [13,
23]. It is a small, low-power digital radio used mainly for
industrial sensor applications. It is designed for embedded
application with low power consumption (it can operate
up to few years on a single AA battery) and low data rates.
Despite the main purpose of ZigBee devices (which is wireless
communication), some devices are equipped with ranging
capabilities. Successful tests of such a device are presented
in [15, 18, 19] AT233RF device presented in those papers

108 POLISH MARITIME RESEARCH, No S1/2017

can measure a distance with decimeter accuracy in the ~300
meters range. The application of such a device next to the ships
GNSS antenna can significantly improve the availability and
accuracy of positioning in the area covered with the ZigBee
signals. The idea of the ZigBee GBAS system is presented in
Figure 1.
Where:
xi, yi, zi coordinates of i-th satellite,
xu, yu, zu coordinates of receiver
Since ZigBee transmitter does not depend on the GPS time,
an additional ground based transmitter will not improve
TDOP parameter. Therefore, we have focused on the PDOP
only.
A major factor influencing satellites visibility (and PDOP
factors) are obstructions of the sky. Knowing the satellites
almanac and obstructions at a certain place, we can determine
the DOP factors and investigate if an additional ground based
transmitter would improve positioning accuracy. At the same
time, the decision about the optimal (from the PDOP point
of view) localization of the transmitter can be made.
In order to verify if the optimal position of a GBAS
transmitter can be selected on the basis of the previous
assumptions, an experiment was conducted.

EXPERIMENT

The experiment took place in Gdansk (Poland) in the

Fig 1. The idea of ZigBee GBAS
To evaluate the impact of transmitters geometry on the
final position error DOP (Dilution Of Precision) factors are
commonly used [10, 25]. Smaller DOP value indicates better
geometry of satellites [3]. Minimum value of PDOP is about
1.63 for four satellites. If design matrix of a GNSS point
positioning solution is defined as:
Gdansk Marina next to the Szafarnia street (Fig.2). The
difficulty that had to be overcome, was to determine the
objects obstructing GPS signals at certain position and time.
The possibility of using the conventional solutions based on
3D or 2.5D maps were rejected because of the generalization
of the description of the height coordinate especially in the
case of buildings.

where
  

and

Matrix A is defined by:

where
Fig. 2 The Marina Gdansk

, ,

One possible solution is the use of a DEM (digital elevation
and model), but even here the process of generalization can
produce some significant errors from the satellite - receiver
line of sight determination point of view. Therefore, the laser

POLISH MARITIME RESEARCH, No S1/2017 109

scanner measurement results were used. The measurements purpose Database Management System (DBMS) PostgreSQL
were performed using Riegl WZ 400 TLS (Terrestrial Laser with PostGIS extension was adapted [28].
Scanner). The measurement data was obtained in horizontal
coordinate systems PL-2000 zone 6 (EPSG:2177) and vertical
PL-KRON86-NH. The resulting point cloud is depicted in
Figure 3.

Fig. 3 The point cloud view

The study area had about 130370 m of a rectangular

shape. At the beginning of the work all points of elements
that did not have a permanent character in the measured space
(floating or moored vessels, shapes of people, etc.) have been
removed. In this case this task was completed manually, but in
more complex situations one can apply one of the automatic
methods of point clouds reduction [11]. The final result was a
cloud of points containing more than 4,000,000 points. The
assumed test trajectory is shown in Figure 4 . This trajectory
consists of 13 points (numbered from one to thirteen from
Fig. 4 The test area overlapped with point cloud, trajectory and possible
north to south) . For each of these points the line of sight to locations of a ZigBee transceiver
every GPS satellite was tested. Figure 4 shows the test area
overlapped with point cloud, trajectory and possible ZigBee
locations. The numbers next to the trajectory points refers The use of PostGIS gives the possibility to perform
to the number of visible satellites at this point. Two points advanced and optimized spatial analysis and transformations
marked in red (point 12 and 13) are the points with the worst of stored objects between coordinate systems and finding the
constellation of satellites. relations between spatial objects described in a variety of
The analysis of the visibility of the receiver-satellite coordinate systems. The use of PostGIS to find a number of
including obstructions described by the cloud of points points contained in the cylinder mentioned above is trivial.
requires all spatial data to be in the same coordinate system. It is limited to application of st_3ddistnace function which
WGS84 ECEF (Earth Centered Earth Fixed) system (EPSG returns the 3-dimensional Cartesian minimum distance
code: 4328) was selected. With the position of the GNSS between two geometries. A 3D section connecting GPS
receiver, satellites and point cloud in this coordinate system, receiver with consecutive satellites and a set of LiDAR points
the satellite visibility was validated. In the validation process were used as an input for this function. As a result, all of the
only two possible situations were determined satellite is distances from LiDAR points to this section were calculated.
visibility or not. No gradation of the quality of the visibility The selection of obstructions were made on the following
was introduced. In this paper, it is assumed that if there is assumption: if there is less than 5 points with distance smaller
more than 300 TLS points in the cylinder of radius 0.5m than 2.5 m the line of sight exists. Otherwise the satellite is
expanded along the receiver satellite line of sight (LOS) obstructed. Tests were performed for a GPS constellation from
section, the satellite is obstructed. With this basic assumption, 2016-05-20 starting from 18:00 UTC depicted in Figure 5.
for each visibility test it is required to analyze the whole point
cloud ( for each LOS). With millions of points this would
be a very time-consuming task. Therefore the data must be
prepared for such analysis by spatial indexation. For this

110 POLISH MARITIME RESEARCH, No S1/2017

 In addition, the cumulative PDOP were calculated and
depicted. We have defined a cumulative PDOP value as:

Where PDOPi is PDOP for i-th trajectory position and

n is the number of points in the trajectory. This parameter
depicts the average value of PDOP for the entire trajectory.
The resulting heat maps show small deviations of the PDOP
value (1.2 2.4). This is caused by many satellites still visible
on the eastern part of the sky.

Fig. 5 The GPS constellation during analyzed test (The elevation and azimuth
values are given in degrees unit. rad = 180)

RESULTS

The number of satellites and resulting PDOP values for

each trajectory point is depicted in Figure 6.

Fig. 7 The heat map for PDOP values

Fig. 6 The resulting PDOP values for each trajectory point To simulate more difficult conditions, free satellites
(namely PG08, PG27 and PG16) where removed from the
constellation. Resulting PDOPs and numbers of satellite in
The location of the ZigBee transmitter impacts the PDOP view without additional ZigBee are depicted in Figure 8.
value. In order to select the position for which the PDOP is
minimal, a grid of possible Zigbee locations was stretched on
the test area. Spacing of grid points was set to 20m. The height
at which ZigBee transceivers were located was set to 18m
above the sea level. For each possible Zigbee location thirteen
PDOP values were calculated assuming receiver position in
consecutive points of the trajectory. As a result, thirteen heat
maps (created for a possible locations of a single transceiver)
were created. In Figure 7 dots represent the possible Zigbee
locations, while underlying color depicts the PDOP values.
PDOP values were calculate for each possible ZigBee
location (taking obstructions in to account) and heat maps
were interpolated (multiquadric radial basis function)[6]. Fig. 8 The resulting PDOP values for each trajectory point (second scenario)

POLISH MARITIME RESEARCH, No S1/2017 111

 It is clear that when the trajectory is getting closer to wharf,
the number of satellites decreases and PDOP value increases.
Without ZigBee the PDOP value at the end of the trajectory
reaches 35 Figure 8. Results of calculated PDOP values for
reduced number of satellites and simulated ZigBee locations
are depicted in Figure 9.
Fig. 9 The heat map for PDOP values (second scenario)
DISCUSSION
Analyzing Figure 9 one can notice possible transmitter
location for which PDOP value is low for each point of
trajectory (point on the blue background). The common point
with low PDOP for each trajectory point is point number 55.
The differences between PDOP values (at point 55) for the
case with and without ZigBee for each point of the trajectory
are depicted in Figure 10.
Fig. 10 The resulting PDOP values for each trajectory point
(scenarios comparison)
112 POLISH MARITIME RESEARCH, No S1/2017
Improvement of PDOP value is much bigger for the points
at the end of the trajectory where the number of satellites in
view decreases significantly.
CONCLUSIONS
Application of ZigBee based GBAS can improve the
geometry of a positioning solution. The improvement is
significant in the most unfavorable environment. When the
number of satellites in view is satisfactory and their geometry
is good, the application of GBAS does not improve the results.
The biggest advantage of ZigBee GBAS is its low cost and low
power consumption. It is important in the task described in
[11, 22, 24]. ZigBee operates on the 2.4 GHz ISM (Industrial,
Scientific and Medical) band which is available for all users for
free. The communication capabilities of the ZigBee protocol
allows GBAS to operate without prior knowledge of GBAS
transmitter location (its position can be transmitted via the
same device). This feature is a major improvement comparing
to other GBAS systems [11]. The main disadvantage is the
necessity to install additional infrastructure on the wharf
and on the vessel. Multipath signal propagation problem may
occur (which can be mitigated using antenna diversity) but
this phenomena is not the subject of this paper [25]
The presented method can be applied for example in multi-
ship trajectory planning [7,9]
Beside the PDOP factor, there are many other issues that
must be taken into account in such system design [8].
REFERENCES
1. Szymkiewicz,R., Finite-element method for the solu-
tion of the Saint Venant equations in an open channel
network.Journal of hydrology122.1 (1991): 275-287.
2. Tejchman, J., Technical concept to prevent the silo honk-
ing.Powder Technology106.1 (1999): 7-22.
3. Tsui, J.,Fundamentals of global positioning system receiv-
ers. New York, NY, USA:: Wiley-Interscience, 2000.
4. Bielewicz, E., Grski, J.,Shells with random geometric
imperfections simulationbased approach. Inter-
national journal of non-linear mechanics37.4 (2002):
777-784.
5. Hegarty, C.J.,Chatre, E. Evolution of the global naviga-
tion satellitesystem (gnss).Proceedings of the IEEE96.12
(2008): 1902-1917.
6. Kowalczyk, K., Rapiski, J.,Mroz, M., Analysis of vertical
movements modelling through various interpolation
techniques.Acta Geodyn. Geomater7.4 (2010): 160.
7. Szlapczynski, R., Evolutionary sets of safe ship trajec-
tories: a new approach to collision avoidance.Journal
of Navigation64.1 (2011): 169-181.
8. Rapiski, J., Pseudolite location error, Environmental
Engineering (2011), vol 3, 1488-1450.
9. Szlapczynski R.,Szlapczynska,J.,On evolutionary com-
puting in multi-ship trajectory planning.Applied Intel-
ligence37.2 (2012): 155-174.
10. Janowski, A.,Rapinski,J., M-split estimation in laser
scanning data modeling.Journal of the Indian Society
of Remote Sensing41.1 (2013): 15-19.
11. Kaliski, K.J., Buchholz,C., HILS for the design of three-
wheeled mobile platform motion surveillance system
with a use of energy performance index.Solid State
Phenomena. Vol. 198. Trans Tech Publications, 2013.
12. Rzepecka, Z. et al.,Effect of additional distance mea-
surements on satellite positioning.Acta Geodaetica et
Geophysica49.1 (2014): 35-44.
13. Leh,L., ZigBee-based intelligent indoor positioning
system soft computing.Soft Computing18.3 (2014):
443-456.
14. Janowski, A., et al.,Mobile indicators in GIS and GPS
positioning accuracy in cities.International Conference
on Rough Sets and Intelligent Systems Paradigms. Springer
International Publishing, 2014.
15. Janicka,J., Rapinski, J., Filtering the results of ZigBee
distance measurements with RANSAC algorithm, Acta
Geodyn. Geomater., vol. 12, No. 3(179), 2015.
16. Wojcik, M.,Tejchman,J.. Simulation of buckling process
of cylindrical metal silos with flat sheets containing bulk
solids.Thin-Walled Structures93 (2015): 122-136.
17. Grski, J.,et al. Effect of geometric imperfections on
aluminium silo capacities.Stahlbau84.1 (2015): 52-57.
18. Rapiski,J.,The application of ZigBee phase shift mea-
surement in ranging, Acta Geodynamica et Geomate-
rialia, v. 12, No. 291780, 2015.
19. Rapinski, J.,Smieja M., ZigBee Ranging using Phase
Shift Measurements.Journal of Navigation68.04 (2015):
665-677.
20. Nowak, A., Dynamic GNSS Mission Planning Using
DTM for Precise Navigation of Autonomous Vehi-
cles.The Journal of Navigation(2016): 1-22.
21. Bharati,B., et al. GPS C/A Code Multipath Error Estima-
tion for Surveying Applications in Urban Canyon.Micro-
electronics, Electromagnetics and Telecommunications.
Springer India, 2016. 135-142.
22. Kaliski,K. J.,Mazur,M. Optimal control of 2-wheeled
mobile robot at energy performance index.Mechanical
Systems and Signal Processing70 (2016): 373-386.
23. Gharghan, S.K., Rosdiadee,N.,Mahamod,I., A Wire-
less Sensor Network with Soft Computing Localization
Techniques for Track Cycling Applications.Sensors16.8
(2016): 1043.
24. Kaliski,K.J., Mazur,M., Optimal control at energy per-
formance index of the mobile robots following dynami-
cally created trajectories.Mechatronics(2016).
25. Shaoping,L., Xu.Ch., Zhong,R., An Active RFID Tag-
Enabled Locating Approach With Multipath Effect
Elimination in AGV.IEEE Transactions on Automation
Science and Engineering13.3 (2016): 1333-1342.
26. Kaur, J.P.,Kaur,R., Mann,G.P., Performance Analysis of
ZigBee Mesh Networks Under Nodes Failure.Proceed-
ings of the International Conference on Recent Cognizance
in Wireless Communication & Image Processing. Springer
India, 2016..
27. Yunlong,T.,Wang,J., Some remarks on PDOP and
TDOP for multi-GNSS constellations.Journal of Navi-
gation69.01 (2016): 145-155.
28. Chrszcz, A., ukasik,P.,Lupa, M. Analysis of Perfor-
mance and Optimization of Point Cloud Conversion in
Spatial Databases.IOP Conference Series: Earth and
Environmental Science. Vol. 44. No. 5. IOP Publishing,
2016.
POLISH MARITIME RESEARCH, No S1/2017 113
 CONTACT WITH THE AUTHOR

Artur Janowski
e-mail: artur.janowski@geodezja.pl
Institute of Geodesy,
Faculty of Geodesy,
Geospatial and Civil Engineering
University of Warmia and Mazury in Olsztyn
Poland;

Jacek Rapinski
e-mail: jacek.rapinski@uwm.edu.pl
Institute of Geodesy,
Faculty of Geodesy,
Geospatial and Civil Engineering
University of Warmia and Mazury in Olsztyn
Poland

114 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 115-124
10.1515/pomr-2017-0029

MODELLING AND SIMULATION OF A NEW VARIABLE STIFFNESS

HOLDER FOR MILLING OF FLEXIBLE DETAILS

Krzysztof J. Kaliski*, Marek A. Galewski*, Micha Mazur *, Marek Chodnicki*

*
Faculty of Mechanical Engineering, Gdask University of Technology, Poland

ABSTRACT

Modern industry expectations in terms of milling operations often demand the milling of the flexible details by using
slender ball-end tools. This is a difficult task because of possible vibration occurrence. Due to existence of certain
conditions (small depths of cutting, regeneration phenomena), cutting process may become unstable and self-excited
chatter vibration may appear. Frequency of the chatter vibration is close to dominant natural frequency of the workpiece
or the tool. One of the methods of chatter vibration avoidance is matching the spindle speed to the optimum phase
shift between subsequent cutting edges passes (i.e. the Liao-Young condition). However, the set of optimum spindle
speeds from the point of view of vibration reduction may be not optimum one from other points of view. For example
milling efficiency or machine tool capabilities cannot be assured. This article presents the idea of a workpiece holder
with adjustable stiffness and discusses a new variant of its realization. In the holder, milling process is performed at
constant spindle speed and feed speed. In order to avoid vibration the holder stiffness is modified. Stiffness changes
modify natural frequencies of the workpiece and thus it is possible to modify dynamic properties of the workpiece in such
a way that arbitrary chosen, constant spindle speed will be optimum, due to the Liao-Young condition performance.
Calculation of the optimum stiffness is performed before milling, based on the workpieces modal identification results
and the finite element model simulations.

Keywords: milling, vibration reduction, optimum spindle speed, optimum stiffness

and, in extreme cases, may lead to the destruction of a tool

INTRODUCTION
Modern industry expectations in terms of milling
operations often demand the milling of the flexible details
(i.e. offshore structures, rotor blades, thin-walled elements)
by using slender ball-end tools. This is demanding task
because it is usually required to achieve surface quality high
enough to treat milling operation final without the required
additional finishing operations. Milling flexible details is also
a challenging problem because such objects may be prone to
vibration especially when the workpiece is difficult to support
or cannot be supported during milling processes. In such
case, tool-workpiece relative vibration plays a principal role
because, in certain conditions, it may lead to loss of stability
and generation of self-excited chatter vibration (Tomkw,
1997). Additionally, vibrations in a machine tool system do
not only reduce surface quality but also increase tool wear
or a workpiece (Nouari et al., 2003).
The regeneration phenomenon is recognised the most
important reason for chatter vibration occurrence (Quintana
and Ciurana, 2011). There are many different methods for
reduction and surveillance of the chatter vibration, i.e.
the use of cutting edge chamfers (Sellmeier and Denkena,
2012), using mechanical dampers (Kim et al., 2006) or smart
materials (Rashid, 2005), robust optimum control (Moradi
et al., 2013), active methods (i.e. active structural control
(Dohner et al., 2004), active holder (Brecher et al., 2010), active
damping (Ganguli et al., 2005; Parus et al., 2009; Kaliski
and Galewski, 2014)), cutting with variable spindle speed
(Soliman and Ismail, 1997; Kaliski and Galewski, 2011).
Another method depends on matching the spindle speed
to the optimum phase shift between subsequent passes of the
tool cutting edges , which was proposed by Liao and Young
(Liao and Young, 1996). This optimum spindle speed can be

POLISH MARITIME RESEARCH, No S1/2017 115

calculated by using the generalised Liao-Young condition
(Kaliski and Galewski, 2015):
zn f pivot joint (Fig. 1).
= , m = 0, 1, 2, , (1)
60 0.25 + m
where:
f workpiece natural frequency of no. [Hz],
n requested spindle speed [rev/min],
z number of cutting edges of the milling tool.
The only unknown element in Eq. 1 is dominant natural
frequency f which may be identified, for example, by
performing the modal test. As the result, a set of optimum
spindle speeds is obtained. From this set, one speed is selected
for performance of milling operation. However, the optimum
spindle speed from the point of view of the Liao-Young
condition, may not be optimum from the point of view of
the other requirements, i.e. cutting process productivity.
For example, if f = 166,6 Hz and z = 2, the set of optimum
spindle speeds is n <2000, 4000, 2222, 1538, ...> [rev/
min]. A large gap between the first and the second speed
can be observed. This may lead to non-optimum utilization
of amilling centre potential. For example, when milling
centre can achieve speeds only up to 18000 rev/min, the
highest optimum spindle speed which can be selected from
the calculated set, is 4000 rev/min.
Moreover, complex-shaped workpieces may have different
dominant natural frequencies for different workpiece zones.
Thus, different spindle speeds are optimum for various
workpiece areas. This leads to the idea of creating the map
of optimum spindle speeds, which is described in details in
(Kaliski et al., 2010; Kaliski et al., 2014a; Kaliski et al.,
2013) together with some examples of successful application.
However, changing spindle speed during tool pass may
reduce surface quality in speed change point and is difficult
to perform in some milling centres due to the properties
of control units. This concerns most of all milling centres
equipped with controllers which do not allow speed changes
without stopping the tool or the workpiece feed. This problem
can be avoided by milling a whole tool pass with only one
spindle speed. However, this also means that in some zones
this speed will not be optimum. Unfortunately, the map
of optimum spindle speeds does not solve the problem of
incongruity of optimum spindle speeds set and accessible
speed range of available machine tool.
A CONCEPT OF A NEW VARIABLE
STIFFNESS HOLDER
Optimal spindle speed depends on dominant natural
frequency of the workpiece (Eq. 1). The frequency depends
on dynamic properties of workpieces. During milling, the
workpiece is mounted in a holder which is usually considered
a rigid support. In one of the previous approaches the authors
proposed to introduce a holder with one degree of freedom
(DOF) and adjustable stiffness. The holder had a form of
116 POLISH MARITIME RESEARCH, No S1/2017
1- DOF vertical column (Kaliski et al., 2014b; Kaliski et
al., 2014c). Purpose of this paper is to investigate another
concept of the holder whose behaviour is based on 1- DOF
Fig. 1. Workpiece holder with adjustable stiffness and 1- DOF pivot joint and
the FEM model of the workpiece
Thanks to the adjustable stiffness it is possible to modify
dynamic properties of the whole system (consisting the holder
and the workpiece) and to modify its natural frequency. This
enables a possibility to reverse the current procedure of
optimum spindle speed selection. Instead of calculating
optimum spindle speed based on natural frequency it is
possible to arbitrary set spindle speed and tune natural
frequency in such a way that chosen spindle speed is optimum
one, in accordance with Eq.1. For example, choosing n =
10000 rev/min implies that natural frequency should be one
of the set : f <83, 416, 750, ...> [Hz]. This is achieved by
suitable adjusting holder stiffness.
HOLDER, WORKPIECE AND CUTTING
PROCESS MODELS
Determination of relationships between instantaneous
cutting layer geometry and the resultant cutting forces
became a subject of various scientific research. For example,
the latter concerned identification of cutting forces for end
milling (Hoffmann et al., 2010) and multi-edge milling (Pajor
et al., 2011), or experimental determination of nonlinear
cutting process parameters in turning (Powaka et al., 2009).
For purpose of the present consideration, cutting process
dynamics can be described by using the updated proportional
model (Kaliski and Galewski, 2015). Based on this model, the
cutting force components for edge no. l depend proportionally
on the instantaneous cutting layer thickness hl(t) and also
the instantaneous depth of cutting al(t); both of them vary in
time. It is assumed that the resultant cutting force lies in the
orthogonal plane. In accordance with directions of the action,
we separate the cutting force component Fyl1 acting along
nominal cutting velocity, and the cutting force component
Fyl2 acting along cutting layer thickness (Fig. 2). The third
cutting force component Fyl3 is neglected.
k a (t )hl (t ), al (t ) > 0 hl (t ) > 0,
Fyl1 (t ) = dl l
0 , al (t ) 0 hl (t ) 0,
k a (t )hl (t ), al (t ) > 0 hl (t ) > 0,
Fyl1 (t ) = l dl l
0 , al (t ) 0 hl (t ) 0,
the effect of first pass of the edge along cutting layer causes
proportional feedback, and the effect of multiple passes
causes delayed feedback, additionally.
Relationships (2)...(4) describe cutting forces of CE no.l
(2) in case of the proportional model. They can be presented
by using matrix notation. Hence, if we neglect nonlinear
geometry change components in general matrix description
(Kaliski and Galewski, 2015), appropriate vector of forces
(3) of CE no. l will have the following form:

Fyl1 kdl a pl (t )hDl (t ) 0 kdl a pl (t ) kdl hDl (t )

Fyl3(t) = 0,
(4)
F = k a (t
yl 2 l dl pl Dl )h (t ) 0 k a (t ) k h (t )
l dl pl l dl Dl
F 0 0 0 0
where: a (t ) = a (t ) a (t ), h (t ) = h (t ) h (t ) + h (t ),
yl 3
  
   

 
l pl pl l Dl l l l 
Fl (t ) Fl0 (t ) D Pl (t ) (5)

qzl (t ) 0 kdl a pl (t ) 0 qzl (t l )
apl(t) desired depth of cutting, apl(t) dynamic change
hl (t ) + 0 l kdl a pl (t ) 0 hl (t l ) ,
in depth of cutting, hDl(t) desired cutting layer thickness;
a pl (t ) 0 0 0 a pl (t l )
hl(.) dynamic change in cutting layer thickness, kd average 

 



   
 
 
dynamic specific cutting pressure, l cutting force ratio w l (t ) DOl (t ) w l (t l )

(quotient of forces Fyl2 and Fyl1), t time-delay between the

same position of subsequent tool cutting edges, f feed per or, with the use of abbreviated notation:
edge, l(t ) current immersion angle of edge no. l.
     
Fl (t ) = Fl0 (t ) DPl (t )wl (t ) + DOl (t ) wl (t l ) , ,(6)

 0
where: Fl (t ) vector of cutting forces of CE no. l, Fl (t )
vector of cutting forces of CE no. l, resulted
 from the desired
cutting geometry and kinematics, DPl (t ) matrix of linear

proportional feedback interactions, DOl (t ) matrix of

linear time-delayed feedback interactions, wl (t ) vector

ofdeflections of CE no. l at instant of time t, wl (t l ) vector
of deflections of CE no. l at instant of time t l, qzl(t) relative
displacement of edge and workpiece along direction yl1, at
instant of time t, qzl(t l) relative displacement of edge
and workpiece along direction yl1, at instant of time t l.
Vector (6) can be also described in six-dimensional space, i.e.:

Fl (t ) = Fl0 (t ) DPl (t ) wl (t ) + DOl (t ) wl (t l ) , (7)

where:

(
Fl (t ) = col Fl (t ), 031 ,) (8)


Fig. 2. Scheme of a slender ball-end milling of one-side-supported flexible w l (.) = col (w l (.), 031 ) , (9)
workpiece in a 1- DOF pivot joint

Dynamic analysis of a slender ball- end milling process (


)
Fl0 (t ) = col Fl0 (t ), 031 , (10)
has been performed, based upon the following assumptions
(Kaliski and Galewski, 2015; Kaliski, 2012): 
D (t ) 0 ,
the spindle together with the tool fixed in the holder, D Pl (t ) = Pl (11)
and the table with the workpiece, are separated from the 0 0 66

machine tool structure; 

D (t ) 0 .
only flexibility of the tool and flexibility of the workpiece D Ol (t ) = Ol (12)
are considered; 0 0 66

coupling elements (CEs) are applied for modelling the

cutting process dynamics;

POLISH MARITIME RESEARCH, No S1/2017 117

 Cutting forces (5) are generated on the basis of the by time-dependent constraint equations (Kaliski and
CEs behaviour. The CEs are attached to appropriate Galewski, 2015; Kaliski, 2012).
nodes lying on the workpiece model surface, and to the Vector of deflections of CE no. l is expressed as a function
top of the bar element representing rotating tool (Fig. 1 of vector of the generalised coordinates q and vector of
and 2). Density of the used workpiece mesh is generally the modal coordinates a. Hence we obtain the following
too small for representing the cutting process properly. relationship (Kaliski and Galewski, 2015; Kaliski, 2012):
Thus, geometric positions of the nearest surface nodes
with respect to actual ball- end mill top position, for time w l (t ) = [Tl (t ) Wl (t )] (13)
instant t, were taken into account. Subsequently, workpiece
vibration along direction perpendicular to its surface, at the where:
current ball- end mill top position, has been determined.
Dynamic change in the cutting layer thickness hl(t) was Tl (t ) = l (t )S l (t ) (14)
calculated by taking into account the slender ball -end mill
vibration along yl2 direction of CE no. l. The reason was transformation matrix of the displacements vector q from
that vibration of the workpiece in its horizontal plane, was the coordinate system xe1, xe2, xe3 of the tool FE model to
neglected. Dynamic change in the depth of cutting apl(t) the coordinate system yl1, yl2, yl3 of CE no. l (Kaliski and
was calculated as relative displacement of the tool along yl3 Galewski, 2015; Kaliski, 2012),
direction and the workpiece along the direction perpendicular

to the machined surface, at top position of ball end mill. l (t ) 0
Influence of the spindle speed on dynamics of the ball- end l (t ) =  (15)
mill and on the other effects, was not included. 0 l (t )
As a result of milling process modelling, a hybrid system
which consists of separated subsystems, is obtained (Fig. 2). It matrix of cosines of angles between the axes : yl1, yl2, yl3
consists of finite element method (FEM) model of workpiece of CE no. l, and xe1, xe2, xe3 of the tool FE model, whereas:
and holder, i.e. stationary model of the one-side-supported
flexible workpiece and the adjustable stiffness holder. The
subsystem moves with the desired feed speed v f. At the cos l t sin l t 0

beginning the workpiece is idealised as a set of isoparametric l (t ) = sin l (t ) cos l (t ) 0 (16)
8-nodes finite elements (Kaliski, 2012). After making the 0 0 1
modal transformation (Kaliski, 2012), behaviour of this
subsystem is described by a vector of its modal coordinates
a. Hence when we consider a finite number of normal modes
mod of the subsystem, we shall define dynamic properties of
it with the use of:
= diag(0i) matrix of angular natural frequencies
of the modal subsystem; i =1, , mod;
= |1 ... mod| matrix of the considered mass normalised
normal modes of the modal subsystem;
i = 1, , mod;
S l t  matrix of coordinates of attachment of CE no. l to the
tool FE model, in the coordinate system xe1, xe2, xe3 (Kaliski
and Galewski, 2015; Kaliski, 2012);
Wl(t) matrix of constraints between the modal displacements
vector a of the workpiece and holder model, and displacements
in the coordinate system yl1, yl2, yl3 of CE no. l (Kaliski and
Galewski, 2015; Kaliski, 2012);
Z diag ] i  matrix of dimensionless damping q
= vector of hybrid coordinates of the whole system.
coefficients of the modal subsystem; a
i = 1, , mod; After transformation of the vector of force interactions of
tool FE model, i.e. non-stationary discrete model of ball- CE no. l (7) to hybrid coordinates, we obtained:
end mill (the Euler-Bernoulli bar having local immovable
coordinate system xe1, xe2, xe3) and the cutting process (i.e.
coupling element (CE) no. l, placed in the instantaneous TlT (t ) TlT (t ) 0
F (t ) = Fl (t ) +
Wl (t ) Wl (t )
l
position of the active cutting edge (Kaliski and T T

Galewski, 2015; Kaliski, 2012)). Cutting edges are

TT (t )D (t )Tl (t ) TlT (t )D Pl (t )Wl (t ) q s
considered active when they have contact with the + l T Pl + (17)
Wl D Pl (t )Tl (t ) Wl (t )D Pl (t )Wl (t ) a m
T
workpiece, while the remaining ones are called inactive.
Behaviour of the subsystem is described by a vector of TT (t )D (t )
+ l T Ol w l (t l ).
Wl (t )DOl (t )
its generalised coordinates q. Dynamic properties of the
Euler-Bernoulli bar are defined by the matrices of inertia,
M, damping , L , and stiffness, K , (Kaliski, 2012); As a result of the hybrid systems consideration, the matrix
abstractive connective subsystem placed in the conventional equation of dynamics of non-stationary model of the milling
contact point S between the tool and the workpiece. Its process in hybrid coordinates obtains the following form
generalised coordinates are related to the other ones (Kaliski and Galewski, 2015; Kaliski, 2012):

118 POLISH MARITIME RESEARCH, No S1/2017

 SIMULATIONS
M 0 L 0 
0 I + 0 2Z + For the purpose of computer simulations of the milling

il il process the holder with the workpiece and slender ball- end

K + Tl (t )D Pl (t )Tl (t ) TlT (t )D Pl (t )Wl (t )
T
mill are idealised with the use of the FEM. Dynamic FE
+ il l = 1 l = 1
il
= (18) simulation is usual practice for acquiring knowledge about


l W T
(t )D Pl (t )Tl (t ) 2
+ Wl
T
(t )D Pl (t )Wl (t ) behaviour of the idealised real ? structures (Iwicki et al., 2014),
l =1 l =1 or performance of real machining processes (Siemitkowski
il T and Przybylski, 2006). Rotating part of the workpiece holder
Tl (t )Fl (t ) + Tl (t )DOl (t )w (t l )
0 T

= li=l 1 is considered to be lumped mass supported by linear spring

W ( )F ( ) + W ( )D ( )w ( ) (Fig. 1). Appropriate degrees of freedom of the workpiece are
l
T 0 T
t l t l t Ol t t l
l =1 fixed with the use of both Single Point Constraints and Multi
Point Constraints, under assumption that deformability of
where il is the number of active coupling elements. the workpiece holder body was neglected. Slender ball-end
In the particular case (Kaliski, 2012) when axis of the mill is idealised by the Euler-Bernoulli bar. Its dimensions
milling tool is perpendicular to the milled surface, the matrix (i.e. length of 123 mm, diameter of 16 mm) and material
of constraints , Wl(t), does not depend on current position of properties are so selected as to represent dynamic properties
CE no. l. Thus we obtain: of real ball-end mills that the authors have utilised before
(i.e. having double bending mode at natural frequency of
Wl (t ) = W(t ) = CW W(t ), (19) 743 Hz). Discrete model of the workpiece ( of EN AW-6101A
aluminium alloy material, mass density of 2.7103 kg/m3,
where: Young modulus of 71010 Pa) consists of about 2800 eight-
node isoparametric finite elements. The mesh resolution is
CW = col (0 0 1 0 0 0 ), (20) about 2.5 mm. The workpiece dimensions are 160 mm
50 mm 5 mm. The FEM model of the workpiece was well
Wt  matrix of constraints between the modal displacements correlated before with the real workpiece installed in a rigid
a, and vertical displacements of the workpiece, at current workpiece holder, with the use of the experimental modal
position of the conventional contact point S. analysis techniques.
Now the matrix of Eq. 18 is simplified to following form: Even if the size of the model is not unexceptionally large, the
number of required computer simulations makes it unusable
in case of direct time-integration methods. Moreover direct
M 0 L 0 
0 I + 0 2Z + time computation requires different damping models which
could be at least partially obtained by means of the commonly
il il
used experimental modal analysis techniques (uczak et al.,
+
K + Tl
T
(t )D Pl (t )Tl (t ) TlT (t )D Pl (t )W (t )
= (21) 2014). Thus it is more convenient to use modal model instead
l =1 l =1
0 2
and modal time synthesis method for the simulation of the
cutting process.
il T
Tl (t )Fl0 (t ) + TlT (t )D Ol (t )w (t l )
=
Basing upon former authors experience the frequency
l =1 range for analysis is limited to 1000 Hz. Within such frequency
0 range four normal modes can be observed, i.e. two modes of
On the basis of the above presented considerations, workpiece holder with the workpiece and double mode of the
it is possible to show that in such particular case also slender ball-end mill (whose natural frequency is over 700
instantaneous change in depth of cutting does not depend Hz). Natural frequencies of the holder with the workpiece, for
on current position of CE no. l. That is to say: different workpiece holder spring settings, are presented in
Tab. 1. Both of the modes for the holder with the workpiece
a pl (t ) = a p (t ) = W(t )a . (22) are presented in Fig. 3 and 4 at the spring stiffness setting
of 5600 N/mm. These modes do not change so much due to
different spring stiffness settings of the workpiece holder.
However both natural frequencies change due to adjustment
of the spring stiffness. It was also noticed that both normal
modes of the holder with the workpiece are well coupled
with the workpiece. Any local modes phenomena were not
observed in this case. If we consider the latter, as well as
effective mass properties of both modes, the first natural
frequency can be expected to be a good approximation of
the chatter frequency (see Eq. 1).

POLISH MARITIME RESEARCH, No S1/2017 119

Tab. 1. Natural frequencies of two first modes of variable stiffness holder
with workpiece A series of simulations of the milling process performance
was done on the basis of the determined matrix equation of
Spring stiffness 1st natural freq. 2nd natural freq. dynamics (Eq. 21), and the adopted direct step by step-
[N/mm] [Hz] [Hz]
integration Newmark method (Kaliski, 2001). Although it
14800 138.62 468.67
is possible to so adjust integration parameters as to make the
11000 131.36 427.48
method unconditionally stable, some calculation problems
8500 124.08 398.47 may occur when cutting edges lose contact with the workpiece
6800 117.16 377.87 (i.e. due to dynamic change in the cutting layer thickness
5600 110.78 362.93 hl(t) caused by high level of the observed vibrations).
4700 104.83 351.54 Therefore it is important to synchronize well integration
4000 99.27 342.60
time steps with the time delay l between the same positions
of subsequent tool cutting edges. In case of a fully developed
3470 94.36 335.80
chatter vibration the obtained results can be only used as an
3050 89.94 330.39 approximation of the milling process and thus conclusion
2670 85.43 325.50 in such case could only be whether the observed process is
unstable [or stable?].
The results of simulation of the workpiece vertical vibration
are obtained for a desired range of spindle speeds and adjusted
values of the holder stiffness. The standard deviation values
of displacements (Tab. 2) and relevant chatter amplitudes
(Tab. 3) are determined. Standard deviation is used as
ageneral index of vibration level and, in fact, represents the
RMS value of displacements but calculated with respect to
non-zero mean displacement value. It is worth noticing that
application of the adjustable stiffness holder reflects good
performance in scope of chatter reduction within the useful
range of spindle speeds from 13000 to 16700 rev/min. In
majority of simulations expected optimum combinations of
spindle speed-holder spring stiffness are obtained (Tab. 3).
Fig. 3. First normal mode (110.78 Hz) of the stiffness holder with workpiece In the case of simulation at the spindle speed n=15745
atspring stiffness of 5600 N/mm rev/min and the optimum holder stiffness k s=11000 N/mm,
vertical displacements of the workpiece vibration (Fig.5a)
and their amplitude spectra (Fig. 5b) are presented. It is
clearly visible that vibration level was low and the chatter
did not occur (there is no spectrum peak in vicinity of the
workpiece natural frequency). In this case vibration level
slightly exceeded 0.01 mm, which is about 10% of the nominal
depth of cutting.

Fig. 4. Second normal mode (362.93 Hz) of the stiffness holder with workpiece
at spring stiffness of 5600 N/mm

120 POLISH MARITIME RESEARCH, No S1/2017

Tab. 2. Standard deviation of displacements [mm] ( expected optimum pairs are distinguished with grey background )

Holder spring stiffness [N/mm]

Spindle speed
[rev/min]
14800 11000 8500 6800 5600 4700 4000
17651 0.002366
16745 0.007657
16651 0.002539 0.005380
15869 0.002711
15745 0.003676 0.002613 0.002671 0.002523 0.002193
15651 0.003866
15284 0.002547
15047 0.002860
14869 0.004322 0.002926 0.002787 0.002927
14745 0.004957
14581 0.002810
14284 0.003170
14047 0.006864 0.003431 0.003163
13869 0.008736
13784 0.003831
13581 0.003441
13284 0.008378 0.004034 0.004537 0.003627
13047 0.010622
12581 0.015531 0.020466 0.031396
12284 0.024176
11932 0.059660 0.059596
11581 0.078220
st
Tab. 3. Vibration amplitude of the 1 mode [mm] (expected optimum pairs are distinguished with grey background and bold numbers)

Holder spring stiffness [N/mm]

Spindle speed
[rev/min]
14800 11000 8500 6800 5600 4700 4000
17651 0.001584
16745 0.007286
16651 0.000310 0.004502
15869 0.001615
15745 0.001045 0.000317 0.001557 0.001632 0.000666
15651 0.001282
15284 0.000942
15047 0.001510
14869 0.002834 0.000472 0.000848 0.001501
14745 0.003316
14581 0.001212
14284 0.001103
14047 0.004760 0.000747 0.005021
13869 0.006904
13784 0.001447
13581 0.000357
13284 0.006728 0.000521 0.002755 0.002057
13047 0.008114
12581 0.005560 0.021472 0.033184
12284 0.029790
11932 0.074891 0.073382
11581 0.082545

POLISH MARITIME RESEARCH, No S1/2017 121

 Fig 5. Displacement , (a) , and its spectrum , (b) , for the optimum pair of the spindle speed n=15745 rev/min and the holder stiffness of 11000 N/mm

Fig 6. Displacement , (a) , and its spectrum , (b) , for the non-optimum pair of the spindle speed n=14745 rev/min and the holder stiffness of 11000 N/mm

Fig 7. Displacement , (a) , and its spectrum , (b) , for the non-optimum pair of the spindle speed n=16745 rev/min and the holder stiffness of 11000 N/mm

Fig 8. Displacement , (a) , and its spectrum , (b) , for the non-optimum pair of the spindle speed n=15745 rev/min and the holder stiffness of 8500 N/mm

Fig 9. Displacement , (a) , and its spectrum , (b) , for the non-optimum pair of the spindle speed n=15745 rev/min and the holder stiffness of 14800 N/mm

122 POLISH MARITIME RESEARCH, No S1/2017

 In the relevant amplitude spectrum only one harmonics
concerning frequency of entering cutting edges into the
workpiece (i.e. that of 525 Hz), is observed. In the case of
non-optimum spindle speeds, vibration level approaches
either over 0.02 mm (Fig. 6a, at lower speed) or even over
0.03 mm (Fig. 7a, at higher speed). The relevant frequency
plots (Fig. 6b and 7b) illustrate, except harmonics concerning
appropriate spindle speeds of 14745 and 16745 rev/min,
appearance of strong chatter amplitudes whose frequency
is close to the first natural frequency equal to 131.36 Hz of the
holder with the workpiece (Tab. 1). Additionally, change of
the holder stiffness, with respect to its optimum value, causes
a noticeable increase in vibration level, as well as significant
chatter amplitudes (Fig. 8 and 9).
CONCLUSIONS
Results of the simulations showed that to modify the
workpiece dynamic properties is possible with the use of
the proposed new workpiece holder. The simulations for
different pairs of holder stiffness and spindle speed show that
vibrations are the lowest only in case of a proper optimum
combination of the two parameters. These results mean that
the proposed idea of a variable stiffness holder has a great
potential. In case of a standard perfectly rigid holder, a method
for vibration reduction was to choose the proper spindle
speed on the basis of the Liao-Young condition. However the
determined optimum spindle speed might not be optimum
from other points of view (i.e. technological ones). The
proposed variable stiffness holder allows to adjust workpieces
dynamic properties in such a way that an arbitrarily given
spindle speed should accomplish the generalised Liao-Young
condition for the desired operational spindle speeds of the
milling machine. However vibration surveillance efficiency
is really evident only in case of some values of spindle speeds.
Hence the proposed approach should be subjected to further
investigations. Despite the latter , the obtained results may
serve as a basis for the optimum performance planning of
machining processes (Deja and Siemitkowski, 2013).
REFERENCES
1. Brecher C., Manoharan D., Ladra U., Kpken H.-G.:
Chatter suppression with an active workpiece holder.
Production Engineering Research and Development,
Vol. 4 (2010), pp.239-245.
2. Deja M., Siemitkowski M. S.: Feature-based generation of
machining process plans for optimised parts manufacture,
, August 2013, Vol. 24, Issue 4, pp. 831846.
3. Dohner J. L., Lauffer J. P., Hinnerichs T. D., Shankar
N., Regelbrugge M., Kwan C.-M, Xu R., Winterbauer B.,
Bridgerf K. : Mitigation of chatter instabilities in milling by
active structural control. Journal of Sound and Vibration,
Vol. 269(2004), pp. 197211.
4. Ganguli A., Deraemaekar A., Horodinca M., Preumont
A.: Active damping of chatter in machine tools - dem-
onstration with a Hardware in the Loop Simulator.
Proceedings of the Institution of Mechanical Engineers,
Part I: Journal of System Control Engineering, Vol. 219
(2005), pp. 359-369.
5. Hoffmann M., Powaka B., Berczyski S., Pajor M.: Iden-
tification of cutting forces in frequency domain in milling.
Advances in Manufacturing Science and Technology,
Vol. 34 (1) (2010), pp. 5-20.
6. Iwicki P., Tejchman A., Chrcielewski J. : Dynamic FE
simulations of buckling process in thin-walled cylindrical
metal silos. Thin-Walled Structures, (2014), pp. 344-359.
7. Kaliski K. J.: Vibration surveillance of mechanical sys-
tems which are idealised discretely (in Polish). Gdask
University of Technology Publishers, Gdask 2001.
8. Kaliski K. J. A surveillance of dynamic processes in
mechanical systems (in Polish). Gdask University of
Technology Publishers, Gdask 2012.
9. Kaliski K. J., Chodnicki M., Mazur M., Galewski M.A.:
Vibration surveillance system with variable stiffness holder
for flexible details milling.[in:] Applied Non-Linear
Dynamical Systems, Springer Proceedings in Math-
ematics and Statistics, J. Awrejcewicz, Vol. 93(2014b),
pp. 175-184.
10. Kaliski K. J., Chodnicki M., Galewski M. A., Mazur M.
: Vibration Surveillance for Efficient Milling of Flexible
Details in Adjustable Stiffness Holder.Vibroengineering
Procedia, Vol. 3 (2014c), pp. 215-218.
11. Kaliski K. J., Galewski M. A. : Chatter vibration sur-
veillance by the optimal-linear spindle speed control.
Mechanical Systems and Signal Processing, Vol. 25 (2011),
pp. 383-399.
12. Kaliski K. J., Galewski M. A.: Vibration surveillance
supported by Hardware-In-the-Loop Simulation in mill-
ing flexible workpieces. Mechatronics, Vol. 24(2014), pp.
1071-1082.
13. Kaliski K. J., Galewski M. A. : Optimal Spindle Speed
Determination for Vibration Reduction During Ball-
End Milling of Flexible Details. International Journal of
Machine Tools and Manufacture, Vol. 92 (2015), pp. 19-30.
14. Kaliski K. J., Galewski M. A., Mazur M.: High Speed
Milling vibration surveillance with optimal spindle speed
POLISH MARITIME RESEARCH, No S1/2017 123
 based on optimal speeds map. Key Engineering Materials,
Vol. 597(2014a), pp.125-130.
15. Kaliski K. J., Mazur M., Galewski M. A. : High speed
milling vibration surveillance with the use of the map
of optimal spindle speeds. [in:] Proceedings of the 8th
International Conference on High Speed Machining,
Metz, France , (2010), pp. 300-305.
26. Rashid M.K. : Simulation Study on the Improvements of
Machining Accuracy by Using Smart Materials. Robotics
and Computer-Integrated Manufacturing, Vol. 21(2005),
pp. 249257.
27. Sellmeier V., Denkena B.: High speed process damping
in milling. CIRP Journal of Manufacturing Science and
Technology, Vol. 5 (2012) , pp. 8-19.

16. Kaliski K. J., Mazur M., Galewski M. A.: The Optimal 28. Siemitkowski M., Przybylski W. : Simulation studies of
Spindle Speed Map for Reduction of Chatter Vibration process flow with in-line part inspection in machining cells.
During Milling of Bow Thruster Blade. Solid State Phe- Journal of Materials Processing Technology , January 10,
nomena, Vol. 198 (2013), pp. 686-691. 2006, Vol. 171, Issue 1, pp. 27-34.

17. Kim M.H., Won D., Ziegert J. : Numerical Analysis and
Parameter Study of a Mechanical Damper for Use in Long
Slender Endmills. International Journal of Machine Tools
and Manufacture, Vol. 46 (2006), pp. 500507.
18. Liao Y.S., Young Y.C.: A new on-line spindle speed regula-
tion strategy for chatter control. International Journal
of Machine Tools and Manufacture, Vol. 36 (1996), pp.
651-660.
29. Soliman E., Ismail F. : Chatter suppression by adaptive
speed modulation. International Journal of Machine Tools
and Manufacture, Vol. 37(1997) , pp. 355369.
30. Tomkw J. : Vibrostability of machine tools (in Polish),
The Scientific and Technical Publications, Warsaw 1997.
Plik : PMR-spec_KK_M..: 33256 zn. norm. [18,5 str], stan
2017-02-06, kor. epw

19. uczak M., Manzato S., Peeters B., Branner K., Berring
P., Kahsin M. : Updating Finite Element Model of a Wind
Turbine Blade Section Using Experimental Modal Analysis
Results, Shock and Vibration, Vol. 2014, issue 1, pp. 71-82. CONTACT WITH THE AUTHORS

20. Moradi H., Vossoughi G., Movahhedy M.R., Salarieh H.: Krzysztof J. Kaliski
Suppression of nonlinear regenerative chatter in milling e-mail: kkalinsk@o2.pl
process via robust optimal control. Journal of Process
Control, Vol. 23 (2013), pp. 631-648. Marek A. Galewski
e-mail: margalew@pg.gda.pl
21. Nouari M., List G., Girot F. : Wear mechanisms in dry
machining of aluminium alloys. International Journal Micha Mazur
of Mechanical Production Systems Engineering, Vol. 4 e-mail: mazur.m.r@gmail.com
(2003) , pp. 2229
Marek Chodnicki
22. Pajor M., Hoffmann M., Marchelek K. : Identification of e-mail: marek@chodnicki.pl
cutting process model parameters for multi-edge rotating
tools (in Polish), Modelowanie Inynierskie, Vol. 10 (2011) Faculty of Mechanical Engineering
, pp. 307-314. Gdask University of Technology
ul. G. Narutowicza 11/12, 80-233 Gdask
23. Parus A., Pajor M., Hoffmann M. : Suppression of Self- Poland
Excited Vibration in Cutting Process Using Piezoelectric
and Electromagnetic Actuators. Advances in Manufactur-
ing Science and Technology, Vol. 33 (2009), pp. 35-50.

24. Powaka B., Pajor M., Berczyski S. : Identification of

nonlinear cutting process model in turning. Advances
in Manufacturing Science and Technology, Vol. 33 (3)
(2009), pp. 17-25.

25. Quintana G., Ciurana J. : Chatter in machining processes:

a review. International Journal of Machine Tools and
Manufacture, Vol. 51(2011) , pp. 363-376.

124 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 125-132
10.1515/pomr-2017-0030

NUMERICAL MODELLING
OF STRUCTURES WITH UNCERTAINTIES

Maciej Kahsin
Gdansk University of Technology, Poland

ABSTRACT

The nature of environmental interactions, as well as large dimensions and complex structure of marine offshore
objects, make designing, building and operation of these objects a great challenge. This is the reason why a vast
majority of investment cases of this type include structural analysis, performed using scaled laboratory models and
complemented by extended computer simulations. The present paper focuses on FEM modelling of the offshore
wind turbine supporting structure. Then problem is studied using the modal analysis, sensitivity analysis, as well
as the design of experiment (DOE) and response surface model (RSM) methods. The results of modal analysis based
simulations were used for assessing the quality of the FEM model against the data measured during the experimental
modal analysis of the scaled laboratory model for different support conditions. The sensitivity analysis, in turn, has
provided opportunities for assessing the effect of individual FEM model parameters on the dynamic response of the
examined supporting structure. The DOE and RSM methods allowed to determine the effect of model parameter
changes on the supporting structure response.

Keywords: Wind Turbine, Offshore, FEA, Modal Analysis, Dynamic Coupling, Sensitivity Analysis, Design of Experiment,
Response Surface Method.

INTRODUCTION to identify parameters of the real object. An attempt has also

The use of numerical techniques [1], [2], [3], [4], [5],
in combination with the development of measurement
techniques [6], [7], [8], [9], [10], [11], and the use of modern
hardware [12], [13], provide opportunities to analyse more and
more complex physical phenomena. Of special importance
here is the progress in techniques used to create virtual
models of real structures which model their behaviour in
the best possible way [14], [15], [16], [17]. For the offshore
structures, especially those founded on the seabed, the
uncertainty of geo-mechanical parameters of the seabed
[18], [19] and various dynamic couplings, which are difficult
to quantify, are the sources of serious complications when
creating a numerical model. In this paper, use is made of
experimental and numerical modal analysis techniques [20]
to develop modal models, the correlation of which will allow
been made, with the aid of the design of experiment (DOS)
and response surface model (RSM) methods [21], to assess
the effect ofunknown dynamic coupling on the structural
response of the offshore wind turbine supporting structure.
EXAMINED OBJECT
The examined object is the scaled (1:40) laboratory model
of the offshore wind turbine supporting structure in tripod
configuration. The model is made of aluminium. Physical
properties of aluminium assumed in the simulations are the
following: Youngs modulus (E) 70 GPa, Poissons ratio ()
0.33, density () 2750 kg/m3. The nominal values of the
geometrical parameters are given in Fig. 1. The height of the
laboratory model is 1,885m, and its mass is 32,5kg.

