Académique Documents
Professionnel Documents
Culture Documents
AC/323(AVT)TP/9
NORTH ATLANTIC TREATY ORGANIZATION
10
Q.
6i-
Papers presented and discussions recorded at the RTO Applied Vehicle Technology Panel (AVT)
Symposium (organised by the former AGARD Fluid Dynamics Panel), held in Amsterdam, The
Netherlands, 5-8 October 1998.
m
DISTRIBUTION STATEMENT A:
Approved for Public Release -
19990315 026
Distribution Unlimited
®
Printed on recycled paper
Copyright © RTO/NAT01999
All Rights Reserved
ISBN 92-837-0004-X
$Q f99-06-/PS-
Fluid Dynamics Problems of Vehicles Operating
Near or in the Air-Sea Interface
(RTO MP-15)
Executive Summary
This Symposium concentrated on issues associated with vehicles operating near the sea surface, issues
common to aeronautics and hydrodynamics, issues involving air-sea interactions; and on providing the
opportunity of bringing together scientists and engineers from Western Europe, North America,
Australia, Russia, and Ukraine.
One-third of the papers were authored by scientists from Russia and Ukraine. The two major topics
covered during the Symposium included: Aerodynamics around Ships, and Non-Classical Aircraft
flying near the air-sea interface i.e. Ekranoplanes. In addition, there were papers on Surface Effect
Ships, Ship Stabilization, Hydrofoil Boats, Underwater Missile Launch, and Ship Bow Waves and
Water Entry.
Aerodynamics around ships is militarily important because ship superstructures create highly unsteady
3-D flows with massive regions of flow separation behind them, and this situation can create severe
problems for the landing of helicopters on the deck. It was very apparent from the presented papers that
current means of calculating these flows are developing, but they are currently very limited. Many
research opportunities exist involving unstable and separated flows and their simulation.
Several papers covered ground effect aerodynamics, specifically Russian technology in this area, as
evidenced by the development of Ekranoplanes. By utilizing ground effect, the gap between slow and
inexpensive ships and fast but expensive aircraft can possibly be filled. However, economic studies,
which include sea state operability, need to be performed. An innovative means of reducing drag during
take-off and thus enabling a reduction in engine power requirements was presented. There were
proposals to use Ekranoplanes in a grand air/sea rescue system, and also a proposal for a future
Symposium on "Marine Vehicles for Rapid Disaster Response".
For vehicles operating in the air-sea interface, load control and vertical control are very difficult and
extraordinarily important. It is necessary to keep vertical accelerations very small, under 1/1 Oth of a G
and, therefore, expectations of 1/4 of a G for some Surface Effect Ships and Hydrofoils present a very
serious problem. A new implementation of an automated moving weight scheme together with
sophisticated fins, which reduced ship motions on an Air Craft Carrier, while ensuring navigational
capability, was presented.
The material assembled in this report was prepared under the combined sponsorship of the NATO
Partnership for Peace Program, the RTO Applied Vehicles and Technology Panel, the United States
Office of Naval Research International Field Office - Europe, and the United States Air Force European
Office of Aerospace Research and Development.
Problemes de dynamique des fluides
des vehicules evoluant dans ou pres
de Pinterface air-mer
(RTO MP-15)
Synthese
Ce symposium a porte essentiellement sur l'exploitation de vehicules evoluant ä proximite de la surface
de la mer. II s'agit de questions relevant ä la fois de l'aeronautique et de l'hydrodynamique, avec des
interactions air-mer. Des scientifiques et des ingenieurs de l'Europe occidentale, de FAmerique du
nord, de l'Australie, de la Russie et de l'Ukraine ont pu etre reunis ä cette occasion.
Un tiers des communications a ete presente par des scientifiques russes et ukrainiens. Les deux
principaux sujets couverts par le symposium comprenaient: l'aerodynamique ä proximite des navires et
les aeronefs non-conventionnels evoluant ä proximite de la surface de la mer, c'est ä dire les
ekranoplanes. En outre, des communications ont ete presentees sur les navires ä effet de surface, sur la
stabilisation des navires, sur les navires hydropteres, le lancement de missiles ä partir de sous-marins,
les vagues d'etrave et la penetration dans l'eau.