POLISH MARITIME RESEARCH, No S1/2017 125

 Fig. 1. Geometrical parameters of the supporting structure.
Three support cases have been analysed: (1) free-free
support configuration, with good correlation observed [2];
(2) model suspended on an elastic rod, and (3) model founded
on rubber blocks. The boundary conditions for the real object
are shown in Fig. 2.
Fig. 2. Conditions of tripod foundation on rubber blocks.
EXPERIMENTAL TESTS
The experimental modal analysis was based on acceleration
measurements at 61 points. 11 accelerometers were mounted
on the central column (TS), and 5 accelerometers on each
remaining supporting structure element (PG, UB, LB).
Electromagnetic inductors fixed to the structure at two points
(marked red in Fig. 2) were used to generate as many vibration
modes as possible.
126 POLISH MARITIME RESEARCH, No S1/2017
Fig. 2. Acceleration sensor locations.
After completing the measuring campaign, natural
vibration modes of the structure were identified within the
range from 66 to 280Hz for the free-free support conditions,
and within 42 to 310Hz for the structure founded on rubber
blocks. The first ten modes and the natural mode of tripod
vibrations were taken into consideration in the further
analysis. A more detailed description of the measuring
campaign and the analysis of the obtained results can be
found in [1] and [2].
FINITE ELEMENT METHOD
Due to the thin-walled structure of the real object, the
same grid topology of two-dimensional finite plate elements
was used in both support cases. Initially, the structure was
attributed with nominal values of material and geometrical
parameters. Despite the symmetry of the structure, each grid
component (pile guide, upper and lower flange, and central
column), was attributed with individual ES properties, which
allowed to control the order of appearance of natural vibration
modes [1]. The only difference between the numerical models
with respect to support conditions for free-free support
configuration and model foundation on rubber blocks was
the use of elastic elements with three degrees of freedom an individual FEM model parameter on the natural vibration
(rotational stiffness was omitted) to model tripod interaction frequency was assessed by analysing local sensitivity, i.e. the
with the rubber blocks, Fig.3. effect of changing a single parameter at fixed values of the
remaining parameters, according to the formulas:

(2)


(3)

Equation (2) describes the sensitivity of the eigenvalue,

Fig. 3. Elastic elements modelling the stiffness of rubber blocks.
while Equation (3) - the sensitivity of the natural vibration
mode vector. [M] is the inertia matrix, [K] is the stiffness
matrix, m is the modal mass, is the resonance frequency,
and q is the analysed parameter. Since the sensitivity is
afunction of model parameters, it had to be calculated in
the iteration loop. The tuning process is schematically shown
in Fig. 4.

It has been assumed that within the tested range the rubber
block material has a linear-elastic characteristic. Applying
the force method to the rubber blocks, their stiffness was
assessed as equal to 19700N/m along the vertical axis, and
to 1000N/m along the transverse axes.
Correlation and tuning of experimental and simulated
models
To assess the quality of the FEM model, its correlation
with the experimental modal model, treated as the accurate
reference model, was estimated. The tuning process made use
of the modal assurance criterion (MAC) matrix, calculated
from the following formula: Fig. 4. Block diagram of the FEM model tuning process.

[) T ) ]2
Test F
E (1) For the tripod in free-free support configuration, the initial
MAC TestFE
[) T ) ] [) T ) ] MAC matrix (for nominal parameters of the FEM model)
Test Test F
E F
E is shown in Fig. 5, while the corresponding relative errors
of natural vibration frequency values are collated in Tab. 1.
where is the natural vibration mode vector. The MAC
matrix represents linear relationship between vectors and
takes values from within the interval <0,1>, where 1 means
full linear relationship between the vectors, while 0 means
that the vectors are fully linearly independent. Due to the
difference in model sizes, the FEM model was to be reduced
to the number of nodes in the experimental model, using the
SEREP (System Equivalent Reduction Expansion Process)
method for instance. Moreover, the degrees of freedom which
were not measured in the experiment, i.e. the rotational
degrees of freedom in this case, were also to be excluded
from the FEM model.
The tuning process was conducted in the iterative
optimisation loop. During these calculations, the values of
the FEM model parameters were changed within an assumed Fig. 5. MAC matrix for free-free support configuration and nominal FEM
range. The goal of the optimisation was to reach the maximum model parameters.
value of MAC, at simultaneous minimisation of differences
between natural vibration frequency values in the modal FEM
model and the experimental model. The quantitative effect of

POLISH MARITIME RESEARCH, No S1/2017 127

Table 1. Relative errors of natural vibration frequency values for nominal FEM Table 3. Final set of FEM model parameters (wall thicknesses of tripod
model parameters. components).

FE [Hz] EMA {Hz] error [%] gr. [mm] diff. [%]

1 62.9 66.13 4.9 TS 3.01 0.33
2 62.9 69.24 9.15 UB_1 3.09 3
3 104.1 91.12 14.24 UB_2 2.99 0.33
4 104.1 94.86 9.74 UB_3 3.008 0.27
5 221.2 210.12 5.27 LB_1 2.995 0.17
6 252.6 247.92 1.88 LB_2 2.998 0.067
7 252.6 258 2.09 LB_3 2.998 0.067

8 262.1 267.1 1.87 PG_1 3.0075 0.25

9 284.2 283 0.42 PG_2 3.02 0.67

10 284.2 287.4 1.11 PG_3 3.015 0.5

The best MAC result obtained in the iterative calculation
process is shown in Fig. 6, while the corresponding relative
errors of natural vibration frequency values are collated
in Tab. 2. The optimisation was performed using only the
wall thickness values of the tripod components, at possible
thickness changes assumed within +/- 3% of the nominal
value. The calculated thicknesses of tripod walls, along
with relative percentage changes in relation to the nominal
parameters, are shown in Tab. 3.
In the case of tuning the model of supporting structure
founded on rubber blocks, it was again the wall thicknesses
which were used as tuned parameters, with their values
changing within +/-3% of the nominal value. But this time the
set of optimised parameters was extended by the stiffnesses of
the rubber blocks, with their permissible variability assumed
within +/- 40% of the assessed stiffness values. The MAC
matrix for nominal values of parameters for the tripod
founded on rubber blocks is shown in Fig. 7.

Fig. 6. Best MAC result obtained in the iteration process. Fig. 7. Initial MAC matrix for the supporting structure
founded on rubber blocks.

Table 2. Relative errors of natural vibration frequencies of the tuned FEM model.
What is noteworthy in Fig. 7 is the absence of the
FE [Hz] EMA [Hz] error [%] experimentally recorded 4th vibration mode in the modal
1 63.5 66.13 4.1
FEM set. In this mode, the tripod behaves as a rigid body
and performs a reciprocating motion along the vertical
2 66 69.24 4.9
axis. None of the examined configurations of FEM model
3 103.41 91.12 11.8 parameters obtained during the iterative procedure indicated
4 105.3 94.86 9.9 the appearance of this mode. At the same time, the obtained
MAC values are relatively small the maximum value
5 220.8 210.12 4.8
obtained during the tests was equal to 0.67. All this lead to
6 251.45 247.92 1.4 a conclusion that the created FEM model does not model
7 253.8 258 1.6 properly dynamic characteristics of the real structure.
8 264.23 267.1 1.08
9 285 283 0.70
10 289.4 287.4 0.69

128 POLISH MARITIME RESEARCH, No S1/2017

DOE AND RSM METHOD BASED ANALYSIS
The effect of element stiffness on the natural vibration
frequency in the FEM model was analysed using the DOE
(Design of Experiment) method. This method makes it
possible to trace the effect of input value changes on the output
values, by introducing a discrete set of input values in which
more than one parameter can be changed. The calculations
were performed on high-performance computers at the
Academic Computer Centre TASK in Gdansk, making use
of the software Noesis Optimus 10.13 and MSC.Nastran.
Due to a significant effect of vertical flexible element
stiffness changes on the results of calculations, the variability
range of this parameter was assumed within a narrow interval, Fig.10. Definition of the output parameter vector.
unlike the transverse stiffness variability range.
The DOE method based modelling process consisted in
creating: the source batch file (bushing_DOE.dat), the source The applied approach made use of the 2nd order model
output file (bushing_doe.f06), the input parameter vector and the 3rd level full factorial DOE method. Taking into
with indicating the place of occurrence of the required consideration all possible combinations of numerical model
parameters in the file (InputArray1 - 9 stiffnesses), and parameters, the total number of experiments was 19683,
the output parameter vector - with indicating the place of Fig. 11.
occurrence of the required parameters in the file (Frequen
- 15 frequencies). Moreover, the action which allowed to
generate the output file from the input file (initiating starting
command for the code MSC.Nastran) and mutual relations
between the above objects were to be defined, Fig. 8.

Fig. 8. Scheme of data use in the code Optimus.

The nominal values of the parameters, their variability

ranges, and the used nomenclature of input/output parameters
are given in Fig. 9 and Fig.10. The stiffnesses BK1, BK4, BK5
of the flexible element (EP) are marked, respectively, using
the following indices: 1, 2, 3 stiffnesses along the x-axis
direction (according to the model coordinate system); 2, 5, 8
- stiffnesses along the y-axis direction, and 3, 6, 9 stiffnesses
along the vertical axis direction.

Fig. 11. Definition of the used 3rd Level Full Factorial method.

Due to a large number of data and the dependence of the

Fig. 9. Definition of the input parameter vector.
analysed state on several parameters (the 2nd order model),
a convenient tool for analysing the obtained results is the
response surface model (RSM) method. This method enables
3D presentation of results of a numerical experiment, and
displays system response changes in the form of surfaces for
two fixed parameters. Figures 12, 13, and 14 show sample
changes of the 3rd natural vibration frequency (corresponding
to the vibration along the vertical axis of the object) for
constant values of stiffnesses EP1 and EP2 in the vertical
direction and changes of the remaining parameters. The

POLISH MARITIME RESEARCH, No S1/2017 129

analysed results reveal small sensitivity of the tripods natural
vibration frequency to stiffness changes in the direction
perpendicular to the vertical axis of the object. The only
stiffness which affects this frequency to a relatively great
extent is the stiffness EP3 in the vertical axis direction.
Decreasing the horizontal stiffness EP makes the tripod
vibrate as a solid body, while increasing the stiffness in those
directions leads to the appearance of frequencies which were
not recorded experimentally (with respect to both, the value
and the eigenvector). This may testify to limited possibilities
to reduce the tripod motion in the directions on its foundation
plane. Moreover, analysing the scatter plots of correlations
between the values calculated using the DOE method and
those estimated by the model, we can see that the non-
compliance range is unacceptably large and, consequently, Fig. 13. Natural vibration frequency changes for constant vertical stiffnesses
the parameter variability ranges should be substantially BK3 and BK6 (the remaining stiffnesses at the maximum level).
decreased. However, this would lead to even smaller effect
of model stiffnesses in the x- and y-axis direction on system
vibration frequency changes. Figure 15 shows a sample scatter
plot for the 3rd frequency. The stiffness in the vertical direction
is definitely too small, and it differs by as much as 12 times
from that recorded experimentally, which testifies that the
stiffness of rubber blocks cannot be determined based only
on the mass of the system and the resonance frequency (even
if the mode represents the undeformed tripod), or on the
force method.

Fig.14. Natural vibration frequency changes for maximal vertical stiffness

values.

Fig. 12: Natural vibration frequency changes for constant vertical stiffnesses
BK3 and BK6 (the remaining stiffnesses at the minimum level).

Fig.15. 3rd natural vibration frequency for maximal vertical

stiffness values.

130 POLISH MARITIME RESEARCH, No S1/2017

 CONCLUSIONS
The test model and the FEM model, which initially
revealed relatively good correlation, became very difficult
to correlate, and thus to identify, after changing the boundary
conditions from the free-free configuration to the elastic
support configuration. Despite resignation from the use of
standard boundary condition, which consisted in reducing the
degrees of freedom at nodes, and replacing it by the support
in the form of flexible elements, the examined structures
exhibited very little similarity to each other in both models.
This effect was likely to be caused by the lack of information
on, or impossibility to determine, parameters describing
quantitatively the structural properties of the supporting
structure. Despite tuning the flexible elements with the aid
of the natural vibration mode which was very similar to the
behaviour of the harmonic oscillator, the attempt to assess
correctly the stiffness of flexible tripod elements in the vertical
direction ended without success. The resultant vibration
frequency of the 3rd natural mode obtained in the numerical
experiments oscillated about 10Hz (at the required value
of 120Hz). The statistical analysis, performed with the aid
of DOE and RSM methods, allowed to notice some trends
in structural responses of the virtual model, as a result of
changing input parameters. Based on the above observations,
negligibly small effect of the analysed parameters on the 3rd
values and modes of other stiffnesses than that in the vertical
direction was identified. Moreover, the DOE results indicate
a possible direction of search for support parameters which
would better reflect the physics of the examined phenomenon.
The stiffness values are to be searched at much higher levels
than the estimated (nominal) vertical stiffness of the blocks
(19700N/m) and, on the other hand, much below the maximal
transverse stiffnesses (1000N/m). The analysis of this type can
also indicate, in general, whether the assumed range of the
searched values is narrow enough to avoid excessive errors.
The above remarks can be important indication for SHM
techniques, for instance. In the case of real offshore wind
turbines founded on the seabed, an issue of high importance
will be correct identification of properties of the seabed-
foundation structure, especially in the case of high gradients
of geological properties in the area occupied by the wind farm.
The presented sample case was relatively easy to analyse,
due to the earlier obtained compliance between the test model
(in the free-free support configuration) and the FEM model.
The inconsistency which appeared after changing the support
configuration was immediately diagnosed as the effect of the
applied form of support conditions which limited the search
area to only 9 parameters. Perhaps, the remaining structural
parameters, solely related with the tripod structure, should
be additionally included to the set of search parameters.
However, this would make their analysis extremely difficult,
if not impossible at all.
BIBLIOGRAPHY
1. Kahsin M., Luczak, M., Peeters, B.: Use and assessment
of preliminary FE model results within testing process
of offshore wind turbine supporting structure, EURO-
DYN 2014: IX INTERNATIONAL CONFERENCE ON
STRUCTURAL DYNAMICSBook Series:EURODYN-
International Conference on Structural Dynamics, pp.
3659-3666Published:2014.
2. Kahsin, M., Luczak, M.: Numerical Model Quality Assess-
ment of Offshore Wind Turbine Supporting Structure
Based on Experimental Data, STRUCTURAL HEALTH
MONITORING 2015: SYSTEM RELIABILITY FOR
VERIFICATION AND IMPLEMENTATION, VOLS.
1 AND 2Book Series:Structural Health Monitoring,
pp.2817-2824Published:2015.
3. Kozicki J., Tejchman-Konarzewski A., Mhlhaus H.:
Discrete simulations of a triaxial compression test for
sand by DEM. INTERNATIONAL JOURNAL FOR
NUMERICAL AND ANALYTICAL METHODS IN
GEOMECHANICS, iss. 18 (2014), pp.1923-1952.
4. Deja M., Siemitkowski M.: Feature-based generation of
machining process plans for optimised parts manufac-
ture// JOURNAL OF INTELLIGENT MANUFACTUR-
ING. -Vol. 24, iss. 4 (2013), pp. 831-846.
5. Niklas K., Kozak J.: Experimental investigation of Steel
ConcretePolymer composite barrier for the ship internal
tank construction// OCEAN ENGINEERING. -Vol. 111,
(2016), pp. 449-460.
6. Ambroziak A., Kosowski P.: Mechanical testing of
technical woven fabrics. JOURNAL OF REINFORCED
PLASTICS AND COMPOSITES. Vol. 32, Iss. 10 (2013),
pp.726-739.
7. Ambroziak A., Kosowski P.: Mechanical properties for
preliminary design of structures made from PVC coated
fabric. CONSTRUCTION AND BUILDING MATERI-
ALS. -Vol. 50, (2014), pp.74-81.
8. Burdziakowski P., Janowski A., Kholodkov A., Matysik
K., Matysik M., Przyborski M., Szulwic J., Tysic P.,
Wojtowicz A.: MARITIME LASER SCANNING AS
THE SOURCE FOR SPATIAL DATA// Polish Maritime
Research. -Vol. 22, Iss. 4(88) (2015), pp. 9-14.
9. Di Lorenzo E. Petrone G., Manzat S., et al.: STRUC-
TURAL HEALTH MONITORING-AN INTERNA-
TIONAL JOURNAL Volume 15, Issue: 3, pp. 289-301
Published: MAY 2016.
10. El-KafafyM., Peeters B., Guillaume P., et al.: MECHANI-
CAL SYSTEMS AND SIGNAL PROCESSING Volume:
72-73, pp. 567-589 Published: MAY 2016.
POLISH MARITIME RESEARCH, No S1/2017 131
11. Litwin W., Olszewski A.: Water-Lubricated Sintered CONTACT WITH THE AUTHOR
Bronze Journal BearingsTheoretical and Experimental
Research// TRIBOLOGY TRANSACTIONS. -Vol. 57, Maciej Kahsin
no.1 (2014), pp.114-122.
Gdansk University of Technology
12. Price, D. C.; Clark, M. A., Barsdell, B. R., et al.: Optimiz- Faculty of Ocean Engineering and Ship Technology
ing performance-per-watt on GPUs in high performance 11/12 Narutowicza St.
computing, COMPUTER SCIENCE-RESEARCH AND 80 - 233 Gdask
DEVELOPMENTVolume31,Issue4,pp.185-193Pub- Poland;
lished:NOV 2016.

13. Locharla G., Rao K., Sudeendra K., Mahapatra K., et al.:
Implementation of MIMO data reordering and sched-
uling methodologies for eight-parallel variable length
multi-path delay commutator FFT/IFFT, IET COMPUT-
ERS AND DIGITAL TECHNIQUES Volume 10, Issue:
5, pp. 215-225 Published: SEP 2016.

14. Iwicki P., Tejchman A., Chrcielewski J.: Dynamic FE

simulations of buckling process in thin-walled cylindrical
metal silos. THIN-WALLED STRUCTURES. Vol. 84
(2014), pp.344-359.

15. Grski J., Mikulski T., Ozibo M., Winkelmann K.:

Effect of geometric imperfections on aluminium silo
capacities// STAHLBAU. -Vol. 84, iss. 1 (2015), pp.52-57.

16. Dymarski C., Dymarski P., Zywicki, J.: DESIGN AND

STRENGTH CALCULATIONS OF THE TRIPOD SUP-
PORT STRUCTURE FOR OFFSHORE POWER PLANT,
POLISH MARITIME RESEARCH Volume: 22, Issue: 1,
pp. 36-46, Published: JAN 2015.

17. Dymarski P., Ciba E., Marcinkowski, T.: EFFECTIVE

METHOD FOR DETERMINING ENVIRONMEN-
TAL LOADS ON SUPPORTING STRUCTURES FOR
OFFSHORE WIND TURBINES, POLISH MARITIME
RESEARCH, Volume 23, Issue 1, pp. 52-60, Published:
JAN 2016.

18. Grelowska G., Kozaczka E.: Sounding of Layered Marine

Bottom - Model Investigations// ACTA PHYSICA
POLONICA A. -Vol. 118, nr. no. 1 (2010), pp. 66-70.

19. Grelowska G., Kozaczka E., Kozaczka S., Szymczak W.:

Sea bottom structure investigation by means of acoustic
methods// POLISH JOURNAL OF ENVIRONMENTAL
STUDIES. -Vol. 19, no. 4A (2010), pp. 35-38.

20. Heylen, W., Lammens, S., Sas, P.: Modal analysis theory
and testing, Katholieke Universiteit Leuven, Belgium,
2009.

21. Pannerselvam, R.: Design and Analysis of Experiments,

ISBN 978-81-203-4499-0, New Dehli, India, 2012.

132 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 133-140
10.1515/pomr-2017-0031

POSSIBILITIES OF HEAT TRANSFER AUGMENTATION

IN HEAT EXCHANGERS WITH MINICHANNELS
FOR MARINE APPLICATIONS

Dariusz Mikielewicz
Jan Wajs
Gdansk University of Technology, Poland

ABSTRACT

In the paper, new trends in the development of microchannel heat exchangers are presented. The exchangers developed
in this way can be applied in marine industry. Main attention is focused on heat exchanger design with reduced size
of passages, namely based on microchannels. In authors opinion, future development of high power heat exchangers
will be based on networks of micro heat exchangers.

Keywords: heat transfer intensification, heat exchangers, pumping power

INTRODUCTION that titanium is relatively expensive and therefore efforts

The weight of the equipment onboard a ship and the
space requirements are nowadays more important than ever,
therefore selection of the right equipment and its size is of
high importance [1]. When investing in new heat exchangers
for todays vessels, ship owners, operators, and shipyards are
looking for ways to reduce costs, boost the performance of
the applied systems, and meet energy efficiency and emission
requirements of contemporary stringent standards. Recent
technological advances contribute to improved performance,
greater reliability, and lower lifecycle costs, all this having
a positive impact on profitability. A typical heat exchanger
design strategy is to obtain a high heat transfer rate using
minimum pumping power and space. When the fluid with
low thermal conductivity passes through the heat exchanger,
there is a necessity to increase the heat transfer rate but in
some of applications, in marine and chemical industries for
instance, the resistance to heat transfer increases cause to
fouling or scaling.
Most cooling systems are made as central cooling systems,
which means that there is only one or two large plate heat
exchangers equipped with titanium plates, which come into
contact with the seawater. Titanium is the only material that
can withstand the aggressive seawater without corrosion
damages on the plates in the heat exchanger. Some analyses
of that issue are presented in [2,3]. It should be borne in mind
must be made to use it responsibly. This means that the
central cooler can operate without problems using seawater
as the cooling medium for the internal freshwater cooling
system. The advantage of the central cooler is that the use
of the corrosive seawater is limited to one or two plate heat
exchangers, and thereby the use of the expensive material
for that exchanger and the seawater pipeline is limited to
a minimum, which will reduce the corrosion damage in
the cooling system. The remaining plate heat exchangers
are normally equipped with stainless steel plates, for which
there is no corrosion risk as the heat exchangers work with
freshwater. In future applications, some effort should be made
to use plastics [4-6], which do not exhibit susceptibility to
corrosion and are much cheaper.
Heat transfer enhancement or augmentation techniques
refer to the improvement of thermo-hydraulic performance
of heat exchangers. Among many techniques (both passive
and active) investigated for augmentation of heat transfer
rates inside circular tubes, a wide range of inserts have been
utilized, particularly when turbulent flow is considered [7].
A lot of methods, such as treated surfaces, porous surfaces
[8], rough surfaces [9], swirling flow devices, coiled tubes,
and surface tension devices have been applied to increase
thermal performance of heat transfer devices. Furthermore, as
a heat exchanger becomes older, the resistance to heat transfer
increases owing to fouling or scaling. These problems are more

POLISH MARITIME RESEARCH, No S1/2017 133

common for heat exchangers used in marine applications than
in other areas. In some specific applications, such as heat
exchangers dealing with low thermal conductivity fluids
(gases and oils) and desalination plants, there is a need to
increase the heat transfer rate [10]. It can be improved by
introducing a disturbance in the fluid flow (breaking the
viscous and thermal boundary layers), but in that case the
process pumping power may increase significantly and
ultimately the pumping cost may become unacceptably
high. Therefore, to achieve a desired heat transfer rate in the
existing heat exchanger at an economical pumping power,
several techniques have been proposed in recent years and
are discussed in the following sections.
For many years, the authors have been conducting
investigations regarding the methods of heat transfer
enhancement to improve the performance of heat exchangers.
In this paper, a new design of plate heat exchanger with
minichannels (MPHE) is proposed. The experimental analysis
of the prototype is described.
CRITERION FOR COMPARISON OF HEAT
INTENSIFICATION EFFECT
A useful method of comparing the performance of smooth
tubes and the tubes with modified surfaces for the purpose
of intensification of heat transfer is to compare them at equal
pumping power. That is expressed by the equation:
(Vp ) = (Vp )
0
(1)
Expressing adequately the volumetric flow rate and the
pressure drop we get the relation between the resistance
coefficient and the Reynolds number:
(Vp ) = (Vp )
0
(2)
The heat transfer intensification coefficient can be
defined as a function of heat transfer efficiency increase and
pumping power at the same flow rate. This parameter is also
used for comparisons between various passive heat transfer
intensification techniques at specified pressure drop. It is
described by an equation arising from Equation (2), [7]:
(Nu / Nu 0 )
E=
( f / f 0 )1 / 3 (3)
where Nu, f, Nu0 and f0 are the Nusselt numbers and resistance
coefficients, respectively, for two channel configurations,
i.e. with and without the turbulising insert. The resistance
coefficient is determined based on the pressure drop or on the
power of the pump. The intensification efficiency (at constant
pumping power) can be expressed as [7]:
D
Ep (4)
D0 SXPSLQJ SRZHU const
134 POLISH MARITIME RESEARCH, No S1/2017
Expressions (4) or (5) can be rather used only for single-
phase flows, as the correlations describing the relation between
the heat transfer coefficient and the Reynolds number, and
between the resistance coefficient and the Reynolds number
are required simultaneously. Finding such correlations for
flow boiling is very difficult, or even impossible in some
circumstances. In that case the heat transfer intensification is
expressed as the heat transferred through the tube in relation
to the pumping power for both cases: smooth and turbulised
tube. Such a methodology is possible and the expression (4)
is based on it.
Heat transfer intensification in the channel with flow
boiling can be described as the ratio of the heat absorbed
by the tube to the pumping power, firstly for the tube with
aturbulising insert, and then for the smooth tube of the same
diameter, subscript 0, respectively. This method can be used
when the qualities in both cases are equal. The expression
is [7]:
Q

V p
Ex = (5)
Q

V p 0 x=const
If Ex > 1, the intensification occurs in the examined
configuration. Equation (5) is the general expression. The
numerator can be interpreted as the net heat which can
be transferred through the channel with intensification.
The denominator describes the same for the smooth tube.
The compared net effect includes both the quantity of the
transferred heat and the expenditure of energy needed for
medium pumping.
An idea of novel plate heat exchanger with minichannels is
presented together with thermal-hydraulic characteristics of
its prototype. This kind of heat exchangers can be prospectively
applied in marine systems, but other applications are also
possible. A simplified routine is presented for design of
micro heat exchangers, such as evaporator or condenser
for instance, with account of pressure drops in the heat
exchanger. Moreover, the comparison analysis between
different models for calculation of two-phase pressure drop
in small diameter tubes is also included. As a result of the
here presented calculations, real heat exchangers with small
dimensions have been obtained.
IMPACT OF HYDRAULIC
DIAMETER OFMICROCHANNELS
ONHEATEXCHANGER SIZE
A common way to increase the heat transfer coefficient
(a in Eq. (6a)) is to reduce the hydraulic diameter (dh) of
the channel. Usually, the laminar flow conditions should be
applied for heat exchange in minichannels as the pressure
drop becomes excessively high. For the laminar flow, such as
that taking place in micro- and minichannels, the value of the
Nusselt number is constant. The rate of heat density, which
is the ratio of heat transfer rate to heat exchanger volume, is:
Q U T A
= ,
V V
where:
d h2 1 Nu
A = n dh L , V = 2 n L, U = = . (6a)
4 1/ 1 + 1/ 2 2 2 d h
As can be concluded from Equation (6), the rate of heat
density is inversely proportional to the hydraulic diameter
of the channel:
Q 1
= 2 . (7)
V dh
The lower the hydraulic diameter of the minichannels in
the heat exchanger, the higher the rate of heat transfer density.
A few methods exist to increase the heat transfer coefficient
in microchannels. One of them, presented above, is to reduce
the hydraulic diameter of the microchannels. Another
method takes advantage of the fact that the developing
flow has a higher heat transfer coefficient than the fully
developed flow, what is schematically presented in Fig. 1.
Hence short micro channels of the length shorter than the
thermal entrance region should be applied to increase the
heat transfer coefficient.
Fig. 1. Change of local and mean heat transfer coefficient along the pipe
Other methods base on the use of corrugated or rough
minichannels.
MINICHANNEL PLATE HEAT EXCHANGER
Concepts to reduce the channel hydraulic diameter,
together with the attempt to break-up the thermal boundary
layer, have been applied in the present work. Figure 2 shows
the prototype plate heat exchanger with minichannels
(MPHE). It consists of a series of gasketed brazed plates with
channels of rectangular cross-section (width a = 1 mm, depth
b = 700 m) and of 40 mm in length. The distance between
the channels is equal to 1 mm, and the heat transfer area is
equal to 0.00694 m2. Three types of heat transfer enhancement
methods were tested:
1) Version I heat exchanger with unmodified minichannels,
2) Version II heat exchanger with minichannels cut
perpendicularly to the flow direction,
(6) 3) Version III heat exchanger with minichannels cut at the
angle of 60 degrees.
The analysis was conducted on the model of heat exchanger
consisting of one hot and one cold passage, Fig. 2. Such an
arrangement was selected to get better insight into heat
transfer and pressure drop in the novel construction. During
the first stage of the experiment the heating fluid (hot water)
prepared by the electric heater was circulating in the system,
while the cold water was a tap water. The heat was transferred
in the counter-current flow of working fluids. In the second
round of the experiment, the heating fluid was water whereas
ethanol was the heated medium.
During the experiments the following parameters were
measured: the hot fluid temperature at the inlet (Th-in) and
outlet (Th-out) of the heat exchanger, the cold fluid temperature
at the inlet (Tc-in) and outlet (Tc-out) of the heat exchanger,
the pressure drops connected with the fluid flow (Pexp), the
volumetric flow rate of hot water, and the volumetric/mass
flow rate of the working fluid. The measured values of these
parameters made the basis for calculating the heat flux (q),
the Logarithmic Mean Temperature Difference (LMTD) in
the heat exchanger, and the overall heat transfer coefficient
(U). The overall heat transfer coefficient was determined with
the aid of the Peclet law based on the heat transfer area and
the average value of the heat rate transferred through the
wall in a given measurement series.
Fig. 2. Minichannel heat exchanger and view of plates with minichannels
ofdifferent types.
EXPERIMENTAL FACILITY
A schematic of the experimental facility is presented in
Fig. 3. The volumetric flow rate of water ( ) was measured
by rotameter V31 (by Heinrichs). Its measurement range
was from 50 l/h to 500 l/h and its accuracy class was 1. The
ethanol volume flow rate ( ) was measured by a calibrated
POLISH MARITIME RESEARCH, No S1/2017 135
rotameter. In the single-phase flow experiments in water-
water configuration the volumetric flow rates of hot and cold
water were varied in the same range, from 50 to 400 l/h. The
temperature of the hot water entering the heat exchanger was
92C, whereas the cold water temperature was equal to 7C in
each measurement series. In the single-phase flow experiments
in water-ethanol configuration the volumetric flow rate of hot
water was varied in the range from 50 to 125l/h, while the
ethanol flow rate was varied in the range from 26 to 78 l/h.
The temperature of the hot water supplying the heat exchanger
was 80 C, whereas the ethanol temperature was equal to
300.5C in each measurement series.
In the case of flow boiling, the volumetric flow rate of
hot water varied in the range from 125 to 175 l/h (which
corresponds to the water mass flux GW ranging from 972
to 1360 kg/(m2s) in a single rectangular channel), while the
ethanol flow rate varied in the range from 14 to 22 l/h (which
corresponds to the range of ethanol mass flux Get from 47 to
102 kg/(m2s) in a single rectangular channel). The temperature
of the hot water supplying the heat exchanger was in the range
91-100.5C, whereas the temperature of the ethanol entering
the heat exchanger was about 80C.
In the boiling f low experiments the ethanol was
preliminarily heated to make the boiling heat transfer inside
the heat exchanger the most effective. It was delivered to
the heat exchanger with subcooling of 3 K in relation to its
saturation temperature controlled by an ultra-thermostat. It
means that the flow rate measurements were done when the
ethanol had temperature near the saturation state and was still
in the liquid state. The flow of water was first directed to the
rotameter and then to the electric heater to obtain the required
parameters at heat exchanger inlet. The heater was controlled
by an autotransformer, which allowed smooth adjustment
of heater power and then precise water temperature setting.
The ethanol was circulating in a closed system equipped with
athermostatic bath, which heated it to a certain level before it
entered the heat exchanger. For the needs of the experiment,
an additional heat exchanger supplied with the tap (cold)
water was provided to the thermostatic bath. It was used to
withdraw from the ethanol the thermal energy gained by it,
thus assuring the stationary state of the analysis.
Fig. 3. Scheme of experimental facility
136 POLISH MARITIME RESEARCH, No S1/2017
SINGLE-PHASE FLOW DATA
Pressure changes as a function of the volumetric flow rate
are presented in Fig. 4. The curve exhibits an exponential
character. Relatively large pressure drops can be observed
in the heat exchanger. The exchanger was tested in
acountercurrent flow configuration with the inlet hot water
having the temperature of approx. 92C. Figure 5 shows the
thermal characteristics of the heat exchanger in the form of
the rate of heat exchanged as a function of the cold water flow
rate. The slope of the heat rate increases with the increase of
the cold fluid flow rate. The values of the overall heat transfer
coefficient are plotted in Fig. 6 as a function of the volumetric
flow rate of cold water. The overall heat transfer coefficient
has been calculated from the Peclet law for the heat transfer
surface equal to 0.00867 m2. It should be emphasized that
the total area taken for these calculations was larger than the
area of 0.00694 m2 mentioned in Section 4. The total area was
calculated from the wetted perimeter of one minichannel,
its length, and the number of minichannels. The area of the
collectors was also taken into account, since it was about
23% of the total heat transfer area and had a great influence
on the heat rate.
Figure 7 shows sample distribution of the heat transfer
coefficient as a function of the logarithmic temperature
difference (LMTD) for a given hot water flow rate (VC = 350
l/h) and at fixed temperatures of hot water (tC) and cold water
(tZ) supplied at the entrance to the heat exchanger. High values
of the mean logarithmic temperature difference result from
the fact that the tested exchanger contained only one plate,
and from high difference, of approx. 80 K, between the heating
and heated water at the entrance to the heat exchanger. In
Tables 1 and 2 the values of heat transfer coefficients are
tabulated.
Fig. 4. Flow characteristics of the Fig. 5. Heat transfer rate vs.
plate minichannel heat exchanger volumetric flow of cold water
Fig. 6. Distribution of overall heat Fig. 7. Distribution of overall heat
transfer coefficient vs. cold water transfer coefficient vs. logarithmic
flowrate mean temperature difference
(LMTD)
Table 1. Heat transfer coefficient values on the cold side of the exchanger
[l/h] 50 400
[l/h] 50 150 250 400
Z [W/m2K] 5655 7823 10017 12686
Table 2. Heat transfer coefficient values on the hot side of the exchanger
[l/h] 50 500
[l/h] 50 150 250 350
C [W/m2K] 5353 10222 12849 15900
The heat transfer results can be presented in the form of
collective charts of heat transfer coefficient as a function of the
Reynolds number for a single channel. Although the obtained
experimental values refer rather to the laminar heat transfer,
the data reduction relation adopted the Dittus-Boelter
correlation, due to the fact that the heat transfer coefficient
exhibits a dependence on the Reynolds number. However, it
appears that the obtained results correlate better when the
dependence of the Nusselt number on the Reynolds number
is considered. The results are shown in Figs. 8 and 9. Figure 8
presents two series of experimental measurements (EXP and
EXP2) and the corresponding values of the Nusselt number
determined from the Dittus-Boelter correlation. The EXP and
EXP2 series correspond to the cold and hot water, respectively.
Some differences in the results are due to a small influence of
the average value of the Prandtl number. The experimental
data show higher values than those obtained from the Dittus-
Boelter correlation. Figure 9 shows the determination of the
constant which would enable to correlate the present results
with the aid of the Dittus-Boelter correlation. We can see that
for the Reynolds number values exceeding 700 the value of
the constant is approximately 0.03, which is higher by 20%
than in the case of the original Dittus-Boelter constant (C
= 0.023). This proves the existence of the turbulent flow at
the Reynolds numbers exceeding 700. The flow may become
turbulent as a result of the existing surface roughness or sharp
inlet edges of the channel.
of water in the channel (at least by 20%) in relation to the
values determined from the Dittus-Boelter correlation.
In the investigations of the water-water configuration, high
500 rates of heat exchanged in the model heat exchanger have been
reported. They reached 6 kW at the volumetric flow rates of
15192
hot and cold water equal to 400 l/h and 500l/h, respectively.
What is worth emphasizing at this point is quite specific
conditions of the experiment, namely high temperature
difference, of approx. 80 K, between the heating water and
heated fluid at heat exchanger entrance. This was reflected in
high values of the logarithmic mean temperature difference
400
in each measurement series (values between 5675 K). High
15917 efficiency of the considered heat exchangers is returned by
the heat transfer coefficient, the values of which attain an
approximate level of 9.6 kW/(m2K).
FLOW BOILING DATA
In case of the flow boiling data, the measurement results
made the basis for determining the heat flux, q, LMTD, and
the overall heat transfer coefficient, U [11]. The overall heat
transfer coefficient was determined with the aid of the Peclets
law based on the heat transfer area equal to 0.00896 m2 and
the average value of the heat rate transferred through the
wall in a given measurement series.
Thermal characteristics of the investigated heat exchanger
are presented in Figs. 10 and 11. These figures show the
transferred heat rate (Fig. 10) and the overall heat transfer
coefficient (Fig. 11) as functions of the ethanol mass flux. Since
the prototype minichannel PHE was designed for the purpose
of heat recovery, therefore the mass flux of the heating fluid
was fixed at certain levels. That is why two heating water mass
fluxes, Gw, were considered (plus an additional value in some
cases) for constant saturation temperature of ethanol (tS_et)
in the heat exchanger. For higher values of the heating water
mass flux, the increase was observed of the transferred heat
rate and the overall heat transfer coefficient. Decreasing of the
transferred rate of heat and overall heat transfer coefficient
values was recorded as a response to the increase of the ethanol
mass flux. This tendency resulted from a specific methodology
assumed for this stage of experiment. For constant water
mass flux and water inlet temperature values, the increase
of the ethanol mass flux caused less intensive boiling (see
Figs.14 and 15). In such a situation, the input heat rate was
not sufficient to keep the boiling at a particular level. The
other reason was water boiling prevention, because of which
the temperature difference between the ethanol and the water
was low and equal to about 11-19 K, what influenced the entire
heat transfer process.

Fig. 8. Comparison of the Fig. 9. Determination of the constant

experimental heat transfer for modification of the Dittus-
coefficient with that determined Boelter correlation as a function
from the Dittus-Boelter correlation ofReynolds number.
(EXP- cold water, EXP2 hotwater).

A unique finding of the study is a significant improvement Fig. 10. Rate of heat vs. ethanol Fig. 11. Overall heat transfer
of the heat transfer coefficient for the single-phase convection mass flux. coefficient vs. ethanol mass flux.

POLISH MARITIME RESEARCH, No S1/2017 137

 Figures 12 and 13 present the heat flux and the overall
heat transfer coefficient as functions of the logarithmic mean
temperature difference (LMTD). The figures were assigned for
two values of water mass flux (Gw =1166 kg/(m2s) and Gw =1360
kg/(m2s)), and for the constant saturation temperature equal
to tS_et=79.6C. In both cases the increase of the analysed
parameters followed the increase of LMTD.
Fig. 12. Heat flux vs. LMTD. Fig. 13. Overall heat transfer
coefficient vs. LMTD.
The influence of the water flow Reynolds number on the heat
flux and the overall heat transfer coefficient is demonstrated
in Figs. 14 and 15. Both characteristics were obtained for the
ethanol mass fluxes equal to Get=60 kg/(m2s) and Get=74kg/
(m2s), based on the Reynolds number of water flow in one
minichannel defined as Rew_1ch = (GwDH)/, where isthe
dynamic viscosity taken from [12] for the average water
temperature, DH is the hydraulic diameter of minichannel
(DH = 0.8235 mm) and Gw is the mass flux described by the
formula Gw = /nab , where is the water mass flow
rate, n is the number of minichannels, and a and b denote
the width and depth of minichannels, respectively. In both
cases an increasing tendency is found.
Fig. 14. Heat flux vs. water Fig. 15. Overall heat transfer
flow Reynolds number for one coefficient vs. water flow Reynolds
minichannel. number for one minichannel.
Fig. 16. Quality of ethanol vapour Fig. 17. Quality of ethanol vapour
at heat exchanger outlet vs. ethanol atheat exchanger outlet vs. inlet
mass flux. water temperature.
The quality of ethanol vapour is the parameter which
represents the intensity of heat transfer, because it depends on
the boiling process and its conditions. Figures 16 and 17 show
the quality of ethanol vapour as the function of the ethanol
mass flux (Fig. 16) and the water inlet temperature (Fig. 17).
The water mass flux and the ethanol saturation temperature
were constant for the corresponding measurement series.
The maximal value of ethanol vapor quality obtained in the
138 POLISH MARITIME RESEARCH, No S1/2017
studies was 0.5. It was not possible to obtain higher values of
quality due to the limitations described in relation to Fig. 10
and 11. Briefly speaking, the assumption of constant water
mass flux and its inlet temperature has led to the restrained
input heat rate and altered the intensity of boiling. Figure
17 shows the results of an additional measurement series
performed for water mass flux of Gw =972 kg/(m2s) to verify
the influence of as many parameters as possible.
A very useful method to determine the heat transfer
coefficient value is the Wilson method [13]. However, it is
mainly dedicated to single-phase heat exchangers. Principles
of this method cannot be fulfilled when analysing two-phase
convection, due to the variable flow rate which, according to
the Wilson method, should be kept constant, for either the
heating or cooling fluid. Due to boiling phenomena it was
not possible to keep the ethanol flow rate constant, therefore
the estimation of the heat transfer coefficients was based
on the known correlations defining the Nusselt number.
According to the literature [14], the application of the Dittus-
Boelter correlation in the analysis of minichannels leads
to high uncertainties. That is why the Hausen correlation
was considered in the present paper as the most suitable for
the considered Reynolds number range, i.e. 2300-6000. The
Hausen correlation for a rectangular channel is described
by following equation:
( ) D 2/3 0.11
Nu w = 0.116 Re w _1ch 2 / 3 160 Prw1/ 3 1 + H , (8)
L w
where Rew_1ch is the single channel Reynolds number for water
flow, Prw is the Prandtl number for water, DH is the hydraulic
diameter of one minichannel, L is the minichannel length,
is the dynamic viscosity at water temperature, and w is
the dynamic viscosity at wall temperature. The necessary
physical and thermal properties were taken from [12] for the
water average temperature.
The calculated values of Nusselt number for the single
water channel Reynolds number are presented in Fig. 18.
They are followed by the calculated values of the heat transfer
coefficient on the water side of the minichannels PHE, shown
in Fig. 19. Both figures are shown to indicate numerical
values representing the heat transfer enhancement and main
parameters causing it.
Fig. 18. Nusselt number vs. water Fig. 19. Heat transfer coefficient
flow Reynolds number for one onheat exchanger water side
minichannel. vs.water flow Reynolds number
forone minichannel.

The heat transfer coefficient on the ethanol side of the
analysed heat exchanger was calculated from Equation (3),
which took the following form:
1
1 1
et = (9)
U w
The calculated results are presented in Figs. 20 and 21.
They show the heat transfer coefficient on the heat exchanger
ethanol side as the function of the heat flux (Fig. 20), and the
ethanol vapour quality (Fig. 21). The maximum value of the
obtained heat transfer coefficient was about 7500 W/m2K for
the ethanol side, and 12000 W/m2K for the water side. The
difference in the mass fluxes should be emphasized here.
The mass flux of ethanol was about 13 times lower than that
of water, and the heat transfer enhancement on the ethanol
side was due to boiling.
Fig. 20. Heat transfer coefficient Fig. 21. Heat transfer coefficient
on heat exchanger ethanol side vs. on heat exchanger ethanol side vs.
heat flux. ethanol vapour quality.
UNCERTAINTIES
The uncertainty analysis of the above experimental
investigations was done in a systematic manner. Taking into
account the low number of measurement repetitions, but
also high data repeatability, statistic uncertainties were not
considered, and the uncertainty analysis focused only on
systematic errors. The analysis was based on the principle of
propagation of uncertainties described by the formula [15]:
2 2
wf wf wf
'y 'x1  'x 2  'x3  ...
wx1 wx 2 wx3
where x is the maximal uncertainty of the measuring
instrument. The uncertainty of the analysed functions
depended on uncertainties of particular variables. In the
present case, the uncertainties related to direct measurements,
indirect calculations, and withdrawal of thermo-physical
properties from tables. The applied uncertainties of various
devices used in the experiment were described in the section
discussing the experimental facility and procedure.
The results of the uncertainty analysis are summarized
in Table 3. The relative uncertainty was calculated from the
following equation:
'y
Gy 100 %
y
Table 3. Summary of the uncertainty analysis
Parameter Relative value [%]
mass flux 3.4-3.8
overall heat transfer coefficient 4.7-9.0
convective heat transfer coefficient (water) 6.1-9.7
convective heat transfer coefficient (ethanol) 8.4-11.7
heat rate 4.7-9.0
heat flux 4.7-9.1
Reynolds number 3.7-4.1
Nusselt number 6.0-9.6
SUMMARY AND CONCLUSIONS
The experimental analysis of the prototype plate heat
exchanger with minichannels under single-phase and flow
boiling conditions of water-ethanol configuration was
described. The determination of thermal performance of
the heat exchanger was the main purpose of this work,
therefore the thermal characteristics of the investigated unit
were presented. Due to the parameters at which the boiling
process was studied, the maximal value of the heat transfer
rate in the considered module was at the level of 0.75 kW
for the water and ethanol mass fluxes equal to 1166 kg/(m2s)
and 47 kg/(m2s), respectively. It should be pointed out that
the above results were obtained for the heat transfer area of
about 0.009 m2, taking advantage from additional significant
rise of hydraulic resistance generated by the initiated ethanol
boiling process. Consequently, the delivery of the ethanol
pump decreased. When the ethanol volume flow rate was
lower, the boiling was intensified and the ethanol vapour
quality increased. The highest value of the overall heat
transfer coefficient was about 4200 W/(m2K), while that of the
transferred heat flux was about 84 kW/m2. The heat transfer
coefficient values presented for the ethanol in boiling state
(up to 6 kW/(m2K) at the heat flux of 62 kW/m2, Fig. 20, have
been confirmed by the fundamental investigations of ethanol
boiling in minichannels [16].
2 In the authors opinion, the proposed construction of the
(10) minichannel plate heat exchanger is promising. It is possible
to consider other materials which would be more suitable for
marine applications, as well as new manufacturing techniques
which can be fully automated [17,18].
(11)
POLISH MARITIME RESEARCH, No S1/2017 139
BIBLIOGRAPHY
1. Korczewski Z., Zacharewicz M.: Alternative diagnos-
tic method applied on marine diesel engines having
limited monitoring susceptibility. Transactions of the
Institute of Measurement and Control, 2012, Vol. 34,
No.8, pp.937-946.
2. Stolarski T.A.: Wear of water-lubricated composite mate-
rials. Wear, 1980, Vol.58, No.1, pp.103-108.
3. Jakubowski M.: Influence of pitting corrosion on fatigue
and corrosion fatigue of ship and offshore structures.
Part II: Load - pit crack interaction. Polish Maritime
Research, 2015, Vol.22, No.3, pp.57-66.
4. Ambroziak A., Kosowski P.: Mechanical properties for
preliminary design of structures made from PVC coated
fabric. Construction and Building Materials, 2014, Vol.50,
pp.74-81.
5. Ambroziak A., Kosowski P.: Mechanical properties
of polyvinyl chloride-coated fabric under cyclic tests.
Journal of Reinforced Plastics and Composites, 2014,
Vol.33, No.3, pp.225-234.
6. Sabik A., Kreja I.: Large thermo-elastic displacement and
stability FEM analysis of multilayered plates and shells.
Thin-Walled Structures, 2013, Vol.71, pp.119-133.
7. Mikielewicz D., Klugmann M., Wajs J.: Flow boiling
intensification in minichannels by means of mechanical
flow turbulising inserts. International Journal of Thermal
Sciences, 2013, Vol.65, pp.79-91.
8. Szymkiewicz A., Modelling water flow in unsaturated
porous media: Accounting for nonlinear permeability
and material heterogeneity, Heidelberg, Springer, 2013.
14. Adams T.M., Abdel-Khalik S.I., Jeter S.M., Qureshi Z.H.:
An experimental investigation of single-phase forced
convection in microchannels. International Journal of
Heat and Mass Transfer, 1998, Vol.41, pp.851-857.
15. Moffat R.J.: Describing the uncertainties in experimental
results. Experimental Thermal and Fluid Science, 1988,
Vol.1, pp.3-17.
16. Zrooga A-B.S.R.: Experimental and theoretical study of
boiling and dryout phenomenon of ethanol in vertical
minitubes. (Ph.D. thesis) Gdansk University of Technol-
ogy, 2010.
17. Deja M., Siemitkowski M.S.: Feature-based generation
of machining process plans for optimised parts manufac-
ture. Journal of Intelligent Manufacturing, 2013, Vol.24,
No.4, pp.831846.
18. Szapczyski R.: Evolutionary Planning of Safe Ship
Tracks in Restricted Visibility. Journal of Navigation,
2015, Vol.68, No.1, pp.39-51.
CONTACT WITH THE AUTHOR
Dariusz Mikielewicz
Gdansk University of Technology,
Faculty of Mechanical Engineering,
Narutowicza 11/12, 80-233 Gdansk
Poland
Jan Wajs
Gdansk University of Technology,
Faculty of Mechanical Engineering,
Narutowicza 11/12, 80-233 Gdansk
Poland

9. Wajs J., Mikielewicz D.: Effect of Surface Roughness

on Thermal-hydraulic Characteristics of Plate Heat
Exchanger. Key Engineering Materials, 2014, Vol.597,
pp.63-74.

10. Cieliski J.T., Mosdorf R.: Gas bubble dynamics experi-

ment and fractal analysis. International Journal of Heat
and Mass Transfer, 2005, Vol.48, No.9, pp.18081818.

11. Wajs J., Mikielewicz D., Fornalik-Wajs E.: Thermal

performance of a prototype plate heat exchanger with
minichannels under boiling conditions. Journal of Phys-
ics Conference Series, 2016, Vol.745, 032063.

12. Refprop v.9.0 National Institute of Standards, 2010.

13. Wilson E.E.: A basis for rational design of heat transfer

apparatus Trans. ASME, 1915, Vol.37, pp.47-82.

140 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 141-148
10.1515/pomr-2017-0032

CALCULATION METHOD FOR FLOW BOILING AND FLOW

CONDENSATION OF R134A AND R1234YF
IN CONVENTIONAL AND SMALL DIAMETER CHANNELS

Dariusz Mikielewicz
Blanka Jakubowska
Gdansk University of Technology, Poland

ABSTRACT

The paper presents the results of calculations performed using the authors model to predict the heat transfer coefficient
during flow boiling for two refrigerants R134a and R1234yf. The experimental data from various past studies have
been collected and the calculations have been conducted for the full range of quality variation and a wide range
ofmass velocity. The aim of the study was to test the sensitivity of the in-house flow boiling and flow condensation
model. The importance of taking into account the surface tension as the parameter exhibiting its importance in case
of the flow in minichannels, as well as the influence of the reduced pressure were analysed. The obtained numerical
results show good consistency with those recorded in the experiments.