Les conditions aerodynamiques ä proximite des navires sont d'un grand interet militaire, car les
superstructures creent, en aval, des ecoulements tridimensionnels fortement instationnaires avec des
zones de decollement tres importantes, ce qui pose parfois de serieux problemes pour l'appontage des
helicopteres. Des communications presentees, il est apparu tres clairement que des methodes de calcul
de ces ecoulements sont en cours de developpement mais restent tres limitees. De nombreuses
possibilites de recherche existent, centrees autour des ecoulements instationnaires et decolies et autour
de leur simulation.
L'aerodynamique des effets de sol a fait l'objet de nombreuses communications. La technologie russe a
bien ete mise en evidence avec le developpement des ekranoplanes. Par le biais des effets de sol, il se
pourrait que l'ecart qui existe entre les navires lents mais peu coüteux, et les avions rapides mais chers,
puisse etre elimine. Cependant, cela impliquerait la realisation d'etudes economiques, entre autres sur la
tenue ä la mer. Un moyen novateur de reduire la trainee au decollage, permettant ainsi de reduire les
besoins en propulsion, a ete presente. Des projets ont egalement ete proposes visant l'utilisation
d'ekranoplanes dans un vaste Systeme de sauvetage en mer, ainsi qu'une proposition de symposium sur
«les vehicules maritimes pour une reaction rapide en situation de catastrophe ».
Pour les vehicules ä proximite de la surface de la mer, le contröle de la charge et le contröle vertical
sont extremement difficiles et importants. II faut reduire les accelerations verticales au minimum
(moins d'un dixieme de g) et, par consequent, les previsions d'un quart de g, pour certains navires et
hydropteres ä effets de surface, posent un probleme serieux. Une nouvelle configuration a ete presentee,
comprenant un Systeme automatise de poids amovible, comportant des derives sophistiquees, qui
permet de reduire les mouvements d'un porte-avions, tout en conservant 1'aptitude ä la navigation.
Ce rapport a ete elabore sous l'egide conjointe du Programme OTAN de Partenariat pour la paix, de la
Commission RTO de technologies appliquees aux vehicules, du bureau exterieur international - Europe,
du Directorat des Etats-Unis de la recherche navale, et du Bureau europeen de recherche et
developpement aerospatial de l'armee de Fair americaine.
Contents
Page
Executive Summary iii
Synthese iv
Recent Publications of the Former Fluid Dynamics Panel vii
Applied Vehicle Technology Panel Programme Committee ix
Reference
CFD Predictions of the Influence of External Airflow on Helicopter Operations when Operating 2
from Ship Flight Decks
by N.H. Wakefield, S.J. Newman and P.A. Wilson
Higher Order Accurate Solutions of Ship Airwake Flow Fields Using Parallel Computers 3
by L.N. Long, J. Liu and A.V. Modi
Modelling and Simulation of Ship Air Wakes for Helicopter Operations - A Collaborative 8
Venture
by CH. Wilkinson, S.J. Zan, N.E. Gilbert and J.D. Funk
Heel Compensation for the Charles de Gaulle Aircraft Carrier: Principles and Control 14
Structure
by S. Kummer, G. Hardier and C. Lambert
Hydrodynamical Characteristics of an Ekranoplane Wing Flying Near the Wavy Sea Surface 18
by V.G. Byelinskyy and P.I. Zinchuk
The First Commercial Ekranoplan "Amphistar" and Prospects for the Development of 24
Passenger Ekranoplans
by D.N. Sinitsyn and A.I. Maskalik
On the Prediction of Nonlinear Free-Surface Flows Past Slender Hulls Using 2D+t Theory: 26
The Evolution of an Idea
by E. Fontaine and M.P. Tulin
The Complex Boundary Integral Equation Method for a Problem of Entry of a 2D Solid Body 27
in an Incompressible Liquid
by D.I. Cherniy
General Discussion GD
vi
Recent Publications of the Former
AGARD Fluid Dynamics Panel
AGARDOGRAPHS (AG)
Turbulent Boundary Layers in Subsonic and Supersonic Flow
AGARD AG-335, July 1996
Computational Aerodynamics Based on the Euler Equations
AGARD AG-325, September 1994
Scale Effects on Aircraft and Weapon Aerodynamics
AGARD AG-323 (E), July 1994
Design and Testing of High-Performance Parachutes
AGARD AG-319, November 1991
Experimental Techniques in the Field of Low Density Aerodynamics
AGARD AG-318 (E), April 1991
Techniques Experimentales Liees ä l'Aerodynamique ä Basse Densite
AGARD AG-318 (FR), April 1990
A Survey of Measurements and Measuring Techniques in Rapidly Distorted Compressible Turbulent Boundary Layers
AGARD AG-315, May 1989
REPORTS (R)
Fluid Dynamic Research on Supersonic Aircraft
RTO Report EN-4, November 1998
Hugh Speed Body Motion in Water
AGARD R-827, February 1998
Turbulence in Compressible Flows
AGARD R-819, Special Course Notes, June 1997
Advances in Cryogenic Wind Tunnel Technology
AGARD R-812, Special Course Notes, January 1997
Aerothermodynamics and Propulsion Integration for Hypersonic Vehicles
AGARD R-813, Special Course Notes, October 1996
Parallel Computing in CFD
AGARD R-807, Special Course Notes, October 1995
Optimum Design Methods for Aerodynamics
AGARD R-803, Special Course Notes, November 1994
Missile Aerodynamics
AGARD R-804, Special Course Notes, May 1994
Progress in Transition Modelling
AGARD R-793, Special Course Notes, April 1994
Shock-Wave/Boundary-Layer Interactions in Supersonic and Hypersonic Flows
AGARD R-792, Special Course Notes, August 1993
Unstructured Grid Methods for Advection Dominated Flows
AGARD R-787, Special Course Notes, May 1992
Skin Friction Drag Reduction
AGARD R-786, Special Course Notes, March 1992
Engineering Methods in Aerodynamic Analysis and Design of Aircraft
AGARD R-783, Special Course Notes, January 1992
ADVISORY REPORTS (AR)
A Selection of Test Cases for the Validation of Large-Eddy Simulations of Turbulent Flows
AGARD AR-345, April 1998
Ice Accretion Simulation
AGARD AR-344, Report of WG-20, December 1997
Sonic Nozzles for Mass Flow Measurement and Reference Nozzles for Thrust Verification
AGARD AR-321, Report of WG-19, June 1997
Cooperative Programme on Dynamic Wind Tunnel Experiments for Manoeuvring Aircraft
AGARD AR-305, Report of WG-16, October 1996
Hypersonic Experimental and Computational Capability, Improvement and Validation
AGARD AR-319, Vol. I, Report of WG-18, May 1996
Aerodynamics of 3-D Aircraft Afterbodies
AGARD AR-318, Report of WG-17, September 1995
A Selection of Experimental Test Cases for the Validation of CFD Codes
AGARD AR-303, Vols. I and II, Report of WG-14, August 1994
Quality Assessment for Wind Tunnel Testing
AGARD AR-304, Report of WG-15, July 1994
Air Intakes of High Speed Vehicles
AGARD AR-270, Report of WG-13, September 1991
Appraisal of the Suitability of Turbulence Models in Flow Calculations
AGARD AR-291, Technical Status Review, July 1991
Rotary-Balance Testing for Aircraft Dynamics
AGARD AR-265, Report of WG11, December 1990
Calculation of 3D Separated Turbulent Flows in Boundary Layer Limit
AGARD AR-255, Report of WG10, May 1990
CONFERENCE PROCEEDINGS (CP)
Missile Aerodynamics
RTO Report MP-5, November 1998
Advanced Aerodynamic Measurement Technology
AGARD CP-601, May 1998
Aerodynamics of Wind Tunnel Circuits and Their Components
AGARD CP-585, June 1997
The Characterization & Modification of Wakes from Lifting Vehicles in Fluids
AGARD CP-584, November 1996
Progress and Challenges in CFD Methods and Algorithms
AGARD CP-578, April 1996
Aerodynamics of Store Integration and Separation
AGARD CP-570, February 1996
Aerodynamics and Aeroacoustics of Rotorcraft
AGARD CP-552, August 1995
Application of Direct and Large Eddy Simulation to Transition and Turbulence
AGARD CP-551, December 1994
Wall Interference, Support Interference, and Flow Field Measurements
AGARD CP-535, July 1994
Computational and Experimental Assessment of Jets in Cross Flow
AGARD CP-534, November 1993
High-Lift System Aerodynamics
AGARD CP-515, September 1993
Theoretical and Experimental Methods in Hypersonic Flows
AGARD CP-514, April 1993
Aerodynamic Engine/Airframe Integration for High Performance Aircraft and Missiles
AGARD CP-498, September 1992
Effects of Adverse Weather on Aerodynamics
AGARD CP-496, December 1991
Manoeuvring Aerodynamics
AGARD CP-497, November 1991
Vortex Flow Aerodynamics
AGARD CP-494, July 1991
Missile Aerodynamics
AGARD CP-493, October 1990
Aerodynamics of Combat Aircraft Controls and of Ground Effects
AGARD CP-465, April 1990
Computational Methods for Aerodynamic Design (Inverse) and Optimization
AGARD CP-463, March 1990
Applications of Mesh Generation to Complex 3-D Configurations
AGARD CP-464, March 1990
Fluid Dynamics of Three-Dimensional Turbulent Shear Flows and Transition
AGARD CP-438, April 1989