Keywords: heat transfer coefficient, flow boiling, conventional channels, minichannels

NOMENCLATURE GREEK SYMBOLS

B blowing parameter heat transfer coefficient, W/m2K

Bo boiling number surface tension
C mass concentration of droplets in two-phase core thermal conductivity, W/mK
Con confinement number density, kg/m3
cp specific heat, J/kgK dynamic viscosity, Pa s
d diameter, m = friction factor
f, fr friction factor
f1, f1z function SUBSCRIPTS
G mass flux, kg/m2s
h heat transfer coefficient, W/m2K 0 reference case
M molecular weight, kg/kmol exp experimental
Nu Nusselt number F Friedel correlation
P empirical correction g, v vapor
p pressure, Pa kr critical
Pr Prandtl number l liquid
q heat flux, W/m2 LO total liquid flow rate
r specific enthalpy of vaporization, J/kg MS Mller-Steinhagen and Heck correlation
R two-phase multiplier Pb pool boiling
Re Reynolds number r reduced
s slip ratio sat saturation
T temperature, K TBP two-phase boiling
x quality th theoretical

POLISH MARITIME RESEARCH, No S1/2017 141

 Table 1: Main thermophysical properties of R134a and R1234yf [7].
INTRODUCTION
R134a R1234yf
In the light of increasing interest in refrigerants possessing
Chemical formula C 2H 2F4 CF3CF=CH2
low global warming potential (GWP), new fluids used in air-
conditioning and refrigeration units cannot be manufactured Boiling point [C] -26 -29
with fluorinated greenhouse gases having GWP greater than Critical point [C] 102 95
150 [1]. Since one of the most popular hydrofluorocarbons
(HFC), namely refrigerant R134a, has a 100-year GWP equal to Molar mass [kg/kmol] 102.03 114.04
1430 in low to medium evaporation temperatures [2], it needs Liquid density at 25 [kg/m3] 1206.7 1091.9
to be replaced by more environmentally friendly refrigerants.
Vapor density at 25 [kg/m3] 32.350 37.925
The need for refrigerant replacements for example in marine
air conditioning and refrigeration systems is very urgent, Liquid enthalpy at 25 [kJ/kg] 234.55 70.076
and R134a should be substituted by other environmentally
Vapor enthalpy at 25 [kJ/kg] 412.33 215.44
friendly refrigerants in the near future. Previous studies have
considered R152a [3] and the natural refrigerant CO2 [4] as Liquid therm. cond. at 25 [kW/mK] 81.134 63.585
possible replacements for R134a in marine air conditioning Vapor therm. cond. at 25 [kW/mK] 13.825 13.966
systems. It should be borne in mind, however, that R152a is
aflammable refrigerant, while for the CO2 system the working Liquid Viscosity [Pa s] 194.89 155.45
pressure is significantly higher than that for the R134a system, Vapor Viscosity [Pa s] 11.693 12.291
which would lead to significant system modifications and
higher costs [5]. Recently, R1234yf has been believed to be Surface tension [N/m] 808u10 -5 616u10 -5
a promising candidate as an alternative for R134a [6]. Its
ODP is equal to 0, whereas GWP 4 and thermophysical In the literature, there are many empirical correlations for
properties are similar to those of R134a. Hence, it meets modelling of boiling heat transfer, and in the case of such
quite well recent environmental requirements and policies. fluids as R134a and R1234yf they have proved rather good
The main thermophysical properties of R1234yf and R134a consistency with experimental data.Several publications,
are summarized in Table 1. namely by Ribatski [27], Tibir and Ribatski [28],
Park and Jung [8] have reported that R1234yf and R134a Sardeshpande and Ranade [29], and Alagesan [30] analyse
have similar nucleate heat transfer coefficients and similar, the experimental data for validation of heat transfer coefficient
low toxicity. R1234yf also has mild flammability, compared predictions using correlations available in the literature. In
to R32 and other hydrocarbon refrigerants [7]. Nevertheless, the present paper, the experimental results collected from
minimizing the refrigerant charge is still important to further the literature are compared with the predictions of the
reduce the risk related to the flammability of R1234yf, model [31,32]. Based on the evidence of comparison with
apromising candidate for successful use in systems currently the abovementioned experimental data, a correction making
designed for R134a. use of the reduced pressure effect has been applied to the
There are many available experimental investigations into authors model to provide feasibly the best consistency of
flow boiling heat transfer of R134a in the literature. Typical the predictions with the experiments.
experimental ranges of these studies have been collected in
Table 2. Unfortunately, the results published for R1234yf are
still limited and inconsistent. For instance, Satioh et al. [11] THE MODEL
studied the boiling heat transfer of the refrigerant R1234yf
flowing in a smooth small-diameter horizontal tube, the The versatile semi-empirical model to calculate flow boiling
inertial diameter of which was 2 mm. From the low to the high and flow condensation due to D. Mikielewicz et al. [31,32]
vapor quality region the difference between the heat transfer has been tested for a significant number of experimental data
coefficient of R1234yf and R134a is small [9]. Del Col et al. and has returned satisfactory results for cases of flow boiling
[10] found that there were no significant differences between processes in numerous fluids. The fundamental hypothesis
the flow boiling performance of R1234yf and R134a. On the of the model is the fact that heat transfer during flow boiling
other hand, Mortada et al. [26] performed an experiment for with bubble generation can be modelled as a sum of two
R1234yf and R134a in a 1.1 mm rectangular channel with contributions constituting the total energy dissipation in
mass flux from 20 to 100 kg/m2 and heat flux varying from the flow, namely the energy dissipation due to shearing flow
2 to 15 kW /m2. Their results showed that the heat transfer without bubbles and the dissipation resulting from bubble
coefficient for R1234yf is lower by about 40% than that for generation.
R134a [9,26]. These results are in opposition to the findings
of [10] and [11].

142 POLISH MARITIME RESEARCH, No S1/2017

Table 2: Range of variation of experimental datafor flow boiling ofrefrigerants R134a and R1234yf,
with the corresponding confinement number Con and reduced pressure.

d G q Tsat
Authors Fluid Con psat/pkr
[mm] [kg/m 2s] [kW/m2] [C]

R1234yf 3.9 200 400 5.93 19.2 10 0.21931 0.12936

Lu et al. [9]
R134a 3.9 200 400 5.67 18.91 10 0.23401 0.10214

30
Del Col et al. [10] R1234yf 1 200 600 50
31 0.73597 0.238

200 6
R1234yf 2 300 12 15 0.41411 0.15086
Satioh et al. [11] 400 24

R134a 2 300 12 15 0.44416 0.12031

Copetti et al. [12] R134a 2.6 240 44 22 0.3281 0.14975

440

300 435 10 10 0.28511 0.12936

20
3 300 370 9 25 11 0.28332 0.13347
Kwang-Il Choi et al. [13] R1234yf
300 10 5 0.29391 0.11026

1.5 500 25 11 0.56664 0.13347

Kundu et al. [14] R134a 7 100 400 3 10 7.417 0.13213 0.09361

725 19 32.198 0.36848 0.20201

2.168
910 28.5 27.197 0.38095 0.17491
Xu et al. [15] R134a
185 18.5 25.533 0.20534 0.16653
4.065 295 28 20.355 0.21193 0.14239
410

10
Mancin et al. [16] R134a 3.4 190 755 25 30 0.23849 0.18974
50

31 0.35018 0.19525
Tibiri and Ribatski [17] R134a 2.3 50 600 7.5 45 22 0.3709 0.14975
41 0.32556 0.25722

Diani et al. [18] R1234yf 3.4 190 940 1 50 30 0.21824 0.23166

10
Owhaib et al. [19] R134a 1.22 200 34 0.64451 0.21251
34

Shiferaw et al. [20] R134a 1.1 200 10 71 31 0.7322 0.19525

Martin-Callizo et al. [21] R134a 0.64 400 5 45 35 1.22388 0.21851

Consolini and Thome [22] R134a 0.51 309 12 71 31 1.57926 0.19525

0.52 280 2.6 61.7 21 1.65038 0.14524

Mahmoud et al. [23] R134a
1.1 320 12.9 102 31.327 0.73058 0.19708

200
Ong and Thome [24] R134a 1.03 400 21.5 111.3 31 0.78196 0.19525
1200

300
1.6 400 30 80 27 0.51684 0.1739
32 0.49997 0.20088
R134a 500
Anwar et al. [25] 0.64 330 30 30 1.26697 0.18974

R1234yf 1.6 300 30 80 27 0.47495 0.21338

400 32 0.45617 0.24447

POLISH MARITIME RESEARCH, No S1/2017 143

 The final version of the model [32] reads: bank for R134a and R1234yf presented in Table 2. Another
modification of Equation (1) wasthe change to the value
TBP C Pb
2
of the exponent m in Equation (2). The sensitivity of the
= RMS
n
+ developed model of two-phase flow multiplier and the non-
LO 1 + P LO (1)
adiabatic effect were tested in calculations. For that purpose,
In Equation (1) C=1 for flow boiling and C=0 for flow two models were introduced into Equation (1), namely the
condensation, whereas LO is the heat transfer coefficient for Muller Steinhagen and Heck correlation [33]. Additionally,
the liquid only case. It may be determined using, for example, the calculation procedure includes the so called blowing
the DittusBoelter equation, (for turbulent flow), or Nu=3.66 parameter B, which is responsible for evaluation of the non-
(for laminar flow). Equation (1) also includes the empirical adiabatic effects appearing due to shear stress modification
correction Pand the modified two-phase multiplier, R MS , due on the liquid-vapour interface [32].
to Mller - Steinhagen and Heck [33]. The modified form of The modified two-phase multiplier R B , inclusive of non-
the two-phase multiplierR MS is: adiabatic effects, has the following form [32]:

1 1
1 B
RMS = 1 + 2 1 x Con m (1 x) 3 + x 3 (2) RMS 1 2 for 0.1 < x 1
f f
1 1z
RB = (6)
2
8 Pb d w
RMS 1 + Re Pr R for 0 x 0.1

It should be noted that the two-phase multiplierR MS in l 0 MS
Equation (2) is raised to the powern, where n= 0.76 for
turbulent flows,andn=2 for laminar flows. Function f1in In Equation (6) the two-phase multiplier should be
Equation (3) is defined as the ratio of the pressure drop in calculated by using the modified Muller-Steinhagen and
the single-phase flow of liquid to the pressure drop in the Heck correlation, Equation (2). The blowing parameter in
single-phase flow of gas, whereas f1z denotes the ratio of the Equation (6) is defined as [32]:
single phase-flow heat transfer coefficient in vapour to the
single-phase heat transfer coefficient for liquid, respectively. Ul
2q
For turbulent flows these functions can be determined from Uv (7)
B
relations: f1=(rL/rG)(mL/mG)0.25, f1z=(mG/mL)(lL/lG)1.5(cpL/cpG), f r G ( s  1) r
while for laminar flows: f1=(rL/rG)(mL/mG)-1 and f1z=(lG/lL).
Furthermore, the exponent m=0 for flow in conventional In Equation (7), s is the slip ratio which can be determined
channels, and m=-1 for flow in minichannels. The empirical from the Zivi relation [32].
correction P in Equation (1) should be calculated as:
Ul
P = 2.5310 3 Re1.17 Bo 0.6 (RMS 1) 0.65 (3) s 3 (8)
Uv
The pool-boiling heat transfer coefficient Pb (1) can
be calculated using a generalized model due to Cooper As a result of application of correction (6), a modified heat
[34]. This model describes the heat transfer coefficient transfer model is obtained. This model has been adopted for
in thefluid in terms of the reduced pressure, molecular calculations in the present work:
weight,DQGDSSOLHGZDOOKHDWX[7KH&RRSHUHTXDWLRQ
which describes the pool-boiling heat transfer coefficient 2
TBP C Pb
has the form: = RBn + (9)
LO 1 + 2.53 10 3 Re1.17 Bo 0.6 ( RMS
*
1) 0.65 pra LO
2

Pb = A p 0.12
r ( log p r ) 0.55
M 0.5
q 3 (4) In the following part, the basic model and its subsequent
modifications, which have been selected for discussion,
The accuracy of model predictions was expected to be is analysed with respect to predictions of heat transfer
improved by some modifications introduced to the empirical coefficient. The exponent a in the modified two-phase flow
correction P. The modified empirical correction P yields: multiplier in Equation (5) was adjusted to the available data
bank for flow boiling of R134a and R1234yf.
a
3 p 0.65
P = 2.53 10 Re Bo (R 1)
1.17 0.6 *
MS sat (5)
p kr
*
RMS in Equation (5) is the two-phase multiplier, calculated
using the original version of Muller-Steinhagen and Heck
correlation [33]. In using Equation (2), the exponent is
always m=0. Exponent a was adjusted to the available data

144 POLISH MARITIME RESEARCH, No S1/2017

 THE RESULTS
To distinguish between the conventional channels and
minichannels the criterion due to KewandCornwell [35]
has been used, which is based on the so-called confinement
number Con, defined as:
1 V
Con
d d (Ul  Uv )
It has been postulated that when the confinement
numberConis greater than 0.5 then the flow exhibits the
properties of the flow in minichannel, in which the surface
tension plays a dominant role. Presented below is a selection
of studies which have been considered in the present study.
Based on the KewandCornwell [35] criterion, the available
data bank was divided into conventional size channels and
minichannels. Amongst the collected data, the criterion of
minichannels, i.e. Con > 0.5, is fulfilled only by the research
due to De Col et al. [10] and Kwang-Il Choi [13] for R1234yf,
while for R134a this criterion is fulfilled by research due to
Owhaib et al. [19], Shiferaw et al. [20], Martin-Callizo et al.
[21], Consolini and Thome [22], Mahmoud et al. [23], and
Anwar et al. [25]. It can therefore be concluded that for both
R134a and R1234yf, the transition from conventional size
channels to minichannels takes place at a channel diameter
smaller than 1.5mm. The relations of the confinement number
Con to the saturation temperature for both refrigerants are
presented in Figures 1 and 2. The values of the confinement
number Con for the analysed refrigerant, together with
the reduced pressure values and the range of variation
of experimental parameters, are collated in Table 2. The
analysis of the parameters in Table 2 indicates the fact that
the experimental research data collected for scrutiny cover
a fullrange of quality variation anda relatively wide range
of mass velocity.
Figure 1: Confinement number Con vs. saturation temperature for R134a
(10)
Figure 2: Confinement number Con vs. saturation temperature for R1234yf
Figures3 to 6 show the results of heat transfer coefficient
calculations for R134a and R1234yf by using the earlier
mentioned method based on Equation (9). In the adopted
formulation, the surface tension effects are included into the
analysis by considering the effect of the confinement number
in equations through Con. The model version applicable to
minichannels was used if Con>0.5, and that applicable to
convectional size channels when Con<0.5. The modification
of the empirical correction described by Equation (3) includes
the effect of reduced pressure (psat/pkr). The new version of the
correction P is applied in Equation (5), where the reduced
pressure is raised to the power a. The value of the exponent
a was adjusted using the regression analysis of functions of
several variables. The accomplished calculations indicate
that best consistency is obtained if a = 4 for each case. The
information about mean average deviation and correlation
factors is given in Table 3. The values of correlation factors
are not very high, which indicates relatively high dispersity
of the experimental data.
Figure 3: Comparison of test results, exp with predictions obtained using
Equation (9), th and a = 4 for R134a
POLISH MARITIME RESEARCH, No S1/2017 145

Figure 4: Comparison of the ratio of experimental values of exp to those
obtained using Equation (9), th as function of quality, and a = 4 for R134a
Figure 5: Comparison of test results, exp with predictions obtained using
Equation (9), th and a = 4 for R1234yf
Figure 6: Comparison of the ratio of experimental values of exp to those
obtained using Equation (9), th as function of quality, and a = 4 for R1234yf
146 POLISH MARITIME RESEARCH, No S1/2017
Table 3: Values of exponent a and MAD
Fluid Model R2 MAD
[%]
R134a Model II 0.1536 25.35
R1234yf Model II 0.4522 16.06
Based on the presented results of calculations, which
were obtained using the versions of the heat transfer model
described by Equation (9), with and without the account
of reduced pressure (psat/pkr)a from Equation (5), it can be
concluded that taking into account the effect of reduced
pressure improves the convergence with the experimental
data, consequently giving smaller values of the mean absolute
deviation and higher of the correlation coefficient.
CONCLUSIONS
The paper presents the analysis of the results of flow boiling
and flow condensation calculations using a model developed
earlier to study experimental data for flow boiling of R134a
and R1234yf. The model was studied in several ways: it was
used either in the original formulation or in a modified
version, which included the reduced pressure effect in the
empirical correction P (value of exponent a was modelled).
The results show that the effect of reduced pressure does not
significantly change the performance of the original model.
The results of calculations show the best compliance with
the experimental data when applying the blowing parameter
in correlation for each case, i.e. for R134a and R1234yf, and
also with and without the account of reduced pressure (psat/
pkr)a from Equation (5).
In authors opinion, the proposed method to calculate
the flow boiling heat transfer coefficient is a reliable tool
in designing heat exchangers. Among other possibilities, it
provides opportunities for analysing other materials which
would be more suitable for marine applications, as well as
new manufacturing techniques which can be fully automated
[36,37]. At present, heat exchangers are normally equipped
with stainless steel tubes. In future applications, some effort
should be made to use plastics [38-40], which do not exhibit
susceptibility to corrosion and are cheaper.
REFERENCES
1. United Nations Environment Program (UNEP) Montreal
Protocol on Substances that Deplete the Ozone Layer,
1997.
2. IPPC, Climate Change, Cambridge University Press,
Cambridge, United Kingdom, 2001.
3. Ghodbane M.: An investigation of R152a and hydrocar-
bon refrigerants in mobile air conditioning, Int. Proceed-
ings of International Congress and Exposition, Detroit,
Paper no. 1999-01-0874, Warrendale, PA.
4. Lorentzen G.: Revival of carbon dioxide as refrigerant,
Int. J. Refrigeration 17 (3), 1994, pp. 292-301.
5. Zhao Y., Liang Y., Sun Y., Chen J.: Development of a
mini-channel evaporator model using R1234yf as work-
ing fluid, 35, 2012, pp. 2166-2178.
6. Minor B., Spatz M.: HFO-1234yf low GWP refrigerant
update, Int. International Refrigeration and Air Condi-
tioning Conference at Purdue, West Lafayette, IN, USA.
Paper no 2349.
7. Refprop v 9.0, National Institute of Standards (NIST),
2010.
8. Park K. J., Jung D.: Nucleate boiling heat transfer coef-
ficients or R1234yf on plain and low fin surfaces, Int. J.
Refrigeration 33 (3), 2010, pp. 553-557.
9. Lu M-C., Tong J-R., W C-C.: Investigation of the two-
phase convective boiling of HFO-1234yf in a 3.9 mm
diameter tube, Int. J. Heat and Mass Transfer, 65, 2013,
pp. 545-551.
10. Del Col D., Bortolini S., Torresin D., Cavallini A.: Flow
boiling of R1234yf in a 1 mm diameter channel, Int. J.
of Refrigeration, 36, 2013, pp. 353-362.
11. Satioh S., Dang C., Nakamura Y., Hihara E.: Boiling
heat transfer of HFO-1234yf flowing in a smooth small-
diameter horizontal tube, Int. J. Refrigeration, 33, 2011,
pp. 1846-1853.
12. Copetti J. B., Macaganan M. H., Zinani F.: Experimental
study on R-600a boiling in 2.6 mm tube, Int. J. Refrigera-
tion, 33, 2013, pp. 325-334.
13. Kwang-Il Choi, Jong Taek Oh, Kiyoshi Satio, Jong Soo
Jeong: Comparison of Heat Transfer Coefficient during
Evaporation of Natural Refrigerants and R-1234yf in
Horizontal Small Tube, Int. J. Refrigeration, 2014, 41,
pp. 210218.
14. Kundu A., Kumar R., Gupta A.: Heat transfer character-
istic and flow pattern during two-phase flow boiling of
R134a and R407C in a horizontal smooth tube, Experi-
mental Thermal and Fluid Science, 57, 2014, pp. 344-352.
15. Xu Y., Fang X., Li G., Li D.: An experimental study of flow
boiling heat transfer of R134a and evaluation of existing
correlations, Int. J. Heat and Mass Transfer, 92, 2016, pp.
1143-1157.
16. Mancin S., Diani A., Rossetto L.: R134a flow boiling heat
transfer coefficient and pressure drop inside 3.4 mm ID
microfin tube, Energy Proceedia, 45, 2014, pp. 608-615.
17. Tibir C. B., Ribatski G.: Flow boiling heat transfer of
R134a and R245fa in 2.3 mm tube, Int. J. Heat and Mass
Transfer, 53, 2010, pp. 2459-2468.
18. Diani A., Mancin S., Rossetto L.: Flow boiling heat
transfer of R1234yf inside 3.4 mm ID microfin tube,
Experimental Thermal and Fluid Science, 66, 2015, pp.
127-136.
19. Owhaib W., Martin-Callizo C., Palm B.: Evaporative
heat transfer in vertical circular microchannels, Applied
Thermal Engineering, 24, 2004, pp. 1241-1253.
20. Shiferaw D., Karayiannis T. G., Kenning D. B. R.: Flow
boiling in a 1.1 mm tube with R134a: experimental results
and comparison with model, Int. J. Thermal Sciences, 48,
2009, pp. 331-341.
21. Martin-Callizo C., Ali R., Palm B.: New experimental
results of flow boiling of R134a in vertical microchannel,
Heat Transfer Conference Proceedings, Edinburgh 10-11
September, 2007, UK.
22. Consolini L., Thome J. R.: Micro-channel flow boiling
heat transfer or R134a, R236fa and R245fa, Microfluid
Nanofluids 6, 2009, pp. 731-746.
23. Mahmoud M. M., Kenning D. B. R., Karayiannis T. G.:
Single and two phase heat transfer and pressure drop in
a 0.52 mm vertical metallic tube, 7th Int. Conference in
Enhanced, Compact and Ultra-compact Heat Exchang-
ers: From Microscale Phenomena to Industrial Applica-
tions, September 13-18, Heredia, Costa Rica, 2012.
24. Ong C. I., Thome J. R.: Flow boiling heat transfer of R134a,
R236f and R245fa in a horizontal 1.030 mm circular
channel, Exp. Therm. Fluid Sci., 33, 2009, pp. 651-663.
25. Anwar Z., Palm B., Khodabandeh R.: Flow boiling heat
transfer, pressure drop and dryout characteristics of
R1234yf: Experimental results and predictions, Experi-
mental Thermal and Fluid Science, 66, 2015, pp. 137-149.
26. Mortada S., Zoughaib A., Arzano-Daurelle C., Clodic
D.: Boiling heat transfer and pressure drop of HFC-134a
and R-1234yf in minichannels for low mass fluxes, Int.
J. Refrigeration, 35, 2012, pp. 962-973.
27. Ribatski G.: A critical overview on the recent literature
concerning flow boiling and two phase flows inside
microscale channels, ECI 8th Int. Conference on Boil-
ing and Condensation Heat Transfer, 3 7 June 2012,
Lausanne, Switzerland.
28. Tibir C. B., Ribatski G.: Flow boiling in micro-scale
channels Synthesized literature review, Int. J. Refrigera-
tion, 36, 2013, pp. 301 324.
POLISH MARITIME RESEARCH, No S1/2017 147
29. Sardeshpande M., Ranade V.: Two phase flow boiling CONTACT WITH THE AUTHOR
in small channels: A brief review, Sadhana, 38, 2013, pp.
1083 1126. Dariusz Mikielewicz,
e-mail: Dariusz.Mikielewicz@pg.gda.pl,
30. Alagesan V.: Flow boiling heat transfer in mini and micro
channels A state of the art review, Int. J. ChemTech Gdansk University of Technology,
Research, 4, 2012, pp. 1247 1259. Faculty of Mechanical Engineering,
Department of Energy and Industrial Apparatus,
31. Mikielewicz D., Mikielewicz J., Tesmar J.: Improved semi- ul. Narutowicza 11/12, 80-233 Gdansk
empirical method for determination of heat transfer coef- Poland
ficient in flow boiling in conventional and small diameter
tubes, International Journal of Heat and Mass Transfer, Blanka Jakubowska
50(19-20), 2007, pp. 3949-3956. e-mail: blanka.jakubowska@pg.gda.pl

32. Mikielewicz D., Andrzejczyk R., Jakubowska B., Mikiele-
wicz J.: Analytical model with non-adiabatic effects for
pressure drop and heat transfer during boiling and
condensation flows in conventional channels and min-
ichannels, Heat Transfer Engineering, 37(13-14), 2016,
pp. 1158-1171.
Gdansk University of Technology,
Faculty of Mechanical Engineering,
Department of Energy and Industrial Apparatus,
ul. Narutowicza 11/12, 80-233 Gdansk
Poland

33. Muller-Steinhagen R., Heck K.: A simple friction pressure

drop correlation for two-phase flow in pipes, Chem. Eng.
Progress, 20, 1986.

34. Cooper M. G.: Saturation nucleate pool boiling; a simple

correlation, Int. Chem. Eng. Symposium 1, 86, 1984, pp.
785 793.

35. Kew P., Cornwell K.: Correlations for the prediction of

boiling heat transfer in small diameter channels, Applied
Thermal Engineering, 17, 8 10, (1997), 705 715.

36. Deja M., Siemitkowski M.S.: Feature-based generation

of machining process plans for optimised parts manufac-
ture. Journal of Intelligent Manufacturing, 2013, Vol.24,
No.4, pp. 831846.

37. Szapczyski R.: Evolutionary Planning of Safe Ship

Tracks in Restricted Visibility. Journal of Navigation,
2015, Vol.68, No.1, pp.39-51.

38. Stolarski T.A.: Wear of water-lubricated composite materi-

als. Wear, 1980, Vol.58, No.1, pp. 103-108.

39. Jakubowski M.: Influence of pitting corrosion on fatigue

and corrosion fatigue of ship and offshore structures.
Part II: Load - pit crack interaction. Polish Maritime
Research, 2015, Vol.22, No.3, pp. 57-66.

40. Ambroziak A., Kosowski P.: Mechanical properties for

preliminary design of structures made from PVC coated
fabric. Construction and Building Materials, 2014, 50,
pp. 74-81.

148 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 149-155
10.1515/pomr-2017-0033

TECHNICAL MONITORING SYSTEM FOR A NEW PART OF GDASK

DEEPWATER CONTAINER TERMINAL

Mikoaj Mikiewicz1
ukasz Pyrzowski1
Krzysztof Wilde1
Oskar Mitrosz2
1
Gdansk University of Technology, Poland
2
Keller Polska Sp. z o.o., Poland

ABSTRACT

Development strategy for the port of Gdask assumes that , up to 2027, it will become the main transport centre in
the Baltic Sea region. To this end , for many years work has been carried out to facilitate access to the port both by
sea and land. One of the elements of extension of the port is the building of another terminal of Deepwater Container
Terminal , Gdask. In this task, geotechnical operations aimed at appropriate preparation of subgrade for storage of
containers and operation of cranes, have been especially important. Effectiveness of the assumed design solution of
land-side girder for gantry crane has been verified by means of a technical monitoring system. This paper presents
main elements of design the land-side girder of a new quay of the container terminal in Gdask, together with its
computational analysis, as well as the results collected during a few months of observation, which have revealed real
work of monitored elements.

Keywords: SHM, monitoring, DCT of Gdask, geotechnics

INTRODUCTION performing appropriate diagnostics [6], [7], [13], [20]. In

Development strategy for Gdansk port assumes that it will
become, up to 2027, the main transport centre in the Baltic
Sea region [3]. The main element which shows importance of
the port is its location as well as the fact that the East Europe
region is the most dynamically developing part of Europe.
For many years investments have been carried out in order
to facilitate access to the port both by sea and from land.
One of the elements of the approved strategy is the extension
of the Deepwater Container Terminal in Gdask, which
has been aimed at the building of a terminal for servicing
container carriers of ULCV class (Ultra Large Container
Vessels) of over 18,000 TEU (Twenty-foot Equivalent Units)
capacity (Fig. 1), that required deepening the basin and
this task, in view of a complex soil - water conditions, all
geotechnical operations constituted a crucial element for
successful finalization of the undertaking. They were aimed
at appropriate preparation of the subgrade over 25 ha area
for ship mooring berths, operation of cranes and storage
of containers. The contract area has been divided into two
main parts [14]: heavy foundation for girder of STS (Ship-to-
shore) gantry crane as well as the deep soilimprovement of
the platform (storage area for containers) and the quay wall
(45m far into land from crane sea-side rail).

POLISH MARITIME RESEARCH, No S1/2017 149


Fig. 1. Bird view of new ship mooring berths and containers storage areas (acc.
DCT Gdask S.A.)
For the reason of a complex character of operations within
the quay it was decided to verify the accepted design solution
of land-side girder foundation (Keller Polska Ltd. Co) by
applying a technical monitoring system installed on steel rods
of micro-piles (Gdask University of Technology).
Fig. 2. Cross-section of the quay
Fig. 3. Top view of the quay, showing location of the monitored micro-piles
150 POLISH MARITIME RESEARCH, No S1/2017
CHARACTERISTICS OF THE MONITORED
STRUCTURE
The land-side girder within which the monitored rods were
placed, was designed as a pile capping beam supported onto
a wall combined with pipes and steel profiles anchored by
means of pair of tie-rods in the rear (land-side) crane girder
which was founded on CFA piles arranged in racking trestle
system (Fig. 2) [3]. Additionally, the land-side girder was
anchored with the use of micro-piles of 300mm diameter,
inclined by 45, and having the length from 31 up to 36 m.
The system rods made of 28Mn6 steel constituted the steel
element of micro-piles. Cyclic monitoring was applied to
axial forces acting in 7 (out of 290 produced) micro-piles
placed over the whole length of the quay (Fig. 3).
The micro-piles were made of high quality steel pipes
whose elements of 1 up to 4 m in length were joined by
means of couplers. Owing to application of a drill bit, 300 mm
diameter was obtained along the whole length of micro-pile.
The reinforcement pipe served as a drilling and injection duct.
The drilling down the pipe was performed under protection
of drilling fluid and in case an obstacle was met the operation
was supported by impact. When the rod was drilled down
up to the designed depth the second phase of injection of
elevated density cement grout followed. The application of
the self-drilled system micro-piles is simple to perform as the
drilling and injecting operations occur simultaneously, and
the drill rod is also used as a final reinforcement without need
of using any protection pipes. Moreover, the irregular- fuzzed
shape of injection body produces excellent connection with
the ground, which results in a high loadcarrying ability of
micro-pile and displacements necessary to activate friction
on side surface amount to only a few millimeters.
interference with the soil medium over pile side surface and
in the bottom, were used. The micro-piles, in view of the
unique structure of hinged connection between micro-pile
and land-side girder (Fig. 5), were defined by means of string
elements capable of transferring tension force only.

DESIGN PROCESS AND LOAD TEST

Computational simulations of behaviour of the structure

was performed in linear elastic range as well as non-linear
one by using the finite elements method [4], [5]. In the context
of applicability of the FEM a formal analysis of the solution
convergence was made. Two independent FEM models were
prepared in order to highlight work of the gantry crane girder
and foundation elements as well (Fig. 4). The first of them
represented global behaviour of the designed structure with
taking into account sea actions, and the other was directed Fig. 5. Detail of hinged connection between micro-pile rod and land-side girder
to proper description of the land-side support, i.e. behaviour
of the pile trestle and micro-piles system.
The performed computational simulations of behaviour
of the most unfavourable sections of the land-side girder
showed uniform work of the designed rods and the expected
force value in ultimate limit state (ULS), with taking into
account design values in load cases, had to amount about
17501950kN, compared with the design value of load-
carrying ability of the rod on the level of 2670kN.
To verify correctness of work of selected micro-piles on
their completion, load tests were performed (Fig. 6) with
the use of mechanical devices [9], [10], [11], [16].

Fig. 4. Schematic diagrams used for calculation of FEM models

For description of the piles, beam elements of elastic

boundary conditions and characteristics reproducing Fig. 6. Production of micro-piles (upper photo), load tests (bottom photo)

POLISH MARITIME RESEARCH, No S1/2017 151


MONITORING SYSTEM
In the technical monitoring systems for structures [12],
[15], [18], [19] it is possible to apply various measuring
technologies such as precise levelling, hydrostatic levelling,
[2], [21], laser scanning [1], photogrammetric methods [24],
advanced geodesic methods as well as specialty measurements
of deformations, pressures, temperatures [23] etc, or also
numerical simulation modules such as in [17], [22]. The
technical monitoring system for micro-piles required to
use technologies which make it possible to fulfill the task at
relatively low costs. Therefore, for the purposes of the designed
system, an innovative system for monitoring axial forces in
micro-pile rods, based on measurement of deformations,
was produced. To this end, 7 modules of 80 cm in length
was built. Each of them was equipped with dedicated string
extensometers. Their calibration was conducted in laboratory
conditions with the use of a strength testing machine (Fig.
7). The final product was fitted with a cover and protection
system to make operation of the measurement module
possible in conditions of elevated pressure resulting from
presence of water as well as other actions coming from the
soil. The complete measurement modules were tested again in
the strength testing laboratory, and their calibration constants
- determined.
Fig. 7. Tests of the measurement module in the strength testing machine
Measurement of deformations (strains) was carried out
with the use of the string extensometers of characteristics
given in Tab. 1. The applied gauges are characteristic of a
high stability of measurements along the time, and make it
possible to measure temperature too.
152 POLISH MARITIME RESEARCH, No S1/2017
Tab. 1. Characteristics of the applied measurement gauges
Measurement range 3000
Resolution 0,1
Accuracy 0.5% F.S.
Nonlinearity <0.5% F.S.
Temperature range -20C 80C
Calibration of the string extensometers installed on the
rods, concerning identification of force-strain relationship,
was of fundamental importance for the preparing of the
innovative measurement technology and the ensuring of
high quality of measurements. The obtained calibration
coefficients made it possible to convert the measured values
of rod deformations into values of forces acting in them.
In order to increase measurement accuracy 2 gauges were
installed on each segment of the rod, then readouts from them
were averaged and on this basis a real value of force acting in
the rod under axial tension was estimated. The experimental
tests were performed in the range 0 400 kN, and their
example results for the rod Z1 and Z7 are presented in Fig.
8. The recorded differences in work of particular gauges on
the same segment of the rod were caused by eccentricity of
tension occurring in the tests.
Fig. 8. Characteristics of work of the gauges installed on the rods Z1 and Z7
The tested measurement modules were attached by means
of couplers to the micro-piles prepared in advance, and
then joined with the zinc-coated heads (Fig. 9). Next, before
commencement to covering the rods with the soil, each of
them was protected against corrosion. Installation of the
measurement modules was conducted stage by stage along
with progressing the building work. The monitoring of a given
rod was started up in the moment of fitting it with gauges at
site of the building work (zeroing measurement).
Fig. 9. Installation of the rod Z2 segment fitted with gauges
MONITORING RESULTS
Taking the readouts from the installed gauges was
commenced on 11.02.2016 and beginning from the moment,
the operation of the strains monitoring gauges in the rods
(in order to then determine forces acting in them) was stable.
During taking the readouts were revealed no disturbances or
effects which would indicate incorrect work of any of installed
extensometers. The only unexpected effect identified during
the readouts taking was occurrence in some building phases
of the quay - of a small bending deformation of the rods
instead of the expected pure axial tension. The phenomenon
was caused by the settling of the quay along with the increasing
number of surface courses and it ceased on completion of the
building work. Beginning from the instant an increase of
tension force acting in each of the rods was only recorded.
The achieved results of the measurements of force in the
monitored rods are presented in Fig.10.
Fig. 10. Changes of forces in the rods
On the basis of the recorded measurements it may be
concluded that work of particular rods is not uniform along
the quay length (656 m). The most strained are the rods in the
middle of the beam, namely Z3 and Z4, whereas the lowest
load were transferred into the extreme elements, i.e. Z1, Z6,
Z7. In the rods Z2 and Z5 average force values were observed.
In the case of the rods Z6 and Z7, a lack of activation of
the micro-piles in the initial phase of the monitoring could
be caused by a greater stiffness of the piles system in the
regions in question, compared with that of the remaining
part of the land-side girder. The longest CFA piles placed
there took a significant part of load from the quay structure.
An influence on the measurements could have the schedule of
dredging (deepening) the basin and installing the tie-rods
which have not yet started to fully work. During taking the
readouts it could be observed that in the zone of the rods Z6
and Z7 a part of dredging work was continued in the basin
just in the time of installing the upper tie-rods, that could
result in the transferring horizontal forces in micro-piles of
smaller values than expected in this building stage.
In the case of Z1 rod placed in the connection with the
existing terminal T1, a small increase in the measured
parameters may be explained by that it was not necessary
to deepen the basin in this area and postponed installation
of upper tie-rods that made micro-piles to cooperate in full
with the quay.
SUMMARY
The application of the innovative technical monitoring
system which makes it possible to monitor forces in the
rods, the made-in-advance analysis of design assumptions
and calculations for the particular work phases of the quay
structure and land-side girder, allowed to show real way
of operation of the micro-piles. The most loaded rod, Z4,
showed the measured force equal to 724kN (Fmeas), whereas
its calculated force (Fcalc) in the representative cross-section
and the corresponding current work phase of the element, was
equal to 807kN (Fmeas/Fcalc = 90%). Therefore, the monitoring
POLISH MARITIME RESEARCH, No S1/2017 153
 Fig. 12. State of the monitored quay in November 2016
results confirmed that the design assumptions concerning the
investigated elements are correct.
Presently, in view of the complex character of the quay
structure and the conducting of the building work phase
by phase, the measurements are still continued (Fig. 12) in
order to observe an influence of the loading generated by
weight and operation of cranes, mooring the ships, changes
in weather conditions, waving, changes in ground water level
etc, on the land-side girder and micro-piles.
BIBLIOGRAPHY
1. Bernat, M., Janowski, A., Rzepa, S., Sobieraj, A., Szulwic,
J.: Studies on the use of terrestrial laser scanning in the
maintenance of buildings belonging to the cultural heritage.
14th Geoconference on Informatics, Geoinformatics and
Remote Sensing, SGEM. ORG, Albena, Bulgaria, 2014, 3,
pp. 307318.
2. Boerez J., Hinderer J., Rivera L., Jones M. : Analysis and
modeling of the effect of tides on the hydrostatic leveling
system at CERN. Survey Review, Maney Publishing, 2012,
44 (327), pp.256264.
3. Buca R., Mitrosz O. : Complex geotechnical engineering for
Port of Gdansk development gateway to Central-Eastern
Europe. Proceedings of 13th Baltic Sea Geotechnical
Conference. Lithuania, 2224 September 2016.
4. Burzyski S., Chrcielewski J., Witkowski W.:
Geometrically nonlinear FEM analysis of 6-parameter
resultant shell theory based on 2-D Cosserat constitutive
model. ZAMM-Zeitschrift fur Angewandte Mathematik
und Mechanik, Vol. 96, No. 2 (2015), pp.191204
5. Chrcielewski J., Sabik A., Sobczyk B., Witkowski W. :
Nonlinear FEM 2D failure onset prediction of composite
shells based on 6-parameter shell theory. Thin-Walled
Structures 105, (2016), pp. 207219.
6. Grelowska G., Kozaczka E., Kozaczka S. : Underwater noise
generated by a small ship in the shallow sea. Archives of
Acoustics. Vol. 38, (2013), No. 3 pp. 351356.
154 POLISH MARITIME RESEARCH, No S1/2017
7. Grelowska G., Kozaczka E.: Underwater Acoustic Imaging
of the Sea. Archives of Acoustics.Vol. 39, (2014), No. 4,
pp.439452.
8. Janowski A., Nowak A., Przyborski M., Szulwic J. : Mobile
indicators in GIS and GPS positioning accuracy in cities. 2nd
International Conference on Rough Sets and Emerging
Intelligent Systems Paradigms (RSEISP) held as part of Joint
Rough Set Symposium (JRS), 2014, Springer International
Publishing.
9. Kaliski K.: The finite element method application to linear
closed loop steady system vibration analysis. International
Journal of Mechanical Sciences 39, 3 (1997), pp. 315330.
10. Kaliski K. J., Galewski M. A.: Chatter vibration
surveillance by the optimal-linear spindle speed control.
Mechanical Systems and Signal Processing, Vol. 25, No.
1, January 2011, pp. 383399.
11. Kaliski K. J., Mazur M.: Optimal control at energy
performance index of the mobile robots following dynamically
created trajectories. Mechatronics 37 (2016), pp. 7988.
12. Kaminski W., Makowska K., Mikiewicz M., Szulwic J.,
Wilde K. : System of monitoring of the Forest Opera in Sopot
structure and roofing, 15th International Multidisciplinary
Scientific GeoConference SGEM 2015, SGEM2015
Conference Proceedings, June 18-24, 2015, Book 2, Vol.
2, 2015, pp. 471482.
13. Kozaczka E., Grelowska G., Kozaczka S.: Detection of
Objects Buried in the Sea Bottom with the Use of Parametric
Echosounder. Archives of Acoustics. -Vol. 38, (2013), No.
1, pp. 99104.
14. Krupka J., Ryszewska R. : The building of a new quay
in the area of Port of Gdask, to serve for extension
of the Gdask Deepwater Container Terminal
(in Polish), Inynieria Morska I Geotechnika, No. 1/2016,
pp. 3845
15. Mariak A., Mikiewicz M, Meronk B., Pyrzowski .
and Wilde K.: Reference FEM model for SHM system of
cable-stayed bridge in Rzeszw. Advances in Mechanics:
 Theoretical, Computational and Interdisciplinary Issues
Kleiber et al. (Eds), Taylor & Francis Group, London,
2016, pp. 383387. CONTACT WITH THE AUTHOR

16. Mikielewicz J., Mikielewicz D. : A simple model of circular Mikoaj Mikiewicz

hydraulic pump. International Journal of Heat and Mass ukasz Pyrzowski
Transfer, Vol. 52(1), 2008, pp.1721. Krzysztof Wilde

17. Mikiewicz M. : Nonlinear FEM analysis and monitoring Gdask University of Technology
the tie-rod structures (in Polish). Wydawnictwo Politechniki Civil Engineering and Environment Faculty
Gdaskiej (Publishing House of Gdask University of 11/12 Narutowicza St.
Technology), 2016. 80 - 233 Gdask

18. Mikiewicz M., Pyrzowski ., Chrcielewski J., Wilde K.: Oskar Mitrosz
Structural Health Monitoring of Composite Shell Footbridge Keller Polska Sp. z o.o. (Co. Ltd)
for Its Design Validation. Proceedings of 2016 Baltic Poland
Geodetic Congress (Geomatics), ed. Juan E. Guerrero Los
Alamitos: IEEE Computer Society Order Number E5972,
2016, pp. 228233.

19. Rucka M., Wilde K.: Experimental study on ultrasonic

monitoring of splitting failure in reinforced concrete. Journal
of Nondestructive Evaluation. Vol. 32, (2013), No. 4, pp.
372383.

20. Szapczyski R.: Evolutionary Sets Of Safe Ship Trajectories:

A New Approach To Collision Avoidance. Journal of
Navigation. Vol. 64, (2011), No. 1, pp.169181.

21. Wilde, K., Meronk B., Groth M., Mikiewicz, M.: Structure
monitoring by means of hydrostatic levelling (in Polish).
27th Scientific Technical Conference on Building failures,
2015, pp. 278284.

22. Wilde, K., Mikiewicz, M., Chrcielewski, J.: SHM System

of the Roof Structure of Sports Arena Olivia. Structural
Health Monitoring 2013.. Vol. II/ ed. Fu-Kuo Chang,
Alfredo Guemes Lancaster, Pennsylvania 17602 U. S. A.:
DEStech Publications, Inc., 2013, pp. 17451752,.

23. Wjcik M., Hrtl J., Ooi J.Y., Rotter J.M., Ding S., Enstad
G.G.: Experimental investigation of flow pattern and wall
pressure distribution in a silo with double-cone insert.
Particle&Particle System Characterization 24, 4-5, 2007,
pp. 296303

24. Ziolkowski P., Nagrodzka-Godycka K., Janowski A. and

Szulwic J.: Remote sensing and photogrammetry techniques
in diagnostics of concrete structures. Computers and
Concrete, Vol. 18, No. 3, 2016, pp. 405420.

POLISH MARITIME RESEARCH, No S1/2017 155

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 156-165
10.1515/pomr-2017-0034

STRENGTH ANALYSIS OF A LARGE-SIZE SUPPORTING STRUCTURE

FOR AN OFFSHORE WIND TURBINE

Karol Niklas
Gdansk University of Technology, Poland

ABSTRACT

The offshore wind power industry is the branch of electric energy production from renewable sources which is most
intensively developed in EU countries. At present, there is a tendency to install larger-power wind turbines at larger
distances from the seashore, on relatively deep waters. Consequently, technological solutions for new supporting
structures intended for deeper water regions are undergoing rapid development now. Various design types are proposed
and analysed, starting from gravitational supports (GBS), through monopiles and 3D frame structures (jackets,
tripods), and ending with floating and submerged supports anchored to the seabed by flexible connectors, including
TLP type solutions.
The article presents the results of examination of an untypical large-size gravitational support intended for waters
with the depth of up to 40 m. Firstly, a general concept of the new design is presented, while the next basic part of the
article describes the support design in detail and provides its strength analysis. The examined support has the form
of a large steel container consisting of conical segments. The strength analysis was conducted using the finite element
method (FEM), in accordance with the standard DNVGL-ST-0126. Modifications introduced to the most heavily
loaded structural node of the support, which was the set of base bottom trusses, is also included. The results of the
performed analysis prove that the presented concept of supporting structure for a 7MW turbine meets fundamental
strength criteria. The nonlinear buckling analysis was performed to evaluate the critical force acting on the support,
which turned out to be 1.44 times as large as the maximum load of the wind turbine. Potentially important issues
for further analyses have been identified as those resulting from the asymmetry of basic loads acting on the support.

Keywords: Renewable energy sources, offshore wind power industry, supporting structure, strength analysis, FEM, computer simulation,
numerical calculations, CFD, conceptual design, buckling, thin-walled structure

1. INTRODUCTION and cumulated power of offshore wind turbines installed in

The offshore wind power industry sector is undergoing
rapid development, as a result of EU policy on energy
production from renewable sources. In two recent decades, the
wind energy conversion technologies have been shifted from
a sort of niche sector to one of most dynamically developing
power industry branches. Following the publication of EU
policy objectives concerning power industry development in
years 2020 to 2030 [1], the development of the wind energy
sector is included to strategic directions, and a substantial
potential for development is seen in offshore installations [2,
3]. The trend in offshore wind power industry development is
illustrated by the diagram in Fig. 1, which presents the annual
EU countries.
Further dynamic development of this sector will be
possible, provided that several technological challenges
concerning production and installation of new wind farms
of large-power turbines (100MW and more) will be solved
[6]. These farms should be situated at a large distance from
the seashore and, consequently, on relatively deep waters. At
present, the technological advancement level provides real
opportunities for commercial investment project execution
only on coastal waters. At the end of 2015, the average
distance of wind farms connected to electric networks from
the seashore was 43.3 km, and the average water region depth

156 POLISH MARITIME RESEARCH, No S1/2017

was 27.1 m [4]. The most recent trends in the offshore wind
power development are shown in Fig. 2.
Fig. 1. Annual and cumulated power of offshore wind turbines installed in EU
countries UE [4, 5]
Fig. 2. Average water region depth and distance from the seashore for wind
farms: connected to the electric networks, accepted, and under construction [4]
For water regions with depths of up to 20 m, the most
frequently used foundation form is a monopile driven into the
seabed, while for the depths exceeding 30 m several support
designs have been adopted from the offshore oil rig industry.
The latter supports usually have the form of 3D frames or
tripods. The offshore wind power industry development
trends clearly lead towards deeper water regions. However,
the main problem here is the lack of effective technical
solutions for supporting structures, although the literature
on supports for offshore wind turbines provides descriptions
of various foundation types, such as monopiles [7-9], tripods
[10,11], jackets [12-13], gravitational supports [14-16], etc. The
solutions proposed for much larger depths, exceeding 50m,
include floating supports, [17-20], partially or fully submerged
supports with flexible connectors [21], and supports with
submerged displacement hulls, a so-called Tension Leg
Platform (TLP) solution [22-23]. What is noteworthy is huge
technological progress in the field of numerical simulations,
supported by experimental tests [24-37]. Visible progress is
also observed in technologies of structure production [38-
40] and structure operation monitoring [41-45]. All these
factors affect considerably the dynamics of offshore wind
power industry sector development.
This article proposes a concept of gravitational type
foundation intended for waters with the depth of up to 40
m. For these waters, neither conventional solutions nor
concepts of new type floating structures have been used so
far. Obviously, the proposed supporting structure can be
adapted to other water depths, but this adaptation would
require new strength analyses. The general solution concept
was proposed within the framework of the research project
AQUILO oriented on verifying various concepts for offshore
wind turbine foundations on the Baltic Sea. In the present
concept, the supporting structure has the form of a large-size
thin-walled steel container, ballasted with so-called pulp
(mixture of sand and water). The strength analysis for the
new support was conducted in accordance with the standard
DNVGL-ST-0126 [46]. In the article, Section 2 presents the
design assumptions, while Section 3, being the main part of
the article, describes the FEM model based strength analysis
of the examined structure. The presented results also include
the assessment of the critical force coming from wind turbine
load, and the presentation of changes introduced to the main
structural node of the design. Section 4 provides conclusions
from the research and indicates possible directions of its
further continuation.
2. DESIGN ASSUMPTIONS
The supporting structure is intended for a 7KW offshore
wind turbine to be operated on North Sea or Baltic Sea waters,
at the depth of up to 40 m. The structure was designed as
a very untypical support of gravitational type, having the
form of a thin-walled steel container. The general plan of
the support with the installed column and turbine is shown
in Fig. 3. The horizontal axis of the turbine is situated 100 m
above the surface level of the 40-meter deep water region. The
design also includes a 70-meter high intermediate column,
similar to that used in monopile designs. This column rests
on the foundation consisting of three conical segments. The
end cross-sections of the first 30-meter high cone have the
shapes of regular dodecagons inscribed in circles of 5.5m
(top) and 7.0m (bottom) in diameters. The next segment,
situated below, increases the diameter of the circle in which
the regular bottom cross-section dodecagon is inscribed to
12.0 m, while the base of the lowest segment, lying on the
seabed, is inscribed in the circle of 40 m in diameter.
While typical gravitational supports are made of concrete
[16, 47], the presented structure is designed in the form of
a thin-walled steel container. The structural shape of the
container is similar to that of stiffened panels used in modern
shipbuilding and offshore industry. The article describes the
research activities concerning the support segment to the
height of 70 m above seabed (underwater part and intermediate
column). The support is designed as a container, playing the
POLISH MARITIME RESEARCH, No S1/2017 157
function of a displacement hull, which is delivered to the
foundation place via towing. Then the support is submerged
by flooding the ballast tanks. Proper stability of the support
standing on the seabed is ensured by ballasting the container
with pulp to the height of 25 m. Other ways of transport and
setting of the support seem to be more difficult and more
expensive. The assumer lifetime of the support is 50 years.
Fig. 3. General plan of the concept of gravitational supporting structure (steel
container with internal ballast)
3. STRENGTH ANALYSIS OF THE
UNTYPICAL OFFSHORE WIND TURBINE
SUPPORTING STRUCTURE
The strength analysis was conducted in accordance
with the standard DNVGL-ST-0126 [46]. The designed
structure was subjected to maximal loads calculated for the
assumed location and time of operation. The calculations
were performed using the FEM model and the ANSYS
software. The environmental loads were calculated with
the aid of computational fluid dynamics (CFD) codes and
the simulation platform STAR-CCM+. Sections 3.1 to 3.4
present the obtained strength analysis results, while the
main results of maximal deflection and stress calculations
are shown in Section 3.5, along with the calculated critical
force which initiates buckling. This force was assessed by
increasing the maximal wind turbine load until the load
capacity limit was obtained. To meet requirements included
in relevant regulations, some changes were introduced to the
most heavily loaded structure fragments. The multi-variant
analysis of the effect of selected design parameters on the
158 POLISH MARITIME RESEARCH, No S1/2017
support strength was also performed to decrease the level of
global column deflections and local stresses in the structure.
3.1 AIM OF ANALYSIS
The strength analysis aimed at verifying the overall
strength of load carrying elements of the untypical offshore
wind turbine supporting structure. An important element
of the analysis was identifying most heavily loaded regions
and introducing changes to the initial design of the structure
to reduce maximal support deflections and stresses at most
heavily loaded structural nodes. The final stage of the analysis
was calculating the critical force and identifying structure
destruction forms.
For the purpose of strength calculations, the FEM model
of the examined support was created. The support structure
was modelled using shell elements, while the intermediate
column was modelled using beam elements, and the nacelle
was modelled using the mass element.
The numerical model of the support having the form of a
large-size thin-walled container was based on shell elements
(shell181), which are four-node elements with 6 degrees of
freedom at each node. The FEM model described the geometry
of the examined object very precisely. The elementary grid
size was 150mm, and respected the principle of at least three
nodes over the span of each structural element. Modelling
low stiffeners with the aid of shell elements provided the
opportunity to connect the stiffener nodes with those of
the truss webs through which they pass. This enabled more
precise modelling of conditions of cooperation between these
elements in the global model. Structural elements which were
not involved in carrying global loads, inner walls in ballast
tanks for instance, were excluded from the analysis. The FEM
model of the examined offshore wind turbine supporting
structure is shown in Fig. 4.
Fig. 4. FEM model of the offshore wind turbine supporting structure (hidden
external plating)
 Tab. 1 Assumptions for CFD simulations performed to determine
3.2 MATERIAL MODEL environmental loads (wind, waves, sea currents)

The structure of the support was designed to be made

Name Value
of normal-strength steel, Category B, according to the
requirements in the standard DNV-OS-B101 [48]. Chemical Type of CFD analysis Implicit unsteady
composition and strength of this material, labelled as S235J0,
meet the requirements of the standard EN 10025-2 [49]. At the Time step dt = 0.02 [s]
same time the most heavily loaded region of the supporting
Turbulence model k-epsilon
structure, which is the passage of vertical trusses into base
bottom trusses, was designed to be made of higher-strength Wave model Irregular wave (Jonswap)
steel NV A27, Category A, for truss webs, and of high-strength
steel EH32, Category E, for truss flanges. Significant wave height H1/3 = 9.01 [m]
The computer simulations performed to verify the strength
Wave period T=11.3 [s]
of the examined structure made use of linear-elastic models,
while critical loads were calculated using isotropic material Frequency f=1200 [-]
models with nonlinear stress/deformation characteristics for
the above steels NVA, NVA27, and EH32. A sample material Peak shape parameter p=4.14 [-]
curve for steel NVA is shown in Fig.5. The Youngs modulus
Spectrum width A A=0.07 [-]
was assumed equal to 2.06e5 MPa, the Poissons number
v=0.3, and the density 7.85 t/m3. The yield points for steels Spectrum width B B=0.09 [-]
NVA, NVA27 and EH32 were assumed equal to, respectively,
235 MPa, 265 MPa and 315 MPa. For all steels, the tangent Water region depth Tw = 40 [m]
plastic hardening modulus was equal to 200 MPa.
The strength calculations were performed taking into
consideration maximal values of environmental loads
obtained from CFD storm simulations, for the irregular
wave and Jonswap wave spectrum. A sample free surface
shape obtained in the environmental load calculations is
shown in Fig. 6, while Fig.7 shows the time history of the
total transverse force. In both CFD and FEM calculations, the
coherent global coordinate system was used with the origin at
the centre of the circle circumscribed about the base bottom.
The assumed direction of hydrodynamic loads was the same
as that of the horizontal load coming from the wind turbine
rotor. The maximal value of the calculated transverse force,
corresponding to the average of 1% of maximal force peaks,
was equal to Fx=6350 kN, while the total moment acting on
the structure, which corresponded to the maximal value of
1% of loads for irregular wave, was equal to My=56800 kNm.