Applied Vehicle Technology Panel
CHAIRMAN:
PROGRAMME COMMITTEE
PANEL EXECUTIVE
Mr. J.K. MOLLOY
Mail from Europe: Mail from USA and Canada:
R.T.A. R.T.A./NATO
BP 25, 7, Rue Ancelle UnitPSC 116
F-92201 Neuilly-sur-Seine Cedex APO AE 09777
France
Tel: 33 (1) 55 61 22 75
Fax: 33 (1) 55 61 22 98
T-l
because of the military importance of the mat- ized by massive regions of flow separation and
ing of helicopters and small naval combatants large unsteady vortical patterns. It is clear that
like frigates, and because of the complexity of these flows create problems for helicopter oper-
the aerodynamics behind the ships structure ations, especially from small ships like frigates,
approaching the deck, and (B) because of the and they do in fact limit helicopter operations
military and commercial interest in the poten- in stormy weather. Problems must also arise in
tial of wind in ground effect vehicles, so-called heavy wind operations from commercial ocean
Ekranoplanes. The Scope of technical coverage structures like drilling platforms, although these
of these two subjects was certainly sufficiently were not discussed at this Symposium.
broad and the presenters sufficiently expert to The training of military helicopter pilots
permit in the case of (A) an assessment of the through the use of flight simulators requires an
overall problem and an understanding of the accurate simulation of the atmospheric environ-
complex fluid dynamics involved and the state ment and this requires effective modeling of the
of present research, and in the case of (B) an ship's wake and especially near the landing deck,
overview of actual vehicle development, espe- which is usually located behind the superstruc-
cially in Russia, a good view of the evolution of ture, a particularly noisy place to operate. As
thinking and techniques, and a fair view of de- stated in Paper # 9, "An air wake model of ap-
tailed aerodynamic problems, especially related propriate fidelity is...an essential element of a
to stability and control. training simulation." From a longer range point
It seemed entirely appropriate to include to- of view, it is clear that effective simulation of
gether with the mainly aerodynamic consider- these complex flows could also lead to improve-
ations mentioned above, a section on hydrody- ment in the wake characteristics through careful
namic aspects with a close connection to aero- alterations in the design of superstructures. In
nautics. These were included in (C) and listed this connection, Russian (TsAGI) work on the
earlier. Many of these have a clear and imme- full scale installation of vortex control devices at
diate connection with aeronautics and its tech- the forward end of aircraft carrier decks is very
niques, as in the paper on underwater missile noteworthy; it is discussed later here.
launch or on hydrofoil boats. It is apparent that An air wake model for simulator use could
the lightness of the coverage of the items un- be derived in pricipal from either full scale or
der (C) do not permit a reasonable evaluation wind tunnel observations and measurements, or
of those subjects. from computer simulations (CFD). Each poses
its own difficulties. Full scale measurements are
difficult and costly to obtain and the conditions
2 Aerodynamics Around cannot be repeated or called up at will; it was re-
Ships counted in Paper # 9 that Australian attempts
at full scale measurements in 1989 on a frigate
This main sector of the meeting has focussed on failed because of "benign" wind conditions dur-
the modelling and simulation of ship air wakes ing much of the test; it is remarkable how often
for helicopter operations. Research in this field this kind of difficulty arises during field work at
is largely carried out as part of an international sea. Extensive wind tunnel experiments have
cooperative program (TTCP) involving four na- been conducted in the UK and in Canada and
tions: the UK, Canada, Australia and the USA. limited model-scale tests " are also conducted on
A very good reportage from participants in the each new ship entering service." Further tests
TTCP was presented, including a Summary of have been conducted in the US. So a wind tun-
the history, role, and present status of the sub- nel data base is growing.