Fig. 5 Stress/deformation curve for the material model of NVA steel

3.3 LOADS AND BOUNDARY CONDITIONS

The performed strength analysis took into consideration

the following loads and boundary conditions:
a) Self-weight of the structure.
b) Environmental loads (dynamic action of wind, waves,
and sea current) corresponding to the maximal loads
generated by the heaviest storm on the Baltic Sea in recent
50 years, for the unprotected support location and water
depth of up to 40 m. The maximal loads were calculated
using CFD codes and the simulation platform STAR-
CCM+. The assumptions adopted in these calculations
are collated in w Tab.1 Fig. 6. Sample free surface of the wave being the support load (CFD
simulation)

POLISH MARITIME RESEARCH, No S1/2017 159

Fig. 7. Time-history of the calculated support load coming from the horizontal
force Fx generated by the action of irregular wave, for Jonswap wave spectrum
(CFD simulation)

c) Loads from the 7KW wind turbine were assumed taking

into account the safety factor k=1.35. The horizontal force
from the examined turbine was equal to Fx = 2.9 MN.
The remaining loads are protected by confidentiality
provisions of the turbine producer and cannot be
published.
d) Hydrostatic loads of seawater (wody = 1.025t/m3) and
inner ballast (pulp, pulpy = 1.7 t/m3). The assumed
submergence of the column at calm water was Tw=40m.
The wave height H1/2=10m was also considered. The
conservatively assumed resultant water head was equal
to Hw=50 meters, while the hydraulic head of the inner
pulp ballast was Hp=25m.
e) The soil stiffness was modelled in accordance with
Annex G to the standard DNVGL-ST-0126 [46]. Each
seabed element within the elliptic area of action was
attributed with stiffness corresponding to the structure/ Fig. 8. Total deflection of the offshore wind turbine supporting structure (a),
seabed interaction. The effective support area was and total support deflection below the intermediate column (b) [mm]
Aeff=896 m2.

3.4 RESULTS

The most important results of the presented offshore wind

turbine supporting structure strength calculations are the
total deflection and the reduced Huber-Mises stresses. The
maximal calculated deflection occurs at the nacelle node
and is equal to 2268 mm (Fig.8a), while the maximal total
deflection of the support part situated under the intermediate
column is observed in the section linking the support with
the intermediate column and is equal to 261 mm (Fig.8b). The
maximal values of the reduced Huber-Mises stresses occur in
main vertical trusses and are equal to 235MPa (Fig.9). The
calculated values are within the design criteria, according
to the standard DNVGL-ST-0126 [46]. Nevertheless, due to
non-uniform stress distribution, some modifications were
introduced to the design of the structure at the concept Fig.9. Reduced Huber-Mises stresses in the offshore wind turbine supporting
development stage see Section 3.6. structure [MPa]

It is noteworthy that the maximal stresses refer to a

relatively small area of the structure. Its part situated on the

160 POLISH MARITIME RESEARCH, No S1/2017

windward side of the turbine is subjected to relatively low
stresses, which results from the nature of loads from waves
and wind turbine, always acting in the wind direction.
Additionally, within the framework of the structure
strength analysis, the fatigue life was assessed for the most
heavily loaded region of the structure. This assessment based
on local stress calculations and the so-called local stress
approach. Although it was beyond the scope of the required
calculations, a conclusion was made that the most heavily
loaded node has sufficient fatigue life.
The last results discussed in this article refer to the
nonlinear buckling analysis. For the horizontal force Fx,
the amplification factor n was assumed which was the ratio
of the set force and the maximal load of the wind turbine
propeller, equal to Fx=2.9 MN. The factor n determines how
many times the maximal force from the wind turbine was
amplified. Figure 10 shows a diagram of maximal turbine Fig. 11. Local plating deformations for the critical force which initiates
node displacement in the horizontal x-axis direction, as buckling
a result of n-fold amplification of the turbine load. The
calculations were conducted at forced displacement input.
The obtained results indicate that at the load amounting to
1.44 of the maximal turbine force, the load carrying ability
of the structure reaches its limit. The calculations were
performed taking into account geometrical and material
nonlinearities. The maximal calculated turbine load which
can be carried by the structure is Fx= 4.18 MN. Further load
increase leads to buckling of the plating and truss webs in
the region of structure passage from column to conical base.
Local deformations for the critical load are shown in Fig. 11
and Fig. 12.

Fig. 12. Maximal truss web deformations for the critical force which initiates
buckling

3.5 GEOMETRICAL SHAPE EVOLUTION OF THE MOST

HEAVILY LOADED REGION OF SUPPORT STRUCTURE

The starting point for developing the concept of the

Fig. 10. Maximal turbine node displacement ux vs. turbine load amplification
factor n[-]
untypical supporting structure for an offshore wind turbine
were structural solutions used in modern merchant vessels.
These structures consisted of plating stiffened with one-
directional low stiffeners, which were in turn supported
by trusses. After preliminary verification calculations, the
column deflection level and the local stress distribution
in the support were modified. The elements loaded with
excessive stresses were reinforced, while those revealing
poor cooperation were removed, at the same time preserving
smooth characteristic of the longitudinal stiffness of the
column. Five calculation variants were analysed for the
self-mass change limit equal to 10%. The final effect of the
design work performed with the aid of the FEM model based
simulation was reducing the maximal total deflections by

POLISH MARITIME RESEARCH, No S1/2017 161

11% and the maximal reduced Huber-Mises stresses by
71%. The total mass of the structure was reduced by 9%. The
geometry change introduced to the most important bottom
truss region is shown in Fig.13. The plating thickness in the
bottom region was 20 mm. The thicknesses of bottom truss
webs and flanges were equal to, respectively, 15mm and 25
mm. Profiles HP430x17 were applied as plating stiffeners.
General arrangement of elements in the final version of the
structure is shown in Fig.14.
Fig. 13. Geometry changes in the bottom truss region: initial version (upper)
and final version (bottom)
Fig. 14. Explanatory figure of the offshore wind turbine supporting structure
(hidden geometry half)
162 POLISH MARITIME RESEARCH, No S1/2017
CONCLUSIONS
The offshore wind power industry is a rapidly developing
sector which already plays an important role in electric
energy production from renewable sources. However, further
technological development of this sector is required to meet
market demands concerning reduction of energy production
costs. What is essentially needed in this area is working out
technical solutions which will provide opportunities for
installation of large-power wind turbines, of 10MW and
more, arranged in wind farms situated at a large distance from
the seashore and, consequently, on relatively deep waters.
A key issue here is turbine immobilisation with respect to
the seabed, which can be solved in many ways. There are
numerous concepts of supporting structures, including
various types of floating platforms, partially submerged
platforms, TLP type columns, etc. A possible and potentially
favourable solution here can also be a gravitational support.
The article presents a concept of a large-size supporting
structure of gravitational type which has the form of a steel
container. The support can be towed to the place of operation
and then submerged by filling ballast tanks with the mixture
of sand and water taken from the seabed. The main part
of the article is the container structure strength analysis,
performed in accordance with the standard DNVGL-ST-0126
[46]. The required numerical calculations were conducted
using the Finite Element Method. The obtained results
reveal that the structure meets basic strength requirements
for the given strength limit. The critical force, which is the
load carrying capacity limit for the structure, was assessed.
Then the structure design was modified. The introduced
modifications allowed to reduce the maximal stress level
by 71%, the maximal column deflection by 11%, and the
total mass of the structure by 9%. A design modification
introduced to the most important bottom region of the
structure is presented. After this modification, the mass of the
structure was 1250 tonnes. The estimated support production
cost is approximately equal to 4 million euro (for the euro/
kg-of-structure coefficient equal to 3.2). It is noteworthy that
further structural optimisation is possible, making use of the
fact that the loads coming from sea waves and wind turbine
are very directional, and not circularly-symmetric as was
assumed in the preliminary support concept. In research
tests of supports intended for deeper water regions, of high
importance is high directionality of maximal design loads.
This factor gains even more in importance when we consider
enormous dimensions and mass of the structure. Hence,
when designing the supporting structure, it is economically
profitable to include in the analysis the direction of action of
maximal environmental loads, assessed based on statistical
hydro-meteorological data for the given water region. This
will remarkably reduce the self-mass of the supporting
structure and, consequently, costs of its production, transport,
installation, and disassembly.
ACKNOWLEDGEMENTS
This research was supported by the National Centre for
Research and Development (NCBR) under the AQUILO
project No. PBS1/A6/8/2012. The numerical calculations were
co-financed by CI TASK in Gdansk. Within the framework of
hydrodynamic load estimation, the research was co-financed
by NCBR under the SmartPS project No. MARTECII/
SmartPS/4/2016. All support is highly appreciated by the
author.
BIBLIOGRAPHY:
1. European Commission, A policy framework for climate
and energy in the period from 2020 up to 2030, impact
assessment, COM(2014) 15 final, SWD(2014) 15 final, Eur-
Lex. 2014
2. European Commission, Communication from the
Commission to the European Parliament, the Council,
the European Economic and Social Committee and
the Committee of the Regions on Launching the public
consultation process on a new energy market design, Eur-
Lex. 2015
3. Wind in power, 2015 European statistics, European Wind
Energy Association, February 2016
4. The European offshore wind industry - key trends and
statistics 2015, A report by the European Wind Energy
Association, February 2016
5. Annual statistics, WindEurope, 2016
6. Wind Energy Scenarios for 2030, A report by the European
Wind Energy Association - August 2015
7. Laszlo Arany, S. Bhattacharya, John Macdonald, S.J. Hogan,
Design of monopiles for offshore wind turbines in 10 steps,
Soil Dynamics and Earthquake Engineering, Volume 92,
January 2017, pp. 126-152, ISSN 0267-7261, http://dx.doi.
org/10.1016/j.soildyn.2016.09.024.
8. Swagata Bisoi, Sumanta Haldar, Design of monopile
supported offshore wind turbine in clay considering
dynamic soilstructure-interaction, Soil Dynamics
and Earthquake Engineering, Volume 73, June 2015,
pp. 103-117, ISSN 0267-7261, http://dx.doi.org/10.1016/j.
soildyn.2015.02.017.
9. A.T. Myers, S.R. Arwade, V. Valamanesh, S. Hallowell, W.
Carswell, Strength, stiffness, resonance and the design of
offshore wind turbine monopiles, Engineering Structures,
Volume 100, 1 October 2015, pp. 332-341, ISSN 0141-0296,
http://dx.doi.org/10.1016/j.engstruct.2015.06.021.
10. Hezhen Yang, Yun Zhu, Qijin Lu, Jun Zhang, Dynamic
reliability based design optimization of the tripod sub-
structure of offshore wind turbines, Renewable Energy,
Volume 78, June 2015, pp. 16-25, ISSN 0960-1481, http://
dx.doi.org/10.1016/j.renene.2014.12.061.
11. B. Yeter, Y. Garbatov, C. Guedes Soares, Fatigue damage
assessment of fixed offshore wind turbine tripod support
structures, Engineering Structures, Volume 101, 15
October 2015, pp. 518-528, ISSN 0141-0296, http://dx.doi.
org/10.1016/j.engstruct.2015.07.038.
12. K.H. Chew, E.Y.K. Ng, K. Tai, M. Muskulus, D. Zwick,
Offshore Wind Turbine Jacket Substructure: A Comparison
Study between Four-Legged and Three-Legged Designs, J
Ocean Wind Energy, 1 (2) (2014), pp. 7481
13. Sebastian Kelma, Peter Schaumann, Probabilistic Fatigue
Analysis of Jacket Support Structures for Offshore Wind
Turbines Exemplified on Tubular Joints, Energy Procedia,
Volume 80, 2015, pp. 151-158, ISSN 1876-6102, http://dx.doi.
org/10.1016/j.egypro.2015.11.417.
14. M.D. Esteban, B. Couago, J.S. Lpez-Gutirrez, V. Negro,
F. Vellisco, Gravity based support structures for offshore
wind turbine generators: Review of the installation process,
Ocean Engineering, Volume 110, Part A, 1 December 2015,
pp. 281-291, ISSN 0029-8018, http://dx.doi.org/10.1016/j.
oceaneng.2015.10.033.
15. W.Z. Lim, R.Y. Xiao, Fluid-structure interaction analysis
of gravity-based structure (GBS) offshore platform with
partitioned coupling method, Ocean Engineering, Volume
114, 1 March 2016, pp. 1-9, ISSN 0029-8018, http://dx.doi.
org/10.1016/j.oceaneng.2015.12.059.
16. Marc Costa Ros, Offshore wind industry review of Gravity
Base Foundations (GBSs), Identifying the key barriers to
large scale commercialisation of gravity based structures
(gbss) in the offshore wind industry, report, The Carbon
Trust, October 2015
17. Simon Lefebvre, Maurizio Collu, Preliminary design of
a floating support structure for a 5 MW offshore wind
turbine, Ocean Engineering, Volume 40, February 2012,
pp. 15-26, ISSN 0029-8018, http://dx.doi.org/10.1016/j.
oceaneng.2011.12.009.
18. Michael Borg, Maurizio Collu, A Comparison on the
Dynamics of a Floating Vertical Axis Wind Turbine on
Three Different Floating Support Structures, Energy
Procedia, Volume 53, 2014, pp. 268-279, ISSN 1876-6102,
http://dx.doi.org/10.1016/j.egypro.2014.07.236.
19. Matthew Hall, Brad Buckham, Curran Crawford,
Hydrodynamics-based floating wind turbine support
platform optimization: A basis function approach,
POLISH MARITIME RESEARCH, No S1/2017 163
 Renewable Energy, Volume 66, June 2014, pp. 559-569, ISSN
0960-1481, http://dx.doi.org/10.1016/j.renene.2013.12.035.
20. M. Collu and M. Borg, 11 - Design of floating offshore
wind turbines, In Offshore Wind Farms, Woodhead
Publishing, 2016, pp. 359-385, ISBN 9780081007792, http://
dx.doi.org/10.1016/B978-0-08-100779-2.00011-8.
21. Yichao Liu, Sunwei Li, Qian Yi, Daoyi Chen, Developments
in semi-submersible floating foundations supporting
wind turbines: A comprehensive review, Renewable and
Sustainable Energy Reviews, Volume 60, July 2016, pp.
433-449, ISSN 1364-0321, http://dx.doi.org/10.1016/j.
rser.2016.01.109.
22. Frank Adam, Thomas Myland, Burkhard Schuldt, Jochen
Gromann, Frank Dahlhaus, Evaluation of internal force
superposition on a TLP for wind turbines, Renewable
Energy, Volume 71, November 2014, pp. 271-275, ISSN
0960-1481, http://dx.doi.org/10.1016/j.renene.2014.05.019.
23. Ali Nematbakhsh, Erin E. Bachynski, Zhen Gao, Torgeir
Moan, Comparison of wave load effects on a TLP wind
turbine by using computational fluid dynamics and
potential flow theory approaches, Applied Ocean Research,
Volume 53, October 2015, pp. 142-154, ISSN 0141-1187,
http://dx.doi.org/10.1016/j.apor.2015.08.004.
24. Ali M. Abdelsalam, K. Boopathi, S. Gomathinayagam, S.S.
Hari Krishnan Kumar, Velraj Ramalingam, Experimental
and numerical studies on the wake behavior of a horizontal
axis wind turbine, Journal of Wind Engineering and
Industrial Aerodynamics, Volume 128, May 2014, pp.
54-65, ISSN 0167-6105, http://dx.doi.org/10.1016/j.
jweia.2014.03.002.
25. J.T. Cieslinski, R. Mosdorf, Gas bubble dynamics
experiment and fractal analysis, International Journal of
Heat and Mass Transfer, Volume 48, Issue 9, April 2005,
pp. 1808-1818, ISSN 0017-9310, http://dx.doi.org/10.1016/j.
ijheatmasstransfer.2004.12.002.
26. Chi-Jeng Bai, Wei-Cheng Wang, Review of computational
and experimental approaches to analysis of aerodynamic
performance in horizontal-axis wind turbines (HAWTs),
Renewable and Sustainable Energy Reviews, Volume 63,
September 2016, pp. 506-519, ISSN 1364-0321, http://dx.doi.
org/10.1016/j.rser.2016.05.078.
27. Krzysztof J. Kalinski, Marek A. Galewski, Chatter vibration
surveillance by the optimal-linear spindle speed control,
Mechanical Systems and Signal Processing, Volume 25,
Issue 1, January 2011, pp. 383-399, ISSN 0888-3270, http://
dx.doi.org/10.1016/j.ymssp.2010.09.005.
28. Dariusz Mikielewicz, Jarosaw Mikielewicz, Joanna Tesmar,
Impr
164 POLISH MARITIME RESEARCH, No S1/2017
29. oved semi-empirical method for determination of heat
transfer coefficient in flow boiling in conventional and
small diameter tubes, International Journal of Heat and
Mass Transfer, Volume 50, Issues 1920, September 2007,
pp. 3949-3956, ISSN 0017-9310, http://dx.doi.org/10.1016/j.
ijheatmasstransfer.2007.01.024.
30. Sebastian Schafhirt, Ana Page, Gudmund Reidar Eiksund,
Michael Muskulus, Influence of Soil Parameters on
the Fatigue Lifetime of Offshore Wind Turbines with
Monopile Support Structure, Energy Procedia, Volume
94, September 2016, pp. 347-356, ISSN 1876-6102, http://
dx.doi.org/10.1016/j.egypro.2016.09.194.
31. K Skalski, A Morawski, W Przybylski, Analysis of contact
elastic-plastic strains during the process of burnishing,
International Journal of Mechanical Sciences, Volume 37,
Issue 5, 1995, pp. 461-472, ISSN 0020-7403, http://dx.doi.
org/10.1016/0020-7403(94)00083-V.
32. T.A. Stolarski, Wear of water-lubricated composite materials,
Wear, Volume 58, Issue 1, 1980, pp. 103-108, ISSN 0043-
1648, http://dx.doi.org/10.1016/0043-1648(80)90215-X.
33. Zhiyu Jiang, Limin Yang, Zhen Gao, Torgeir Moan,
Numerical Simulation of a Wind Turbine with a Hydraulic
Transmission System, Energy Procedia, Volume 53, 2014,
pp. 44-55, ISSN 1876-6102, http://dx.doi.org/10.1016/j.
egypro.2014.07.214.
34. Niklas, Karol. Calculations of notch stress factor of a thin-
walled spreader bracket fillet weld with the use of a local
stress approach.Engineering Failure Analysis 45 (2014),
pp. 326-338.
35. J.A. Sainz, New Wind Turbine Manufacturing Techniques,
Procedia Engineering, Volume 132, 2015, pp. 880-886, ISSN
1877-7058, http://dx.doi.org/10.1016/j.proeng.2015.12.573.
36. C. Dymarski, P. Dymarski, Developing methodology for
model tests of floating platforms in low-depth towing tank,
Archives of Civil and Mechanical Engineering, Volume 16,
Issue 1, January 2016, pp. 159-167, ISSN 1644-9665, http://
dx.doi.org/10.1016/j.acme.2015.07.003.
37. XF]DN00DQ]DWR63HHWHUV%%UDQQHU.%HUULQJ
3.DKVLQ08SGDWLQJ)LQLWH(OHPHQW0RGHORID:LQG
7XUELQH%ODGH6HFWLRQ8VLQJ([SHUimental Modal Analysis
Results, Shock and Vibration. -Vol. 2014, iss. 1 (2014),
pp.71-82
38. A. Cichaski, The influence of mesh morphology on the
SCF in 2D FEM analysis of flat bars with opposite V-notch
under tension in Proceedings of 22nd International
Conference on Engineering Mechanics 2016, edited by
Igor Zolotarev and Vojtech Radolf, Engineering Mechanics
2016, pp. 110-113 (Svratka, Czech Republic, 2016)
39. Deja, M. & Siemiatkowski, M.S., Feature-based generation of 49. DNV-OS-B101, OFFSHORE STANDARD, METALLIC
machining process plans for optimised parts manufacture, MATERIALS, Det Norske Veritas 2009
J Intell Manuf (2013) 24: 831. doi:10.1007/s10845-012-0633-x
50. EN 10025-2 Non-alloy structural steels, European
40. Jiahai Yuan, Chunning Na, Yan Xu, Changhong Zhao, Wind Committee for Standardization, 2005
turbine manufacturing in China: A review, Renewable and
Sustainable Energy Reviews, Volume 51, November 2015,
pp. 1235-1244, ISSN 1364-0321, http://dx.doi.org/10.1016/j.
rser.2015.07.048.

41. A. Zieliski, H. Smoleska, W. Serbiski, W. Koczewicz,

A. Klimpel, Characterization of the Co-base layers CONTACT WITH THE AUTHOR
obtained by laser cladding technique, Journal of Materials
Processing Technology, Volumes 164165, 15 May 2005, Karol Niklas
pp. 958-963, ISSN 0924-0136, http://dx.doi.org/10.1016/j.
jmatprotec.2005.02.093. Gdask University of Technology
11/12 Narutowicza St.
42. Magdalena Mieloszyk, Wiesaw Ostachowicz, An 80 - 233 Gdask
application of Structural Health Monitoring system Poland
based on FBG sensors to offshore wind turbine support
structure model, Marine Structures, Volume 51, January
2017, pp. 65-86, ISSN 0951-8339, http://dx.doi.org/10.1016/j.
marstruc.2016.10.006.

43. D.J.M. Fallais, S. Voormeeren, E. Lourens, Vibration-

based Identification of Hydrodynamic Loads and
System Parameters for Offshore Wind Turbine Support
Structures, Energy Procedia, Volume 94, September 2016,
pp. 191-198, ISSN 1876-6102, http://dx.doi.org/10.1016/j.
egypro.2016.09.222.

44. M. Siemiatkowski, W. Przybylski, Simulation studies of

process flow with in-line part inspection in machining
cells, Journal of Materials Processing Technology, Volume
171, Issue 1, 10 January 2006, pp. 27-34, ISSN 0924-0136,
http://dx.doi.org/10.1016/j.jmatprotec.2005.06.051.

45. Binita Shrestha, Martin Khn, Adaptation of Controller

Concepts for Support Structure Load Mitigation of
Offshore Wind Turbines, Energy Procedia, Volume 94,
September 2016, pp. 241-248, ISSN 1876-6102, http://
dx.doi.org/10.1016/j.egypro.2016.09.231.

46. Wojciech Litwin, Influence of local bush wear on water

lubricated sliding bearing load carrying capacity,
Tribology International, Volume 103, November 2016,
pp. 352-358, ISSN 0301-679X, http://dx.doi.org/10.1016/j.
triboint.2016.06.044.

47. Standard DNVGL-ST-0126, Support structures for wind

turbines, DNV GL AS, April 2016

48. VON DER HAAR, C and MARX, S. Design aspects of

concrete towers for wind turbines. J. S. Afr. Inst. Civ. Eng.
2015, vol.57, n.4, pp. 30-37. ISSN 2309-8775. http://dx.doi.
org/10.17159/2309-8775/2015/v57n4a4.

POLISH MARITIME RESEARCH, No S1/2017 165

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 166-173
10.1515/pomr-2017-0035

ENVIRONMENTAL ASPECTS OF COASTAL EARTH STRUCTURES

MADE OF SOIL-ASH COMPOSITES

Rafa Ossowski, PhD

Gdansk University of Technology, Poland

ABSTRACT

This paper presents an innovative technology of building the coastal dams with the use of soil-ash composites. A
natural- scale test stand and results of the performed environmental tests are presented ; they have been compared
with standards for purity of soils and waters. Also, an analysis of the obtained results and discussion of trends and
anomalies in the observations, are given.

Keywords: marine environment protection, soil-ash composites, construction monitoring

INTRODUCTION In this paper is presented an innovative technology for

Soil structures in coastal zone generally fulfil protective
functions they are aimed at direct protection of coast or/
and a greater area of land , e.g. in case of depression terrains.
In Poland are more often used the soil structures ( also those
reinforced ) to serve as one of coast protection elements (storm
embankments etc), however in some EU countries such as
Germany, Denmark, the Netherlands, a relatively large part
of coastal structures constitute protection embankments
which protect firm land against water thrust resulting first
of all from sea tides and, in the second order , from storms.
Building material for erecting soil coastal structures is won
from sites located possibly near the building site, because of
transport cost. However in case of soil work of a large volume,
local resources of crush stone aggregates do not constitute
sufficient reserves and then one reaches for material from
more distant sources. On the other hand , in coastal zone
the regular dredging of water fairways is carried out from
which a large supply of dredge winning is available , they
may serve as a substitute of natural aggregates in erecting
the soil structures.
building soil embankment with the use of a composite soil
[1]: a mixture of silt-up material won during dredging work
and bottom ash obtained from coal combustion in a heat &
power and station. An important aspect in utilizing waste and
secondary materials for building the coastal structures is to
ensure appropriate protection for the marine environment.
Therefore the issue discussed further in this paper are selected
environmental investigations carried out on a real soil structure.
A part of them was conducted in the frame of preliminary
experiments on selection of structural material , and another
part in the frame of monitoring the ready- to- use structure.
The presented results are merely a part of the research project
in which physical , chemical and mechanical aspects of
composite soils were investigated [2], as well as mechanical
and hydraulic qualities of structures [35]. It is obvious
that such research does not cover all the possible spectrum
of analyses associated with the monitoring of coastal and
offshore objects , which connects such distant domains like
the investigation of non-linear permeability features of soil
medium itself [6], laser coast scanning [7] or acoustic test
methods [8-11].

166 POLISH MARITIME RESEARCH, No S1/2017

 EXPERIMENTAL DIKE dike (protection against erosion). All the object was covered
by plants ( grass).
Investigations of composite soils were performed on an
experimental test stand ( in natural scale) placed close to
Gdask and built in the frame of the EU project DredgDikes
[1,5] which has been aimed at the developing of innovative
technology of building soil structures. A superior task of
the project was to investigate usability of fine-grain winning
materials as well as ash-sand composites for erecting sea - coast
and river embankments.
For this reason the body of the experimental dike at Gdask
was designed on the basis of a composite of a sand dredged out
of the river bed of the Dead Vistula, and a bottom ash from
EC2 heat & power station at Gdask.
Various materials forming the composite were tested in
a laboratory both in the phase of designing and during
the erecting of dike structure [2]. The tests were aimed at
optimization of material in the sense of its mechanical Fig.1. Model 3D model of the test stand ;marking colours stand for : green
dike body, bronze tight walls, blue - water, black bottom tightening (on
properties ( compactibility and stability of the body) hydraulic ground base)
ones (permeability , leaching the toxic substances ) as well
as resistance of the structure against long-lasting hydraulic
loading and erosion in case of a water flow over crown of the
body.
The preliminary laboratory tests were aimed at
determination of optimum proportion of ashes and sand
to form a soil composite for dike core. The tests resulted in
knowledge of mechanical properties of the composite , a. o. its
compressibility, strength parameters, as well as compactibility
and water permeability. The tested parameters were determined
for component materials (sand, ash) as well as their mixtures
in appropriate proportions changed step by step by10% [2].
As a result of the preliminary tests the optimum composite
Fig. 2. The ready-to-use test stand aerial view
70/30 ( ash/ sand) was selected to be the basic material for
building the experimental dike. As a matter of fact , such
mixture has a relatively high filtration ratio (1.3510-5 m/s),
however , a decrease in hydraulic conductance resulting from CHARACTERISTICS OF STRUCTURAL
consolidation of the dike body as well as cementation due to MATERIALS
puzzolan features of ashes, were observed. The tests conducted
on specimens taken out from the experimental dike showed The composite soil for building the dike consisted of
a drop in filtration rate by 40% per 180 days [2]. This was the mixture of the winning dredged out of the bed of the
the composite used for building the test stand in the form of Dead Vistula ( river sand ), and the by-product of coal
the experimental dike of 3 m in height , placed at Trzcinisko combustion in the heat & power station (bottom ash).
village close to GdaVN [1,3]. Slope of its scarps was 1:2, breadth Short characteristics of the materials are presented below.
of crown - 3 m, full footing breadth - 15 m and length - 24 A. Dredged winning
m. The test stand itself covered 4 m segment extracted from The dredged winning is a deposit material which has been
the dike by means of tight walls which permit to obtain plane won from the bed of basins or rivers/canals . The natural
flow necessary for filtration and overflow investigations at sedimentation process of drifting materials and aggregates leads
controlled water level (Fig.1, 2). to forming the sediments and, consequently, decreasing the
As mentioned above, the experimental dike is relatively depth of reservoirs and waterways. In case of sailing waters the
homogenous , built of the composite made from bottom ashes sediments are to be removed regularly by dredging. The dredged
in 70% and sand in 30% ( marked : 70/30 composite ). Under winning is consisted of organic parts ( micro-organisms and
the test stand an isolation clay layer of 0,5 m in thickness was remains of plants and animals) and non-organic ones (mainly
placed. Additionally, the coating layers were applied : of 50 cm sand and dust). Content and characteristics of a sediment and,
thick clay on the water-exposed side and 30 cm thick mixture consequently, dredged winning , depends on geomorphology of
of dusts and Tefra binder ( produced on the basis of volatile a given reservoir or drainage basin, as well as on hydrodynamic
ashes ) to form the upper layer on the air-exposed side of the and hydro-geologic processes which occur in surface waters.
A separate issue is salinity of deposit and its chemical content.

POLISH MARITIME RESEARCH, No S1/2017 167

Dredging operations provide large amount of soil materials
which may be utilized for building soil structures. In Poland
the yearly balance of dredging output from the river and sea
waterways amounts to about 1,1 mln m (acc. data of Maritime
Administration Office in Gdynia), however in Germany about
41 mlnPRIGUHGJHGZLQQLQJLVUHDFKHGIURPWKHGUHGJLQJof
sea waterways and about 5 mlnPIURPLQODQGZDWHUZD\V
[12].
In many cases the dredged winning may be considered
a material which is capable of substituting or lowering the
demand for natural raw materials ( including aggregates) that is
conducive to environment protection. A crucial factor deciding
on its possible application is lack of contamination of bed
sediments , however the bettering of a slightly contaminated
winning is also economically justified.
B. Coal combustion by-products ( CCP)
The by-products of coal combustion come from power plants
firing hard or brown coal. Main admixtures of hard and brown
coal are of mineral kind such as quartz , silt , calcite and slates.
Hence, amount of the CCP produced by a power plant depends
ENVIRONMENTAL INVESTIGATIONS
A. Legal regulations
Permissible concentration of contaminations in soil
structures are not subjected to separate legal regulations
in Poland. One may however relate to regulations dealing
with purity of soils. The act in force in this matter is the
Decree of Ministry of the Environment , of September
1 , 2016 on the method of carrying out assessment of
soil surface contamination (Dz.U.2016 poz.1395) [22].
It distinguishes soil categories depending on their use.
Tab. 1 shows permissible contents of substances which cause
risk of contamination to soils ( given in the most restrictive
requirements) , for selected indices of contamination.
Tab. 1. Selected permissible values of contaminations contained in soil (for the
highest class of purity) [22]

Benzo(a) pyrene [g/kg]

Oil index [mg/kg]
on a kind and volume of coal fired in boilers as well as on

WWA [g/kg]
Substance

Zn [mg/kg]
Hg[mg/kg]
Cd[mg/kg]
Cu[mg/kg]
Pb[mg/kg]
content of minerals in coal ( combusting technology is of a
secondary importance) . CCP amount produced worldwide
is estimated to reach about 550 mln t /year , out of which 95
mln t /year is covered by EU countries [13].
The CCP may be classified according to mineral content of
coal as well as technology of its combustion. The main part of
it is consisted of : fly ash , bottom ash , boil slag as well as ash
Permissible

available
no data
value

from fluidal boilers (FBC) . Apart from the above mentioned
components also products of sulfur removal from exhaust
gases such as plaster , wet desulfurization products (FGD) as
well as semi-dry absorption products (SDA). Fly ash is formed
from fine grain particles of 1 up to 100 microns in size (average
granulation is in the range of 20 - 25 microns). Bottom ash is
a granular material consisted of bigger particles ( dusty and
sandy ) of 0.1 to 50 mm in diameter.
Chemical content of the ashes is determined by chemical content
of coal fuel and its combustion technology. Ash is composed
of typical minerals which form most of the earth crust rocks.
Al-silicates , Ca , Fe and Mg - oxides form more than 90
% of mass of typical ashes. The remaining compounds of
the elements such as : Na , K , Ti, and S cover about 8 % of
ash mineral content, and the trace elements such as As , Kd
, Pb and Se no more than 1 % of the content. In opinion of
naturalists these are the elements which potentially endanger
the natural environment in case of their uncontrolled release
into e.g. the water environment. Basic chemical properties
and reactivity of the CCPs are determined by content of Ca,
Fe and S compounds. Ashes produced in combustion process
30 100 100 100 2 2 300
Another source which soil contamination level may be referred to are
the recommendations of the Institute of Soil Cultivation, Fertilization
and Science on Soils in Puawy [14]. The recommendations propose
classification of soils into 6 categories , beginning from the level 0-
non-contaminated soils having natural content of heavy metals , and
ending on the level V very heavy contaminated soils , generally
excluded from agriculture production. However it should be taken into
account that the recommendations concern only a 20 cm layer of soil
in the context of agriculture production hence any comparison with
a soil for building purposes is not fully authoritative ( reliable). Tab. 2
presents a comparison of the requirements for light non-contaminated
soils (of the level 0) with those for soils of an elevated content of metals
( of the level 1).
Tab. 2. Natural content of heavy metals in soil ( a non-contaminated soil) [14]
Cd [mg/kg]
Cu [mg/kg]

Zn [mg/kg]
Pb [mg/kg]

Hg [g/kg]

of brown coal and bitominous shales have a relatively high

Substance
content of calcium (usually over 15 % of CaO), that generates
seepages of alkaline character (pH = 11 12). Ashes from hard
coal combustion contain generally much less Ca and produce
solutions of in between slightly acid and slightly alkaline lack of
Level 0 30 15 0,3 50
recommendation
character (pH = 5 10) [1].
lack of
Level 1 70 30 1 100
recommendation

168 POLISH MARITIME RESEARCH, No S1/2017

 As far as contamination of filtering water seepage through the Before the starting of building operations, specimens of both
dike body is concerned , the measured values may be compared with component materials, i.e. the ash from the heat & power plant
the recommendations for purity of water. Such recommendations and the sand from dredging work , were tested to control their
are contained in The Decree of Ministry of the Environment , of contamination level. After the completing of the test stand (in
November 9 , 2011, on the method of classification of homogenous the frames of the planned monitoring of the environment),
parts of surface water areas and the environmental quality standards were performed analyses of contamination of water samples
for priority substances (Dz.U.2011.257.1545) [23], selected permissible taken directly from the test stand. The samples were taken
values of which are given in Tab. 3. from six testing points : the upper reservoir , four piezometric
holes and bottom seepage water (Fig. 3). Two series of the
Tab. 3. Selected permissible values of contaminations for coastal waters of the tests were performed in simulated hydrodynamic conditions
highest category of purity [23].
of water flow through the dike body. The chemical analyses
were conducted by accredited laboratories with the use of the
methods of atomic and emission spectrometry ( analysis of
Benzo(a)pyrene [g/l]
Phenol index [mg/l]
Oil index [mg/l]

heavy metals ) , gas chromatography (hydrocarbons) as well

WWA [g/l]

Cu [mg/l]

Zn [mg/l]
Substance

Hg [g/l]
Pb [g/l]

Cd [g/l]

as spectrophotometry.
Permissible value

no data

no data

Fig. 3 Cross-section of the dike with the indicated points from where the
0,2 0,01 0,05 7,2 0,05 0,2 0,05 water samples were taken for the analysis: UR - upper reservoir, P1... P4
piezometric holes, LR - lower reservoir

Tab. 5 presents the results of the laboratory tests performed

Another important document is The Decree of Ministry of before starting the building work. Specimens of the sand tilt
the Environment , of November 27 , 2002, on the requirements and bottom ash were tested separately.
for surface water used for supplying potable water to population
(Dz.U.2002.204.1728) [24]. Tab. 4 shows selected permissible values Tab. 5. Collected results of environmental analyses of composite component
materials
of contaminations according to the above mentioned decree, contained
in the water intended to be supplied as potable for population.
Benzo(a)pyrene [g/
Name of specimen

Phenol index

Tab. 4 Selected permissible values of contaminations contained in water

Oil index
[mg/kg]

[mg/kg]

[mg/kg]

[mg/kg]

[mg/kg]
[g/kg]

[g/kg]
WWA

intended to be supplied as potable [24]

kg]
pH

Hg
Cd
Cu
Pb
Benzo(a)pyrene
Phenol index
Substance

Zn [mg/l]
Oil index

WWA
[mg/l]

[mg/l]

[mg/l]
[g/l]

[g/l]

[g/l]

[g/l]

[g/l]
Hg
Cd
Cu
Pb

sand 7.98 106 0.08 43,0 1.2 16 <2 <0.5 1.8

ash 8.90 47 0.02 12,3 <0.5 23 30 <0.5 14.9

nno dataavailable
Permissible value

no data available

Comparing the obtained results with the requirements

0,001

0,05
0,2

5,0

3,0
1,0
50

of the Decree of Ministry of Environment issued in 2016

[22] (Tab.1) one can observed that heavy metals contained
in both the components satisfy the soil purity criteria , not
yet approaching the upper limit of the standard. As far as
The presented regulations are used as the references for the organic contaminations are concerned , this is benzo(a)
investigations conducted on the test stand described below. pyrene which shows concentrations much lower than the
permissible levels , however oil index exceeds the standard
B. Results of the laboratory analyses level significantly. The comparison of concentrations of the
tested metals with the levels recommended in the guidelines
The investigations of contaminations in the composite [14] (Tab. 2) indicates that both the soil composite components
soil ( silt + ash ) as well as in the water filtering through the satisfy the requirements of the level 1, i.e. they can be
dike were an element of environmental control of the project. numbered among the soils of elevated content of heavy metals

POLISH MARITIME RESEARCH, No S1/2017 169

 Tab. 7. Collected results of chemical analyses of the water filtering through the
( however they are still considered not contaminated soils ). dike - the experiment 2 (blue marked values - below detection limit)
In the second phase of the project , after erection of the test
stand , two series of water contamination tests were conducted
in hydrodynamic unsteady flow conditions so as to check in Indices Unit UR P1 P2 P3 P4 LR
natural scale whether there is any hazard of contamination
leaching from the dike body. The entire test stand was isolated
pH mg/l 7,88 7,73 9,00 9,41 9,49 8,32
from parent ground by means of 0,5 m layer of clay in order to
eliminate seepage into natural soil. The specimens were taken
in two test series at 20 days interval. Tab. 6 and 7 present the Oil index mg/l 0,1 0,2 0,3 0,8 0,6 0,2
obtained values of selected concentrations and contamination
Phenol
indices in particular measurement points indicated in Fig. index
mg/l 0,01 0,01 0,01 0,01 0,01 0,01
3. The concentrations of all the contaminations in water are
significantly higher in comparison with contaminations of the Benzo(a)
J/l 0,0025 0,0140 0,0170 0,0120 0,0110 0,0025
pyrene
soil composite components. In Fig. 47 are directly compared
more interesting test results from both experiments , i.e. pH WWA J/l 0,010 0,036 0,038 0,024 0,024 0,010
reaction, WWA content , concentration of Cu and phosphates
. From pH reaction results it may be concluded that water
Benzene J/l 0,5 0,5 0,5 0,5 0,5 0,5
inflow to the piezometer 1 installed relatively near the upper
reservoir, runs very fast ( there is no increase in pH reaction
as a result of alkaline character of ash-soil composite ) . The Zn mg/l 0,036 0,016 0,15 0,17 0,013 0,24
pH reaction values in the piezometers 2, 3 and 4 are elevated (
due to pH reaction of the composite) and drop to their initial Cd J/l 0,1 0,1 0,36 0,44 0,1 0,77
values in the lower reservoir.
Cu J/l 2,3 3,6 24,0 69,0 2,2 76,0
Tab. 6. Collected results of chemical analyses of the water filtering through the
dike - the experiment 1 (blue marked values - below detection limit)
Pb J/l 1,5 1,0 11,0 35,0 1,0 81,0

Indices Unit UR P1 P2 P3 P4 LR
Hg J/l 0,01 0,04 0,03 0,08 0,09 0,05

pH mg/l 7,92 7,77 9,55 9,47 9,60 7,61 Phosphates mg/l 1,61 1,53 0,15 0,09 0,09 0,2

Oil index mg/l 0,1 0,1 0,1 0,1 0,2 0,1

Phenol
mg/l 0,01 0,01 0,01 0,01 0,01 0,01
index
Benzo(a)
J/l 0,0025 0,003 0,0025 0,0025 0,0025 0,0596
pyrene

WWA J/l 0,010 0,010 0,010 0,010 0,010 0,172

Benzene J/l 0,5 0,5 0,5 0,5 0,5 0,5

Zn mg/l 0,92 1,30 0,34 2,30 1,30 1,10

Fig. 4. Comparison of pH reaction of water in the experiments 1 and 2

Cd J/l 0,64 0,1 0,2 1,9 0,1 0,1

Cu J/l 11,0 4,3 10,0 13,0 3,3 4,9

Pb J/l 1,5 1,0 1,0 3,2 1,0 1,0

Hg J/l 0,02 0,01 0,01 0,03 0,01 0,01

Phosphates mg/l 1,32 1,44 0,28 0,16 0,18 0,35

Fig. 5. Comparison of WWA concentration in water in the experiments 1and 2

170 POLISH MARITIME RESEARCH, No S1/2017


Fig. 6. Comparison of Cu concentration in water in the experiments 1and 2
Fig. 7. Comparison of phosphate concentration in water in the experiments
1and 2
The results of the tests on organic contamination may
be considered somewhat astonishing, namely : the sudden
jump of WWA values (Fig . 5) and benzo(a)pyrene content in
seepage in the test 1, as well as the detecting some ( minute
) concentrations of both the indices inside the dike in the
test 2. The first above mentioned issue may be explained by
a random external fuel contamination (from a motor pump
delivering water to the upper reservoir), especially as the
effect does not occur in the test 2. As far as the second issue is
concerned , i.e. a rise in concentration of both contaminations
over the detection limit , its source is hard to identify and
its cause should be search for outside the dike body, taking
into account their lack in the preceding test. An analysis of
meteorological data makes it possible to form a hypothesis
on migration of the above mentioned compounds by air
from a nearby oil refinery (about only 8 km distant in WNW
direction). Apparently, during the fortnight preceding the
test 1 no significant rainfall (of total precipitation of about 5
mm) was observed, however during the fortnight preceding
the test 2 the rainfall of about 50 mm occurred at prevailing
west and north- west winds (exactly from the direction of the
oil refinery ). Hence , probably , the seeping water introduced
the organic contaminations contained in the rainfall ( whose
drops absorbed contaminations from the refinery area) into
the dike body, which were revealed in the test 2. Their lack in
the water delivered to the upper reservoir ( UR) results from
their lack in the water taken from the Dead Vistula to fill the
reservoir. Worth stressing that the measured quantities of
contaminations of the water inside the dike are on the safe
side of the standard permissible values.
It is also hard to explain the discrepancy in measurements
of concentration of Cu compounds ( and also of Pb ones )
in both the experiments (Fig.6, Tab. 6 and 7). Whereas the
first of them shows relatively small variations of the above
mentioned concentrations (in the range of 3,311,0 g/l), the
other shows much greater variation of the concentrations (from
2,2 to 76,0 g/l) and the peaks in the points P3 and LR. Inverse
trends are observed in the case of Cd and Zn - here the bigger
concentrations were found in the experiment 1. In comparison
with the reference values (Tab. 3 and 4), the measured values
generally satisfy the standards and only in a few cases they
slightly exceed the permissible concentrations. It should be
mentioned that the comparison concerns the standards for
potable and clean surface waters, hence the obtained results
may be considered very satisfactory , not indicating occurrence
of any environmental hazards.
A favourable factor in the context of low content of heavy
metals in seepage, is a slightly alkaline reaction of the dike
body , and , consequently, the water inside the dike (pH 910).
The laboratory tests confirm that such conditions do not allow
for excessively fast rate of leaching such elements as : Pb, Cd,
Cu from the composite soil [15].
An interesting observation is also the effect of purifying
influence of the dike as far as phosphates are concerned. In
both the experiments a decreasing concentration of phosphates
can be observed during flow of the water through the dike
(Fig. 7), which is favourable in the context of lowering the
eutrophication.
SUMMARY
The performed tests on the experimental dike showed that it
is hard to obtain repeatability of test results at a lack of control
over all boundary conditions ( possible infiltration of rain water
which may change chemical properties of building material
of the dike, and the filtering of river water through the dike).
On the other hand , it was possible to verify , in natural scale,
a lack of environmental hazards resulting from the use of the
materials, tested in advance, and classified as by-products (silt
winning, ash). However it should be stressed that the conducted
investigations constitute only a contribution to discussion on
the environmental aspects of the materials of the kind and
therefore they should be continued in the future.
A subsequent research step may be directed toward the
developing of a numerical model of the phenomenon of water
filtering through the dike , which would take into account
the flow in conditions of not entire saturation of the soil [6]
as well as the phenomenon of transportation at accounting
for adsorption of contaminations on soil particles. Such task
should not be counted among trivial ones, and a significant
potential for its solving may be attributed to the tools based
on the meshless methods such as e.g. the Meshless Local
Petrov-Galerkin method [16],or the Smoothed Particle
Hydrodynamics [1720]. Such approach will be used in the
next steps of the research in question.
POLISH MARITIME RESEARCH, No S1/2017 171
 Obviously, soil ash composites, like other composite
materials used in other domains of building industry[21],
constitute a valuable structural material whose potential has
not been so far thoroughly tested and exploited.
ACKNOWLEDGEMENT
This paper has been prepared on the basis of the research
conducted in the frames of the DredgDikes project financially
supported by EU resources in the frames of the South Baltic
research fund (ERDF) for the years 20102015.
The scientific work has been partly supported from the
financial resources, for the years 20112014 , intended for
realization of a scientific project in the frames of international
cooperation.
BIBLIOGRAPHY
1. www.dredgdikes.eu
2. Baachowski, L.; Sikora, Z. : Mechanical properties of bottom
ash dredged material mixtures in laboratory tests, Studia
Geotechnica et Mechanica, 2013, 35(3) , pp. 311.
3. Zbigniew Sikora , Rafa Ossowski: Geotechnical Aspects of
Dike Construction Using Soil-Ash Composites. Procedia
Engineering , 2013, pp. 1029 1035
4. Olschewski J., Cantr S., Saathoff F., Duszyski R.,
Ossowski R: Overflowing tests at the Polish DredgDikes
research dike stability of the dike surface against erosion.
Engineering Structures and Technologies, 2014 , pp.
159168
5. Saathoff, F.; Cantr, S.; Sikora, Z. : South Baltic Guideline
on the Application of Dredged Materials, Coal Combustion
Products and Geosynthetics in Dike Construction, 2015,
100 p.
6. Szymkiewicz A.: Modelling water flow in unsaturated
porous media: Accounting for nonlinear permeability and
material heterogeneity. Springer , Heidelberg, 2013 , 237p
7. Burdziakowski P., Janowski A., Kholodkov A., Matysik K.,
Matysik M., Przyborski M., Szulwic J., Tysic P., Wojtowicz
A.: Maritime laser scanning as the source for spatial data.
Polish Maritime Research, 2015, Vol. 22, No. 4(88) pp.914.
8. Grelowska G., Kozaczka E.: Underwater Acoustic Imaging
of the Sea. Archives of Acoustics , 2014, Vol. 39, No. 4 , pp.
439452
9. Grelowska G., Kozaczka E.: Nonlinear properties of the
Gotland Deep Baltic Sea. Archives of Acoustics , 2015 ,
Vol. 40, No. 4, pp. 595-600
172 POLISH MARITIME RESEARCH, No S1/2017
10. Kozaczka E.: Acoustics waves in the sea. Archives of
Acoustics , 2013 , Vol. 38, No. 3 pp. 441442
11. Kozaczka E., Grelowska G., Kozaczka S.: Detection of
Objects Buried in the Sea Bottom with the Use of Parametric
Echosounder , Archives of Acoustics, 2013,Vol. 38, No. 1
, pp. 99104
12. Kothe H.: Dredging and the environment international
guidance for best practice, PIANC , 125th Anniversary
Celebration in Asia, Nagoya 2010.
13. vom Berg W., Feuerborn H-J.: Coal Combustion Products
in Europe valuable raw materials for the construction
industry. Concrete Plant International , 2006 , Special Print
04/06
14. Kabata-Pendias A. et al.: Assessment of degree of
contamination of soils and plants by heavy metals and
sulphur ( in Polish). General guidelines for agriculture
industry , IUNG, Puawy, 1993
15. Komonweeraket K., Cetin B. et al.: Effects of pH on the
leaching mechanisms of elements from fly ash mixed soils,
FUEL , 2015 ,Vol. 140 , pp. 788802
16. Ossowski R.: Local heaviside weighted mlpg meshless
method approach to extended flamant problem using radial
basis functions , Studia Geotechnica et Mechanica , 2008,
pp. 173 180
17. Liu G. R. and Liu M. B.: Smoothed particle hydrodynamics:
a mesh-free particle method, World Scientific, 2003
18. Szewc K., Taniere A., Pozorski J., Minier J.: A study on
application of Smoothed Particle Hydrodynamics to multi-
phase flows, International Journal of Nonlinear Sciences
and Numerical Simulation, 2012 ?
19. Szewc K., Pozorski J., Minier J.: Analysis of the
incompressibility constraint in the smoothed particle
hydrodynamics method , Int. J. for Numerical Meth. in
Eng., 2012, Vol. 92, No. 4
20. Szewc K., Pozorski J., Taniere A.: Modeling of natural
convection with Smoothed Particle Hydrodynamics: Non-
Boussinesq formulation, International Journal of Heat and
Mass Transfer , 2011 , Vol. 54, pp. 48074816
21. Mikiewicz M, Daszkiewicz K., Ferenc T. Witkowski W
and Chrcielewski J.: Experimental tests and numerical
simulations of full scale composite sandwich segment of a
foot-and cycle-bridge, Advances in Mechanics: Theoretical,
Computational and Interdisciplinary Issues Kleiber et al.
(Eds), Taylor & Francis Group, London, 2016, pp. 401404.
22. The Decree of Ministry of the Environment , of September
1 , 2016, on the method of conducting assessment of soil
surface contamination ( in Polish) , (Dz.U.2016 poz.1395)

23. The Decree of Ministry of the Environment , of November

9 , 2011, on the method of classification of the state of
homogenous areas of surface waters and environment
quality standards for priority substances ( in Polish) ,
(Dz.U.2011.257.1545)

24. The Decree of Ministry of the Environment , of November

27 , 2002, on the requirements which are to be satisfied by
surface waters intended for the delivering as potable to
population ( in Polish) , (Dz.U.2002.204.1728)

CONTACT WITH THE AUTHOR

Rafa Ossowski

Gdask University of Technology

11/12 Narutowicza St.
80 - 233 Gdask
Poland

POLISH MARITIME RESEARCH, No S1/2017 173

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 174-181
10.1515/pomr-2017-0036

INTERNET PHOTOGRAMMETRY FOR INSPECTION

OF SEAPORTS
Zygmunt Paszotta1
Malgorzata Szumilo1
Jakub Szulwic2
1
Univeristy of Wamia and Mazury in Olsztyn, Poland
2
Gdansk University of Technology, Poland

ABSTRACT

This paper intends to point out the possibility of using Internet photogrammetry to construct 3D models from the
images obtained by means of UAVs (Unmanned Aerial Vehicles). The solutions may be useful for the inspection of
ports as to the content of cargo, transport safety or the assessment of the technical infrastructure of port and quays.
The solution can be a complement to measurements made by using laser scanning and traditional surveying methods.
In this paper the authors recommend a solution useful for creating 3D models from images acquired by the UAV using
non-metric images from digital cameras. The developed algorithms, created and presented software allows to generate
3D models through the Internet in two modes: anaglyph and display in shutter systems. The problem of 3D image
generation in photogrammetry is solved by using epipolar images. The appropriate method was presented by Kreiling in
1976. However, it applies to photogrammetric images for which the internal orientation is known. In the case of digital
images obtained with non-metric cameras it is required to use another solution based on the fundamental matrix
concept, introduced by Luong in 1992. In order to determine the matrix which defines the relationship between left
and right digital image it is required to have at least eight homologous points. To determine the solution it is necessary
to use the SVD (singular value decomposition). By using the fundamental matrix the epipolar lines are determined,
which makes the correct orientation of images making stereo pairs, possible. The appropriate mathematical bases and
illustrations are included in the publication.