ject (Paper #9). This paper is extremely im- A variety of serious efforts have been made at
portant for anyone interested in this subject at numerical simulations of these flows, which in-
this time. volve unstable shear regions leading to large un-
The flow around ships is induced both by their steady vortical structures of distinctly three di-
forward motion, their seaway induced motions, mensional character. These originate from all of
and by winds near the sea surface. The shape of the bluff body features above the deck. Finally,
the ship itself, and particularly its superstruc- behind the last large deck structure before the
ture leads to highly unsteady flows character- tanking area a large separated flow region forms
T-3
of finite length, no doubt partially stabilized by which must be imposed. Calculated veloc-
the presence of the horizontal landing deck sur- ity perturbations around a deck structure were
face. The simulation is complicated by uncer- shown, but no comparisons with experiments
tainties in the wind field over the ocean, which were made; in these calculations; the viscous
is unsteady and effected by atmospheric strati- driving terms were neglected. This approach
fication. A further complication arises from un- to unsteady simulation seems worthwhile, al-
certainties due to the motion of the ship in a though the influence of the initial disturbances
seaway. is very worrisome. In these calculations, wind
Finally, the action of the helicopter rotor it- turbulence was used to drive the unsteady deck
self can have a large influence on the deck flow flow. However, it seems highly unlikely that the
and vice versa, as the rotor induces substantial deck flow unsteadiness is, in essence, controlled
velocities directed downwards over an area often by wind turbulence. As in the Karman vortex
not very much larger than the helicopter. Most strut, it seems likely that an inherent instabil-
of the simulations were made in the absence of ity of the shear flows produced by separation
the rotor flow and these are first discussed be- from the deck structure reaches some equilib-
low. rium quasi-oscillatory state through the balance
How effective is the state of the art in CFD of non-linear processes, with little effect from
simulation for these turbulent flows? Five pa- small ambient disturbances.
pers reflect on this question (#2, 3, 5, 6, 8 One of the most interesting approaches to un-
and 9). A major limitation at this time is that steady separated flow calculation, paper #6, in-
current RANS codes (Reynolds averaged Navier volved the use of the incompressible Euler equa-
Stokes) are constituted to handle steady, i.e. tions, allowing for the distortion and propaga-
mean flows only. How well do these RANS code tion of vorticity fields, once created, in the ab-
simulate mean flows? How well do they compare sence of viscosity. In this connection we can
with each other? The latter question cannot think of the simulation of vortex sheet roll-up
be answered at all since comparsions were not by inviscid theory, or of von Karman's predic-
made, although similar separated region topol- tion of vortex street geometry by an inviscid
ogy can be seen among several papers. The for- analysis of stability. In the present case, the
mer question can be answered only roughly. In question arises as to the origin of the vorticity
#3, oil flow streaklines on deck are compared which appears behind the deck structures. In
with flow vector simulations in a qualitatively fa- the absence of viscous boundary layers here, the
vorable way. In #8 favorable comparisons of the vorticity which presumably appears here in the
topology of the aft deck separation region are form of shear layers at separation points must
made. However, it was noted that the measured be an artifice of the numerical calculation, and
shear regions were much heavier than in the sim- must presumably depend upon the roundness of
ulations. It is recommended that priority the corners where separation occurs, the grid
be given to careful comparisons between size, and the numerical scheme. It is never-
the numerical results of different research theless a most interesting phenomena and one
groups and with the results of a bench- to be taken advantage of, as was done here.