Keywords: inspection of wharf, security of seaports, registration of cargo ships, UAV (unmanned aerial vehicles)

INTRODUCTION spatial observations. A tool like a photogrammetry may be

During recent years the use of unmanned aerial vehicles
(UAVs) has increased significantly. UAVs with the capability
of photogrammetric data acquisition opens various new
applications in aerial and terrestrial photogrammetry
and also introduces low-cost alternatives to the classical
photogrammetry [25, 26]. An extensive overview of the
evolution and the state-of-the-art of photogrammetry by
using UAV is given by Eisenbeiss and Unger [4, 31]. The
authors describe early developments and present a review
on UAVs for photogrammetric topics like flight planning,
image acquisition and orientation and data processing. In
this paper the authors focus on generating 3D images from
images taken from UAVs through the Internet.
The solution is based on the website photogrammetry and
can be used for inspection of ports, wharves, coasts, and ships.
Construction of stereoscopic model, optionally enriched
with scale imaging, allows for a correct identification and
useful for evaluating and measuring tanks, silos and special
constructions in the ports [1, 7, 12-15, 17, 28, 30]. The main
limitation of the method is range of UAV and availability of
satellite positioning system [11].
There are also used groups of drones which take images
synchronously, can analyze filling ?, moving the autonomous
objects and inspect moving and vibrating parts of vessels.
Emergency service might be another area of the potential
application where rescue teams might need quick and portable
access to UAV data. The examining stereoscopic model for the
purpose of identifying objects and judging their significance
is much easier than plane image interpretation [21, 22].
In traditional photogrammetry (which is based on
photogrammetric images) generating stereoscopic models
is a public process. The method of analogue spatial
visualization in stereoscopic entertainment mediums such
as Kaiserpanorama (consisted of sets of stereo slides and
amulti-station viewing apparatus) has been known since the

174 POLISH MARITIME RESEARCH, No S1/2017

mid 19th century. In the literature source [20] four conditions THEORETICAL FUNDAMENTALS
are mentioned for the analogue way of obtaining the 3D
effect. They refer to: The concept of a fundamental matrix is introduced,
parallactic angle, assuming that homologous points lie on the epipolar plane
the difference of scales of the left and right image, as the starting point.
separation of transmission channels for left and right According to Fig. 1, the following equation must be
images (sending the right eye image to the right eye satisfied:
and the left eye image to the left eye),
watching pictures on the epipolar plane. rt "(b r') = 0 (1)
When all the above mentioned conditions are met, to
obtain 3D images from non-metric images will be possible. where:
The first condition is satisfied if the images acquired during
the photogrammetric flight have appropriate forward overlap. i j k 0 bz ry '+b y rz ' 0 bz b y rx '

The images taken when the UAVs have only turned around, b r'= bx by bz = bz rx '+0 bx rz ' = bz 0 bx ry ' = B r ' (2)
are improper. In this case the base which represents the rx ' ry ' rz ' b y rx '+bx ry '0 b y bx 0 rz '
distance between the projection centres of both images is
too small. Moreover, these are not overlapping images. If Q is the rotation matrix for vector r" to the object
The second condition (scale stability) should be guaranteed coordinate system (O1,x,y,z), i.e.:
for in-flight plane. The next condition depends on the
selection of 3D display techniques. In this paper the anaglyph (3)
rt "= Q r" ,
method and active shutter 3D systems are implemented. The
theoretical basis of the fourth condition is described in detail then the scalar product is given as :
in this work beneath.
A breakthrough in the process of 3D image generation was (rt ")T Br'= (r")T QT Br'= 0 . (4)
made in 1976 when Kreiling developed a method of generating
epipolar images [3, 9, 27, 33]. However, it is possible to project Let the two cameras be characterized by two calibration
stereogram images onto the common plane if the camera matrices A 1 and A 2 (containing only the principal distance):
constants and the elements of relative orientation are known.
Generating and superimposing component digital images is rp '= A1 r' rp "= A 2 r" . (5)
also a public process.
Many functions and actions which perform measurements
and computations, must be programmed. It appears that it
is possible to generate 3D images (the pair of images which
relative orientation is known and which could be visualized
as a stereoscopic model) with the use of the idea of anaglyphic
images on the Internet, interactively. It is possible to reach
such solution on an ordinary computer, but the quality of
the so obtained three-dimensional image is not sufficient.
The generation and visualization of 3D images by using
shutter glasses is a more advanced approach. Both of the
mentioned solutions work as an Internet application in JAVA
and use client-server technology, which in practical terms
means communication between applets and the servlet.
This paper presents the theoretical fundamentals for the
adopted solution, together with a plan for its implementation.
Since a few years ago (from about 2001) up to now these
authors have developed interactive Internet photogrammetric
applications [22, 23]. The described research is a summary
of work performed on non-metric images.
Due to the use of JAVA software , the presented solution Fig. 1. Geometrical relationships on a stereogram
might be used in most of the contemporary web browsers.
The authors propose a solution which is available through
web browsers as a portable tool and convenient alternative to therefore:
standalone desktop applications. The use of Internet solution
allows to make advantage of all the merits of this technology, (rp " )T AT2 QT B A11rp ' 0 (6)
such as universal access to the data, simultaneous use by
multiple users.

POLISH MARITIME RESEARCH, No S1/2017 175

 Let F AT2 QT B A11 to be called a fundamental matrix. It
helps to determine epipolar lines. For the point with the
coordinates rp " on the right image, the epipolar line on the left
image is described by the following equation (in homogeneous
coordinates):
rp "u"= 0 , where u"= Frp '
number of non-zero elements on the diagonal of diagonal
matrix V. In order to obtain rank (M)=8, or rather rank
(M)=8, the least singular value should be equal to zero.
If our measurements of the coordinates of homologous points
were correct, there should be a sharp jump between the eighth
and the ninth singular value. The corrected matrices V and
(7) M will be marked V and M, respectively , where :

The shape of the matrix F AT2 QT B A11 is 33. There are M ' S' V ' D'T (11)
nine coefficients to determine. One of them is connected
with scale factor and hence it could be reduced, which
makes that only eight coefficients have to be calculated. To
determine them, at least eight homologous points are needed.
This algorithm was introduced by Longuet-Higgins [16, 24].
Therefore Eq. (6) can be written as follows :

(rp " )T F rp ' 0 (8)

All the epipolar lines on a given image plane, either left

or right, go through one epipole. They make up a pencil of
lines which meet in the epilpole (epipolar pencil). If they
are marked e p ' and e p " , respectively, then the following
relationships are satisfied [6]:

F e p '= 0 e p "F = 0 (9)

,

IMPLEMENTATION OF THE APPLICATION

Fig. 2. Activity diagram in the process of generating anaglyphic 3D images
Implementation of the method in question requires the
proper software and execution of various operations both on
the part of the client and of the server. Subsequent stages of The order of matrix M is now 8, and the system of
elaboration of the non-metric stereo pair ( beginning from equations has the following form:
measurement of homologous point to visualization of the 3D
image) are presented in the complete activity diagram Fig. 2. M' f 0 .(12)
Determination of a fundamental matrix is the first, decisive
and most important point in the process of creating 3Dl Adopting one unknown as a parameter, we obtain a
images from non-metric images. Therefore in this section we homogeneous system which will be solved with the use
focus on the method of its calculation. Lets assume that the of SVD. To achieve an unambiguous solution, 8 points
coordinates of at least eight homologous points are measured. are needed. For a larger number of points, the condition
Then Eq. (8) can be applied and written as follows : M ' f  c 0 is replaced with: r 2 (M ' f  c) 2 min which will
be noted as follows :
Mf 0 (10)
M' f | c (13)
where: f [ f11 , f12 , f13 , f 21 , f 2 , f 23 , f 31 , f 32 , f 3 ]T vectors of
ordered elements of matrix F and each row of matrix M On the basis of Eq. (11) and Eq. (13) we have :
equals >x 2 y 2 , y 2 x1 , x1 , x 2 y1 , y 2 y1 , y1 , x 2 , y 2 ,1@ for rp " >x2 , y2 ,1@
and rp ' >x1 , y1 ,1@ . V ' D'T f | S'T c (14)
A system of homogeneous equations with nine unknown
elements of matrix F is obtained. If the rank of matrix M, By marking
equal to k, is lower than nine then it has an infinite number
of non-zero solutions, depending on k parameters. To S'T c = g , D'T f = p , (15)
find the solution of the system we apply singular value
decomposition (SVD). It has a particularly useful property, we have
namely, that the rank (M) is equal to the number of non-
zero values of particular matrices, and so on [= also ? ] the V'p g . (16)

176 POLISH MARITIME RESEARCH, No S1/2017

 Finally, the following solution is obtained The selection of images is the first issue. Data which are
necessary for constructing 3D image , obtained from several
f = D'p . (17) sources:
-images and their fragments - from a directory of images on
The Internet application is implemented in a client-server the data server,
architecture where the client is a web browser. The presented parameters of images (size, resolution)
software allows to generate three-dimensional images from from the database and other data,
non-metric images over the Internet in two modes: anaglyph coordinates of homologous points and
and shutter system. The detailed description of selected 3D initial parameters of process of 3D image generation,
display techniques is presented beneath in this paper. have been delivered to the client.
The flowchart of the application operation is based on the The application has started working on the client side after
possibility to download wirelessly (e.g. by using the Internet downloading the above mentioned website. The user has had
connection through the cellular network, Wi-Fi) the non- to select the stereo pair and appropriate image pyramid level,
metric digital images that are stored on the application server. indicate the area of interest on the chosen images and send
It is possible to automate the process of selecting images for , to application server , request for preparation of a part of
stereo pairs, but this condition is not a subject matter of this the image. Later on, the web browser windows looked like
publication. The authors suggest that it is even purposeful in Fig. 3.
to use the photogrammetric operators (human) experience, It is possible to install ones stereogram on the server
who, by the selection of the proper image, makes it possible- and run the whole process. However, it is a more profitable
for the server application - to generate 3D models with a option to start with using the images which are already there
better quality and thus suitability for the end user. With and the attached text data (set of homologous points). There
this solution, the user of client application will be able to are images taken by OLYMPUS PEN E-P2 camera from
obtain a 3D model useful for analyses related to verification microdrones md4-1000 UAV. The size of each image is 3084
of maritime space. rows by 4096 columns.
The concept of Internet photogrammetry used to develop The camera specification is available on the website http://
3D models from digital images obtained from UAVs , has goo.gl/Q8J27l and http://goo.gl/sa1v8t. Therefore, we select
a significant economic advantage over the solutions based a stereogram and data as given in Fig. 3. The data can be
on a license assigned to the end clients computer. In the modified by using a series of reduced-resolution versions
proposed solution the server application is critical, while of an image formed by smoothing and subsampling of the
client application ran through a web browser is the so-called original left and right image, and stored on the server.
thin client and may be used by a broader group of users Subsequently, we determine the matrix M from formula (10).
(non-professionals in the field of photogrammetry, who may The values of its elements are displayed in a separate window.
stay in different places away) who can simultaneously assess They are given in Tab. 1.
, in terms of another, their own expertise and knowledge. For inspection in the ports, other UAVs equipped with
e.g. a thermal camera and based on different configurations
[2], can be used.
THE EXAMPLE OF ONLINE 3D IMAGE Tab. 1. Elements of the matrix M=[mij]
GENERATION i j 1 2 3 4 5 6 7 8 9
1 531604 153200 383 229020 66000 165 1388 400 1
2 2718729 785166 1223 791388 228552 356 2223 642 1
3 4109767 642001 1601 236164 36892 92 2567 401 1
4 997326 1007883 621 2211462 2234871 1377 1606 1623 1
5 2484746 2305182 1166 3684499 3418233 1729 2131 1977 1
6 5262180 3265020 1870 4159092 25805888 1478 2814 1746 1
7 166625 560170 155 3623825 12182794 3371 1075 3614 1
8 2037110 3515708 1042 6205170 10709076 3174 1955 3374 1
9 5886370 7233877 2051 9712080 11935368 3384 2870 3527 1

Tab. 2. Elements of the orthogonal matrix S=[Sij]

i j 1 2 3 4 5 6 7 8 9
1 -0.0136 0.0504 -0.0540 0.0540 0.0703 0.6944 0.5970 -0.3567 0.1535
2 -0.0593 0.2553 -0.3233 0.0346 0.3502 -0.2050 0.2053 -0.5470 -0.5653
3 -0.0563 0.3586 -0.6906 -0.2317 0.1023 -0.1970 0.1189 0.3923 0.3469
4 -0.1280 0.0358 0.0888 0.6021 0.1723 -0.1015 0.5083 0.5125 -0.2257
5 -0.2205 0.1709 0.1079 0.5024 0.0532 -0.2630 0.0635 -0.4260 0.6329
6 -0.2461 0.4975 -0.1514 0.3154 -0.4869 0.2520 -0.4624 0.0293 -0.2329
7 -0.4247 -0.6424 -0.5021 0.1071 -0.1953 -0.2600 0.0945 -0.1668 0.0287
Fig. 3. A detailed description of elements in the application window 8 -0.4860 -0.2120 0.1059 0.0199 0.2776 -0.5475 -0.4797 0.2687 -0.1678
9 -0.6716 0.2608 0.3315 -0.4625 -0.0027 0.2290 0.3182 -0.0418 0.0415

POLISH MARITIME RESEARCH, No S1/2017 177

Tab. 3. Elements of the orthogonal matrix D=[dij]
x1 a 0  a1 x 2  a 2 y 2 (18)
I j 1 2 3 4 5 6 7 8 9
1 -0.2785 0.7076 -0.6486 0.0325 -0.0009 0.0002 -0.0001 0.0000 0.0000
2 -0.3141 0.3696 0.5022 -0.7159 0.0017 0.0000 0.0013 0.0000 0.0000
3 -0.0001 0.0003 -0.0003 0.0005 0.2581 -0.9571 -0.0792 -0.1052 0.0018
4 -0.5000 0.2219 0.4907 0.6782 -0.0014 0.0000 -0.0013 0.0000 0.0000
5 -0.7575 -0.5599 -0.2938 -0.1627 0.0002 0.0000 -0.0001 0.0000 0.0000
6 -0.0002 -0.0001 -0.0001 0.0014 0.0499 -0.1221 0.6800 0.7212 -0.0049
7 -00002 0.0003 -0.0006 0.0018 0.9420 0.2529 -0.1714 0.1392 -0.0019
8 -0.0003 -0.0001 -0.0002 0.0018 0.2088 0.0714 0.7085 -0.6703 0.0041
9 0.0000 0.0000 0.0000 0.0000 0.0007 0.0013 -0.0003 0.0068 1.0000

Tab. 4. Values on the diagonal of the matrix V=[vij]

v11 v22 v33 v44

26517003.4290 8860122.3556 3241135.9293 742286.8409

v55 v66 v 77 v88 v99 Fig. 4. Epipolar images

1432.5705 178.1029 35.8178 1.5759 0.0092

The matrix M is distributed against the singular values. subsequent points are set by moving along the row in
Tab.2, 3, 4 present the solution of singular value decomposition the first epipolar line in the right image. The line is
(SVD) of the matrix M. determined by the selected starting point and the epipole.
Let the element v99 be equal to zero; according to formulae The corresponding point on the relevant epipolar line of the
(13) (17) we have: left image is selected on the basis of the jump a1 determined
Tab. 5. Elements of the matrix F=[fij] by means of the previously calculated relationship between
the coordinates. 3D images are generated from the pyramid
i j 1 2 3 images without any additional change of scale. The levels
1 -0.00023 -0.00551 17.79343 of brightness of selected basic colour (blue and red) are
2 0.00688 0.00030 -49.46823 interpolated by using the method of the nearest neighbour
3 -18.51439 41.08829 10000.00000 (Fig. 5). Resampling is to be avoided as this would deteriorate
the sharpness of anaglyph and would take a longer time to
The element f33 is a parameter on which the other elements generate it on the server.
of the matrix F depend. After adopting its value the solution
which satisfies the condition (8), is obtained. By doing this, the
second stage shown in the diagram of Fig. 2 , is completed.

3D IMAGE GENERATION WITH THE USE OF THE

ANAGLYPHIC METHOD

A fundamental matrix does not immediately determine the

component images of a 3D image. According to formula (9) it
is possible to determine the epipolar lines which correspond
to each other (Fig. 4). We will not discuss epipolar lines now,
and the reader is referred to the relevant literature [8, 9, 10,
18, 19, 29, 32].
The web-based solution does not generate a 3D image
for the whole area of the joint coverage, but only for the
area selected by the user. This is done by indicating the area
centre in the right and left images and by selecting its size,
e.g. 500x500 pixels. The problem is that it is convenient to
start the generating process from the corner. Let us determine
such a corner for the right image. From the formula (7) we
define the equation of the corresponding epipolar line in the
left image and find the starting point in it. The coordinate x1,
i.e. the number of the column, will be determined in aclose
relationship between the coordinates of the homologous
points which have been identified earlier and used to establish
a fundamental matrix. Adopting its form as: Fig. 5. An example of anaglyph displayed in Internet browser window.

178 POLISH MARITIME RESEARCH, No S1/2017

THE VISUALIZATION OF 3D IMAGES WITH THE USE
OF SHUTTER GLASSES METHOD.
As the mentioned previously, the anaglyph 3D image
is the easiest way of three-dimensional visualization.
However the use of filters diminishes the quality of colour
sensations. Using the active shutter 3D system to visualize
is better solution. To implement this method a computer
with an appropriate graphic card, image display with 120
Hz frequency and shutter glasses , might be essential. Such
hardware configuration is now also available in laptops.
When we want to create the software for 3D visualization,
the algorithms to display alternately colourful epipolar images
for the left and right eye , should be used. Depending on
what the graphics card and its driver enable, one uses the
OpenGL or DirectX tools displaying images in full screen
or window mode. It should be noted, however, that not every
video card with a stereo option accepting the OpenGL has
functions to buffer the stereoscopic images. If we do not
write such computer program, the 3D browser normally
installed together with the graphics card, can be used. In the
case of the NVIDIA GeForce, GTX 460 M is the 3D Vision
Photo Viewer. This solution, however, requires generating
astereoscopic image as a jp file. The user of the program has
three possibilities of visualization:
as stereo-pair images, the left and right images side by
side,
as an anaglyph which has to have glasses with filters,
one may also watch three-dimensional images through
shutter glasses with the use of active shutter method
controlled by the graphics card.
Images can be displayed in full screen or window mode.
Among other programs supporting the jp format in the
Windows system are: the VRex DepthStudio, StereoPhoto
Maker 3DP Stereo Slideshow, Logipole Konvertor, XnView,
3DFantastic 3D Stereo Producer. More detailed information
can be found at http://extension.nirsoft.net/jps. Another
hardware solution for visualizing the three-dimensional
image of the jp files is a TV set equipped with 3D function.
Generally speaking, the jps file is nothing but a modified
jpg file. Images of the left and right stereo pair are included in
a common jpg file. In the initial header part of the collection
and a marker with information about the characteristics
and location of the image components are placed. Full
specifications of this form are available on the Internet
at http://paulbourke.net/dataformats/stereoimage/spec.
pdf. However, such type of collection might be created in
asimplified manner.
We put the left and right images next to each other in the
collection of any graphic format without compression such
as the BMP. Each line of this set contains the first line of the
first image and the second line of the second image. If the
number of pixels in a line of the left and right images is equal,
the total number of pixels in a line should be an even number.
Then, the BMP format is converted to jpg format. In the end,
the name of the extension from jpg to jps should be changed.
Activity diagram for the image generating concept presented
in this work will be similar to images like the one shown in
Fig. 2. However in this case the procedure is modified in such
a way that colourful epipolar images are built, the horizontal
pair are assembled from them and left on the server. The user
draws the image through the Internet in a separate process
and displays it in his/her 3D browser.
CONCLUSION
The paper presents a method of generating epipolar images
based on the method using the fundamental matrix. This
method is applicable in case of using images taken with
non-metric cameras. Epipolar images were used to produce
stereoscopic images by using anaglyph methods and shutter
system. The developed methods and algorithms were utilized
in the web client-server application, which is available at
http://stereo.uwm.edu.pl. The generated 3D images in the
jps format can be found at http://stereo.uwm.edu.pl/stereo/.
The above mentioned website contains applications that
allow to test various solutions regarding the processing and
transmitting of images over the Internet. The Web Server
runs on request, as this solution is non-profit, i.e. given for
scientific and educational purposes. The solutions in question
are not prepared for distribution or subject to other marketing
activities. The presented solution - as a concept and practical
implementation - can be an excellent enrichment of the set
of tools for the inspection of ports, including cargo ships
and cargo handling.
The solutions are actual achievements of these authors
and , besides practical usefulness , have also high educational
merits. It is hoped that they contribute to the development
and popularization of photogrammetric technology.
REFERENCES
1. Bielewicz, E., Gorski, J.: Shells with random geometric
imperfections simulation - based approach. International
Journal of Non-Linear Mechanics, 2002, Vol. 37, Iss. 4-5,
pp.777-784, DOI: 10.1016/S0020-7462(01)00098-1
2. Burdziakowski, P., Szulwic, J.: A commercial of the shelf
components for an unmanned air vehicle photogram-
metry. 16th International Multidisciplinary Scientific
GeoConference SGEM 2016, www.sgem.org, SGEM2016
Conference Proceedings, ISBN 978-619-7105-59-9 / ISSN
1314-2704, June 28 July 6, 2016, Book 2 ,Vol. 2, pp.
739-746, DOI: 10.5593/SGEM2016/B22/S10.095
3. Cho, W., Schenk, T. & Madanim, M.: Resampling Digital
Imagery to Epipolar Geometry. Proceedings of XVIIth
ISPRS Congress Technical Commission III: Mathemati-
cal Analysis of Data, Washington, D.C., USA, ISPRS
Archives, 1992, Volume XXiX , Part B3, pp. 404-408.
POLISH MARITIME RESEARCH, No S1/2017 179
4. Eisenbeiss, H.: UAV photogrammetry. Diss. ETH
No.18515, Institute of Geodesy and Photogrammetry,
Zurich, Switzerland, Mitteilungen , 2009, No.105. p. 235.
5. Elias, R.: Projective Geometry for Three-Dimensional
Computer Vision. Multiple View Geometry in Computer
Vision. Proceedings of Seventh World Multi-conference
on Systemics, Cybernetics and Informatics, SCI03.
Orlando, USA , 2003, Vol. V, pp. 99-104.
6. Frstner, W.: New orientation procedures. The Interna-
tional Archives of Photogrammetry and Remote Sensing,
2000, Vol. XXXIII-B3A, pp. 297-304.
7. Grski, J., Mikulski, T., Ozibo M., Winkelmann K.:
Effect of geometric imperfections on aluminium silo capaci-
ties. Stahlbau , 2015, Vol. 84, iss. 1, pp.52-57
8. Hartley, R., Zisserman, A.: Multiple View Geometry in
Computer Vision. Cambridge University Press, Cam-
bridge, UK, 2000.
9. Hartley, R.I.: In defense of the eight-point algorithm. IEEE
Transactions on Pattern Analysis and Machine Intel-
ligence, , 1997, Vol. 19, Iss. 6 , pp. 580-593.
10. Janowski, A., Nagrodzka-Godycka, K., Szulwic, J.,
Ziolkowski, P.: Remote sensing and photogrammetry tech-
niques in diagnostics of concrete structures. Computers
and Concrete, 2016, Vol. 18, Iss. 3, pp. 405-420, DOI:
10.12989/cac.2016.18.3.405
11. Janowski, A., Nowak, A., et al.: Mobile Indicators in GIS
and GPS Positioning Accuracy in Cities. Lecture Notes
in Computer Science, 2014, Vol. 8537, pp. 309-318, DOI:
10.1007/978-3-319-08729-0_31
12. Kaliski, K. J., Buchholz, C.: HILS for the design of three-
wheeled mobile platform motion surveillance system with a
use of energy performance index. Solid State Phenomena,
2013 , Vol. 198, pp. 90-95.
13. Kaliski, K., Mazur, M.: Optimal control of 2-wheeled
mobile robot at energy performance index. Mechani-
cal Systems and Signal Processing, 2016 ,Vol. 70-71, pp.
373-386.
14. Kozak, J., Tarelko, W.: Case study of masts damage of the
sail training vessel Pogoria. Engineering Failure Analysis.
, 2011, Vol. 18, Iss. 3, pp. 819-827
15. Litwin, W.: Water lubricated marine stern tube bear-
ings - attempt at estimating hydrodynamic capacity.
ASME/STLE 19-21.10.2009 , Proceedings of the ASME/
STLE International Joint Tribology Conference, 2010.
DOI: 10.1115/IJTC2009-15068
180 POLISH MARITIME RESEARCH, No S1/2017
16. Longuet-Higgins, H.: A computer algorithm for recon-
structing a scene from two projections. Nature , 1982 ,Vol.
293, pp. 133-135.
17. Luczak, M., Manzato, S., Peeters, B., Branner, K., Berring,
P., Kahsin, M.: Updating finite element model of a wind
turbine blade section using experimental modal analysis
results. Shock and Vibration , Vol. 2014, iss. 1, pp.71-82
18. Luong, Q.T.: Fundamental matrix and self-calibration.
PhD Thesis, University of Paris, Orsay, 1992.
19. Luong, Q.T., Faugeras, O.D.: The fundamental matrix:
Theory, algorithms, and stability analysis. International
Journal of Computer Vision, 1996, Vol. 17, Iss. 1, pp.
43-75, DOI: 10.1007/BF00127818.
20. McGlone, J.C., Mikhail, E.M, Bethel, J. & Mullen, R.:
Manual of Photogrammetry. Fifth edition. American
Society for Photogrammetry and Remote Sensing, Mary-
land, USA , 2004.
21. Nejadasl, F.K., Lindenbergh, R.: Sequential and
automatic image-sequence registration of road areas
monitored from a hovering helicopter. Sensors, 2014,
Vol. 14, Iss. 9, pp. 16630-16650, DOI:10.3390/s140916630.
22. Paszotta, Z., Szulwic, J., Szumilo, M.: Internet photogram-
metry as a tool for e-learning. 8th International Conference
of Education, Research and Innovation, 2015, ICERI2015,
ISBN: 978-84-608-2657-6, pp. 4565-4573
23. Paszotta, Z., Szumilo, M.: A web-based approach for
online digital terrain model and orthoimage generation.
International Archives of the Photogrammetry Remote
Sensing and Spatial Information Sciences, 2010, Vol. 38-4,
Iss. W13, WebMGS 2010: 1st International Workshop on
Pervasive Web Mapping, Geoprocessing and Services,
Como, Italy, 2010.08.26-27, ISSN 2194-9034.
24. Press, W., Flannery, B., Teukolsky, S. & Vetterling, W.:
Numerical recipes in C: The Art of Scientific Computing.
2nd ed., pp. 59-70, Cambridge University Press, UK. 1992.
25. Przyborski, M., Szczechowski, B., Szubiak, W., Szul-
wic J. & Widerski, T.: Photogrammetric development of
the threshold water at the dam on the Vistula river in
Wloclawek from unmanned aerial vehicles (UAV). 15th
International Multidisciplinary Scientific Geoconference,
SGEM 2015, Albena, Bulgaria, June 18-24, 2015. DOI:
10.5593/SGEM2015/B31/S12.063
26. Ruzgien, B., Aksamitauskas, C., Daugla, I., Prokopimas,
S., Puodiukas, V. & Rekus, D.: UAV photogrammetry for
road surface modeling. The Baltic Journal of Road and
Bridge Engineering, Vilnius , Technika, , 2015, Vol X,
No. 2, p. 151-158 DOI: 10.3846/bjrbe.2015.19.
27. Soheilian, B., Paparoditis, N. & Vallet, B.: Detection CONTACT WITH TE AUTHORS
and 3D reconstruction of traffic signs from multiple view
color images. ISPRS Journal of Photogrammetry and Prof. Zygmunt Paszotta
Remote Sensing , 2013,Vol. 77, pp. 1-20, DOI:10.1016/j. e-mail: paszotta@uwm.edu.pl
isprsjprs.2012.11.009. Univeristy of Wamia and Mazury in Olsztyn
Faculty of Geodesy, Geospatial and Civil Engineering,
28. Sondej, M, Iwicki, P.; Wojcik, M. et al.: Stability analyses of Oczpowskiego 2, 10-718 Olsztyn
a cylindrical steel silo with corrugated sheets and columns. Poland
Steel and Composite Structures, 2016, Vol. 20, Iss. 1, pp.
147-166, DOI: 10.12989/scs.2016.20.1.147 Ph.D. Malgorzata Szumilo
e-mail: malgorzata.szumilo@uwm.edu.pl
29. Szeliski, R.: Computer Vision Algorithms and Applica- Univeristy of Wamia and Mazury in Olsztyn
tions. Springer-Verlag, London, UK , 2011. Faculty of Geodesy, Geospatial and Civil Engineering,
Oczpowskiego 2, 10-718 Olsztyn
30. Tejchman, J., Wojcik, M.: Modelling of shear localization Poland
during confined granular flow in silos within non-local
hypoplasticity. Powder Technology, 2009,Vol. 192, iss. 3, Ph.D. Jakub Szulwic
pp.298-310, DOI: 10.1016/j.powtec.2009.01.021 e-mail: szulwic@pg.gda.pl
Gdansk University of Technology
31. Unger, J., Reich, M. & Heipke, C.: UAV-based photogram- Faculty of Civil and Environmental Engineering,
metry: monitoring of a building zone 2014. The Interna- Narutowicza 11/12, 80-233 Gdansk
tional Archives of the Photogrammetry, Remote Sensing Poland
and Spatial Information Sciences, Vol. XL-5, 2014 ISPRS
Technical Commission V Symposium, 23-25 June 2014,
Riva del Garda, Italy.

32. Wang, L., Liu, Z. & Zhang, Z.: Efficient image features
selection and weighting for fundamental matrix estima-
tion. IET Computer Vision, 2016, Vol. 10, Iss. 1, pp. 67-78,
DOI: 10.1049/iet-cvi.2014.0436.

33. Zhang, Z.Y.: Robust wide-baseline stereo from maximally

stable extremal regions. International Journal of Com-
puter Vision, 1998, Vol. 27, Iss. 2, pp. 161-195.

POLISH MARITIME RESEARCH, No S1/2017 181

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 182-187
10.1515/pomr-2017-0037

THE EFFECT OF FISHING BASIN CONSTRUCTION ON THE

BEHAVIOUR OF A FOOTBRDGE OVER THE PORT CHANNEL

ukasz Pyrzowski
Mikoaj Mikiewicz
Jacek Chrcielewski
Gdansk University of Technology, Faculty of Civil and Environmental Engineering

ABSTRACT

The paper analyses possible causes of failure of the rotating footbridge over the Ustka port channel. In July, 2015,
strange behaviour of this object was observed in the form of excessive vibrations of bridge platform suspension rods,
with the accompanying acoustic effects. A preliminary geotechnical analysis has revealed that this destructive effect
was caused by the nearby construction works, namely construction of a fishing basin and communication routes in
the area close to the bridge, which affected the bridge lashing rod foundation settings. Ground vibrations generated
by certain construction activities were likely to have direct impact on decreasing the bearing capacity of these rods
and increasing the susceptibility od the piles to extraction. After detecting the above problems in bridge operation, its
geodetic monitoring was started. The data recorded during this monitoring, along with the results of force measurements
in the rods, have made the basis for a series of numerical simulations, performed in the Finite Element Method (FEM)
formalism. The bridge structure was analysed in the conditions defined as the emergency state. Extreme efforts of bridge
elements and its dynamic characteristics were examined. A possible source of strange behaviour of the footbridge during
its operation which was recognised during these simulations was the coincidence of the global natural frequency of the
entire bridge structure with local vibrations of suspension rods, at the frequency approximately equal to 1Hz. This
situation was likely to lead to the appearance of the so-called internal resonance phenomenon. As a final conclusion
of the research, recommendations were formulated on possible object oriented corrective actions.

Keywords: footbridge, port basin, failure

INTRODUCTION the object guarantor, the contractor immediately asked for

temporary closing of the bridge.
The analysed footbridge (Fig.1.) is situated in the area
of Port Morski w Ustce Sp.zo.o., Pomorskie voivodeship.
Its function is to pass the pedestrian and cycle traffic over
the port channel being the mouth of the Supia river. The
rotating structure of the bridge makes it possible to close
the passage temporarily to allow the watercraft to enter or
leave the port [19-22].
The footbridge was constructed in accordance with the
design [11] and positively passed load tests [14], which testified
that all design assumptions had been fulfilled. The object Fig. 1. Rotating footbridge for pedestrians over the Ustka port channel
was opened for public use in December, 2013, and until (configuration W1a)
June, 2015, it operated without failure. On July 3, 2015, the
CEO of Port Morski w Ustce Sp. z o.o. informed HYDRO- As a result of experts activities [2], it was concluded that
NAVAL Sp. z o.o., the contractor, about worrying symptoms the investments carried out by the Ustka Commune in the
in footbridge behaviour, observed as excessive vibrations of area close to the footbridge had a considerable effect on its
bridge suspension rods accompanied by acoustic effects. As state. It was recognised that the construction works related to

182 POLISH MARITIME RESEARCH, No S1/2017

the construction of the fishing basin and the communication
routes, which were carried out in close proximity of the bridge,
had changed ground parameters in the area surrounding the
bridge foundations. This effect related to dynamic immersion
of sheet pile walls at a distance of only 13 m from one of bridge
lashing rod anchor blocks. The final effect of the carried-out
construction works was the increased susceptibility of the
piles to extraction [5,6].
The article presents results of the analyses which aimed at
clarifying possible causes of the footbridge failure. The data
coming from geodetic measurements and those recorded in
force measurements in the rods made the basis for a series
of numerical simulations performed in the Finite Element
Method (FEM) formalism. Taking into account the measured
data has made it possible to evaluate the current static effort of
the bridge and the dynamic behaviour of both its individual
elements and the structure as a whole. Then, based on the
obtained results, calculations were made to indicate possible
causes of bridge behaviour change. Finally, after identifying
the most likely failure causes, a simulation was performed
to suggest corrective actions which would allow to reinstate
correct operation of the bridge.
CHARACTERISTICS OF THE OBJECT
The object of analysis is a steel rotating suspended
footbridge structure with the span in the form of truss (Fig. 2).
The pylon jest fixed in the base and stabilised by two lashing
rods anchored by their foundations in the ground. From the
pylon side, the support for the suspended span is provided
by the rotating mechanism, while from the other side it is
supported by a system of hydraulic cylinders taking the form
of sliding articulated supports. The bridge platform is a steel
orthotropic plate. Beside the truss girders, steel elliptic clamps
welded to the truss chords are used as lateral stiffeners of the
span. Span suspension and lashing rods are made of a system
of structural bars.
with open bottom. Rotating the bridge structure between the
operating state (open for pedestrians) and the closed state
(open navigation channel) is executed using the mechanism
installed under the span platform [9,13].
The footbridge was designed for the load caused by a crowd
of pedestrians, q=4kN/m2 acc. to [15], and ambulance travel.
The supporting structure was made of steel S355J, while the
suspension and lashing rods were made of steel DS520.
Basic parameters of the footbridge are: span 52,3m, pylon
height 23,96m, total width 5,38m, total height 4,43m, slope
angle with the obstacle 90.
GEODETIC MEASUREMENTS
In the case of three independent hard construction works,
i.e. construction of the footbridge, the fishing basin, and
the communication routes, carried out in the areas close
to each other, the zones of their mutual influence had to
be determined [5,6]. Unfortunately, this task was not done.
In particular, initial geodetic measurements and object
monitoring in the area influenced by the construction
works were not performed. The geodetic monitoring of the
bridge was only started after its incorrect behaviour had been
reported. The parameters recorded in the measurements
included: pylon deviation from vertical, distances between
lashing rod anchor points, and height ordinates of pylon and
lashing rod foundations. These parameters were measured
in all four positions of the bridge span: opened/locked -W1,
opened/unlocked -W1a, closed/locked -W2, and closed/
unlocked -W2a.
From the point of view of bridge operation analysis, it
was the measurements performed at points situated at pylon
top (Tab.1) which turned out to be most valuable, as similar
data recorded on December 12, 2013, i.e. before the failure
were also available. The calculated differences between them
made the basis for evaluating the cumulated effect of all
construction works. It turned out that during this time, in
similar environmental conditions, the top of the pylon in
position W1a moved towards the span by the distance Ux of
about 10mm (Tab. 1).
Tab. 1. Deviation of pylon top from vertical and changes of foundation
height ordinate. Points situated at south side

Deviation of pylon top Change of lashing rod

from vertical, foundation height ordinate,
Span Ux [mm] Uz [mm]
position
2013 2015 2015 2015 2013 2015 2015 2015
10.12 07.08 09.23 11.24 10.12 07.08 09.23 11.24
Fig. 2. General view of the footbridge in configuration (W1) locked, open
for pedestrian traffic: a) cross section, b) south-side view; XYZ the reference W1 - 87 81 79 - 2334 2339 2340
coordinate system used in the analysis
W1a 75 - 88 85 - - - -

W2 - 72 61 57 - 2334 2339 2340

At the top of the pylon, pull rod anchors are fixed to the
rotating head, which protrudes with respect to the bridge W2a 54 - 64 61 - - - -
opening axis in such a way as to ensure passive movement
of suspension rods to follow the motion of the span. The Due to the fact that they were not monitored from the very
pylon lashing rods are fixed in anchor blocks embedded in beginning of bridge operation, the changes of the pylon and
the ground. The bridge is set on steel pipe piles 508/10 mm lashing rod foundation height ordinate are time-limited and

POLISH MARITIME RESEARCH, No S1/2017 183

only include the values recorded in a relatively short time
interval, from July to November, 2015. It is noteworthy that
construction works in the area close to the footbridge were
conducted long before formal failure reporting, i.e. before the
first monitoring measurement. The measurements performed
in bridge position W1 have revealed that during this short
time of monitoring the foundation of the right lashing rod
moved up (Uz) by 6mm (Tab. 1).
FORCE MEASUREMENTS IN RODS
In order to evaluate current effort states of individual
footbridge elements, forces were measured in the pull rods
composing the bridge suspension and in the lashing rods.
The measurements were made using the vibration method
[23] and the 15-channel set for vibration monitoring in
building structures WE3001a, cooperating with three-axis
accelerometers MEMS. The forces in the rod elements were
evaluated by measuring their free vibrations. Along with
local vibration frequencies, the measurements performed on
selected rods made it possible to work out the characteristics
corresponding to global movements of the entire structure.
The adopted rod labelling in the footbridge structure is given
in Fig.3.
FEM CALCULATIONS
Computer simulations of the state of the structure were
performed in the FEM environment within the linear-elastic
range of material and the geometrically nonlinear range of
the structure with imperfections [1,3,4,7,8,10,17,18]. In the
adopted FEM models (Fig. 4) the following finite elements
were used:
a) 1-dimensional, 2-node spatial rod (beam) finite elements of
Timoszenko type with eccentricity - used to model: pylon,
grate, elliptic clamps, ribs and traverses of orthotropic
platform plate, piles (8632 elements in the model);
b) 2-dimensional, 4-node sheet elements of Timoszenko-
Reissner type used to model: the orthotropic platform
plate (7627 elements in the model);
c) 1-dimesional, 2-node spatial elements of rod-grate type
used to model: the footbridge structure suspension system
(360 elements in the model).
The boundary conditions for the support took into account
the foundation/ground cooperation and the conditions
resulting from locking/unlocking of the bridge span structure
on the river bank opposite to the pylon.

Fig. 3. Top view of the footbridge structure rod labelling

The identified absolute values of normal forces N in the

lashing rods and the longest suspension rods are given in Fig. 4. Visualisation of the FEM model used in the analysis
Tab.2, along with the reference to their design values.
Tab. 2. Comparing measured and designed normal forces N in rods

Span
position W1 W1a W2 W2a

Force Measurement Design Measurement Design Measurement Design Measurement Design

Rod N [kN] N [kN] N [kN] N [kN] N [kN] N [kN] N [kN] N [kN]
OL 426 768 522 748 74 124 72 97
L5 85 145 113 140 78 33 --- 140
OP 483 768 579 748 181 213 223 326
P5 68 145 92 140 78 44 --- 136

Symbol --- means that the measured result (in span
position W2a closed/unlocked) ambiguously indicates
the free vibration frequency. It is noteworthy that the forces
recorded in the measurements differ much from those
assumed by the designer [11] and introduced to the structure
by the rod rigging producer.
The parameters calculated in the steady-state variant
included: extreme internal forces, efforts in complex
state according to the HMH hypothesis, and structure
displacements, as functions of the acting loading scenarios.
The bridge behaviour was verified under the load state as
in [15]: characteristic with respect to the useless state of the
structure, in order to determine basic (P) and additional

184 POLISH MARITIME RESEARCH, No S1/2017

(PD) forces in the rods. The dynamic calculations allowed the in-situ measurements and the numerical simulations
to evaluate global and local (with respect to particular rods) performed in FEM environment indicate that for the unlocked
free vibration frequencies and modes. structure, schemes W1a and W2a, global free vibration
frequencies are observed at the approximate frequency level
of 1Hz.
CURRENT STATE OF THE FOOTBRIDGE Tab. 5. Three first global free vibration frequencies of the footbridge structure
STRUCTURE results of numerical calculations

Span position f1 [Hz] f2 [Hz] f3 [Hz]

The collected measurement data set was used, along with
W1 3,03 3,41 5,31
model updating techniques, to work out a numerical model
reflecting the current state of the footbridge structure. The W1a 0,96 1,42 2,64
compliance of the theoretical description with the really W2 3,05 3,38 5,23
measured rod forces and pylon displacements was analysed. W2a 0,96 1,15 2,66
The identified differences between the forces recorded in the
measurements and those obtained from FEM calculations It is noteworthy that some local free vibration frequencies
were considered acceptable from the point of view of the range of rods, which were measured for force evaluation, are also
and goal of the performed analysis. The adopted numerical situated close to 1Hz. Putting those two facts together
model was positively validated as correctly modelling the justifies formulating the conclusion that it is the compliance
real behaviour of the footbridge structure. Tab. 3 collates of these above frequencies which most likely leads to the
representative pylon displacements. appearance of the internal resonance phenomenon. This
Tab. 3. Measured and calculated pylon top displacements (in useless state) internal resonance, in turn, can lead to coupling of global
free vibration frequencies of the footbridge structure with
Measurement Measurement local frequencies of the longest platform suspension rods. This
Span position [mm] [mm] Calculation
2015.09.23 2015.11.24 [mm] coupling can cause large-amplitude vibrations, accompanied
W1 81 79 81 by characteristic acoustic effects observed on the object during
W1a 88 85 86 footbridge opening/closing.
W2 61 57 57
W2a 64 61 61
THE EFFECT OF LASHING ROD ANCHOR
BLOCK DISPLACEMENT ON FOOTBRIDGE
As the next step, extreme efforts of footbridge elements in OPERATION PARAMETERS
the current state were evaluated. For this purpose, calculations
were performed for a combination of loads, according to the Mistuning of the system of tension forces in the suspension
basic and additional load system [15]. The results are collated and lashing rods, along with the measured slipping of the
in Tab. 4. They should be related to the parameters of steel lashing rod foundation block out of the ground, suggest the
S355J used as material for the structure of the footbridge. need for performing a numerical simulation of parametric
Tab. 4. Extreme efforts of footbridge structure elements in current state, analysis type to assess the effect of possible further foundation
stresses acc. to HMH [MPa] displacement on the redistribution of internal forces in the
structure. This simulation was made for the bridge position
Quantity \ Scheme W1 W1a W2 W2a
variant W1 and the so-called useless state of the structure
Pylon 101 96 187 249
as the initial phase. The controlled variable in the analysis
Lattice girder 169 55 164 166 was the displacement parameter which represented vertical
Span rings 322 53 205 181 upward movement of the right lashing rod foundation. The
Ring stringers 354 45 194 75 control parameters which were considered as representative
Platform, traverse 94 24 78 81
for problem illustration were the following: force in the right
lashing rod, force in the longest span suspension rod, and
Platform, ribs 96 39 68 64
pylon displacement.
Platform, plate 153 60 150 150 The vertical displacement of the foundation which was
recorded during geodetic monitoring increased by 6 mm.
The simulation results reveal that despite the anomalies However, the results of the performed simulation suggest
observed in the dynamic behaviour of the footbridge, the that the real displacement which possibly took place from
static effort state of its individual elements, expressed by the beginning of footbridge operation is of the order of 19
stresses in complex effort state, is correct and does not cause mm, as only for this value the calculated control parameters
anxiety with respect to their potential failure. reach the levels corresponding to the in-situ measurements
To illustrate the dynamic response of the structure in its (see Fig. 5). It is also noteworthy that in the real situation, the
current state, global free vibration frequencies (see Tab. 5) analysed foundation was likely to move in an uncontrolled
and modes were determined numerically. The results of both

POLISH MARITIME RESEARCH, No S1/2017 185

manner also in the horizontal direction, which could affect
the obtained results.
Fig. 5. Changes of control parameters: force in the right lashing rod (OP), force
in the longest right suspension rod (P5), and horizontal pylon displacement
(Ux) as functions of vertical upward displacement of the right lashing rod
foundation
CONCLUSIONS
Restoring correct operation and safe use of the footbridge
over the Ustka port channel requires corrective actions,
which should aim at eliminating damages caused by the
construction of the fishing basin and communication routes
in the area close to the bridge. These corrective actions should
consist, first, in stabilising the most vulnerable lashing rod
foundations and then, in adjusting tension forces in the rods
to bring them to the target state with the vertical axis of the
rotating head at the top of the pylon. After force adjustment
in the rods, the internal resonance phenomenon, being
the effect of coupling of global and local frequencies at the
level of about 1 Hz, should disappear. However, due to the
theoretical nature of the performed analysis, the process of
force adjustment in the rods should be followed by relevant
examination, making a basis for formulating final conclusions
relating to the dynamic response of the object. Additionally,
due to incorrect operation of the system, being the result of
changes in ground conditions in the footbridge foundation
area, during the first year of further bridge operation after its
release to service, permanent Structural Health Monitoring
System (SHT) is to be installed, according to [16], along with
the numerical simulation module which will monitor safe
operation of the footbridge [12,24].
BIBLIOGRAPHY
1. Bielewicz E, Grski J.: Shells with random geometric
imperfections simulation - based approach. International
Journal of Non-Linear Mechanics, 37(4-5), 2002, pp.
777-784.
2. Bolt A., Horodecki G., Cudny M., Wilde K., Chrcielewski
J., Mikiewicz M., Pyrzowski .: Technical expertise con-
cerning possible causes of failure of the openable footbridge
structure over the Ustka port channel (in Polish). Gdansk
University of Technology, 2015.
186 POLISH MARITIME RESEARCH, No S1/2017
3. Burzyski S., Chrcielewski J., Witkowski W.: Geometri-
cally nonlinear FEM analysis of 6-parameter resultant
shell theory based on 2-D Cosserat constitutive model.
ZAMM-Zeitschrift fur AngewandteMathematik und
Mechanik 96 (2), 2015, pp.191-204
4. Chrcielewski J., Witkowski W.: Four-node semi-EAS ele-
ment in six-field nonlinear theory of shells. International
Journal for Numerical Methods In Engineering 68(11),
2006, pp. 1137-1179.
5. Horodecki G.A.: Environmental interactions of deep
excavations in urbanised areas (in Polish). Inynieria
Morska i Geotechnika 3, 2016.
6. Horodecki G.A., Bolt A.F., Dembicki E.: Geotechnical
problems in design and execution of enclosed excava-
tions (in Polish). Inynieria i Budownictwo 12, 2002, pp.
683-686.
7. Iwicki P., Tejchman A., Chrcielewski J.: Dynamic FE
simulations of buckling process in thin-walled cylindrical
metal silos. Thin-Walled Structures 84, 2014, pp. 344-359.
8. Kaliski K.: The finite element method application to linear
closed loop steady system vibration analysis. International
Journal of Mechanical Sciences 39 (3), 1997, pp. 315330.
9. Kaliski K. J., Mazur M., Kaliski K. J., Mazur M.: Opti-
mal control at energy performance index of the mobile
robots following dynamically created trajectories. Mecha-
tronics 37, 2016,pp. 79-88.
10. Kreja I., Schmidt R.: Large rotations in first-order shear
deformation FE analysis of laminated shells. International
Journal Of Non-Linear Mechanics 41(1), 2006, pp.101-123.
11. ukaszewicz A.: Design of openable footbridge over the
Ustka port channel (in Polish). HYDRO-NAVAL Sp.zo.o.:
2013.
12. Mariak A., Mikiewicz M, Meronk B., Pyrzowski .,
Wilde K.: Reference FEM model for SHM system of
cable-stayed bridge in Rzeszw. Advances in Mechanics:
Theoretical, Computational and Interdisciplinary Issues,
Taylor & Francis Group, London 2016, pp. 383-387.
13. Mikielewicz J., Mikielewicz D.: A simple model of circular
hydraulic pump. International Journal of Heat and Mass
Transfer52(1), 2008, pp. 17-21.
14. Mikiewicz M., Wilde K., Chrcielewski J., Pyrzowski .
et al.: Report of test loads of the openable footbridge over
the Ustka port channel (in Polish). Gdansk University of
Technology, 2013.
15. PN-85/S-10030 Bridge objects. Loads (in Polish).
16. Ordinance of the Minister of Maritime Economy of CONTACT WITH THE AUTHORS
23 October, 2006, on technical conditions of use and
detailed scope of examination of marine hydro-technical ukasz Pyrzowski
structures (in Polish).
Mikoaj Mikiewicz
17. Sabik A., Kreja I.: Large thermo-elastic displacement and Jacek Chrcielewski
stability FEM analysis of multilayered plates and shells.
Thin-Walled Structures 71, 2013, pp.119-133. Gdansk University of Technology
Faculty of Civil and Environmental Engineering
18. Siemitkowski M., Przybylski W.: Simulation studies of 11/12 Narutowicza St.
process flow with in-line part inspection in machining 80 - 233 Gdask
cells. Journal of Materials Processing Technology 171 Poland
(1), 2006, pp. 27-34.

19. Szapczyski R.: Evolutionary Sets of Safe Ship Trajectories

Within Traffic Separation Schemes. Journal of Navigation
66 (1), 2012, pp. 65-81.

20. Szapczyski R.: Evolutionary Planning of Safe Ship Tracks

in Restricted Visibility. Journal of Navigation 68 (01), 2015,
pp. 39-51.

21. Szapczyski R., Szapczyska J.: Customized crossover

in evolutionary sets of safe ship trajectories. International
Journal of Applied Mathematics and Computer Science
22 (4), 2012, pp. 999-1009.

22. Szapczyski R., Szapczyska J.: On evolutionary comput-

ing in multi-ship trajectory planning. Applied Intelligence
2011, pp. 1-20.

23. Topolewicz K., Topolewicz M., Rucka M., Wilde K.:

Vibration method for evaluating forces in suspension
members illustrated by the case Gdansk overbridge (in
Polish). Wrocawskie Dni Mostowe, 2014, pp.439-446.

24. Wilde, K., Mikiewicz, M., Chrcielewski, J.: SHM

System of the Roof Structure of Sports Arena Olivia.
Structural Health Monitoring 2013 Vol. II, Pennsyl-
vania 2013, pp. 1745-1752.

POLISH MARITIME RESEARCH, No S1/2017 187

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 188-194
10.1515/pomr-2017-0038

UNDERWATER WET REPAIR WELDING OF API 5L X65M

PIPELINE STEEL

Grzegorz Rogalski
Dariusz Fydrych
Jerzy abanowski
Gdansk University of Technology, Poland

ABSTRACT

This paper presents results of the research of effect of polymer insulation of pipeline made of API 5L X65M steel
as well as underwater wet welding parameters on properties of joints made by covered electrodes. Effect of heat
input on structure and hardness of joints during repair of underwater pipeline was analyzed. Welding defects like
microcracks, micro-lacks of fusion, slag inclusions, as well as HAZ hardness increase over an assumed acceptance
criterion for welded joints in pipes without anticorrosion polymer insulation, were identified. A significant effect of
polimer insulation on structure and properties of welded joints, was found.

Keywords: underwater welding, wet welding, underwater pipeline, API 5L X65M steel

Tab.1. Grades of steels applicable to pipelines acc. API 5L rules and

INTRODUCTION ISO3183 standard

For the mining and operating of crude oil and natural Standard Notation of steel grades
gas resources located under seabed drilling platforms of
various design solutions depending on many factors are used. API 5L A25 A B X42 X46 X52 X56 X60 X65 X70

Operation of such platforms is always connected with necessity

ISO 3183 L175 L210 L245 L290 L320 L360 L390 L415 L450 L485
of drilling the borehole and transporting the stuff. It is possible
due to application of pipes of appropriately high mechanical Re min
175 210 245 290 320 360 390 415 450 485
properties. In case of the mining in deep waters both casing MPa
column of boring rig and underwater transporting pipelines
are usually made of a high strength steels in compliance with Underwater pipelines are very often protected against
rules of American Petroleum Institute (API) as well with ISO aggressive water environment by means of anticorrosive
standards. These are weldable C-Mn steels or C-Mn steels with polymer coating. Such protection increases their service
small alloying addition of niobium, vanadium or titanium life. An excellent example of this technology are pipelines
(Nb+V+Ti0,15%) delivered in normalized, quenched and used for pressing water into deposits located 1450 m deep
tempered condition or after thermomechanical processing. under Baltic Sea bed and exploited by Baltic Beta platform of
Depending on chemical composition and delivery condition, Lotos Petrobaltic firm. Because of low pressure in the deposit
the pipes show yield strength within the range of 175830 a water pressing system which increases the pressure and
MPa. Tab. 1 presents the comparison of basic grades of steels extracts crude oil, was installed. To build such pipelines pre-
for pipelines inaccordance with API 5L rules and ISO 3183 insulated pipes made of X65 API 5L high strength steel are
standard. commonly used.

188 POLISH MARITIME RESEARCH, No S1/2017

 In spite of many applied precaution systems ocean by applying low-alloy electrodes containing nickel. 2% Ni
engineering structures may undergo failures [1,2]. The content in deposited metal is considered optimum value [24].
main causes of the failures are the following: corrosion Authors [25, 26] described repair work carried out on
wear, collisions with vessels, excessive operational loads, a pipeline of API 5L X65M steel in underwater conditions
material defects, fatigue of materials, improper construction with the use of the Friction Taper Plug Welding (FTPW)
as well as workmanship [3-6]. It is usual practice to remove method consisting in drilling the surface layer of damaged
used or failed elements of pipelines and substitute them pipe and friction welding of holes by using appropriately
with new ones. However it is often necessary to repair fitted steel pins.
locally a pipeline by using welding processes. All the repair This work concerns a research which has been undertaken
operations on pipelines with the use of welding or overlaying to evaluate if it is possible to conduct underwater repair of
welding are to meet determined quality criteria. Such criteria a pipeline by using the covered electrode welding method
concerning underwater welding are given in AWS D3.6M: (MMA). Moreover, possible influence of anticorrosive coating
2010 requirements. The document determines 3 classes for of pipeline on weld quality was investigated.
welded joints made underwater. The class A covers joints with
properties close to those produced in the air and is usually
applicable to joints produced in hyperbaric dry conditions. EXPERIMENTAL
The class B was introduced for less-responsible welded joints
of a lower permissible value of toughness, greater porosity For the experiments two pipe segments of the dimensions:
with possible presence of inclusions in weld applicable to =102.9 5.7 250 mm were taken from the water pressing
welded joints produced with a wet method. The class O for pipeline installed at the underwater crude-oil field (Fig. 1).
welded joints operating under low mechanical loading; it The specimen C was covered with an anticorrosive polymer
contains lower demands in relation to the class B [7]. coating of 2.5 mm in thickness. The specimen W had metallic
The wet welding method with the use of covered electrodes surface after mechanical removal of the anticorrosive coating.
is the most commonly applied to underwater welding [8-13]. Over 75% circumference of the pipes 1.5 mm-deep grooves
The technique is characterised by a relatively low cost and simulating the repair - demanding defects (cracks, corrosive
possibility of carrying out the welding in situation where wear) on pipeline surface, were cut.
application of other methods, e.g. a local dry chamber or
dry welding, is either unprofitable or impossible at all [9].
An advantage of wet welding is possible easy operation of
welding torch and its drawback is a limited water depth
at which a diver- welder is permitted to carry out welding
operations. Direct contact of joined elements and electric
arc with water contributes to occurrence of welding defects.
The greatest problems are due to susceptibility to cracking,
especially cold cracking, of welds made underwater [3,8,10,13-
17]. High cooling rate of welded joint and associated with
it tendency to forming hard quenching structures within
the HAZ and increased amount of diffusible hydrogen in
deposited metal is conducive to the phenomenon [18-22].
Cold cracking is one of the forms of hydrogen-generated
failures of welded joints [10,17,23].
Nevertheless, underwater welding has been for many years
applied successfully to repair structures made of low-carbon
steels. With appropriately selected welding technology, welds Fig. 1. Specimens prepared to welding. C specimen covered with anticorrosive
and overlying welds of relatively good mechanical properties polymer coating; W specimen without the coating.
can be obtained. More troublesome is the underwater welding
of high strength steels with a higher value of carbon equivalent
(Ce). Based on many investigations, approximate safety
criteria for the wet welding of non-alloy or low-alloy steels, MATERIALS USED FOR THE EXPERIMENTS
have been determined as follows [21]:
carbon equivalent for base material: Ce < 0,30%, The pipeline was made of API 5L X65M steel and to its
heat input: higher than 1,5 kJ/mm. welding Barracuda E 46 2 1 Ni RR 51 (AWS A5.1: E7014)
In case of application of austenitic electrodes value of covered electrodes intended for the underwater welding
carbon equivalent of welded steel may be even higher - up were used. Chemical composition of the pipes and electrode
to 0,40 %. In such situation possibility of the occurrence of deposit is given in Tab. 2 and their mechanical properties -
hot cracks in weld should be taken into account [21]. An in Tab. 3. Carbon equivalent was determined according to
increase in weld plasticity at wet welding can be reached the formula(1).

POLISH MARITIME RESEARCH, No S1/2017 189

 No. Welding current Arc voltage Travel speed Heat input* Remarks
I [A] U [V] Vsp [mm/s] QL [kJ/mm]
(1)
C3 194 28.8 3.6 1.24 Insulated
specimen
Tab. 2. Chemical composition of API 5L X65M steel and weld deposit
ofE462 1 Ni RR 51 electrodes W1 215 28.3 3.2 1.52 Non-insulated
specimen
Percent weight content of components
Material W2 200 27.5 3.3 1.33 Non-insulated
C Si Mn P S Cr Ni Cu CE IIW specimen
API 5L X65M Max. Max. Max. Max. Max. - - -
Max.
0.12 0.45 1.60 0.025 0.015 0.43 Non-insulated
Acc. to the W3 194 28.8 3.6 1.24 specimen
certificate 0.13 0.30 0.78 0.005 0.005 0.57 0.16 0.23 0.41
E 46 2 1 Ni RR 51 0.05 0.45 0.50 0.025 0.025 - 0.3 - - *
Heat input was calculated according to PN-EN 1011-1:2009 standard

Tab. 3. Mechanical properties of API 5L X65M steel and weld deposit

ofE462 1 Ni RR 51 electrodes
Mechanical properties
Material Yield point Tensile strength Elongation
RESULTS OF THE TESTS
[MPa] [MPa] [%]
API 5L X65M 450-600 535-760
E 46 2 1 Ni
RR 51 min. 560 500-640 min. 20 NON-DESTRUCTIVE TESTS

E 46 2 1 Ni RR 51 electrodes have granted approval for The visual and penetrating tests were carried out in
underwater welding, issued by ABS and LRS classification compliance with the recommendations of PN-EN ISO 17637
societies. Because of a specific character of water environment and PN-EN ISO 3452-1 standards, 72 h after completion
the electrodes are additionally covered with a layer which of welding. As assumed, the welded joints have to meet
protects electrode coating against absorbing humidity. The acceptance criteria for B - quality level according to the
electrodes facilitate slag removal after welding, that makes requirements of PN-EN ISO 5817 standard, excluding defects
it possible to avoid forming non-metallic inclusions in weld of 5th group acc. PN-EN ISO 6520-1, e.g. an excessive weld -
during multilayer welding. back reinforcement. No welding defects characteristic for such
The experiments were carried out with the following plan: joints and welding conditions, e.g. weld or HAZ cracks as well
execution of technological tests of welding in order as craters or pores, were observed. Local spots of incompletely
to determine welding current parameters as well as filled groove were found on the insulated specimen C2 (Fig. 2).
optimization of angular position of electrode against the
pipe, DESTRUCTIVE TESTS
underwater wet repair welding,
non-destructive, visual and penetrating tests, Macroscopic examinations
macroscopic and microscopic examinations, The macroscopic examinations were conducted according
hardness measurements by Vickers method. to the requirements of PN-EN ISO 17639 standard. Fig. 2
One-layer surface circumferential welds were made on and3 present macro cross-sections of the surface layer welds.
test stand intended for the underwater welding up to the
depth of about 0.5 m. The surface welds were produced by
using covered electrodes of 4.0 mm diameter, in flat welding
position, supplied with direct current electrode negative
(DCEN) polarity. The test stand chamber was filled with
the artificial sea water (of 13 salinity) in 20C temperature.
Six welded specimens were produced, 3 of them on the
pipes insulated with the polymer coating which was removed
over the width of about 20 mm, and 3 on the pipes without
insulation. Three values of heat input corresponding with
the minimum and maximum values of welding current
appropriate for the applied type of electrodes, as well as
the mean value of welding current, were used. The welding
parameters are presented in Tab. 4.
Tab. 4. Welding parameters

No. Welding current Arc voltage Travel speed Heat input* Remarks
I [A] U [V] Vsp [mm/s] QL [kJ/mm]

C1 215 28.3 3.2 1.52 Insulated

specimen

C2 200 27.5 3.3 1.33 Insulated

specimen

190 POLISH MARITIME RESEARCH, No S1/2017


Fig. 2. Macro cross-section of the welded specimens C1, C2, C3 (insulated
specimen C). Etch. Nital
Fig. 3. Macro cross-section of the welded specimens W1, W2 i W3 (specimen
without insulation W). Etch. Nital
The specimens presented in Fig. 2 and 3 show full depth of
fusion and fully filled groove, only the specimen C2 exhibited
undercuts in face of the weld. In case of underwater welding
such defects are caused by an improper angular position
of electrode against welded element. The remaining joints
have a correct, symmetrical structure. On all specimens
the wide heat affected zone caused by influence of the water
present inside and outside the pipe, can be observed. In the
insulated specimens (C1, C2, C3), the HAZ covers a smaller
area than in the specimens without insulation (W1, W2, W3).
It can be explained by a lower cooling rate of the joint due
to presence of polymer coating on pipe surface. The largest
fusion penetration was observed in the specimen without
polymer coating (W1) where the largest heat input, equal to
1,52 kJ/mm, was applied.
Microscopic examinations
The microscopic examinations were performed to reveal
structures in characteristic areas of the joints as well as
possible microscopic welding defects. Fig. 4 presents of base
and weld materials structures. In the base material fine-grain
ferrite structure with a low share of pearlite, characteristic
for this grade of steel, is observed, whereas the weld has a
dendritic structure with column arrangement of ferrite and
quasi-pearlite.
a) b)
Fig. 4. Microstructures: a) base material of API 5L X65M steel, b) weld area.
Magn. 100x
In the welded joints in insulated pipes no cracks were
found. Fig. 5 shows microstructures along fusion line of the
joints. There are visible characteristic areas: the dendritic
structure of the weld, coarse-grain structure of the HAZ,
normalized area of the HAZ, as well as the structure of the
base material. As can be observed, in HAZ microstructure
of the C1 specimen made with the highest value of heat input
the biggest grains were formed close to the fusion line, Fig.
5a. In the remaining joints (C2 and C3) size of the grains in
the HAZ is significantly smaller, Fig. 5b,c.
a) Microstructure of the specimen b) Microstructure of the specimen
C1, magn. x200, C2, magn. x200,
c) Microstructure of the d) Microstructure of the specimen
specimen C3, magn. x200, W3, magn. x200,
Fig. 5. Microstructure of HAZ in the joints where no cracks were found;
a) the specimen C1, b) the specimen C2, c) the specimen C3;
d) the specimen W3, magn. x 200
POLISH MARITIME RESEARCH, No S1/2017 191
 The tests revealed presence of trans-crystalline cracks in
the HAZ microstructure of the specimens without insulation
(W1 and W2), Fig. 6. They are characteristic cold cracks which
are formed at high cooling rate of the joint in the conditions
of stress and diffusible hydrogen occurrence. In both the
specimens the cracks propagated along the fusion line: in
the specimen W1 across the area of bainitic structure (Fig.
6a), and in the specimen W2 (Fig. 6b) the crack occurred
in the coarse-grain area of the HAZ. In the specimen W3
welded with the lowest value of heat input no cracks were
found (Fig. 5d).
a) b)
an excessive increasing of the maximum hardness in the
HAZ, up to the level of 386 HV10 and 385 HV10 for the upper
and lower measurement line, respectively, was observed. In
HAZ of the subsequent specimens without insulation coating
(W2 and W3) the maximum hardness values equal to 343
and 353 HV10, respectively, i.e. close to that of the selected
criterion, were measured.
a)

Fig. 6. Cracks occurred in the HAZ of the specimens : a) W1, magn. x 200; b)
b) W2, magn. x 100

MEASUREMENTS OF HARDNESS

The hardness measurements were conducted on the

cross-section of tested joints by using the Vickers method in
accordance with the recommendations of PN-EN ISO 9015-1
standard. Value of the loading force was set on the level of
98 N (HV10). The measurements were performed at 2 mm
distance from the upper and lower edge of the specimen, Fig. 7. c)
The distance between centres of indents was not smaller than
1 mm. Fig. 8 and 9 show results of the tests.

Fig. 8. Distribution of HV10 hardness values along the measurement lines 1

and 2 for the welded joints of the pipes with insulation coating : a) C1, b) C2,
c) C3

Fig. 7. Distribution of hardness measurement points on the cross-section a)

ofwelded repair joints

The hardness acceptance criterion for the tested joints was

taken equal to 350 HV10 on the basis of the DNV-OS-C401
Offshore Standard (Fabrication and testing of offshore
structures). Maximum hardness values of the HAZ in the
insulated pipe specimens amounted from 300 to 318 HV10,
hence the values were distinctly lower than the acceptance one
equal to 350 HV10. In the specimen W1 (without insulation)

192 POLISH MARITIME RESEARCH, No S1/2017

b)
c)
Fig. 9. Distribution of HV10 hardness values along the measurement lines
1and 2 for the welded joints of the pipes without insulation coating:
a) W1, b) W2, c) W3
SUMMARY
The results of the conducted research show that it is
possible to perform repair of cracks and fill-up corrosion
wear in the pipeline made of API 5L X65M steel, by means
of wet welding with the use of covered electrodes both in the
case of insulated pipes and those without insulation coating.
The visual and penetrating tests showed that the applied
welding procedure made it possible to obtain welded joints
of a correct and symmetrical structure and relatively smooth
face. The macroscopic examinations did not disclose any
welding defects typical for underwater wet welding with the
use of covered electrodes, such as non-metallic inclusions,
cracks, pores, incomplete fusion.
During microscopic examinations, a growth of grain size
in the HAZ of the joints, proportional to the heat input, was
observed. In the HAZ of two specimens without insulation
(W1 and W2) cold cracks were found. The cold cracks run
across crystals through brittle areas of the HAZ. It was caused
due to the following factors : the relatively high value of carbon
equivalent of the steel (0,41%), high cooling rate of the joint as
well as high level of diffusible hydrogen in the welded joint.
On the basis of the earlier conducted tests [16,21] it may be
assessed that the diffusible hydrogen content in the deposited
metal exceeded 35 ml/100g. Presence of polymer insulation
was conducive to the lowering of width of the HAZ. Any
distinct effect of heat input during welding (within the range
of 1.24-1.52 kJ/mm) on maximum hardness in the HAZ of
the joints, was not observed.
It is worth stressing that changes in amount of introduced
heat are limited for the reason of necessity of obtaining
correct geometry of the joints. However, a distinct decrease
in maximum hardness of the HAZ (below assumed in the
DNV-OS-C401 standard criterion) was revealed in the joints
of polymer insulated pipes, It may be expected that in the case
of welding elements of a greater thickness, which requires
to use multi-pass technique, a drop in maximum hardness
of the HAZ, resulting from the phenomenon of tempering,
would occur [27].
CONCLUSIONS
On the basis of an analysis of the results of the performed
research, the following conclusions may be drawn:
1. It was proved that it is possible to repair cracks and fill-up
corrosion wear in a pipeline made of API 5L X65M steel
by using wet welding with covered electrodes.
2. Within the scope of the research no distinct effect of heat
input on maximum hardness in the HAZ of the tested
joints was observed.
3. Removal of the polymer insulation resulted in increasing
of maximum hardness in the HAZ and occurrence of cold
cracks in the HAZ.
BIBLIOGRAPHY
1. Ersdal G., Sharp J., Galbraith D.: Ageing accidents: sug-
gestion for a definition and examples from damaged
platforms. ASME 2014 33rd International Conference on
Ocean, Offshore and Arctic Engineering, OMAE 2014-
23674, pp. 1-9.
2. Ismail Z., Kong K.K., Othman S.Z., Law K.H., Khoo
S.Y., Ong Z.C., Shirazi S.M.: Evaluating accidents in the
offshore drilling of petroleum. Regional picture and reduc-
ing impact. Measurement, 51, 2014, pp. 18-33.
3. Rowe M., Liu S.: Recent developments in underwater wet
welding. Science and Technology of Welding and Joining,
6(6), 2001, pp. 387-396.
4. Kozak J., Grski Z.: Fatigue strength determination of
ship structural joints. Polish Maritime Research, 18(2),
2011, pp. 28-36.
5. Szlapczynski R., Szlapczynska J.: An analysis of domain-
based ship collision risk parameters. Ocean Engineering,
126, 2016, pp. 47-56.
6. Jakubowski M.: Influence of pitting corrosion on fatigue
and corrosion fatigue of ship structures. Part I Pitting
corrosion of ship structures. Polish Maritime Research,
21(1), 2013, pp. 62-69.
POLISH MARITIME RESEARCH, No S1/2017 193
7. AWS D3.6M: 2010: Specification for underwater welding.
American Welding Society, Miami, USA 2010.
8. Kong X., Li C., Zou Y., Hu Y.: Measurement and analysis
of the diffusible hydrogen in underwater wet welding joint.
MATEC Web of Conferences 39, 2016, pp. 1-5.
9. Rogalski G., abanowski J., Fydrych D., Tomkw J.: Bead-
on-plate welding on S235JR steel by underwater local dry
chamber process. Polish Maritime Research, 21(2), 2014,
pp. 58-64.
10. abanowski J., Prokop-Strzelczyska K., Rogalski G.,
Fydrych D.: The effect of wet underwater welding on cold
cracking susceptibility of duplex stainless steel. Advances
in Materials Science, 16(2), 2016, pp. 68-77.
11. Gao W., Wang D., Cheng F., Di X., Deng C., Xu W.: Micro-
structural and mechanical performance of underwater
wet welded S355 steel. Journal of Materials Processing
Technology, 238, 2016, pp. 333-340.
12. Garai I., Kralj S., Kouh Z., Pacak M.: Analysis of
underwater repair technology on the jack-up platform
spud can. Brodogradnja, 61(2), 2010, pp. 153-160.
13. Arias A.R., Bracarense A.Q.: Velocidade de propagao
de trinca por fadiga de soldas subaquticas molhadas:
avaliao fora da gua. Soldagem & Inspeo, 20(4), 2015, 25. Yin Y., Yang X., Cui L., Cao J., Xu W.: Microstructure and
pp. 403-411.
14. Silva L.F., Dos Santos V.R., Paciornik S., Mertens J.C.E.,
Chawla N.: Multiscale 3D characterization of discontinui-
ties in underwater wet welds. Materials Characterization,
107, 2015, pp. 358-366.
15. Gao W., Wang D., Cheng F., Deng C., Liu Y., Xu W.:
Enhancement of the fatigue strength of underwater wet
welds by grinding and ultrasonic impact treatment. Jour-
nal of Materials Processing Technology, 223, 2015, pp.
305-312.
16. Fydrych D., wierczyska A., Rogalski G.: Effect of
underwater wet welding conditions on the diffusible
hydrogen content in deposited metal. Metallurgia Italiana,
107(11/12), 2015, pp. 55-60.
17. Maksimov S.Y.: Underwater arc welding of higher strength
low-alloy steels. Welding International, 24(6), 2010,
pp. 49-454.
18. Padhy G. K., Ramasubbu V., Murugesan N., Remash C.,
Albert S.K.: Effect of preheat and post-heating on diffusible
hydrogen content of welds. Science and Technology of
Welding and Joining, 17(5), 2012, pp. 408-413.
194 POLISH MARITIME RESEARCH, No S1/2017
19. Pacikiewicz K., Zieliska-Lipiec A., Tasak E.: Cracking
of high-strength steel welded joints. Advances in Materials
Science, 13(3), 2013, pp. 76-85.
20. Pandey C., Saini N., Mahapatra M.M., Kumar P.: Hydro-
gen induced cold cracking of creep resistant ferritic P91
steel for different diffusible hydrogen levels in deposited
metal. International Journal of Hydrogen Energy, 41(39),
2016, pp. 17695-17712.
21. Fydrych D, abanowski J., Rogalski G.: Weldability of high
strength steels in wet welding conditions. Polish Maritime
Research, 20(2), 2013, pp. 67-73.
22. Nowacki J., Sajek A., Matkowski P.: The influence of weld-
ing heat input on the microstructure of joints of S1100QL
steel in one-pass welding. Archives of Civil and Mechani-
cal Engineering, 16(4), 2016, pp. 777-783.
23. Zieliski A., Sobieszczyk S.: Hydrogen-enhanced degra-
dation and oxide effects in zirconium alloys for nuclear
applications. International Journal of Hydrogen Energy,
36(14), 2011, pp. 8619-8629.
24. Guo N., Liu D., Gu W., Li H., Feng J.: Effect of Ni on
microstructure and mechanical properties of underwater
wet welding joint. Materials & Design, 77, 2015, pp. 25-31.
mechanical properties of underwater friction taper plug
weld on X65 steel with carbon and stainless steel plugs.
Science and Technology of Welding and Joining, 21(4),
2016, pp. 259-266.
26. Wang F., Yang X., Cui L., Yin Y.: Fabricating defect-free
API X65 steel welds in underwater wet condition with
friction taper plug welding. Materials and Manufacturing
Processes, 31(16), 2016, pp. 2123-2129.
27. Fydrych D., wierczyska A., Rogalski G., abanowski
J.: Temper bead welding of S420G2+M steel in water envi-
ronment. Advances in Materials Science, 16(4), 2016, pp.
5-16.
CONTACT WITH THE AUTHORS
Grzegorz Rogalski
Dariusz Fydrych
Jerzy abanowski
Gdask University of Technology
Faculty of Mechanical Engineering
Poland
Corresponding author: grzrogal@pg.gda.pl
 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 195-202
10.1515/pomr-2017-0039

LIVING AND WORKING BENEATH THE SEA NEXT APPROACH

Lech Rowiski
Gdask Univerity of Technology, Poland

ABSTRACT

The idea of living beneath the sea is very new if compared with millennia of shipping activity. In fact, ocean surface
was considered mainly as medium suitable for transport of persons and goods as well as aggression and robbery. More
practical attempts to live on the water surface are limited to well protected internal waters.
The presented concept of an underwater-above water accommodation, especially for residential purposes proposes
a possible solution. It was developed under the name of Water Discus. To some extent it is a response to expressed
requirements of potential operators as well ideas of European Commision regarding development of the blue economy
(doc COM (2014) 254). Several basic research and design experiments were required to achieve current understanding
of the concept requirements..

Keywords: living beneath the sea, blue economy, underwater accommodation

INTRODUCTION Beside manned submersibles, several hyperbaric habitats

The idea of living beneath the sea is very new if compared
with millennia of shipping activity. In fact, ocean surface
was considered mainly as medium suitable for transport
of persons and goods as well as aggression and robbery.
More practical attempts to live on the water surface are
limited to well protected internal waters. Idea of men living
beneath the sea was born during dark days of World War
II when scuba diving technology was developed practically.
Tether-less diving apparatus for everybody allowed for
mass scale contact with underwater environment and for
widely understood exploration and pleasure. New capabilities
crated also new expectations. The following next two decades
witnessed fierce development of manned marine technology.
were developed and tested at sea. Finally, whole submarine
towns were proposed. The process proved that available
technology is adequately advanced to assure safety. At the
same time it has become apparent that proposed solutions
are too expensive to assure commercial feasibility. It has also
indicated importance of psychological issues that need to be
considered while developing a manned system to be used in
strange environments. A very comprehensive summary of
these achievements can be found in literature [1]. Fifty years
later, ideas regarding living and working beneath the sea are
even more futuristic than they were before. The presented
concept of an underwater-above water accommodation,
especially for residential purposes proposes a possible

POLISH MARITIME RESEARCH, No S1/2017 195

solution. It was developed under the name of Water Discus.
To some extent it is a response to expressed requirements of
potential operators as well ideas of European Commision
regarding development of the blue economy (doc COM
(2014) 254). Several basic research and design experiments
were required to achieve current understanding of the concept
requirements.
GENERAL CHARACTERISTICS OF THE
WATER DISCUS CONCEPT
The Water Discus (WD) is a novel-type underwater
accommodation designed for coastal, shallow water areas.
The WD is a manned, floating structure (barge) that can
be self-propelled or towed and temporarily anchored to the
bottom or a bottom fixed foundation. The WD is not a
passenger submersible craft. Its submerged accommodation
part (Water Disc) has a constant connection with the part
located above the water level. The solution assures permanent
atmospheric air ventilation and several independent escape
routes. The main idea behind the concept of the WD is to
create an environment that will allow users to stay underwater
for an unlimited period of time and enjoy the views of
the surrounding marine life (the coral reef ) through the
panoramic windows. In addition, WD aims to offer facilities
for various underwater research activities.
The WD complex consists of three basic components:
1. A movable unit called Float that contains an underwater
accommodation called the Water Disc,
2. A fixed unit called the Air Disc,
3. A Support of the Air Disk composed of a few legs fixed
into a foundation structure.
The components are separate structures connected by
means of flexible hoses and cables to allow relative vertical
movements.
a b
Fig. 1 Configurations of the WD units depending on water depth and space requirements.
a Basic configuration for 10 m water depth, surfacing capability; b - Basic configuration for 10-30 m water depth, surfacing capability; c Reduced configuration
for 20-30 m water depth, no legs, surfacing capability; d Minimal configuration for 10 m water depth, no legs, no surfacing
196 POLISH MARITIME RESEARCH, No S1/2017
DESCRIPTION OF THE WD UNIT
THE FLOAT
The Float is the basic element of the WD concept. It is
composed of the Water Disc, a central vertical communication
shaft and a ballast tank. The Water Disc provides underwater
accommodation to the inhabitants or visitors that are interested
in exploration of external marine environment. It is being
submerged using water ballast pumped into the ballast tank.
The water ballast is used to add as much weight to the Water
Disc as needed to exceed its buoyancy, allowing it to submerge
to the sea bottom. It is not submerged permanently as in a case
of hydro-technical structure. The Water Disc can be either
surfaced or submerged depending on the amount of water
in the ballast tank. Whenever necessary, release of the water
ballast will make the Water Disc positively buoyant, allowing
it to automatically rise to the surface. An option to rotate
the underwater disc 360 degrees is another innovative and
interesting solution considered in the WD concept. Thanks
to the rotation capability, every user of the underwater
accommodation can admire the scenery changing behind
the windows. The rotational movement is not quick enough
to be actually felt by those staying in the Water Disc.
In principle, the float can be operated on its own. When
installed at a location, the ballasted Float is sucked into
sediments or a cavity prepared in a rock (concrete foundation).
However, in such a mode of operation it would require
substantial support during installation and maintenance.
Large barges and heavy cranes are required. It is mainly due
to stability problems during towing and submerging.
In the basic configuration the two-deck underwater Water
Disk has an external diameter of approximately 30m. It can
be utilized for various purposes. In the considered example,
it contains rooms of an underwater hotel, daily visitor space,
lobby, bars and theme exhibition.
Below the lowest deck of the Water Disc, extends a heavy
stiffened structure that forms a suction skirt or caisson. The
c d
suction skirt can be sucked into bottom sediments or placed
in the prepared cavity. It is important to seal the space below
the suction skirt to stop penetration of the pressure field into
the caisson. Positive pressure below the suction skirt can
move the Float up-wards.
At the center of the Water Disc structure there is a
cylindrical shell of the communication tower (shaft). It begins
some 3 m below the sea bottom (inside the caisson structure)
and extends to 30 m above sea bottom (Deck 9) and 20 m
above the water surface.
Structure of the communication tower provides mechanical
connection and vertical separation of the Water Disc and the
ballast tank. It also assures space for transfer of people between
Air Disc and Water Disc by means of two personnel lifts and
two separated and thermally insulated spiral staircases. Pipes
and air ducts and cable trays built into the tower structure
are used to transfer fluids and ventilation air, electric power
and signals.
The ballast tank in form of a hexagonal structure is
assembled on top of the communication tower. Various
technical rooms are arranged inside the tank structure. The
ballast tank structure of the self-operated Float can also be
extended horizontally to form additional accommodation
spaces that can be utilized for different purposes.
THE AIR DISK AND THE SUPPORT
The essential function of the Air Disk is to support of the
operation of the float (Water Disc). In this role, the Air Disc
is a mechanical guide for the float that allows for its safe
vertical movement.
Fig. 2 Example General Plan of the Water Discus as proposed for the Ocean Science Exhibition Centre in Gdynia, Poland.
The Air Disc is suspended on the legs and provides
stabilization of the Float at a selected location. It also houses
majority of technical facilities required for the WD operations.
Personnel rooms as well as additional guest rooms and other
facilities can be arranged within the Air Disc also. Additional
space can be added to the basic structure by application of the
satellites, i.e. smaller discs that can accommodate a variety
of facilities (bar, restaurant, spa), that are connected to the
air disc.
Such a defined functions of the Air Disc can be extended to
meet user requirements and imagination. This remark regards
architectural appearance of the Air Disc also. Its shape can be
arbitrary. In the case of the example application, the Air Disk
consists of a three deck accommodation structure. Counting
from the sea bottom, Deck 6 contains spaces available for
visits, exhibitions, entrances to internal and external lifts,
and visitor rest rooms. Deck 7 is devoted for a restaurant and
bar with 200 seat. Space above Deck 8 contains a galley and
various technical rooms and offices. It is covered with a roof
shaped according to an architectural vision. In its center, the
Air Disc structure large hexagonal cavity is arranged. This
cavity contains the ballast tank of the float in basic operational
condition.
To secure the Air Disc at a required level above the water
surface (6m approximately) the support unit is required.
It is composed of the legs and the foundation. The legs,
penetrating and sliding inside the structure of the Air Disc,
are permanently integrated with the foundation structure.
The foundation, is a circular steel structure, rectangular in the
cross section. It is divided by vertical walls to several caissons
that are opened at the bottom. The foundation caisson begins
POLISH MARITIME RESEARCH, No S1/2017 197
it service to the WD unit from the moment of the launch. It
is assumed, that the WD unit is assembled in a dry dock and
towed out after assembly. To make it possible, the WD unit
needs to be positively buoyant and have limited draft. The
foundation caissons, while they are empty of water, provide
buoyancy and stability of the WD unit while it is under tow to
a given location or shipyard. Otherwise, additional buoyancy
units of comparable size would be required. During short
distance transport of the WD to an exploitation site location,
the legs are in the up-rise position. While on the site, the legs
are lowered using the support ballast system and by filling
the support caissons with sea water. If local condition allows
for, the foundation is than sucked into the bottom sediment.
In a case of location with natural rocky bottom or a concrete
foundation, the basement is sucked into the prepared cavity
of required depth (3 m approximately). The foundation can
be additionally fixed to the bottom or the concrete foundation
to resist extreme weather conditions.
DESIGN ASSUMPTIONS
Safety of peoples involved in operation of the WD are of
primary importance as for every manned system. Physical
and mental comfort are also important but not critical in
the case professional properly trained crew members. The
mental comfort becomes a critically important factor when
commercially manned activity (tourism) is considered. It
was confirmed during experiments with the hyperbaric
underwater habitats [1] that an underwater accommodation
needs to be noiseless and motionless to allow guests or visitors
fall asleep. Such high level of mental safety is an apparent
condition (requirement) to achieve a commercial success
tourism industry. Safety of a technical system requires safe
design, safe procurement, safe exploitation. To achieve high
level of safety, the WD concept is developed and designed with
extensive support of a ship classification society. It will also
be procured according to classification rules and procedures.
Accidental safety issues considered for the WD regard
mainly the following phenomena:
1. Safety during relocation ( when the WD unit is
floating and under tow)
2. General damage due to collision with large ships
3. Local water leaks
4. Catastrophic flooding due to collapse of a large
window
5. Fire and smoke
6. Typhoon (wind waves and current)
7. Tsunami or freak wave
8. Earthquake
Majority of the threats listed above can be assessed using
statistical methods due to availability of the measurement and
observation data (weather). The other ones need to be treated
as accidents that cannot be predicted but must be considered
in design and operation of the WD unit.
198 POLISH MARITIME RESEARCH, No S1/2017
LIMITING ENVIRONMENTAL
CONDITIONS
A specific purpose (application) and a considered location
of the device are other important factors while considering
commercial investment in this branch of the blue economy.
They indicate environmental constraints and define technical
form of an object to be designed and operated safely.
Considering living and working beneath the sea, it is
important to find reasonable ranges of environmental
conditions that make such activities feasible without excessive
costs, threats and discomfort to persons involved. After long
evaluation, it was found that feasible compromise between
various factors indicate the following ranges of environmental
parameters for defined above underwater accommodation:
Water depth range 8 - 30 m
Water temperature range 0 - 36 C
Air temperature range -20 - 40 C
Sea current speed (maximum) 1 m/s
Maximum wave height
(hundred year wave) 6m
Maximum tide amplitude 2m
Wind speed up to 250 km/h
The parameters listed above, together with the compromise
regarding minimum feasible dimensions define basic
configuration of the WD unit. It is possible to operate the
technology outside these limits but it increases costs beyond
commercially acceptable level.
PRINCIPAL PROBLEMS AND AREAS OF
RESEARCH
Ecological issues can be most serious obstacles in practical
application of the WD technology as in any case of investment
in a coastal area. For this reason, the WD technology is
developed with widely understood ecology in mind. Minimum
disturbance to the environment during the installation and
exploitation as well as minimal feasible consumption and
emission during exploitation are primary objectives of the
developed configuration. Capability of compensation of
possible disturbances to the local environment by means
of structural and operational modifications is also studied.
High buoyancy of the submerged structure is a principal
and does not always appreciate technical problem in design
of a large underwater accommodation compartment. It is well
illustrated by the figure valid for the basic configuration of
the WD. It needs to be stressed, that the basic configuration
is considered to be almost the smallest feasible for an
underwater hotel application. Displacement of the submerged
components of the basic configuration (the Water Disc of
30 m DIA) amounts to 5750 m3. The whole Float structure,
optimized for expected mechanical loads, and outfitted with
all required equipment weights 1700 tons (only). It means a
force of some 5000 tons is required to submerge the Water
Disc to the operational depth and stabilize it on the bottom. In
the developed concept of the WD, the gravity force generated
by a ballast weight is utilized. Sea water ballast and a solid
ballast are used for this purpose. Amounts of both types of
the ballast are optimized to allow the ballasting procedure
that is described below.
When the Water Disc is finally submerged, its general
strength and stability on the bottom become an issue.
During calm weather it is mainly long term exposure to
external pressure. Wind, wave and current loads dominate
in adverse weather conditions. There are no established rules
regarding design of a large volume structures located on
bottom in coastal areas. In majority of cases, large marine
structures that are designed to be operated at rough sea are
raised high above the water surface. This well-established
solution, in form of a platform elevated on slim legs, minimize
hydrodynamic loads imposed by waves and currents. The
idea to locate a large accommodation structure in shallow
water is somehow an opposite approach. It means, that the
majority of design, installation, operational rules assuring
such a specific seaworthiness need to be developed in a full
featured R&D process.
Third area of technical problems that were successfully
solved in the development process regards exploitation
without the need for extensive operational assistance. In a
series of design experiment, a configuration allowing self-
installation and refloating was searched for. Finally the device
(WD) composed of the elements characterized above was
developed. It assures operation without the need for the heavy
cranes as a support. This substantially reduces installation
and operational costs and reduces environmental impact of
installation activity .
Other important subjects of research in WD development
process are as follows:
 Psychological issues that require research regarding
influence of considered solution are of mental com-
fort of visitors, crew and personnel.
 Optimization of the system for energy harvesting,
transformation and storage.
 Selection of a materials and structure for underwa-
ter windows to maximize safety and reduce long
term operational costs.
INSTALLATION (BALLASTING) SEQUENCE
OF THE WD UNIT AT THE LOCATION
The following sequence of the ballasting concept of the
WD allows to understand necessities and capabilities of the
WD ballasting concept. It is illustrated on fig 2.
1. The WD unit is floating at the surface using buoyancy
of the Water Disc and foundation caissons connected
to the legs that there are in raised position. In this state,
the WD unit can be towed to a location site by tugs,
relocated or towed to a shipyard for overhaul.
2. While at the location, the legs are lowered to the bottom
and the leg caissons are sucked into the sediments. In
this phase, the Air Disc is slightly immersed also so it
needs to be watertight at the bottom part.
3. The Water Disc and part of the communication tower
are submerged by pumping sea water into the ballast
tank. The communication tower and the Air Disc are
locked together.
4. The Air Disc is elevated to the required level by a buoy-
ancy force as result of releasing of the ballast water. The
Air Disc and the legs are locked together and the float
can be released and submerged using water as ballast
again.
5. If required the sequence can be repeated to elevate the
Air Disc in 3 m steps to any higher level, say up to 12 m
above water if required. Longer legs and higher vertical
shaft are required accordingly.
6. While the Air Disc is blocked with the legs at a required
height above the water surface, the floater (and the Water
Disc) can be submerged or surfaced using the ballast.
At the surfaced position personnel can leave the Water
Disc accommodation using emergency hatches provided.
7. If the ballast tank and the caisson of the Water Disc
are empty, the window frames of the Water Disc are
above the water surface exposed for cleaning, repair
or replacement.
EMERGENCY ESCAPE ROUTES AND
EQUIPMENT
Submersion of the human occupied compartment below
water surface combined with threats indicated above calls
effective escape routes. These are required both to fulfill legal
rules (SOLAS, local building safety laws) and to give safety
awareness (filings) to the unit visitors.
These issues are the principal reason for the arrangement
of the Water Disc and the communication tower. It assures
possibly short escape routes from the submerged spaces in
a case of emergency.
The design offers several alternative solutions available
in various situations. They provide both real safety as well
as psychological comfort for the visitors that need to feel
perfectly safe.
In the case of operational configuration when the Water
Disc is submerged, two thermally insulated personal lifts
are used for vertical transport of peoples between the Air
and the Water Discs. For emergency cases, two independent
staircases are always available inside the Communication
Tower. The staircases are sealed with fire doors, thermally
insulated and ventilated (de-smoked) with separate blowers.
As an alternative the external lifts operating between Deck 2
and Deck 6 can be utilized independently of above mentioned
internal lifts. If the external lift is docked to the Water Disc it
is also possible to use the ladder located inside the suspension
shaft of the external lift as an emergency escape route.
In a case of extensive emergency, the Water Disc will be
surfaced by release of the water ballast and blowing-out water
from the suction caisson. This position allows for escape of
the visitors and the staff using emergency hatches provided in
ceilings of the Water Disc compartments. The hatch covers are
POLISH MARITIME RESEARCH, No S1/2017 199
 Fig. 3 Ballasting sequence of the Water Discus

200 POLISH MARITIME RESEARCH, No S1/2017

operated from both sides and can be opened when the hatch
is above the water surface (when the Water Disc is surfaced).
DESIGN METHODOLOGY
Due to lack of established design rules of the WD,
technology is developed using the first principle method.
Various considered technical solutions and exploitation
procedures are developed using value analyses. The value
analyses are performed using criteria indicated in the sample
table below. The criteria are indicated at the first line of table
1. The criteria given are the weight values ranging from 1 to 5.
The solutions proposed for every matter considered are given
a score between 1 and 5. The results of the weight values and
scores for all the criteria are summed and used for comparison
of these solutions. In the considered example, the solutions
are shown on fig 1. In this case, solution 3 seems to be most
appropriate for the assumed criteria and considered. The results
of such a consideration are listed in the solution matrices for
further study.
The sample given above regards a high level layer with
qualitative evaluation. In fact, to compare all the features,
detailed calculations or sub-analyzes are made, based on
various possibly well-defined configurations.
Fuzzy logic algorithms are investigated as a tools that can
facilitate a multi-criteria optimization of solutions that are
considered within different layers.
Table 1 The sample value analysis matrix as used to study WD unit location
SUMMARY: WD - FRIEND OF THE
COASTAL ENVIRONMENT
After more than five years of the WD concept development,
the eco-oriented advantages of the WD technology can be
summarized as follows:
1. The Water Discus unit is designed to be environmentally
friendly (minimum impact, during installation - no
extensive engineering work required, low discharge,
no spills).
2. The WD proposed for low latitudes is almost self-sup-
plying (electrical energy) due to application of PV panel
that cover the roof.
3. The Water Discus unit is designed to be the ultimate
place of marine and ecological education (direct view of
natural environment, exposition of solar panels and heat
pumps at work), marine ecology oriented exhibitions.
4. The Water Disc is a perfect place for long term biological
studies of the reef/coastal environment.
5. The Water Discus is a perfect place for long term studies
of protection, restoration and creation of artificial reefs
(corals).
6. Due to relocation capability, the Water Discus unit can
be located at a given position and relocated if results of
research performed indicate a better position (for exam-
ple for better view or protection of natural coral reefs).
7. The externally stiffened structure of the Water Discus
is a perfect artificial reef itself.

Considered matter WD-WDR arrangement

Considered Personnel Equipment Through life

Functionality Aesthetics Investment cost Total score
solution safety safety cost

Weight(W)
5 4 4 3 3 5
value (1-5)

Score (S)
5 3 4 4 5 5
Solution 1 - (1 5)
1 WDR to land 102
WD to land
Result (WxS) 25 12 16 9 15 25

Score (S)
5 4 3 3 5 5
Solution 2 - (1 5)
2 WD to land 102
WDR to land
Result (WxS) 25 16 12 9 15 25

Score (S)
Solution 3 - 5 5 5 5 5 5
(1 5)
3 WDR to land 136
WD to WDR
Result (WxS) 25 20 20 16 15 35

Score (S)
Solution 4 - 5 4 4 5 5 5
(1 5)
4 WD to land 127
WDR to WD
Result (WxS) 25 16 16 15 15 25

WD - Water Discus, WDR - Water Discus with reduced Air Disc

POLISH MARITIME RESEARCH, No S1/2017 201

8. Such complicated structures are appreciated by several
fish spices. With plenty of cavities, it is a much better
marine habitat than smooth rock or empty sea bottom.
9. The Water Discus being a substantial obstacle to waves
can shelter natural reefs and other features if properly
located.
10. The Water Discus is considered to be a typhoon shelter
as it is glazed with double safety glass windows.
LITERATURE
obsugi sportw wodnych (Device for vertical transport
and water sports), Patent appl. P-411494 dated 05.03.2015.
11. Rowiski L.: Dach ruchomy przestrzeni mieszkalnej lub
technicznej (articulated roof for human occupied and
technical spaces, Patent appl. P-411699 dated 23.03.2015.
12. Rowiski L.: Statek do obserwacji podwodnych i
nadwodnych (A ship for underwater/ above water
observation , industrial design Wp-23204, 05.03.2015,
valid to 05.03.2020

1. Miller J. W., Koblick Ian G. Working and living at sea,
Best Publishing Company, 1995
2. Project report: Przeprowadzenie bada oraz stworzenie
konstrukcji obiektu podwodno - nadwodnego dla celw
naukowych (Research and development of a structure of
the underwater-above water device for scientific purpose),
Project (IX.B.16)-Operation Program: Innovative Economy
2007-2013, PO IG 1.4.
13. Rowiski L., Zagrski M.: Urzdzenie do transportu
pionowego osb midzy pomieszczeniem nadwodnym
i pomieszczeniem podwodnym (Device for vertical
transport of persons between underwater compartment
and above water compartment), Patent appl. P-417805,
dated 01.07.2016

3. Rowiski L., Zdrojewski J.: Pomieszczenie podwodno-

nadwodne, zwaszcza do celw mieszkalnych
(Underwater-above water accommodation for residential
purposes), Patent appl. P-392432 dated 17.09.2010;
CONTACT WITH THE AUTHOR
4. Rowiski L., Zdrojewski J.: Underwater-above water
accommodation for residential purposes, United States Lech Rowiski
Patent 9,010,266, dated 21.04.2015;
Gdask University of Technology
5. Rowiski L., Zdrojewski J.:Urzdzenie do nurkowania 11/12 Narutowicza St.
zwaszcza do wd pytkich (Device for diving, especially 80 - 233 Gdask
for shallow water), Patent PL222384 dated 13.09.2012, Poland

6. Rowiski L., Zdrojewski J.: Tytu: Pomieszczenie

podwodno-nadwodne, zwaszcza do celw mieszkalnych
(Underwater-above water accommodation for residential
purposes), Patent 224208 dated 17.04.2014, Patent appl.
P-40076 dated 25.11.2013;

7. Rowiski L.: Tytu: Pomieszczenie podwodno-nadwodne,

zwaszcza do celw mieszkalnych (Underwater-above
water accommodation for residential purposes), Patent
appl. P-406227 dated 25.11.2013;

8. Rowiski L., Porala B., Waamaniuk R. : Urzdzenie do

transportu osb pod powierzchni wody (Device for
transport of persons below water surface) Patent appl.
P-406871 dated 17.01.2014.

9. Rowiski L.: Izobaryczne pomieszczenie podwodne,

mieszkalne lub techniczne (Isobaric underwater
accomodation for human occupancy or technical
equipment) Patent appl. P-411357 dated 23.02.2015.

10. Rowiski L.: Urzdzenie do transportu pionowego i

202 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 203-212
10.1515/pomr-2017-0040

TECHNICAL STATE ASSESSMENT OF CHARGE EXCHANGE SYSTEM

OF SELF-IGNITION ENGINE, BASED ON THE EXHAUST GAS
COMPOSITION TESTING

Jacek Rudnicki
Ryszard Zadrg
Gdansk University of Technology, Poland

ABSTRACT

This paper presents possible use of results of exhaust gas composition testing of self - ignition engine for technical
state assessment of its charge exchange system under assumption that there is strong correlation between considered
structure parameters and output signals in the form of concentration of toxic compounds (ZT) as well as unambiguous
character of their changes. Concentration of the analyzed ZT may be hence considered to be symptoms of engine
technical state. At given values of the signals and their estimates it is also possible to determine values of residues which
may indicate a type of failure. Available tool programs aimed at analysis of experimental data commonly make use
of multiple regression model which allows to investigate effects and interaction between model input quantities and
one output variable. Application of multi-equation models provides great freedom during analysis of measurement
data as it makes it possible to simultaneously analyze effects and interaction of many output variables. It may be also
implemented as a tool for preparation of experimental material for other advanced diagnostic tools such as neural
networks which , in contrast to multi-equation models, make it possible to recognize a state at multistate classification
and - in consequence to do diagnostic inference. Here , these authors present merits of application of the above
mentioned analytical tools on the example of tests conducted on an experimental engine test stand.

Keywords: diagnostic model , self-ignition engine, exhaust gas components, artificial neural networks

INTRODUCTION Correct run of working cycle (especially run of combustion

During execution of engine running process, parameters
of its structure undergo changes. It is not without any impact
on its performance described by a set of output parameters.
Mutual relation between engine structure parameters
and output parameters makes it possible, in determined
conditions , to take output parameters as engine technical
state symptoms measured without its disassembling , because
physical- chemical processes taking place during running
process and quantities describing them can be generally
observed and measured from outside. Amount of emission
of exhaust gas components belongs a. o. to the quantities in
question [2, 5, 6, 8].
in engine cylinder) depends to a large extent on correctly
working charge exchange system which has to ensure first
of all effective discharging the combustion products and
filling the working space with an required amount of air of
appropriate quality.
Technical state identification of elements of charge
exchange system is practically impossible in view of existing
standard measurement equipment of engine. Known
diagnostic methods are first of all based on the assessing of
data concerning flow of working medium volume or mass
and its losses in engine cylinders [11].
The following elements were taken to represent charge
exchange system of typical ship engine (Fig. 1):

POLISH MARITIME RESEARCH, No S1/2017 203

 - air filter; INITIAL ASSUMPTIONS
- air compressor;
- air cooler; On the basis of analysis of the set of quantities
- engine cylinders together with elements of air inlets characterizing the charge exchange system an initial version
and exhaust gas outlets; of diagnostic table ( Tab. 1) and the systems topological model
- gas turbine. ( Fig. 2) was prepared [13].
In view of occurrence of many cases of intuitive
determination of changes as well as due to rather not strong
relations of the exhaust gas toxicity indices shown in Tab.1
with the engine running parameters and these with structure
parameters ( failures), it is necessary to verify data of Tab. 1
empirically and determine the existing relations by simulation
investigations.

Fig. 1 Schematic functional diagram of engine charge exchange system. 1 air

filter, 2 compressor, 3 electric drive motor of air compressor, 4 air cooler,
5 engine cylinders together with air inlet and exhaust gas outlet elements

Tab. 1 Initial diagnostic table for charge exchange system of self-ignition engine

Diagnostic parameter Failure location and type

DLUOWHU air compressor air cooler engine cylinder

contamination of

contamination of

contamination of

untightness of

untightness of
RZSDUW

valves
cooler
wear,

TPC
OWHU

3UHVVXUHGURSRQDLUOWHU SF + 0 0 0 0
Compressor compression ratio s 0 - 0 0 0
&RPSUHVVRUDGLDEDWLFHIFLHQF\ ad 0 - 0 0 0
Supercharging air pressure drop Sba 0 0 + 0 0
Air temperature drop in cooler 7ba 0 0 - 0 0
&RROHUHIFLHQF\FRROLQJHIIHFWLYHQHVV ch 0 0 - 0 0
$LUH[FHVVFRHIFLHQW - - - - -
6SHFLFIXHOFRQVXPSWLRQ b + + + + +
Supercharging air pressure Sba - - - - -
Compression pressure in engine cylinder Sc - - - - -
Maximum combustion pressure Smax - - - - -
Exhaust gas temperature 7g1 - - - - -
Turbo-compressor rotational speed nt - - - - -

Concentration of nitrogen in collector CNOx(k) - - - - -

oxides in casing CNOx(s) - - - - 0

Concentration of carbon in collector CCO(k) + + + + +

oxide in casing CCO(s) 0 0 0 + 0

Concentration of in collector CHC(k) + + + + +

hydrocarbons in casing CHC(s) 0 0 0 + 0
Exhaust gas smokiness. in collector D + + + + +
Notation:
- decrease in value of a parameter; + - increase in value of a parameter; 0 without any change.
Comment: The diagnostic parameters pF, s, ad, pba, Tba, ch, b are complex quantities determined on the basis of results of direct measurements.

204 POLISH MARITIME RESEARCH, No S1/2017


Fig. 2 Topological model of engine charge exchange system
1 air filter 2 compressor, 3 air cooler, 4 engine combustion chamber, 5 fresh water
cooler, 6 injection apparatus, 7 engine crankshaft casing, 8 exhaust gas collector
SET OF EMPIRICAL METHODS FOR
SIMULATION INVESTIGATIONS ON
CHARGE EXCHANGE SYSTEM
Simulation investigations are mainly aimed at determining
whether it is possible to identify state of charge exchange
system on the basis of assessment of changes in exhaust gas
toxicity indices, including the following [12]:
verification of the parameters given in Tab. 1;
verification of changes in values of diagnostic
parameters, depending on location and type of
unserviceability of elements of charge exchange
system;
determination ( detection) of connections (relations)
of structure parameters of elements of charge exchange
system with engine working parameters, and these
with diagnostic parameters exhaust gas toxicity
indices;
possible selection , out of exhaust gas toxicity indices
, crucial diagnostic parameters , and , out of engine
running indices, auxiliary ( verifying) diagnostic
parameters of state of elements of engine charge
exchange system.
The unserviceability types selected as a result of the
performed analysis of elements of charge exchange system
were simulated in the following way:
contamination of air filter by throttling partly
(shadowing) its active surface pf;
POLISH MARITIME RESEARCH, No S1/2017
wear or contamination of compressor flow
part - by throttling partly air at inlet ps;
untightness of piston rings cylinder
unit by grinding partly sealing surface of
rings as well as enlarging the piston-ring joints
to get assumed values of the cross-section area of
untightness Su;
untightness of inlet and outlet valves by
grinding partly fying face to get assumed values
of the cross-section area of untightness Sz.
The approximate wear values given in
Tab. 2 served as a basis for preparation of the
specified unserviceability events to be used in
basic simulation investigations. The way of their
preparation consisted in the following :
removal (e.g. by grinding or machining)
given volumes of material out of surfaces of
cooperating elements of combustion chamber space
, which correspond to preliminarily calculated
wear values or result from experimental data (
verifying data from repair workshop),
execution of preliminary tests in respect to
unserviceability events which decide on quality of
filling. The percentage changes in the air flow drag
ps (0;33,3 %) result from the air volume fluxes at
clean (V = 0,12 m3/s) and contaminated (V=0,08
m3/s) flow part of air compressor. And, the changes
in the air flow drag pf (0;50 %) result from the
difference between the air filter active surface
of clean flow (4071,5 mm2) and that contaminated
(2035,75 mm2).
In view of a vast range of necessary experimental
measurements , resulting from the aim of simulation
investigations , it was assumed , like in case of fuelling system
testing, to carry out laboratory tests aided by experiment
planning theory [4, 7] as well as the computer software
STATISTICA PL.
1. The set of input data X:
x1 - engine rotational speed n [rpm];
x2 engine torque Ttq [ Nm];
x3 air filter contamination pF [%];
x4 contamination of compressor flow part ps [%];
x5 untightness of TPC unit Su [mm2];
x6 untightness of valves S z [mm2].
2. The set of output data Z :
z1 - engine power Pi [kW];
z2 air excess ratio ;
z3 hourly fuel consumption B [g/h]
z 4 - maximum pressure in injection pipe pwtr(max);
z5 maximum compression pressure pc(max) [MPa];
z6 - cylinder exhaust gas temperature Tg1 [K];
z7 mean indicated pressure pmi [MPa];
z8 compression pressure in the instant of fuel injection
pc [MPa];
z9 maximum combustion pressure pmax [MPa];
z10 carbon oxide concentration within exhaust gas outlet
collector CCO(k) [ppm];
205
 Tab. 2. Input data in the plan of experimental test on charge exchange system
No. Input quantity Notation
1
Rotational speed n rpm
2
Torque Ttq
3 Untightness of rings of TPC
Su
unit
4 Untightness of inlet and
Sz
outlet valves.
5 Flow drag increase within
pF
filter
6 Flow drag increase within
ps
flow part of compressor
z11 carbon oxide concentration within crankshaft casing
CCO(s) [ppm];
z12 concentration of hydrocarbons within exhaust gas
outlet collector CHC(k) [ppm];
z13 concentration of hydrocarbons within crankshaft
casing CHC(s) [ppm];
z14 nitrogen oxides concentration within exhaust gas outlet
collector CNOx(k) [ppm];
z15 nitrogen oxides concentration within crankshaft casing
CNOx(s) [ppm];
z16 oxygen concentration in exhaust gas CO2 [%];
z17 pressure increase rate in cylinder (p/)s;
z18 indicated specific fuel consumption bi;
z19 engine indicated efficiency i.
3. The set of the constants, C, which may occur during
measurements and have influence on changes in output
quantities Z (e.g. resulting from a used kind of fuel
oil , lubricating oil ) . As assumed , the experimental
investigations will be short-lasting and carried out
in similar conditions , i.e. they will be not taken into
account in the experiment plan and it should deal only
with one measurement block.
4. The set of the disturbing quantities, H, which may
undergo changes during execution of the experiment:
h1 ambient temperature ta [oC];
h2 - atmospheric pressure pa [hPa];
h3 relative humidity of air in the laboratory M [%].
STATISTICAL ANALYSIS OF RESULTS OF
THE EXPERIMENT
In view of the assumed linear model of influence of charge
exchange system structure parameters on engine working
indices , especially concerning exhaust gas components, a
simplified statistical analysis of results of the experiment
206 POLISH MARITIME RESEARCH, No S1/2017
Values of input quantities
Dimension Remarks
min (-1) max (+1)
850 1100
Nm 0 77
grinding sealing surface and
mm2 0,57 1,178
joint
Groove in fying face of inlet
mm2 0 0,75
and outlet valve
% 0 50 Flow throttling
% 0 33 Flow throttling
was performed. As a result of the analysis the following was
obtained:
approximating polynomials (functions of the
experiments object) describing functional model of
charge exchange system;
assessment of main effects and interactions of the
determined ( assumed) model;
correlation of input and output quantities describing
the experiments object;
assessment of adequacy of function of the experiments
object.
Based on an analysis of the models possible to be used and
the data given in Tab. 3, was assumed a model which takes
into account two-factor interactions, despite more favourable
values of the basic coefficients (R 2, MS) for estimation of
representation (approximation) of relations between input
and output quantities of the experiments object , with taking
into account interactions of a higher order, e.g. three-factor
ones [3, 14].
As results from the data of Tab. 3 , the model with
three-factor interactions covers entirely all the relations
( variations) between input and output quantities
(the coefficient of determination R 2 = 1), which simultaneously
means that no residues are rejected and they are not taken
into account in the model (MS = 0). However ,as results
from mathematical description of any empirical relation ,
its accuracy increases along with number of interactions
which contribute simultaneously to increasing confounding
of approximated relations. Hence, confounding degree
influences clarity of subject-matter analysis of describing
problem ( phenomenon ) and , consequently, may lead to
incorrect conclusions and decisions. In such cases , as a result
of the confounding ( especially multi-factor one), obtained
calculation results are often incorrect and inconsistent with
both theoretical physical principles and empirical reality of
a phenomenon under investigation. Therefore , taking into
account minor differences between values of the analyzed
statistical measures , one assumed the model of two-factor
interactions for subject-matter analysis. For the models , were
determined approximating polynomials for output quantities
 Tab. 3 Set of values of selected statistical measures for empirical functional models of charge exchange system

Model with three-factor

Output Model without interactions Model with two-factor interactions
interactions
quantities R2 MS R2 MS R2 MS
B 0,009662 0,009663 0,99766 0,00114 1 0
A/F 0,93843 18,2333 0,97808 29,2124 1 0
0,93752 0,08526 0,97774 0,13666 1 0
CCO(k) 0,87006 97291 0,99916 2816,31 1 0
CCO(s) 0,56833 16,583 0,87708 21,2500 1 0
CHC(k) 0,81543 1501,49 0,99677 210,125 1 0
CHC(s) 0,70837 1126,9 0,99998 0,5000 1 0
CNOx(k) 0,88088 1102,3 0,95341 19,400 1 0
CCO2(k) 0,97955 0,13079 0,99512 0,14051 1 0
CO2 0,97314 0,38387 0,99593 0,26153 1 0
tg1 0,95357 757,25 0,99786 157,250 1 0
pc 0,7297 0,0244139 0,99775 0,0009125 1 0
pc(max) 0,84452 0,0292583 0,99272 0,0061625 1 0
pmi 0,88763 4562,319 0,98978 3321,125 1 0
pmax 0,85422 0,0258563 0,99811 0,0015062 1 0
bi 0,50217 690,7553 0,98854 127,2012 1 0

describing state of investigated engine with unserviceable
charge exchange system as well as statistical correlations
between input and output variables of the experiments
object. The determined approximating polynomials make
it possible to determine any relation between particular
variables and also to calculate and estimate influence of
introduced (simulated) failures (wear) of elements of charge
exchange system on engine running and toxicity indices.
It may be also assumed that it would be possible to determine
mutual relations between structure parameters and exhaust
gas toxicity indices directly or indirectly through engine
running indices. As assumed, it will be this way possible to
reveal, out of exhaust gas components, diagnostic parameters
for given engine fuelling elements or units . Correctness of
the assumption is confirmed by the existing relationships (
correlations) between wear of elements of charge exchange
system and concentration of toxic components in engine
exhaust gas.
CONCEPT OF A NEURAL DIAGNOSTIC
SYSTEM
Since the form of approximating polynomials for output
quantities describing state of investigated engine with
unserviceable charge exchange system is complex , an
applicable alternative may be a neural approximating model
which, based on experimental results as well as determined
coefficients of approximating polynomials , may be applied to
the modelling of arbitrary nonlinearity and which is highly
resistant against disturbances [1, 9, 10].
A condition for reaching correct responses from used
neural networks is appropriately numerous set of learning
data. To work out such a set can be made possible by
using the above described approximating polynomials.
For the purposes of the simulation investigations whose
results are presented in the further part of this paper, a general
schematic diagram of neural failure detection system was
worked out under the following assumptions :
the system identifies the following classes of engine
technical states:
- the class of states S0 technically serviceable engine,
- the class of states S1 untightness of sealing rings,
- the class of states S2 untightness of inlet and/or
outlet valves,
- the class of states S3 increase of flow drag within
air filter,
- the class of states S4 increase of drag within flow
part of air compressor,
for each of the class of states of technically
unserviceable engine will be worked out a separate
neural sub-network (4 sub-networks in total) which
generates , at its outlet , the value True (1) in case
of identification of a dedicated class of states or the
value False (0) in the opposite case. The schematic
diagram of separate sub-network is shown in Fig. 3
the engine torque Ttq and engine rotational speed
n were assumed to be independent variables ( input
parameters),
parameters under diagnostic surveillance:
- exhaust gas temperature tg1,
- concentration of nitrogen oxides within exhaust gas
collector , CNOx(k),
- concentration of carbon oxide within exhaust gas
collector ,CCO(k),
- concentration of hydrocarbons within exhaust gas
collector , CHC(k),
- concentration of oxygen in exhaust gas , CO2,
for each of the parameters a neural model will be
worked out and all the models prepared this way

POLISH MARITIME RESEARCH, No S1/2017 207

 Fig. 3 Schematic diagram of the neural subsystem for detection of the state Si, (i = 0, 1, 4)

will form the so-called bank of neural observers
[1, 8], intended for the modelling of values of the
considered parameters in normal ( free of failures )
engine running state,
by comparing output signals from the model and
diagnosed engine, residues, i.e. signals representing
discrepancies between model and engine, will be
determined,
the obtained vector of residues r = [r1 ... r5] will be
analyzed by a neural classifier of residues whose task
will be to determine if a failure occurred in such
case the value True (1) will be generated at its output,
otherwise the system will generate the value False
(0),
the vector a = [w1 ... w4] (wi = 0 1) achieved as a
result of action of the entire system (4 sub-networks),
will constitute a diagnostic information source (e.g.
in case of obtaining the vector [0 0 0 0] there exist
premises to consider the current state of engine as
that belonging to the class of states S0 etc)
The results presented in the further part of the paper was
limited to only one sub-network in this case - to the network
which identifies occurrence of the class of states S1
In the case in question the form of the polynomials is as
follows :
(1)
where :
[y i] column vector of parameters under diagnostic
surveillance (i = 1, 2, 5) :
y1 exhaust gas temperature tg1,
y2 concentration of nitrogen oxides in exhaust gas- CNOx,
y3 - concentration of carbon oxide in exhaust gas - CCO,
y4 - concentration of hydrocarbons in exhaust gas - CHC,
y5 - s concentration of oxygen in exhaust gas - CO2.
[Bi] column vector of constants appearing in
approximating polynomials:
[x k] row vector of input variables:

RUN AND RESULTS OF SIMULATION

INVESTIGATIONS where:
Ijk regression coefficients concerning interactions y j yk
The results obtained in the form of approximating
polynomials of output quantities describing state of considered
engine with unserviceable charge exchange system, were
used for working out the set of learning data for particular
sub-networks.

208 POLISH MARITIME RESEARCH, No S1/2017

 [Aj,k] matrix of regression coefficients of approximating
polynomials
In connection with that five parameters subjected to real
time control were selected, it become necessary , in line with
the assumed concept , to form the same number of neural
models representing relationship between the input variables
: Ttq and n , and variables subjected to diagnostic surveillance
in the state considered normal , i.e. the state s0.
The first stage of the work consisted in running preliminary
investigations aimed at selection of type and optimum
structure of neural networks of particular models. To this
end, were applied automatic tools of the software STATISTICA
Neural Networks v. 7.0 aiding in the testing of neural networks
used for data analysis and prediction issues [15].
The aim of training the networks was to reach such state of
them , which determines correct reaction in the broad range
of input functions which , in this case, are various values of
engine load (in the range of 077 Nm) and rotational speed
(850 1100 rpm), introduced to inputs. The worked out
by using approximating polynomials - learning set covered
19578 cases in relation to each of the five parameters. The
example realizations of changes in values (cases : CNOX and
tg1) in function of engine torque and rotational speed are
presented in Fig. 4 and 5.
Fig. 4. The functional relationship between CNOX and (Ttq, n)
Fig. 5 The functional relationship between tg1 and (Ttq, n)
As a result of the performed simulations and analysis of
their results, the network of the type : multilayer perceptron
with one hidden layer, was selected.
The stage of preliminary investigations made it possible
to perform the basic training of networks for each case of
neural observer modelling changes in selected variables. The
training and final architecture of the networks was realized
by means of the software MATLAB 2015b and its dedicated
extension Neural Network Toolbox [16].
Application of the basic quality measure for
the worked - out neural models , i.e. values and
distribution of residues and percentage errors between
values expected at output from networks and its real
response, showed adequate quality of representation
and negligible , from practical point of view, differences.
The next stage was aimed at the working out of structure and
training of residues generator. For each of the considered class
of states, the task of the residues generator is to determine
values of differences between monitored output signals
of the diagnosed engine, y i = f(Ttq, n), and values of
responses, corresponding with them, from the side of
the worked out neural observers bank , y is = f(Ttq, n).
For the presented case of the class of states S1:
exhaust gas temperature tg1 - y1,
concentration of nitrogen oxides in exhaust gas, CNOx -
y2,
concentration of carbon oxide in exhaust gas , CCO - y3,
concentration of hydrocarbons in exhaust gas , y4,
concentration of oxygen in exhaust gas, CO2 - y5,
neural model response, tg1s - y1s,
neural model response CNOxs- y2s,
neural model response CCOs - y3s,
neural model response CHCs - y4s,
neural model response CO2s y5s,
The vector of residues r = [r1, r2, r3, r4, r5] , obtained this way
, may be considered a signal which contains information on
occurred failures ( in this case untightness of sealing rings).
During the engine running in the state taken as normal (the
class of states S0), components of the obtained vector of residues
POLISH MARITIME RESEARCH, No S1/2017 209
should be close to zero , but in the case of failure occurrence
the difference becomes significantly greater.
Fig. 6 The functional relationship between tg1 and (Ttq, n) class of states S1
(Su <0,57, 1,178> [mm2])
The learning sets worked out with the use of approximating
polynomials, contained about 20000 cases with reference
to each of the five parameters. For the presented case of the
class of states S1 , the set of model responses y is=f(Ttq,n)
was worked out for random changes in the values Su within
the range of 0,7 1,178 mm2, n values within the range of
850 1100 rpm, Ttq - values within the range of 0 77 Nm.
Fig. 6 exemplifies the relationships y1s=f(Ttq,n) for the class
of states S1 .
Fig. 7 Values of the residues r2 (NOx concentration ) the class of states S1.
210 POLISH MARITIME RESEARCH, No S1/2017
Based on experience dealing with optimum network
structure and type , it was decided to select a linear neural
network which models relations between its ten inputs (y1, y2,
y3, y4, y5, y1S, y2S, y3S, y4S, y5S) and five outputs (r1, r2, r3, r4, r5).
The results achieved in the case of the class of states S1 , are
shown for the parameter CNOx in Fig. 7.
Like in the case of the neural observers bank , the test of
the network and application of its quality measure in the form
of values and distribution of modules of the residues between
values expected at networks output and its real response,
showed very good quality of representation and negligible ,
from practical point of view , differences. Fig. 8 presents an
example result for the class of states S1 .
Fig. 8 Distribution of modules of absolute errors of residues generator the
class of states S1
The task of the last element of failure detection and
location system , i.e. the state classifier (block of residues
evaluation), is to analyze residues vector in order to determine
whether a selected failure has occurred or not. This is a typical
classification issue consisting in the matching of vectors
symptoms to one of the distinguished classes of states.
Based on the 1st stage of investigations, to solve the
presented issue , the network of the multilayer perceptron type
with one hidden layer , was designed and trained, see Fig. 9.
Fig. 9 Topological diagram of neural classifier of states
 Tab. 4 Results of testing the neural system for detection and location of errors
The classifier , sending to networks input, the vector r=[r1,
r2, r3, r4, r5] , triggers activation of one neuron of output
layer the value True (1) in the case of identification of
the dedicated class of states , or the value False (0) in the Correctly classified Erroneously
opposite case. Subset of testing state classified state
set (number of cases (number of cases
The learning sets intended for training the classifiers were (%) (%)
worked out on the basis of the assumption that a change ,
by less than 10% , in values of particular classifiers , which State s1 99,8 0,2
results in transition of engine to a given class of states Si
(Tab.2, row 3 9), triggers the response False at classifiers State s2 98,4 1,6
output . Otherwise, the response True is obtained at neural
State s3 99,3 0,7
networks output.
The training, validation and test of the classifier, based on
State s4 98,9 1,1
the learning set , showed its very good matching and negligible
number of erroneously classified cases (below 1% during the
State s5 99,5 0,5
testing phase).
Mean 99,18 0,82

TEST RESULTS OF NEURAL MODELS

The working-out of particular networks and positive SUMMARY

completion of their training made it possible to test the
presented system with the use of the testing set of simulated,
determined cases of values of the exhaust gas temperature
tg1 and the concentration of nitrogen oxides in exhaust
gas CNOx , concentration of carbon oxide in exhaust gas,
CCO, concentration of hydrocarbons in exhaust gas , CHC ,
and concentration of oxygen in exhaust gas , CO2.
The set of test cases containing 5000 sets of values (1000
cases for each of the class of states) were worked out on the
basis of the prior obtained approximating polynomials which
were randomly changed in each of the set by using a generator
of pseudo-random numbers. The change concerned values of
all the parameters appearing there, within the range of 5%;
The presented procedure had to show first of all
susceptibility of the system on disturbances , and
consequently, its potentially lower or greater usefulness
for implementation in real ship power plant conditions.
Fig. 10 shows an error matrix for the case of the class of states
S1 , and Tab. 4 summarizes generally results of the tests.
The example investigation results, i.e. the values
obtained during active experiment , indicate that the
proposed on-line system for detection and location of
failures correctly identifies a given class of engine states.
From practical point of view its quality may be called excellent.
However it may also seem that specificity of use of especially
responsible power systems such as ship power plant and its
functional subsystems require higher percentage values of
correct classification than those shown in Tab. 4 .
Another issue is empirical verification of the presented
system. In the considered cases the set of data was relatively
homogenous because it was obtained by means of algebraic
relationships.
In case of investigations carried out directly on a real
object, obtained results may appear quite different , that
obviously requires to carry out further investigations whose
main condition is to realize in practice , as to its program
solutions and instrumentation, the system discussed in this
paper.

BIBLIOGRAPHY

1. Korbicz J., Kocielny J.M., Kowalczuk Z., Cholewa W.:

Diagnostics of processes. Models. Artificial intelligence
methods. Applications. (in Polish), Warsaw, WNT 2002.

2. Kropiwnicki J., Kneba Z.: Carbon dioxide potential

reduction using Start-Stop system in a car. Key Engineering
Materials, Vol. 597 (2014), s. 185-192.

3. Kufel T.: Econometrics. Solving Problems Using GRETL

Software, in Polish, Polish Scientific Publishers PWN,
Warszawa. 2007.
Fig. 10 State classifiers error matrix the class of states S1

POLISH MARITIME RESEARCH, No S1/2017 211

4. Kukieka L.: Basics of Engineering Research, in Polish,
Polish Scientific Publishers PWN, Warszawa 2002.

5. Markowski J., Pielecha J., Jasiski R., Kniaziewicz T.,

Wirkowski P.:Development of alternative ship propulsion CONTACT WITH THE AUTHOR
in terms of exhaust emissions. 1st International Conference
on the Sustainable Energy and Environment Development Jacek Rudnicki
(SEED 2016), E3S Web of Conferences 10, 00140 (2016)
Ryszard Zadrg
6. Piaseczny L. Zadrg R.: The influence of selected damages
of engine SI type on the changes of emission of exhaust Gdask University of Technology
gas components, Diesel Engines, Opole 2009. Faculty of Ocean Engineering and Ship Technology
11/12 Narutowicza St.
7. Polaski Z.: Design of Experiments in Technology, Polish 80 - 233 Gdask
Scientific Publishers PWN, Warszawa 1984. Poland

8. Rudnicki J., Zadrg R.: Problems of modelling toxic

compounds emitted by a marine internal combustion
engine for the evaluation of its structure parameters.
Combustion Engines No.3/2015(162), ISSN 2300-9896,
Pozna 2015.

9. Skoundrianos E.N., Tzafestas S.G.: Fault diagnosis via

local neural networks. Mathematics and Computers in
Simulation 60 (2002) 169-180. Elsevier Science 2002.

10. Tadeusiewicz R.: Neural networks (in Polish). Warszawa,

Akademicka Oficyna Wydawnicza RM 1993.

11. Zadrg R.: Criteria for the selection of the diagnostic

parameter for diagnosis of marine diesel engine,
LOGISTYKA No. 4/2010, ISSN 1231-5478, Pozna 2010.

12. Zadrg R.: The Multi-equational models of leakproofness

of charge exchange system of ship engine, (in Polish), in
monograph Gaseous engines selected issues edited by
Adam Duyski, University of Czestochowa Publishing,
ISBN 978-83-7193-461-2, ISSN 0860-501., Czstochowa
2010.

13. Zadrg R. et al.: Identification models for the technical

condition of the engine on the basis of exhaust component
emissions, in Polish, The report of the research project no.
4T12D 055 29, AMW, Gdynia 2008.

14. Zadrg R., Zellma M.: The usage of multi-equation models

in analysis of dynamic process in marine diesel engine
research. JOURNAL OF POLISH CIMAC, Vol.7, No 1,
ISSN 1231-3998, str. 295-304, Gdask 2012.

15. STATISTICA Neural NetworksTM. Prze-wodnik

problemowy. StatSoft, Krakw 2001.

16. Neural Network Toolbox. Matlab. Users guide. The

MathWorks, Inc. 2014.

212 POLISH MARITIME RESEARCH, No S1/2017

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 213-223
10.1515/pomr-2017-0041

THE INFLUENCE OF WATER AND MINERAL OIL ON VOLUMETRIC

LOSSES IN A HYDRAULIC MOTOR

Pawel liwiski
Gdansk University of Technology, Faculty of Mechanical Engineering, Poland

ABSTRACT

In this paper volumetric losses in hydraulic motor supplied with water and mineral oil (two liquids having significantly
different viscosity and lubricating properties) are described and compared. The experimental tests were conducted using
an innovative hydraulic satellite motor, that is dedicated to work with different liquids, including water. The sources
of leaks in this motor are also characterized and described. On this basis, a mathematical model of volumetric losses
and model of effective rotational speed have been developed and presented. The results of calculation of volumetric
losses according to the model are compared with the results of experiment. It was found that the difference is not more
than 20%. Furthermore, it has been demonstrated that this model well describes in both the volumetric losses in the
motor supplied with water and oil. Experimental studies have shown that the volumetric losses in the motor supplied
with water are even three times greater than the volumetric losses in the motor supplied with oil. It has been shown,
that in a small constant stream of water the speed of the motor is reduced even by half in comparison of speed of motor
supplied with the same stream of oil.

Keywords: volumetric losses, satellite motor, water, oil

comparison to mineral oil water has a very low viscosity

INTRODUCTION
Hydraulic motor in the hydraulic system is the executive
element. Its purpose is to convert hydraulic energy into
mechanical one. An energy carrier in the hydraulic systems
is liquid. The type of liquid is conditioned by the requirements
for these systems. In most hydraulic systems mineral oil is
commonly used [3,4,5,6,24]. However, in some industrial
sectors, the requirement is a non-flammable liquid (mining,
steel mills, etc.) or non-toxicity for the environment and
human health (food industry) [13]. The liquid which is non-
flammable and non-toxic and certainly suitable for energy
transfer in hydraulic systems is water.
There is a growing trend in the world towards research and
development of components and hydraulic systems supplied
with water [3,7,13,23].
Such studies are also of particular importance in marine
technology, where hydraulic power circuits are used a lot,
and the water is a generally available working liquid [2]. In
and low lubricating properties [8]. These features adversely
influence the efficiency of energy conversion in hydraulic
systems [21,22]. Despite this, attempts are made to develop
innovative components and hydraulic systems supplied with
water [13,14].
So far, each hydraulic device is dedicated to a specific type
of working liquid. For example, a hydraulic motor dedicated
for oil systems should not be used in systems where the
working medium is water. In other case this motor has low
efficiency and very little durability [22].
Both the pump and the hydraulic motor are components
of the hydraulic system, in which there are large energy losses
[1,9,10,11,21,22,24]. The biggest influence on the volumetric
losses in hydraulic displacement machines has a leakage in the
clearances of working mechanism [1,9,11,12,14,17,19,20,21,22,
24]. It has been proved that in a motor working mechanism
clearances that a not-fully developed turbulent flow takes
place. This is particularly well visible in the case of supplying
a hydraulic motor with low viscosity liquid [14,16,20]. Besides,

POLISH MARITIME RESEARCH, No S1/2017 213

in the hydraulic motor, there are other sources of volumetric
losses, in addition to the clearances. The sources of these
losses are described later in this article.
Considering water as a working fluid one should expect
particularly high volumetric losses in machines, greater than
the ones with mineral oil. So far there is a lack of results of
research comparing the influence of water and mineral oil on
volumetric losses in pumps and hydraulic motors. There is
also no specific information about the design of pumps and Fig. 1. General view and axial cross section of satellite motor (SM type):
motors, which can be supplied with both water and mineral C stator, S satellite, R rotor,1 shaft, 2 casing, 3 front casing,
oil. There is rich literature on flows in different flat gaps, 4 rear casing, 5 and 6 inlet and exit manifold, 7 and 8 compensation
(commutation) plates [14,16,17]
orifices and other simple openings. Furthermore, there is no
rich studies in literature that compare the influence of type
of liquid on flow characteristics in these gaps. In literature The toothed unit, shown in Fig. 2, is the satellite working
the volumetric losses have been describe in general without mechanism of the motor. It consists of a toothed rotor R
specifying the type of clearances [1,9,11,20,21]. In addition, (4humps), toothed stator C (6 humps) and ten S wheels
the leakage of low viscosity liquid is described by a different (satellite).
model than the leakage of high viscosity liquids (mineral oil,
rapeseed oil) [20,21].
Therefore, from the scientific and cognitive point of view,
the research and description of water and mineral oil influence
on volumetric losses in a hydraulic motor is appropriate and
justified. The development of a universal mathematical model
describing the volumetric losses in the motor supplied either
by liquid of high viscosity or low viscosity is also appropriate
and reasonable. Therefore, the issue of the influence of the Fig. 2. The working mechanism of a satellite motor: C stator, R rotor,
type of liquid on the volumetric losses in the hydraulic motor S satellite, 110 working chambers, LPC low pressure working
is a new issue and represents an important scientific problem. chamber, HPC high pressure working chamber, Vk-min working chamber
with minimum area Amin (dead chamber), Vk-max working chamber with
Consequently, the following objectives were stated in this maximum area Amax [12,14,16,17,18,20]
article:
a) indication and description of sources of volumetric losses
in the motor;
b) description of the mathematical model of the volumetric
losses;
c) comparison of volumetric losses in motor supplied with
mineral oil and water;
d) comparison of results of experimental research with
mathematical model.
The experimental research of the influence of the type of Fig. 3. Distribution plate: OC inflow/outflow hole [12,14,16,17,18,20]
liquid on volumetric losses was carried out on a prototype of
a hydraulic satellite motor, developed by the author.
The stator is toothed inside and consists of six humps. The
rotor is toothed outside and consists of four humps. Gear
SATELLITE MOTOR wheels called satellites work with the stator and the rotor.
Spaces, called working chambers, are formed between the
For experimental tests a prototype of satellite motor was satellites, the stator and the rotor. The number of working
selected marked with the symbol SM-0,75/25. The design of chambers is equal to the number of satellites.
this motor is presented in Fig. 1. The working mechanism During the rotor rotation the working chambers:
of the satellite motor is a specific gear mechanism in which a) change their volume from minimum Vk-min to maximum
the rotor revolves around the shaft axis and the revolving Vk-max forming a high pressure chamber HPC. The chamber
motion is done by satellites which are in gear with the stator is filled with water;
and the rotor (Fig. 2). b) change their volume from maximum Vk-max to minimum
Vk-min forming a low pressure chamber LPC. The chamber
is emptied of liquid.
The working chambers in the satellite mechanism are
closed by the compensation plates (Fig. 1 elements 7 and
8, and Fig. 3), which also play the role of distribution plates.

214 POLISH MARITIME RESEARCH, No S1/2017

 The number of filling and emptying cycles of the working Dk the coefficient depends on stiffness of axial clearance
chambers per one rotation of the shaft is the product of the compensation unit;
number of humps of the rotor and the stator. Thus, 24 cycles DC the coefficient depends on stiffness of stator.
correspond to one shaft revolution.
The results of research, described in the following section The formula (1) is true if (Fig. 4):
concern the satellite motor type SM-0,75/25 with the following
parameters: (3)
theoretical displacement qt=32,94 cm3/rev.;
teeth module m=0,75mm; and
height of working mechanism H=25mm;
axial clearance of satellites hS=5,35um; (4)
axial clearance of rotor hR=6,38um.
The values of E, A1, A2, K, Dk and DC are given in Table 1.

SOURCES OF VOLUMETRIC LOSSES

In the positive displacement machines the sources of
volumetric losses are:
a) the liquid flow rate in flat clearances in satellite mechanism
(Fig. 4);
b) the liquid flow rate in clearances in commutation unit
(short clearances) in satellite mechanism (Fig. 5);
c) the liquid flow rate in the spaces between the teeth of the
working mechanism (tip clearances Tc and backlashes G)
(Fig. 6);
d) the liquid flow rate caused by the cyclic elastic deformation
of the working chambers, mainly due to the cyclic Fig. 4. The flows in flat clearances [17]
deformation of the stator as the element with the smallest
stiffness;
e) the liquid flow rate depending on its compressibility.
In all satellite motors the internal leakage is discharged FLOW RATE QC IN COMMUTATION UNIT
into the low-pressure channel (outflow channel). CLEARANCES (SHORT CLEARANCE)

FLOW RATE QLFG IN FLAT CLEARANCES Fig. 5 presents the process of passage of the low-pressure
OF WORKING MECHANISM working chamber LPC through the dead chamber V k-min into
the high-pressure working chamber HPC. Two commutation
Flow rate QLfg in flat clearances (Fig. 4) of working unit clearances are formed during this process. There is
mechanism are described by equation [14,17]: leakage in these clearances which results from the pressure
difference in the IH and OH channels:

a) QCm1 from channel IH to chamber Vk-min, caused by the

(1) pressure difference p1;

b) QCm2 from chamber Vk-min to channel OH, caused by the
where: pressure difference p2.
 E the degree of laminarity of the flow [17]; A similar process occurs during the transition of chamber
 Q kinematic viscosity of liquid; HPC to LPC.
 U density of liquid; The leakage occurs in the range of a very small shaft
m teeth module; rotation angle D=(ab;+ab). For further considerations, it is
A1, A2, K coefficients; assumed that:
h equivalent axial clearances of rotor and satellites, a) the angular shaft position D=0o corresponds to chamber
described as [14,17]: Vk-min or chamber Vk-max;
b) angle ab is the critical value of angle D, for which QCm1=

(2) QCm2=0.

hR and hS respectively: axial clearance of rotor and

satellites;
H height of working mechanism (equal to height of
curvature);

POLISH MARITIME RESEARCH, No S1/2017 215

Fig. 5. Process of transition of chamber LPC to HPC via Vk-min [14,15]: a) final Assuming that:
phase of reducing the LPC chamber volume (=(-b,0)); b) chamber Vk-min
(=0); c) initial phase of increasing the HPC chamber volume (=(0,b))
(7)

The flow rate in commutation unit clearances is described and

by equation [14,15]:

(8)

(5) and transforming the formula (25) one obtains:

The value of coefficients C1, C2, C3 i J for motor SM-0,75/25 (9)

are given in Table 1.
where:
FLOW RATE DQFC DEPENDS ON COMPRESSIBILITY t an increment time;
OF LIQUID n rotational speed;
Cfc the coefficient.
The compressibility of liquid l is defined as: The theoretical displacement qt depends on teeth module
m and high H of satellite working mechanism. For the

(6) mechanism as shown in Fig. 2 qt can be expressed by the

formula [14]:
where:
V, V respectively: volume, change of the volume of the (10)
liquid;
p pressure increase. where (Fig. 2):
Amax maximum area of working chamber;
 Amin minimum area of working chamber.

216 POLISH MARITIME RESEARCH, No S1/2017

 Thus:
(11)
If in the commutation unit is zero overlap or positive
overlap, then there is a compression of fluid in the death
volumes. Furthermore, a compression of the liquid in the
spaces between the teeth can occur. These phenomena have
no effect on the volumetric losses. The increase of pressure
in the working chamber starts from the supply pressure pi-1
in the working chamber increasing in volume to a pressure
pi-max in the maximum volume V k-max. Then, the pressure
decreases to a value of pi-2 in the working chamber reducing
its volume. The value of pressure pi-max depends on the axial
clearances of the satellites and the rotor and depends on
the rotational speed. At low speeds, the pressure pi-max does
not reach large values. It is estimated that as a result of the
flow of liquid from the maximum volume V k-max, through
clearances, to the low pressure chamber, the pi-max is close
to the pi-1. And at high rotational speeds the liquid does not
have time to flow through the clearances to the low pressure
chamber. As a result, depending on the type of liquid and its
compressibility, the pressure pi-max is considerably higher than
the pressure pi-1. Then the value of pi-max has an effect only on
the torque of mechanical losses in the motor.
In death volume Vk-min, there is a minimum pressure pi-min
similar to the pi-2.
FLOW RATE Q DCH CAUSED BY THE ELASTIC
DEFORMATION OF THE WORKING CHAMBERS
Elastic deformation of the working chambers in satellite
mechanism is caused by the pressure difference Dpi in these
chambers and the change in pressure in the volumes Vk-min and
Vk-max. As a result, the displacement of the motor is changing
about Dqdch. The flow rate DQdch is proposed to be express with
a simplified equation [14]:
(12)
density and compressibility of the liquid). The impact of
where:
Cdch the coefficient;
Kch the stiffness of working mechanism.
FLOW RATE DQB DEPENDS ON TIP CLEARANCE AND
BACKLASH
The spaces between the teeth in satellite mechanism (tip
clearance and backlash Fig. 6) is the cause of volumetric
losses DQb. The value of DQb is proportional to the motor
speed n. The increase of Dpi causes an increase in elastic
deformation of:
the teeth, that is the spaces between the teeth increase;
the rotor and stator, that is the backlash increase.
Fig. 6. Tip clearance Tc and backlash G [19]
Thus, the flow rate DQb is proposed to be express with the
simplified formula [14]:
(13)
where:
qb the change of displacement caused by the movement
of satellites in spaces between the teeth;
Cb the coefficient.
FLOW RATE DQID DEPENDS ON THE INERTIA OF
THE SATELLITES AND THE LIQUID IN WORKING
CHAMBERS AND DEPENDS ON THE DAMPING
PROPERTIES OF LIQUID
If increasing the rotational speed of the operating
mechanism increases, then the acceleration of satellites
increases and the acceleration of the liquid in the chambers
increases also. As a result, the inertia forces of the liquid and
satellites increase. At the time of change the commutation unit
phase (e.g. at the time of opening the inflow and cut-off the
outflow from the working chamber) these forces are opposed
to changing the position of satellites in the space between
the rotor teeth and in the space between the stator teeth. In
addition, the speed of the jump teeth of satellites, when
changing commutation unit phase, depends on the damping
properties of the working liquid (that is: the viscosity, the
above factors on the volumetric losses DQid is proposed to
be describe with the simplified formula [14]:
(14)
where Cid is the coefficient.
EXTERNAL LEAKAGE QEX
External leakage Qex in the hydraulic motor is unacceptable,
that is Qex=0.
POLISH MARITIME RESEARCH, No S1/2017 217
MATHEMATICAL MODEL OF VOLUMETRIC

LOSSES (22)

but for laminar flow of emulsion:

KNOWN MODEL OF VOLUMETRIC LOSSES
(23)


The literature describes the volumetric losses Qvl as follows:
a) in motor supplied with oil [1,21]: mE the dynamic viscosity of HFA-E emulsion (for atmospheric
pressure) in temperature T1 in the inflow port of motor;
(15) Qex,O the external leakages of oil, expressed by the formula;

b) in motor supplied with low viscosity liquid (like water or (24)

HFA-E emulsion) [20,21]:
p2,O pressure in outflow port in motor supplied with oil;
Bh, kex, Cs,O, Cs,E, Cex,E, Cl,O, Ct,O the coefficients.







(16)

The pressure drop pi,O in working chambers of a motor supplied
In the above formulas [1,20,21]: with oil is [21]:
Q vl,O, Q vl,E the component of volumetric losses
depends on drop pressure pi,O in motor working chamber (25)
and depends on angular velocity of motor shaft. In a motor
supplied with oil this component is expressed by the formula: but the pressure drop pi,E in working chambers of a motor supplied
with emulsion is described by the equation:

(17)


(26)

and in a motor supplied with emulsion:
Where [1,21]:



pich,O the pressure drop in internal channel of a motor supplied

(18) with oil:

pi,O drop pressure in working chamber of motor supplied

with oil; (27)

pi,E drop pressure in working chamber of motor supplied
with low viscosity fluid (HFA-E emulsion); Cich,O the coefficient;
 Z the angular velocity of the motor shaft; k H the coefficient described by:
kc the coefficient depends on the compressibility module
of oil KO and emulsion K E, expressed by the equation:
(28)


(19)

E the density of emulsion.
Ql,O the component of volumetric losses with laminar
nature in a motor supplied with oil: The Paszota Z. and Maczyszyn A. [9,11] description of
volumetric losses is dependent on the type of machine (pump

(20) or motor) and does not show the sources of these losses. The

volumetric losses according to Paszota Z. and Maczyszyn
O the dynamic viscosity of oil (for atmospheric pressure) A. are as follows [9,11]:
in temperature T1 in the inflow port of a motor; a) for a motor with constant displacement:
Qt,O the component of volumetric losses with turbulent
nature in a motor supplied with oil:

(29)


(21) b) for a pump with constant displacement:

O the density of oil;

ks the coefficient. For not fully developed turbulent flow (30)

of emulsion:

218 POLISH MARITIME RESEARCH, No S1/2017

where:
k1, k9, apv, avv, anv coefficients;
np speed of pump;
nt, n theoretical and actual speed of motor;
pn nominal pressure in hydraulic circuit;
pi pressure drop in working chambers of a pump or
motor;
qPt theoretical displacement of pump;
QPt theoretical capacity of pump.
n related kinematic viscosity (n = 35cSt).
The inconvenience of these models is that they describe
the volumetric losses in general, without specifying the
type of gaps and other sources of these losses. In addition,
the leakage of a low viscosity liquid (like water or HFA-E
emulsion) is described by a different mathematical formula
than the leakage of a high-viscosity liquid (like mineral oil
or vegetable oil). Therefore, it was advisable to develop anew
method to describe the volumetric losses in a satellite motor
a universal method for both high viscosity and low viscosity
liquids.
N E W M AT H E M AT IC A L DE SCR I P T ION
VOLUMETRIC LOSSES
Volumetric losses Qvl in the hydraulic motor are proposed
to be express as (Fig. 7):
(31)
Fig. 7. Characteristic of components of volumetric losses in hydraulic motor
The components DQfc, DQdch, DQb are proportional to
both the pressure drop Dpi in the working chambers and
the effective engine speed n. But, the component DQid is only
dependent on n. That is:
(32) experimental data according the formula:
From the equations (11),(12),(13),(14) and (35) it follows
that:
(33)
All components of volumetric losses depend on the drop
pressure pi in working chambers. The pressure in the
working chambers is difficult to measure. In normal engine
operating conditions the pressure is measured at its ports.
Thus, it is convenient to use the pressure drop Dp in the motor.
The relationship between the pressure drop in the motor p
and pressure drop pi in its chambers, is as follows:
(34)
Internal channels in motors have a complex geometry. In
these channels, there is a turbulent, not fully developed flow.
To describe the pressure drop pich in the internal channels,
it is proposed to adopt a simplified relationship as [18]:




(35)
where Cl and Ct are coefficients depend on geometrical
dimensions of internal channels.
OF RESULTS OF EXPERIMENTAL RESEARCH
AND SIMULATION
The motor SM-0,75/25 was tested using:
the Total Azolla 46 (=40cSt, =873kg/m3) oil;
pure tap water (=0,853cSt, =996kg/m3).
The motor parameters directly measured on the test stand
are:
pressure p1 in the motor inlet port and pressure p2 in the
motor outflow port;
torque M on the motor shaft;
motor absorbency Q (the flow rate to the motor);
the rotational speed of the motor shaft n.
The flow rate Q in the motor is:
(36)
where q is the change of displacement coming at one rotation
of motor shaft.
The volumetric losses Q vl are a function of pressure
drop pi in motor working chambers. In this connection
it is necessary to designate the pressure drop pich in the
internal channels. The methodology of measuring the pich
is described in [18]. The pich is described by formula (35)
and values of coefficient are given in Table 1.
The volumetric losses Q vl were determined from
(37)
The characteristics of volumetric losses Qvl in the motor
can be presented as a function of pi when:
a) the flow rate Q is constant (Q is the parameter independent
of the motor) Fig. 8. Then when load M of the motor
POLISH MARITIME RESEARCH, No S1/2017 219
 (thereby pi) increases, the speed n of motor shaft
decreases;
a) the speed n of motor shaft is constant Fig. 9 and Fig. 10.
Then when load M of the motor (thereby pi) increases,
the absorbency Q of motor increases also.
In Fig. 9 and Fig. 10 the characteristics of QLfg+QC are shown
obtained experimentally and as a result of the calculation
accordance with the model (1) and (5). The methodology
for determining the components QLfg and QC are described
in [14,17].
In Fig. 9 and Fig. 10 are shown the characteristics of
Qvl obtained experimentally. But experimentally obtained
characteristics of q as a function of pi are shown in Fig.
11 and Fig. 12.
The values of coefficients C and Cid of model (33) can be
calculated from the equation of the trend line of experimental
characteristics of q=f(pi). These values are given in Table1.
Characteristics of q, calculated according to formula (33),
are plotted in Fig. 11 and Fig. 12.
Whereas the volumetric losses, calculated according to
equations (32) are shown in Fig. 9 and Fig. 10 (the dotted line).

Fig. 8. Characteristics of Qvl =f(Dpi) for Q=const. in motor supplied

withoiland water
Fig. 11. Characteristics of q=f(pi) in motor supplied with oil

Fig. 9. Characteristics of Qvl =f(Dpi) for n=const. in motor supplied with oil

Fig. 12. Characteristics of q=f(pi) in motor supplied with water

Table 1. Values of coefficients in volumetric losses model for motor SM-0,75/25

[14,17]

A1 A2 C Ct Cl Cid C1

[-] [-] [-] [-] [-]

Oil 0,077
0,145 0,17 1,33
Water 1,25
C2 C3 DC Dk K

[-] [-] [-] [-]
  
Oil 1,26 0,684 1,141 0,76 0,996
-0,131
Water 1,05 0,183 1,0 0,55 0,655

Fig. 10. Characteristics of Qvl =f(Dpi) for n=const. in motor supplied with water

DISCUSSION
From the formula (36) it follows that (after the adoption
of Qex=0): The test results have shown that the volumetric losses in
the motor supplied with water are at most several time greater

(38) than the volumetric losses in the motor supplied with oil.

The greatest differences was observed for motor supplied

220 POLISH MARITIME RESEARCH, No S1/2017

with asmall stream of liquid. Furthermore, for Q = const.,
the ratio of is variable (Fig. 13). The value of pi, for
which the ratio of reaches the maximum, depends
on the Q. The smallest value of pi is observed for the smallest
values of Q. An impact on this phenomenon has:
 change of the nature of the flow (with the increase of pi)
of water and oil in the clearances in the motor working
mechanism;
 change of q under the influence of pi;
 change of engine speed n.
a) for high speed (n=1500rpm) 10% for oil and 25% for
water;
b) for low speed (n=100rpm) 8% for oil and 20% for water.
However, the results of calculations of q, according to
the above model, for high pi are heavily biased. Therefore, it
is proposed to describe the q generally by the formula (33).
By way of exception, in order to increase the accuracy of the
simulation of volumetric losses in the motor operating at low
load, it is permitted to accept model (39).
The effect of volumetric losses in the hydraulic motor is a
drop of speed nl. The real engine speed n can be calculated
from the relationship (36) taking into account the equation
(33). Then:

(40)

The loss of speed nl may be calculated from the equation:

(41)

The characteristics of nl=f(pi) are shown in Fig. 14 and

Fig. 13. Characteristics of for motor supplied with oil and water Fig. 15.

The mathematical model of volumetric losses has been

developed based on the analysis of sources of these losses. It
has been shown that on the basis of experimental results, it is
possible to calculate the coefficients of the model. Based on the
proposed mathematical model, characteristics of volumetric
losses were plotted and compared with the results of the
experiment. It has been found that the simulation results
differ from the results of the experiment:
a) for high speed (n=1500rpm):
for pi=2MPa about 35% both for oil and water; Fig. 14. Characteristics of nl=f(pi) result of experiment and calculation
for pi=25MPa about 4% both for oil and water; according to formulas (40) and (41). The motor supplied with oil
b) for low speed (n=100rpm):
for pi=2MPa about 20% for oil and 35% for water;
for pi=25MPa 7% both for oil and water.
It turns out that the greatest impact on accuracy of
calculation of the volumetric losses has the accuracy of
calculation of the change of motor displacement q. The
Dq calculated according to (33) gives the biggest mistake in
the range of small values of pi (the biggest differences are
observed for pi<6 MPa Fig. 11 and Fig. 12). The smaller
error in the lower range of pi is obtained if q is described by
the following empirical formula (this model does not reflect
the physical phenomenon):
Fig. 15. Characteristics of nl=f(pi) result of experiment and calculation
according to formulas (40) and (41). The motor supplied with water

(39)

Thus, it is easy to estimate how much the motor speed will
where the values of constants are: be reduced if the motor will be supplied with a low viscosity
for water: C1=10,6 i C2=0,1; liquid (like water). In Fig. 16 are shown characteristics of
for oil: C1=11,7 i C2=0,1. speed ratio of the motor supplied with water (nw) and the
Then for low pressure drop (pi=2MPa) the calculations motor supplied with oil (no).
differ from the results of the experiment:

POLISH MARITIME RESEARCH, No S1/2017 221


Fig. 16. The ratio of motor speed supplied with water (nw) and supplied
withoil (no)
Thus, with a small constant stream of water (Q=5 l/min)
and at high pressure drop in motor (pi=25MPa) the speed
of motor is reduced even by half (in comparison to speed of
motor supplied with oil)!
In conclusion, the mathematical model of volumetric
losses, presented in this paper, describes quite accurately
these loses in the motor supplied with both a mineral oil or
water. Therefore, the volumetric losses in the motor supplied
with various liquids can be assessed by comparison of the
model coefficients.
The mathematical model presented in this article, is
acomponent of a model of overall efficiency of the hydraulic
motor. This model is also suitable to describe volumetric
losses in a satellite pump.
ACKNOWLEDGMENT
This article was developed in the framework of the project
LIDER/35/102/L-2/10/NCBiR/2011, entitled: New study of
hydraulic satellite machines for drives with bioliquids and
non-flammable liquids. This project is funded by the National
Centre for Research and Development in Poland.
REFERENCES
1. Balawender A.: Physical and mathematical model of losses
in hydraulic motors. Developments in mechanical engi-
neering, Gdansk University of Technology Publishers.
Gdansk 2005.
2. Dymarski C., Dymarski P.: Developing Methodology for
Model Tests of Floating Platforms in Low-Depth Towing
Tank. Archives of Civil and Mechanical Engineering, No
1/2016, DOI: dx.doi.org/10.1016/j.acme.2015.07.003
3. Guzowski A., Sobczyk A.: Reconstruction of hydrostatic
drive and control system dedicated for small mobile plat-
form. American Society of Mechanical Engineers, 2014
doi: dx.doi.org/10.1115/FPNI2014-7862.
4. Jasinski R.: Problems of the starting and operating of
hydraulic components and systems in low ambient tem-
perature (Part I). Polish Maritime Research, No 4/2008.
222 POLISH MARITIME RESEARCH, No S1/2017
5. Jasinski R.: Problems of the starting and operating of
hydraulic components and systems in low ambient tem-
perature (Part II). Polish Maritime Research, No 1/2009.
6. Jasinski R.: Problems of the starting and operating of
hydraulic components and systems in low ambient tem-
perature. Part III.Methods of determining parameters for
correct start-ups of hydraulic components and systems
in low ambient temperatures. Polish Maritime Research,
No 4/2009.
7. Litwin W., Olszewski A.: Water-Lubricated Sintered
Bronze. Journal Bearings - Theoretical and Experimental
Research. Tribology Transactions, vol. 57, No 1/2014.
8. Lubinski J., Sliwinski P.: Multi parameter sliding test
result evaluation for the selection of material pair for wear
resistant components of a hydraulic motor dedicated for
use with environmentally friendly working fluids. Solid
State Phenomena, No 225/2015.
9. Maczyszyn A.: Energy analysis of rotary positive displace-
ment machines used in hydrostatic transmissions. PhD
thesis. Gdansk University of Technology, 2014.
10. Osinski P., Deptula A., Partyka M.: Discrete optimization
of a gear pump after tooth root undercutting by means of
multi-valued logic trees. Archives of Civil and Mechanical
Engineering, No 4/2013, DOI: 10.1016/j.acme.2013.05.001.
11. Paszota Z.: Energy losses in hydrostatic drive. LABERT
Academic Publishing, 2016.
12. Patrosz P.: Deformation in the axial clearance compen-
sation node in the satellite pump unit. Hydraulics and
Pneumatics 1/2014, Poland.
13. Pobedza J., Sobczyk A.: Properties of high pressure water
hydraulic components with modern coatings. Advanced
Materials Research. Trans Tech Publications Ltd,
849/2014. doi: 10.4028/www.scientific.net/AMR.849.100.
14. Sliwinski P.: Satellite displacement machines. Basis of
design and analysis of power loss. Gdansk University of
Technology Publishers, 2016.
15. Sliwinski P.: The basics of design and experimental tests
of the commutation unit of a hydraulic satellite motor.
Archives of Civil and Mechanical Engineering, No
16/2016, DOI: 10.1016/j.acme.2016.04.003.
16. Sliwinski P. Satellite pump and motor. Machines Tech-
nologies Materials 9/2014.
17. Sliwinski P.: The flow of liquid in flat gaps of satellite
motors working mechanism. Polish Maritime Research,
No 2/2014.
18. Sliwinski P.: Pressure losses and power balance in the CONTACT WITH THE AUTHORS
unloaded satellite pump. Hydraulika a Pneumatika, No
1-2/2013. Pawel Sliwinski
e-mail: pawel.sliwinski@pg.gda.pl
19. Sliwinski P.: New satellite pumps. Key Engineering Mate- Gdansk University of Technology,
rials, No 490/2012. Faculty of Mechanical Engineering
Poland
20. liwiski, P.: Influence of oil and emulsion HFA-E on
flow characteristics in gaps of hydraulics satellite motors
(in Polish). Hydraulika i Pneumatyka, No 5/2007.

21. Sliwinski, P.: Comparison of phenomena in hydraulic sat-

ellite motors supplied with oil-in-water emulsion and oil.
PhD thesis. Faculty of Mechanical Engineering, Gdansk
University of Technology, 2006.

22. Urbanczyk J.: Research of hydraulic motors for needs of

small mechanization in the mining industry. PhD thesis.
AGH University of Science and Technology, 1999.

23. Walczak P., Sobczyk A.: Simulation of water hydraulic

control system of francis turbine. American Society of
Mechanical Engineers, 2014. doi: dx.doi.org/10.1115/
FPNI2014-7814

24. Zloto T., Nagorka A.: An efficient FEM for pressure

analysis of oil film in a piston pump. Applied Mathemat-
ics and Mechanics, vol.30, No 1/2009.

POLISH MARITIME RESEARCH, No S1/2017 223

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 224-229
10.1515/pomr-2017-0042

INVESTIGATIONS ON WATER CIRCULATION

IN ANIMAL SEA-WATER BASINS ON THE EXAMPLE OF SEALS
BREEDING POOLS

Piotr Zima
Gdansk University of Technology, Poland

ABSTRACT

This paper presents general comments concerning investigations on water circulation in animal breeding pools
containing sea water. As an example are given results of computer simulation of water circulation in seals breeding
pools situated in Marine Station at Hel, belonging to Oceanographic Institute , Gdansk University. A mathematical
model of three main pools was prepared with taking into account their inflow and outflow water supply points. Next,
the object indication ( tracer) tests were done with the use of mathematical modelling as well as in-situ measurements.
For description of flow field in steady conditions a simplified model of 2D flow in the form of Helmholtz biharmonic
equation of stream function , recalculated then into velocity vector components, was used. The equation , supplemented
with appropriate boundary conditions , was solved numerically by using the finite differences method. The spreading
of a substance dissolved in water (tracer) was analyzed by solving 2D equation of transient advecting - dispersing
transport. To solve it the finite volumes method was applied. The applied model was verified by conducting the
indication tests with the use of the rhodamine WT as a tracer. The obtained results made it possible to reconstruct water
circulation within the seals pools and identify stagnation zones in which water circulation may be made difficult.

Keywords: circulation of water in animal breeding pool, water flow through the pools, computer simulation,

simulations and indication tests in situ). Unfortunately,

INTRODUCTION
Circulation of water in animal breeding pools is an
important factor which influences its quality. It is especially
important in case of basins where warm-blooded animals live.
The animals live in a limited space and influence parameters of
water in the pools during feeding and by excreting metabolism
products which contaminate the water. To prevent it, pools of
the kind are fitted with water exchange systems. Their main
task is to ensure an appropriate water change rate. They should
also ensure an appropriate circulation to prevent forming the
stagnation zones which may lead to growing concentration
of contamination or microbes. (Lehnert et al., 2014). The
detrimental phenomena can be counteracted by extending
the zone of active exchange of water as well as by forming
appropriate hydraulic conditions aimed at elimination of
stagnation zones. This can be done by a few ways. One of
them is the mathematical modelling of flow field (Zima,
2014) as well as the conducting of tracer tests (numerical
mathematical modelling has one important drawback
consisting in that to solve converse problems is not possible
(when demands are imposed on problem solution and
considered system geometry has to be solved, and/or limit
conditions which make it possible to satisfy the demands).
It allows only to analyze the accepted solutions and try to
find a solution which satisfies the assumed criterion the best.
As far as the sea-water pools are concerned, additional
problems should be expected. Circulation of water in such
objects consists in delivering sea water during its exchange.
It may cause problems dealing with moving contaminations
from open water area and deposing them within closed
ecosystem of animal breeding pools. It concerns all kinds
of contaminations (biological, chemical and physical ones).
Among physical ones, the more and more often present
microplastic forms should be mentioned (Setl et al, 2014).
On the other hand, an aggressive impact of sea water on
elements of water exchange systems should be remembered.
This is especially dangerous in case when changes of pump

224 POLISH MARITIME RESEARCH, No S1/2017

working parameters occur along with time (Puzyrewski et al.
2010, Szantyr et al. 2011). Therefore even at a water exchange
system correctly designed and tested it is necessary to control
periodically operation of pump systems, and, at changed
parameters of their operation to analyze their impact on
water circulation within pools.
METHODS USED FOR INVESTIGATIONS
INQUESTION
As already mentioned, various methods may be applied
to investigate water circulation in animal breeding pools.
Good results can be achieved by means of mathematical
modelling with the use of computer simulations. In order
to perform mathematical modelling tasks it is necessary to
prepare data for such model (a mathematical abstraction)
which represents a real object (Meerschaert 2007) and to solve
appropriate mathematical physics equations, which describe
water flow (Szymkiewicz 1991, 2010, Szewc et al. 2012a), and
in case of tracer tests behaviour of the tracer in the form of
dissolved substance in the calculated velocity field (or a two-
phase medium Szewc et al. 2012b). The first phase, i.e. the
forming of a mathematical model of real object, requires to
define a calculation area, i.e. geometry of a pool in question.
In case of animal breeding pools filled with sea water, various
techniques of performing measurements which allow to use
the more and more exact methods, may be implemented,
depending on their size. In case of large pools, techniques
used in sea-bed bathymetry are applied (Kozaczka et al.
2012, Grelowska et al. 2013, Kozaczka 2013, Grelowska and
Kozaczka 2014), for small pools geodesic measurements are
satisfactory (Burdziakowski et al., 2015). Apart from area of
apool, it is necessary to know parameters which describe
water circulating system, namely : pump capacity and location
of water inflows to and outflows from the pool .
The next step is to perform computer simulations. General
motion equations of non-homogenous liquid represent an
extremely powerful tool at contemporary development level
of numerical methods and today capabilities of computers.
The tool consists in far -leading formalization exemplified
by commercial computer software. Its application becomes
formally more and more simple, that results in growing
popularity of applying Computational Fluid Dynamics in
environment engineering (Lipeme Kouyi et al. 2003, Zima et al.,
2009). However, application of such programs for real objects,
at satisfying an appropriate accuracy level, requires huge
resources associated with preparation of a calculation model
(generation of calculation area), and running the simulations,
appears often very time-consuming. For this reason,
simplified models which allow for utilizing specific features
of an object and reaching good quality results, are commonly
used . In this paper an example of such approach is presented.
To determine water velocity field in seals breeding pools,
2D kinematic model in the form of Helmholtz biharmonic
equation was applied. It allows for reproduction of velocity
field in steady- state conditions (that is equivalent to specificity
of water circulation in objects of the kind) as well as to draw
trajectories of particles indicating stagnation zones. However
it does not provide a full picture of changes in concentration
of a factor which contaminates water in the pool. For this
reason also tracer tests should be performed by simulating
introduction of a tracer in the form of a substance dissolved
in water in trouble spots of the pool. Migration of the tracer
can be revealed by solving the dissolved substance transport
equation. In the case in question, 2D equation of unsteady
advecting- dispersing transport of dissolved substance was
applied (Sawicki and Zima, 1997).
The tracer tests should be conducted by numerical
simulating as well as indication tests on the object in situ.
To the indication tests, substances harmless for animals
living in the pools, should be applied (Zima, 2014). In the
presented example the indication tests were conducted with
the use of the tracer in the form of the rhodamine WT.
Portions of an appropriately dissolved tracer were delivered
into characteristic spots, and then measurements of changes
in rhodamine concentration along with time were made in
selected points of the pools. Two Cyclop 7 sounders of Terner
Designs were used for the measurements.
MATHEMATICAL MODEL
In producing the mathematical model a schematization
of real object should be first of all made. In the model an
idealization of real conditions should be built in order to
accept appropriate assumptions simplifying mathematical
description of the phenomenon in question (Meerschaert,
2007). In this paper results of the investigations performed
in the seals breeding pools of Marine Station at Hel, Institute
of Oceanography, Gdask University, are presented as an
example. The arrangement and basic data concerning the
pools are shown in Fig. 1 where water inflows and outflows are
also depicted. The analysis covered water flow through three
main pools (No. 3, 4 and 7), in which seals live. Depth of the
pools is different and changes from h = 1.0m up to h = 2.9m.
Point outflow Line outflow
of water of water
Pool No. 3 Point outflow
Pool No. 4
of water
Pool No. 7
Submerged inflow of water
Su
Submerged inflow of water
Submerged inflow of water
Fig.1. The assumed physical model of the seals breeding pools
The next stage of building the mathematical model is to
accept a mathematical description of run of the phenomenon
. First, the description model of steady flow field at different
water depth was assumed (the particular pools have different
depth values and are connected to each other). As already
mentioned above, was taken the kinematic model in which,
POLISH MARITIME RESEARCH, No S1/2017 225
apart from the satisfying of mass conservation equation, where: n = [n x ,ny] is the velocity directional vector:
also kinematic condition concerning fluid element rotation
is fulfilled (Anderson, 1995). The model assuming changing ux uy
nx , ny
rotation of fluid element is commonly in use. In case of 2D u u
vector velocity: u=[u x,uy] the condition can be written as
follows: The longitudinal, DL , and transverse, DT, coordinates of the
dispersion tensor (6) are described by the following formula
rot u = (x, y) (1) (Elder, 1956):

where: (x, y) the assumed rotation distribution function.

DL = h v * 30 < < 3000 (7)
If the Stokes assumption is accepted (on neglecting inertia DT = h v * 0.15 < < 0.30
forces) for steady flow conditions and constant temperature
of water, with taking into account Eq. (1), the following is where: v * dynamical velocity. In this paper values of the
obtained: coefficients and are determined on the basis of available
literature sources. Finally, = 300 and = 0.23 was assumed.
= 0 (2) Eq. (5) constitutes mathematically an initial boundary
value problem, which, to be solved, requires assuming initial
It is the equation for the harmonic function , called condition and boundary ones.
Helmholtz equation. If the stream function (x, y) is defined Eq. (4) and (5), in general cases, can be solved only by
as follows: using numerical methods. In this work the finite difference
method was used for biharmonic equation (Anderson, 1995)
and the finite volume method for dissolved substance
w\ w\
h ux ,  h u y (3) transport (LeVeque, 2002). To apply the above mentioned
wy wx methods the assumed solution area was split into square
cells of x=y=0.25 m (Fig.2). For solving the problem the
(where h stands for water depth), Eq. (2) (biharmonic one) authors original programs were used.
can be expressed in the following form:

1 1
= = 2 h = 0 (4)
h h

Eq. (4) may be interpreted to be a generalization of

kinematic models in which velocity rotation is a harmonic
function described by Laplace equation. In the general case
Dirichlet problem should be solved for biharmonic operator. Fig.2. The assumed calculation area
Solution of the problem in 2D space (x,y) is the stream function
(x, y). With the use of Eq. (3) velocity vector components
can be calculated over the whole area .
Next, was assumed the migration model for dissolved COMPUTER SIMULATIONS OF WATER
substance (tracer), which, in unsteady conditions, at FLOW THROUGH THE POOLS
changeable depth as well as 2D space, may be written in the
following form (Sawicki and Zima, 1997): To obtain a solution of the biharmonic equation (4) it
is necessary to take into account water supply to the seals
hc (hu x c ) (hu y c ) 1
breeding pools. In Fig. 1 location of water inflow and outflow
c c
+ + = hD xx + hD xy spots in particular pools, is indicated. In these spots, in the
t x y h x x y
(5) mathematical model, the boundary condition simulating
1 c c water inflow and outflow was assumed. Next, computer
+ hD yx + hD yy
h y x y simulations were carried out in steady motion conditions.

where: t time, c concentration of dissolved substance;

and, coordinates of the dispersion tensor D may be defined
as follows:

D xx D L n x2  DT n y2 , D xy D yx DL  DT n x n y , Dy D L n y2  DT n x2 (6)

226 POLISH MARITIME RESEARCH, No S1/2017


Fig. 3. Distribution of the stream function (x, y) in steady motion conditions
in the seals breeding pools
The obtained results in the form of distribution of the
stream function (x, y) are presented in Fig. 3 (from 0 up to
100% m3/s per unit of depth). The zones inside closed lines
of the stream function indicate areas of closed circulation
of water. Then, by using Eq. (3), velocity distribution was
calculated over the entire defined area. Results of the
calculation are presented in Fig. 4. Maximum velocity values
(in the spots of water flow into the pools) were equal to 0.5m/s
(the longest vectors shown in Fig. 4).
Fig. 4. Distribution of the water velocity field in steady motion conditions
inthe seals breeding pools
Next, simulations of flow of the tracer introduced to inflow
spots of each of the seals breeding pools, were conducted
under assumption on continuous delivery of the tracer (100%
amount). Their results are given in Fig. 5. The performed
calculations made it possible to estimate duration time of
flow of the tracer up to outflow spot. For all the pools the
time was equal to about 68 min.
Fig. 5. Distribution of tracer concentration after reaching the spots of water
outflow from the seals breeding pools
a) b)
Fig. 6. Tracer tests in the pool No. 7 :
a) input of dosage of the rhodamine WT at the spot of water inflow to the pool,
b) continuous measurement ofconcentration of the rhodamine WT at the spot
of water outflow from the pool
In order to verify the achieved results, the tracer tests were
conducted in situ in the seals breeding pools. 20% rhodamine
WT was applied as a tracer. 100 ml dosage was introduced
to the spot of water inflow to each of the pools, and in the
outflow spots continuous measurements were conducted by
means of two Cyclop 7 sounders (Fig.6 tracer tests in the
pool No. 7). The tests were carried out in each of the pools
separately (during the tests no seals were inside the pools).
The example results, in the form of changes of concentration
along with time, observed in the spot of outflow from the
pool No. 3, are presented in Fig. 7 (where continuous line
stands for measurement, broken line result of numerical
simulations) .
40
35 Tracer study
measurement
30 numerical simulations
25
CRWT [%]
20
15
10
5
0
-5
0 1 2 3 4 5 6 7
t [min]
Fig. 7. Measurement results of concentration of the rhodamine WT at outflow
from the pool No. 3, and results of numerical simulations measurement
oftime of reaching, by the tracer, the outflow spot from the pool No. 3
DISCUSSION OF THE OBTAINED RESULTS
AND CONCLUSIONS
The performed computer calculations made it possible to
draw the stream function field (x, y) and obtain values of
the components ux(x, y) and uy(x, y) of the velocity field u in
the tested seals breeding pools. The distribution and shape
of isolines of the stream function showed the arrangement
of main directions of water circulation as well as location of
outflow-free (stagnation) zones within the pools. In the pool
No. 3 such zones occur mainly in its corners, and the largest
one is situated just behind the inflow delivering water to the
pool. In the pool No. 4 water flows in the form of uniform
streams from inflow to outflow and builds the stagnation
POLISH MARITIME RESEARCH, No S1/2017 227
zone in the channel connecting the pools No. 4 and 7. In the
pool No. 7 the main stream of water circulation is formed
along the streams connecting inflow and outflow spots. In
the pool was built a vast stagnation zone in which exchange
of water is made difficult and which does not take part in
circulation of water flowing through the pool. The highest
velocities occur at inflows and outflows (between 0.48 and
0.51 m/s), and next, inside the pool, they drop to the value of
a few centimeters per second only . All the stagnation zones
are characteristic of closed water circulation in which velocity
values are on the level below one centimeter per second.
Next, the mathematical modelling of spreading the tracer
was performed for each of the pools. The problem consisting
in simulation of introduction of the tracer into inflow
spots to each of the pools was solved and its spreading was
calculated . The computer simulations were verified by means
of the tests in situ. The conducted tracer tests showed that
duration time of arriving the outflow by the tracer particles
was between 6 and 8 min for all the pools. The calculation
results confirmed the conducted measurements. Therefore it
seems that application of the proposed approach in the stage
of formulating the design guidelines for the animal breeding
objects of the type is justified, hence it is also suggested to
use the gained experience in case of modernization of other
objects intended for the breeding of water animals so as to
obtain such form of water exchange in an object, which could
ensure an appropriate change rate of the whole water inside
the pool (of at least a few times per day)
BIBLIOGRAPHY
1. Anderson J. D.: Computational Fluid Dynamics. The
Basics with Applications, McGraw-Hill Inc., New York,
USA (1995), pp. 1547
2. Burdziakowski P., Janowski A., Kholodkov A., Matysik
K., Matysik M., Przyborski M., Szulwic J., Tysic P.,
Wojtowicz A.: Maritime Laser Scanning As The Source
For Spatial Data, Polish Maritime Research, Vol. 22, No.
4(88) (2015), pp. 914
3. Elder J.W.: The dispersion of marked fluid in turbulent
shear flow, J. of Fluid Mech., Vol. 5, No. 4 (1956), pp.
544560
4. Grelowska G., Kozaczka E., Kozaczka S.: Gdansk Bay
sea bed sounding and classification of its results, Polish
Maritime Research, Vol. 20, No 3 (2013), pp. 4550
5. Grelowska G., Kozaczka E.: Underwater Acoustic Imaging
of the Sea, Archives of Acoustics, Vol. 39, No. 4 (2014),
pp. 439452
6. Kozaczka E., Grelowska G., Kozaczka S., Szymczak W.:
Processing data on sea bottom structure obtained by means
of the parametric sounding, Polish Maritime Research,
Vol. 19, No 4(76) (2012), pp. 310
228 POLISH MARITIME RESEARCH, No S1/2017
7. Kozaczka E.: Acoustics waves in the sea, Archives of
Acoustics, Vol. 38, No. 3 (2013), pp. 441442
8. Lehnert K., Mller S., Weirup L., Ronnenberg K., Pawlicz-
ka I., Rosenberger T., Siebert U.: Molecular Biomarkers
in Grey Seals (Halichoerus grypus) to Evaluate Pollutant
Exposure, Health and Immune Status, Marine Pollution
Bulletin, Vol. 88, No. 12 (2014), pp. 311318.
9. LeVeque R. J.: Finite Volume Method for Hyperbolic
Problems, Cambridge University Press, New York, USA
(2002), pp. 1558
10. Lipeme Kouyi, G.,Vazquez, J., Poulet, J.: 3D free surface
measurement and numerical modelling of flows in storm
overflows, Flow Measurement and Instrumentation,
Vol.14, No. 3 (2003), pp. 7987
11. Meerschaert M. M.: Mathematical Modeling, 4 ed. Elser-
vier Science (2013), pp. 365
12. Puzyrewski R., ochowski K., Flaszyski P.: A Method
for Analyzing Ram Pressure Characteristics of Impeller
Pump Rotor, Polish Maritime Research, Vol. 17, No. 2(64)
(2010), pp. 5257
13. Sawicki J.M; Zima P.: The Influence of Mixed Derivatives
on The Mathematical Simulation of Pollutants Transfer,
4th International Conference on Water Pollution, Bled,
Slovenia (1997), pp. 627635
14. Setl O., Fleming-Lehtinen V., Lehtiniemi M.: Ingestion
and Transfer of Microplastics in The Planktonic Food Web,
Environmental Pollution, Vol. 185 (2014), pp. 7783
15. Szantyr J., Flaszyski P., Tesch K., Suchecki W.,
Alabrudziski S.: An Experimental and Numerical Study
of Tip Vortex Cavitation, Polish Maritime Research, Vol.
18, No. 4 (2011), pp. 1422
16. Szewc K., Pozorski J., Minier J.: Analysis of The Incom-
pressibility Constraint in The Smoothed Particle Hydro-
dynamics Method, International Journal for Numerical
Methods in Engineering, Vol. 92, No. 4 (2012a), pp.
343369
17. Szewc K., Taniere A., Pozorski J., Minier J.: A Study on
Application of Smoothed Particle Hydrodynamics to Multi-
Phase Flows, International Journal of Nonlinear Sciences
and Numerical Simulation (2012b), pp. 383395
18. Szymkiewicz R.: Numerical Modeling in Open Channel
Hydraulics, Book Series: Water Science and Technology
Library, vol. 83 (2010), pp. 1-419;
19. Szymkiewicz R.: Finite-Element Method for the Solu-
tion of the Saint -Venant Equations in an Open Channel
 Network. Journal Of Hydrology, Vol. 122, No. 14 (1991), CONTACT WITH THE AUTHOR
pp. 275287
Piotr Zima
20. Zima P.: Numerical Simulations and Tracer Studies as a
Tool to Support Water Circulation Modeling in Breeding Civil Engineering and Environment Faculty
Reservoirs, Archives of Hydro-Engineering and Environ- Gdask University of Technology
mental Mechanics, Vol. 61 (2014), No 34, pp. 113123 11/12 Narutowicza St.
80 - 233 Gdask
21. Zima, P., Mkinia, J., Swinarski, M., Czerwionka, K.: Poland
Combining Computational Fluid Dynamics with a
Biokinetic Model for Predicting Ammonia and Phos-
phate Behavior in Aeration Tanks, Water Environment
Research, Vol. 81, No. 11 (2009), pp. 23532362 .

POLISH MARITIME RESEARCH, No S1/2017 229

 POLISH MARITIME RESEARCH Special Issue 2017 S1 (93) 2017 Vol. 24; pp. 230-241
10.1515/pomr-2017-0043

DESIGN OF STRUCTURE OF TENSION LEG PLATFORM FOR 6 MW

OFFSHORE WIND TURBINE BASED ON FEM ANALYSIS

Jdrzej ywicki
Pawe Dymarski
Ewelina Ciba
Czesaw Dymarski
Gdansk University of Technology, Poland

ABSTRACT

The article presents the calculation and design stages of the TLP platform serving as a supporting construction of a
6 MW offshore wind turbine. This platform is designed to anchor at sea at a depth of 60 m. The authors presented
the method of parameterization and optimization of the hull geometry. For the two selected geometry variants, the
load and motion calculations of the platform subjected to wind, wave and current under 50-year storm conditions
were performed. The maximum load on the structure was determined in these extreme storm conditions. For these
loads, the MES calculation of the designed platform was performed for the selected variant. Authors have presented a
method for calculating maximum wind, wave and current stresses on the structure during the worst storm in the past
50 years. For these loads the MES endurance calculations of the designed platform were made. Based on the results of
these calculations, the required structural changes and recalculations have been made in succession to the structural
design of the platform, which meets the design requirements and has the required ad hoc strength. The article contains
stress analysis in difficult nodes of constructions and discusses ways of solving their problems. The work is part of the
WIND-TU-PLA project from the NCBR Research Agreement (Agreement No. MARTECII / 1/2014).

Keywords: offshore wind turbine, TLP, support structure, FEM

INTRODUCTION
In recent years, more and more rapid development in
the field of offshore wind turbines has been observed. More
attention is being paid to reliability, durability and energy
efficiency. For this purpose, specialized software using FEM
and CFD methods [1, 2, 3, 4, 5, 6, 7] is used. With estimates
of the analysis results, that for waters deeper than 50 meters,
cost-effective solution is the use of floating structures. There
are already solutions of floating structures, which are going to
be used on an industrial scale, mainly buoyancy stabilized [8]
or ballast stabilized [9, 10] structures. In recent years, there
are more and more publications including computational
analysis of TLP-type of foundation systems for offshore wind
turbines [11, 12, 13]
The works presented in this article were carried out under
the WIND-TU-PLA acronym project from the MARTEC
II Era-Net program. The projects objective is to develop
foundation systems for the southern Baltic basin where
areas have been pre-designated where wind turbines are to
be installed.
This article presents a design dedicated for 50m + depth
water. The subject of analysis is the Tension Leg Platform.
According to the assumptions, it is a buoyancy structure,
designed for a depth of 60 m, fixed to the bottom by tight ties
and allows the installation of a turbine of 6 MW.
The available articles [11, 12, 13], as well as the classification
societies [14], show that many issues related to the design of

230 POLISH MARITIME RESEARCH, No S1/2017

floating platforms (and in particular TLP structures) still HULL GEOMETRY DESIGN. PARAMETRIC
require improvement or refinement. OPTIMIZATION
Due to the fact that in industry magazines hardly
any detailed information about the design and strength The visualization of the model and the main geometrical
calculations of this type of construction can be found, the assumptions are shown in Figure 1. The design assumption
authors undertook to solve this task within the framework was to install a 6MW turbine, pre-modelling on the SENVION
of the above project. The calculations take into account all 6.2M turbine [15]. Table 1 shows the basic technical parameters
the significant loads of an object, namely the weight of the of the reference turbine as well as the 5MW (NREL Reference
structure, including the turbine tower, as well as hydrostatic Wind Turbine) [16] and the values that were used for the test
forces and aerodynamic and hydrodynamic forces acting calculations.
on the whole object during extreme operating conditions, Tab. 1. The main characteristics of the wind turbines
restoring forces of the anchorage system and the inertia
force acting on the mass of the structure during accelerated
Design [18] NREL 5MW Test case
movement.
This article presents the stages of designing a TLP structure Rotor Type Senvion 6.2M NREL 5MW NREL 5MW
based on an ad hoc strength analysis. The analysis of fatigue
strength, due to a much wider range of work, will be presented Rated Power 6.15 MW 5 MW 5 MW
in a separate publication. Cut in Wind Speed 3.5 m/s 3.0 m/s -
Rated Wind Speed 14.0 m/s 11.4 m/s 11.4 m/s*
Cut out Wind Speed 30 m/s 25 m/s 25 m/s**
Rotor Diameter 126 m 126 m 126 m
Rotor Operational
7.7 12.1 rpm 6.9 12.1 rpm 12.1 rpm
Interval
Thrust coefficient
Not available 0.730 0.730*
(rated) CT,11.4 m/s
Thrust coefficient (cut
Not available 0.074 0.074**
out) CT,25 m/s

Rotor+Nacelle Mass 460 t 350 t 460 t

Tower Height 95 m 90 m 95 m

* Calculations using the simplified (parametric) method, 1 DoF

** Calculations were made using a method based on the Morison equation, in
six degrees of freedom, taking into account the stiffness / elasticity of the leg (s)

In order to determine the optimal dimensions of the

platform, a simplified parametric analysis was carried out,
influenced by the shape parameters: the height of the pontoons
Ha, the volume displacement Vb and the length of the tendons
for the main parameters describing the dynamics of the
platform. The main geometry parameters of the platforms
are shown in Figure 2.

Fig. 1. Main design dimensions. Fig. 2. Parameters describing the geometry of the platform.

POLISH MARITIME RESEARCH, No S1/2017 231

 The computational model used for parametric analysis is
based on the equation of motion along the horizontal axis Ox,
which coincides with the direction of the force (wind direction
and wave direction) [17]. Equation (1) is based on the second
principle of dynamics, with the mass of the structure being
increased by the mass of added water. On the right side of
the equation there is the force of resistance, proportional to
the square of the velocity of the body relative to water, and
a part representing the force arising from the wave action
of the body.
(1)
- Changing the length of the Lt strings (by appropriate
pontoon formation or changing the depth of the water area
d) will result in a change in the restoring force characteristics
(equation (2)), and hence the change in 0.
Calculations were made for a wave corresponding
to a 50-year storm: Hs = 9.01m, Tp = 11.3s [20], assuming
unfavourable wind force Uw = 11.4 m / s, because the thrust
coefficient of the turbine was CT, 11.4m / S = 0.73 (see table 1).
These are the operating conditions at which the thrust on
the turbine reaches its maximum value.
As a result of parametric analysis, the maximum amplitudes
of the acceleration (longitudinal) and amplitude of the forces
in the legs were obtained.

where: ma added mass, density of sea water, CD drag

coefficient of submerged part of the structure [17], Ap projected
Area of the structure on a plane normal to the motion direction
(Ox), U velocity of surrounding water (based on Airy wave
theory), Fx,aero total aerodynamic force, Fx,rest restoring force
due to the mooring system, given by the following formula:

(2)

where g is a gravity.
Added mass ma and drag coefficient CD has been
approximated using simple empirical formulas for 2D sections Fig. 3. Maximum amplitudes of horizontal acceleration as a function of Ha
[17]. Added mass for the pontoon ma,a and for a cylindrical and Vb. Lt=45m; rt=25m; d=60m; T0=20m.
column ma,c are calculated from the formulas:

(3a)

(3b)

where: Dc - column diameter, c1 coefficient dependent on

the proportion of rectangular pontoon section, for a section
close to the square c1 = 1.51.
la,p the total length of the pontoon, projected on a plane
perpendicular to the direction of inflow, hc,s the height of
the immersed part of the column.
Parameters have been selected in such a way that, by
changing them, they have a specific effect on the dynamics of
the object. By analysing the parts of equation (1), and equation Fig. 4. Minima of tendons tensions as a function of Ha and Vb . Negative
(2) and equation (3a), one can conclude that: values, mean compressive strength of tendons (loss of stability). Lt=45m;
rt=25m; d=60m; T0=20m
- Changing the hull volume V b will result in an increase in
the forcing force (the second part to the right of the equation
(1)) and the increase in the restoring force (equation (2)). Figures 3 and 4 show graphs of accelerations and
Changing the restoring force will change the frequency of minimum forces in legs as a function of the height of the
the oscillation 0 of the platform, arms Ha and volume displacement V b. The optimal solution
- Changing the height of the pontoon Ha will change the is characterized by minimum amplitudes of acceleration (less
mass of the accompanying water (equation (3a)) and change construction loads). The correctness of the solution is that the
the resistance force by changing the surface Ap (the first part forces in the legs are always stretching, because in the case
of the right side of equation (1). The change in the force of of compression, as shown in Figure 7c, the platform will lose
resistance will also change the force of the oscillation of the its stability and begin to turn over. Taking into account the
object. above limitations, two solutions were chosen from the area
of permissible solutions (Table 2, Figure 5), with the weight

232 POLISH MARITIME RESEARCH, No S1/2017

of both variants estimated on the basis of the pre-selected
plating thickness assuming that the mass of the stiffeners
and girders is about 20% of the plating weight.
Tab. 2. Main parameters of selected platform variants
TLP 1_5
Volume of submerged part of the platform,
5360 m3
Vb
(Mean) height of arms (pontoons), Ha 7.64 m
Length of tendons (legs), Lt 42.4 m
Radius of tendons distribution, r t 25m
Mass of the structure (first approximation),
1800 t
m coordinate system and to the system associated with the
Fig. 5. Visualization of the discussed platform variants.
Both TLP platform variants have been subjected to detailed
computational analysis using Morisons [17, 18] software,
Boundary Element Method, and RANSE-CFD.
A quick and relatively accurate method is to calculate
the motion using the Morison equation. The structure is
divided into segments, (based on the literature [17]) for which
the coefficients of resistance and the added mass coefficient
are determined. The simplified hydrodynamic model
(segmentation) is shown in Figure 6. The hydrodynamic
force y1 component in the local coordinate system (Figure
6), on the i-th segment, is calculated using the formula [21]:
(4)
TLP 2_1
5500 m3 where CD drag coefficient on the y1 (local coordinate
system), fluid density, Ap - the surface segment on the
8.97 m direction y1, w the velocity component for the water on
the direction y1, V b segment volume, CA ratio of mass of
47.4 m accompanying water, - velocity and acceleration of
the segment in direction y1.
25m
The composition of the hydrodynamic force z1 is calculated
similarly.
Designated forces are transformed into the global
1900 t
object. Based on the knowledge of the forces and positions of
the segments, the total moment of these forces is calculated
for the whole object. Knowing the forces and moments allows
to solve the system of equations describing the movement of
an object (in six degrees of freedom).
Fig. 6. Division of structure (incl. Turbine) into segments for which forces are
calculated (see equation 1) with the local system shown.
Calculations, like the above, were performed for 50-year
storm conditions [18]. However, due to the fact that, at high
amplitudes of motion, the platform lost stability at the
maximum rated wind speed (Uw = 11.4m/s; CT,11.4/s = 0.73). It
was decided that the calculations would be performed for the
conditions Uw = 25m/s, with thrust coefficient CT,25m/s =0.074.
It is envisaged that under extreme wave conditions, the wind
turbine will operate at a reduced power setting of ~ 8m/s to
~ 25m/s in order to reduce the thrust.
Motion charts of both platforms (position versus time)
are shown in Figures 7a, 7d. They show the waveform,
acceleration, and force functions of the platform during the
first 1200 seconds of a one-hour simulation. It is clear from
the above that the amplitudes of these values are smaller for
POLISH MARITIME RESEARCH, No S1/2017 233
 Fig. 7. Graphs of displacement, acceleration and forces of the tendons to the TLP 15 of the (a, b, c) and TLP 2_1 (d, e, f). Graphs were made for 1200 seconds of
simulation.
platform 2_1 than for l5. Therefore, geometric parameters
of variant 2_1 were chosen as assumptions for the design.
For the design of the platform construction, it was
necessary to select a load system for which the design would
be maximized.
The graphs shown in Figure 7 include the first 1200 seconds
of the simulation. Based on the results for the full hour, a
point was chosen for the time at which the reaction forces in
the legs of the platform (one leg) would reach the maximum.
The resulting load distribution along the vertical axis of
the structure is shown in Fig. 10.
DESCRIPTION OF THE PRELIMINARY
DESIGN
The illustrated support structure was designed as a Tension
Leg Platform. The main part of the structure is an underwater
tricot fuselage with a centrally located column on which a
tower with a wind turbine is attached. The underwater volume
234 POLISH MARITIME RESEARCH, No S1/2017
of about 5500m3, the height of the arms, the length of the
anchor rods were chosen so that despite the assumed loads the
construction of the tendons was always stressed - stretched.
The draft is 20m. The hull height of the regular platform is
10 m. At the ends of the arms 25 m apart from the centre of
the main column there are string fixing nodes with which the
platform is fixed to the bottom. A conical column with a base
diameter of 8m is ended with a 5.5 m diameter fixing flange
at a height of 15 m above the free surface area. The whole
structure is made of steel. The weight of the initial variant of
the platform itself is 1430t. Figure 8 shows the prefabrication
system of the construction variant together with the detail
of the thicknesses of the individual components.
As the tendons, pipes having an outer diameter of 800
mm and a wall thickness of 40mm are proposed. This type
of tendon offers benefits in terms of relatively low prices
compared to e.g. chains and ease of transport by towing
them as displacement elements. One of the problems faced
by the concept makers was to fix rigid strands to the platform
hull. Based on preliminary analyses, it was determined that

Fig. 8. Preliminary design of the structure.
Figure 9. Platform solid model.
attaching the pipes to the hull should allow the tendon to tilt
at an angle of 17 from the vertical. As part of this task, the
concept of a plain bearing has been developed to allow the
platform to move in any direction. The fixation assumes the
construction of a bearing with a polymeric shell that is part
of the sphere. The materials used allow the surface pressure
of 70 MPa, the contact surface is selected so as not to exceed
this value even under extreme operating conditions. The
final part of the platform arms has been designed to allow
the assembly of the bearing components and
ultimately enable the anchoring process of the
platform. Figure 9 shows the solid model of the
platform showing the fragment of the spherical
canopy that is luminaire for the polymer bearing
bush. From the bottom it connects to a specially
shaped cone allowing the pipe to move freely
and to pivot at a predetermined angle. From the
upper deck, access to the bearing allows a steel
sleeve with a diameter of 2400mm. The whole
node was reinforced by radial stiffeners. The
method of bearing and anchoring process has
been developed for internal project reports, but
is not discussed in more detail in this article.
LOAD AND OPERATING
CONDITIONS OF THE SUPPORT
STRUCTURE
The following load and operating conditions
of the support structure are the same for each
stage of the modification of the construction
geometry shown in the work. Presented results
of individual simulations were obtained for the
following assumptions. As previously mentioned,
the authors assume that the structure will be
located in the Baltic Sea area at a depth of 60
m. The top flange to which the column will be
mounted is located at a height of 15 m above
the water surface. Structural loads derived from
environmental conditions were calculated by
hydromechanical simulation taking into account
the most unfavourable conditions corresponding
to the 50-year storm in the Baltic Sea [18]. The
most unfavourable conditions are considered
to be the load system that causes maximum
stress in one of the anchor anchoring strings
of the platform, while in the opposite legs the
stresses reaches the minimum. This situation
is most often the case when the forces of inertia
caused by the accelerated motion of the platform
acting on the massive elements (nacelle, rotor
and tower) have a vector worthy of the direction
of the aerodynamic load generated mainly on
the turbine.
The technical characteristics of the turbine
used to determine the foundation load are shown
in Table 3:
Based on the hydromechanical simulation described in
Chapter 2, the course of horizontal load derived from currents
and waves and the wind as a function of the height of the object
is determined. For computation, the MES model was divided
into segments that are the same as for hydromechanical
simulation. Thanks to this solution, the model can be easily
loaded by applying the average values of the horizontal load
obtained from the hydromechanical calculations to the given
POLISH MARITIME RESEARCH, No S1/2017 235
area of construction. The zero altitude was assumed at the
level of the water mirror. The value of the continuous load
acting on the height of the structure is shown in Figure 10.
Tab. 3. Technical parameters of the turbine
Turbine power:
The height of the turbine hub MASL
Turbine diameter
Mass of nacelle and rotor
Column mass
MES MODEL
The surface model of the structure shown in Figure 8
was created in Autodesk Inwentor. Next, the geometry was
imported into HyperMesh, where the FEM model was built.
The hull, together with the internal design elements, was
6 MW mapped with first-order quadrilateral and quadrilateral
elements with a mean size of 90 mm. Due to the fact that
95 m
the entire structure is modelled, the platform superstructure
with the tower was modelled with 200 mm elements, which
126 m
significantly reduced the total number of components. The
elements used for the first iteration are summarized in Table
462 t
4, the grid fragments shown in Figure 11.
458 t Tab. 4

A significant factor affecting load is the acceleration of the
horizontal movement of the structure due to aerodynamic
and hydrodynamic forces and forces from the anchorage
system. The maximum acceleration from a one-hour
hydromechanical simulation determined from the motion
equation for an irregular wave was 3.41 m/s2, which was read
from the graph shown in Figure 6e.
Hydrostatic pressure of seawater acts on the structure
from +10 m from the average water level (wave crest) to the
bottom of the structure (z = -20 m). The sea water density
was taken as 1026 kg/m3. The adopted load model is designed
to simulate conditions where the wave crest reaches a height
of approximately 10 m at high waves, which results in
asignificantly higher hydrostatic pressure on the structure
than in the absence of waves.
Figure 10 shows the forces from the turbines thrust and
the inertial forces acting on the nacelle and rotor masses
that are present in the discrete model (FEM) in the form
of concentrated force. The drawing does not show forces
from the anchoring system, as they appear in the model as
responses on the supports.
Degrees of
2D elements 1D elements
freedom
Perfectly
Square Triangular
Rigid
Number 694579 10133 2129 4144783
Average size 90/200 90/200 2750 -
In order to apply forces from the turbines thrust and the
mass of the nacelle and rotor on the top of the column - 95m
above the level of the free surface, a central node was created.
It has been modeled by connecting nodes on the circumference
of the top edge of the column in one central node with rigid
elements. Below is the model of support. At the end of each
of the platform arms there is a fragment of the sphere that is
ultimately a spherical ball joint that connects the platform
with a tight string. In the centre of the sphere with a radius of
1400 mm a central node has been formed, where all the nodes
of the elements on the surface of the sphere are connected by
rigids. Thanks to this solution, three nodes were created, in
which the degree of freedom of the model was appropriately
removed. The arm node in the Y direction has been locked
so as to prevent movement in the three directions X, Y, Z and
rotating movement relative to the vertical
axis, while the other two nodes have been
disabled to move in the Z direction. The
solutions described are shown in Figure 11.

MODEL THE MATERIAL

A linear-resilient material model

has been adopted with the following
properties:

x Youngs module E = 210 GPa

x Poissons number = 0.3
x Density = 7860 kg/m3

Fig. 10. Load acting on platform and tower.

236 POLISH MARITIME RESEARCH, No S1/2017

 calculation data. This load generates a bending
moment, which is a significant load on the
structure. The assumed load distribution has
no significant effect on the local strength of
the structure, and significantly facilitates the
preparation of the computational data.
Earth acceleration is applied to the opposite
redirection in the Z direction. In addition, in
the Y direction, all the elements of the structure
have a predetermined acceleration of 3.41 m/s2.
The values of generated inertial forces acting
on a given segment of a structure are shown in
Figure 10.
Hydrostatic pressure was applied in the
normal direction to the plating elements of the
platform and the column in accordance with the
assumptions in Chapter 4.

MES CALCULATIONS
Structural strength calculations have been
made in the linear range. HyperWorks v13
software, manufactured by Altair was used.,
HyperMesh software was used as a preprocessor.
Calculations were made using the Optistruct
Fig. 11. Model fixation.
solver, while results were generated using the

LOAD IMPLEMENTATION

According to the above mentioned, the MES

model was put under load at the centre point
at the top of the wind tower with the following
forces:
vertical force acting on the reverse Z
direction and 4.53MN from the mass of
the nacelle and rotor;
a transverse force acting on the Y direction
of 0.45 MN being the value of the turbines
thrust;
Transverse force acting on the Y direction
of 1.57 MN as the inertia mass of the
nacelle and rotor mass for the calculated
acceleration of the structure of 3.41 m/s2.

The hydrodynamic load of the structure

according to the assumptions presented in the
previous chapter was applied to the nodes in the
Y direction. The model grid was split at the level
of the segments from which the node numbers
were read. Each of the segment nodes applied
with the force that was determined by dividing
the CFD simulation by the total load of the
segment by the number of nodes. The accepted
way of implementing the load has no real
influence on the local strength of the structure,
Fig. 12. Equivalent stresses [MPa]
and significantly facilitates the preparation of the

POLISH MARITIME RESEARCH, No S1/2017 237

HyperView postprocessor. The simulation was performed x The passage of the lateral plating of the central part
on a PC equipped with a quad-core 64-bit Intel i7 processor of the platform and arm has been thickened over
with 3.40 GHz, 16 GB of DDR3 RAM and an 840 PRO SSD. the whole circumference. An increase in the sheet
Calculations for construction in the final stages of geometry thickness to 40 mm included the plating band
modification, where the number of surface finite elements was stretched between the two frame stiffeners shown
about 0.8 million lasted about 10 minutes. in Figure 13.
x The thickness of the plating in the centre column
area was increased to 30 mm.
CALCULATION OF THE STRENGTH OF
THE PRE-CONSTRUCTION
The calculations made for the initial form of the structure,
shown in Figure 8, were meant to indicate the critical areas
of the proposed structure. It should be added that for the
previously described method of loading the structure, the
most stressed part of the structure is the arm in the direction
of the transverse force Fy. In this case, the highest bending
moment is generated on the indicated arm of the platform
in comparison to the other arms.
In the above figure it can be seen that the region of
occurrence of the greatest stress, which reaches 448 MPa is
the portion of the hull at the junction node of the primary
arm mounting structure (det. A on FIG. 12). The stress
concentration is also visible on the web of the main girder
connecting the arm portion of the regular fixing location of
the tendon. In this case, the cause of high stress is too rapid
change in the shape of geometry of such a large load which
are local forces in the tendons. Moreover, considerably higher
than the yield strength of the carbon steel (235 MPa) present
in the webs and agents frames in the regions of changing the
direction of the stiffener.

Fig. 13. Construction changes - first stage.

MODIFICATIONS OF CONSTRUCTION
The calculation results for the modified support structure
Because of the high stress values at the points of the are shown in Figure 14. The scale has been set so that the red
analyzed structure, there has been a need for modifications colour indicates the stresses above 235 MPa.
in the most stressed areas where the stresses exceed the
permissible values for normal strength steel. The geometric
changes in the model introduced in the first modification
step are described below. The modified design is shown in
Figure 13.
x Based on the results obtained, it was found that
the design at the transition site of the regular arm
shape into the bearing seat should be significantly
redesigned. It was decided that the sloping section
of the inclined plating to the horizontal support
plate for the stiffening of the bearing shell must be
rounded to avoid high stress concentrations at the
node. This plating was joined together through a
rounded sheet of 40 mm thick and with a bending
radius of 1500 mm.
x The shape of the central web is adjusted to the
rounded shape of the above-mentioned plating. In
addition, the thickness of the web at this point was
increased from 20 mm to 40 mm.
Fig 14. Equivalent stresses after changes [MPa].

238 POLISH MARITIME RESEARCH, No S1/2017

 The changes made reduced the maximum stress reduced
to about 316 MPa, also, the area of stresses beyond the yield
strength decreased. In the figure above, it can be observed
that, as expected, modifications introduced in the structure
restriction region have worked very favourably, distributing
the stresses in a mild manner.
Maximum stresses occurred in the corner connecting
the side wall plating and the bottom of the platform. The
cause of the concentration in this area is the sharp change
of shape at the point of contact of the falling platelets of the
side plating of the regular platform with the lateral arm of
the arm. Visible areas of the high stress concentration are
still the frame parts. On the basis of the results obtained, the
geometry was to be transformed further, thus reducing the
stresses in excess of 235 MPa.
analysis, it can be stated that the highest stresses of 286 MPa
occur only on the individual finite elements of the model in
the region of joining the top shell to the column. It should be
noted here that calculations using coating elements (2D) in
areas of sudden shape change may exhibit unnatural stresses.
As a result, the authors allow for a slight excess of stresses
in the belt of one element width from the edge joining the
two parts of the structure. In order to obtain more reliable
results at the junctions of major structural parts, additional
local analyses of selected regions using 3D elements should
be performed. Ensuring structural strength in areas where
the stress exceeds 235 MPa, for the assumed load system,
will be achieved by using high strength steels. The maximum
displacement that occurred on the turret mount flange was
115mm.

FINAL STAGE OF MODIFICATION.

VERIFICATION CALCULATIONS.
The last modification shown in Figure 15 has led to a
structure that meets the criteria in terms of strength. Due
to still high stress in the area of the central column passage
through the top plating, it was decided that this section of
the column would be a 70 mm thick ring and 2 m high -
deposited 1 m on either side of the plating. Thickness of the
conical column has been increased to 60mm. Enclosures in
more stressed areas (blue) were given a thickness of 36mm.
The brackets of the frame stiffeners were modified. In areas
of rounded passages, inserts of thickness 50 mm were used,
in addition, these places were supported by brackets. In the
longitudinal girders, the relief openings have been made in
the less stressed webs significantly reducing the mass of the
longitudinal stiffeners. The total weight of the structure after Fig.16. Reduced stress [MPa]
the variations shown was 1456 t.

Fig.17. Displacement [mm]

Fig. 15. Support structure after optimization.

Figures 16 and 17 show the stresses and deformations of

the final version of the support structure. Based on their

POLISH MARITIME RESEARCH, No S1/2017 239

 SUMMARY
The present article deals with current and complex issues
related to the design of marine facilities under high power
wind farms. The authors undertook the task of designing a
preliminary steel variant of the TLP platform designed for
anchoring on the Baltic area with a depth of about 60 m. The
next steps in the design of the platform design were presented,
which was to provide ad hoc strength at maximum load that
could occur during marine operation. Preliminary strength
calculations indicated the critical structural nodes that had
to be modified. After the changes described in Chapter 10 on
structural modification, the stress areas exceeding the limit
values for normal strength steel decreased considerably. The
next step was the changes in the thickness of the plating and
stiffeners to larger ones in the stressed areas and the reduction
of the thickness in the unloaded areas. The authors are aware
that for this type of construction designed as a marine wind
turbine support structure, fatigue calculations are required
to take account of the impact of environmental loads. Such
calculations will be carried out in the next stage of the project.
In the presented platform more than 90% of the weight
of the structure can be made of normal steel hull strength
(class NV B - Re235 in accordance with [19]). It raises the
advantages of this project, which has already attracted interest
among representatives of investors of wind farms. In the
most demanding areas, NV-class steel, E 36 - Re 355 as per
[19] must be used.
According to the authors, this article presents a number of
issues related to the design of steel naval constructions very
useful for designers, researchers and investors of this type of
objects. In addition, it allows to estimate the level of technical
and material costs related to their execution.
ACKNOWLEDGMENT
FEM analysis has been performed with HYPER WORKS
software. Calculations were carried out at the Academic
Computer Center in Gdask (TASK).
This research was supported by The Polish National Centre
for Research and Development (NCBR) under the project
WIND-TU-PLA ERA-NET MARTEC II (Agreement No.
MARTECII/1/2014)
REFERENCES
1. M. Deja, M. S. Siemitkowski: Feature-based generation of
machining process plans for optimised parts manufacture.
Journal of Intelligent Manufacturing August 2013, Volume
24, Issue 4, pp 831846.
2. Niklas K., Kozak J.: Experimental investigation of Steel
ConcretePolymer composite barrier for the ship internal
tank construction. OCEAN ENGINEERING. Vol. 111,
(2016), s.449-460
240 POLISH MARITIME RESEARCH, No S1/2017
3. Hirt ., Lampart P.: Complex multidisciplinary
optimization of turbine blading systems. ARCHIVES OF
MECHANICS. Vol. 64, nr. 2 (2012), s.153-175
4. Sabik A., Kreja I.: Large thermo-elastic displacement
and stability FEM analysis of multilayered plates and
shells. THIN-WALLED STRUCTURES. Vol. 71, (2013),
pp.119-133.
5. Kahsin M., XF]DN M.: NUMERICAL MODEL
QUALITY ASSESSMENT OF OFFSHORE WIND
TURBINE SUPPORTING STRUCTURE BASED
ON EXPERIMENTAL DATA. Structural Health
Monitoring 2015: System Reliability for Verification and
Implementation: Proceedings of the 10th International
Workshop on Structural Health Monitoring.- Vol. 1/ ed.
Fu-Kuo Chang, Fotis Kopsaftopoulos 439 North Duke
Street Lancaster, PA 17602-4967, U.S.A. : DEStech
Publications, Inc., 2015, s.2817-2824
6. Kahsin M., Luczak M., Peeters B.: Use and
assessment of preliminary FE model results within
testing process of offshore windturbine supporting
structure. EURODYN 2014: IX INTERNATIONAL
CONFERENCE ON STRUCTURAL DYNAMICS, %RRN
6HULHV(852'<1,QWHUQDWLRQDO&RQIHUHQFHRQ6WUXFWXUDO
'\QDPLFV, 3DJHV
7. XF]DN M., Manzato S., Peeters B., Branner K., Berring
P., Kahsin M.: Updating Finite Element Model of a Wind
Turbine Blade Section Using Experimental Modal Analysis
Results. SHOCK AND VIBRATION. -Vol. 2014, iss. 1
(2014), s.71-82
8. Brommundt M., Krause L.,Merz K., Muskulus M.: Mooring
system optimization for floating wind turbines using
frequency domain analysis. Energy Procedia 24 (2012)
289296
9. Karimirad M., Moan T.: A simplified method for coupled
analysis of floating offshore wind turbines. Marine
Structures 27 (2012) 45-63
10. Jeon S.H., Cho Y.U., Seo M.W., Cho J.R., Jeong W.B.:
Dynamic response of floating substructure of spar-type
offshore wind turbine with catenary mooring cables. Ocean
Engineering 72 (2013) 356364
11. Bachynski E.E., Moan T.: Design considerations for tension
leg platform wind turbines. Marine Structures 29 (2012)
89-114
12. Adam F.,Myland T., Schuldt B., Gromann J., Dahlhaus
F.: Evaluation of internal force superposition on a TLP for
wind turbines. Renewable Energy, Volume 71, November
2014, Pages 271275
13. Adam F., Ritschel U., Plumridge E., Gromann J.:Pre-
Design of a TLP steel-concrete composite substructure for CONTACT WITH THE AUTHOR
a 6 MW wind turbine as a way to essential cost-reduction.
Conference proceedings RENEW, 2016, Lisbon Jdrzej ywicki

14. DNV-OS-J103 Design of Floating Wind Turbine Structures. e-mail: jedzywic@pg.gda.pl

JUNE 2013
Pawe Dymarski
15. https://www.senvion.com/global/en/wind-energy- Ewelina Ciba
solutions/ wind-turbines/6xm/62m126/ Czesaw Dymarski

16. Jonkman J., Butterfield S., Musial W., Scot G.: Definition Gdask University of Technology
of a 5-MW Reference Wind Turbine for Offshore System 11/12 Narutowicza St.
Development. Technical Report NREL/TP-500-38060 80 - 233 Gdask
February 2009 Poland

17. Sarpkaya T.: Wave forces on offshore structures, Cambridge

University Press, 2010

18. Dymarski P., Ciba E., Marcinkowski T.: Effective method

for determining environmental loads on supporting
structures for offshore wind turbines. POLISH MARITIME
RESEARCH 1(89) 2016 Vol. 23; pp. 52-60 10.1515/
pomr-2016-0008

19. DNV-OS-J101 Design of Offshore Wind Turbine

Structures. MAY 2014

POLISH MARITIME RESEARCH, No S1/2017 241