mark experiment. This has proved effective Once created, the distortion and propagation
in other applications, as for instance in the cal- of the inviscid vorticity field will occur accord-
culation of ship stern flows. Finally, it is possi- ing to the Euler equations. It is not surprising
ble, as mentioned by one of the authors, that the that highly unsteady flows appear once vorticity
RANS codes are insufficient for these separated is introduced at separation points, since vortex
flows and that large eddy simulation or other sheets tend to be inherently unstable. The in-
more sophisticated approaches are necessary. teresting question occurs as to the connection
The possibility of extending the RANS mean between these unstable, inviscid, vortical fields
flow calculation to the simulation of unsteady and the real flow fields with viscosity. In sim-
components was discussed in #3. This involved ple separated wakes at high Reynolds number it
the solution of the so-called non linear distur- is known that the length of the separated flow
bance equations, (NLDE), for the unsteady flow behind the body depends on Reynolds stresses
as driven by a known mean flow (RANS). The acting on the boundary of the separation cavity,
unsteady flow depends upon initial disturbances and the success of an inviscid simulation would
T-4
riety of Russian vehicles as well as some others of the best aircraft figure, 5, shown by Aframeev,
abroad, coupled with a high current interest in Fig. 8 of #29. Of course, actual values must
increasing the speed of marine vehicles, there take into account the necessity of operation in
seems a high interest in Ekranoplanes, and their reasonable seaways.
future. For example, the list of References in Do special missions exist fitting the technical
paper #30 refers to a variety of relevant confer- strength of the Ekranoplane, which is to oper-
ences in the last ten years. ate near the sea at speeds up to 400 km/h and
The papers presented here, notably but not in weights up to, perhaps, 1000 t? The use of
only the Russian contributions offer a good op- Ekranoplanes to provide a launching and land-
portunity for an evaluation, and particularly ing base for reuseable aerospace planes or Space
#32 (a review of the present state of research), Shuttles was discussed in paper #27. The re-
and #29 (an objective and critical look at quirement of high launch payload (600 t) and
technical performance and possibilities by E.A. high speed (220 m/sec) could, they suggested
Aframeev from St. Petersburg). be carried out by a large Ekranoplane.
Did the Ekranoplanes in actual practice re- A clearcut vision for the future development
alize an improvement in the aerodynamic ef- of Ekranoplanes was provided by E.A. Eframeev
ficiency (L/D) of modern aircraft? The an- from the Kylov Shipbuilding Research Institute,
swer is no improvement, only similar per- paper #29. He discounted competition with air-
formance, depending on the degrading effect craft which seem, if anything, technically supe-
of increasing sea states, Fig. 1 of #29. In rior to the Ekranoplane. He stressed the ability
some other performance aspects however the of the latter to operate in two-modes both as a
Ekranoplane is quite inferior: payload effi- ship and as an airplane. He therefore proposed
cience, thrust/weight requirements, and struc- a new direction of development in which large
tural weight of both hull and wing, see figures Ekranoplanes would be designed for full and sus-
4, 5, 8 of #29. The main reason for this infe- tained marine operation at cruising speeds up
riority lies in requirements arising from marine to, say, 15 kts, and flight speeds of, say 250 kts in
operation, for instance the necessity to augment seas up to 3.5 m significant wave height. Loaded
take-off and to provide for water loads. weight of 750 t and payload of 250-300 t with
Can present performance shortcomings be a range of 3-4000 km were projected. For this
overcome? The Russians had augmented lift for two-mode vehicle he imagines use for commer-
take-off by diverting the engine jet to stagnate cial passenger and cargo transportation, includ-
underneath to wind. In paper #30, progress on ing car-passenger ferries, rescue, and tourism
a hybrid hydrofoil-Ekranoplane was described. (perhaps, cruise ships). He stressed a special
The prognosis does not sound good as it was application to create a global international sea
stated that "the resulting operational efficien- rescue system. This idea and this paper de-
cies puts this vehicle in the helicopter class." serve careful study. At minimum, Aframeev
This is not surprising as structurally and sea- has provided what seems to be a critical evalu-
way effective hydrofoils at speeds over 100 kts ation of Russian Ekranoplane technical perfor-
will have efficiencies for below those of wings mance, as already touched upon in preceeding
in air. A more promising approach to improve paragraphs.
take-off is being taken by Fisher-Flugmechanic, The systematic technical development of
paper #37, who propose to use the propellers Ekranoplane aerodynamics was carried out fol-
to pump air underneath the wind and seal lowing their introduction in the 70's. Consid-
it partially underneath with a hovercraft type ering the high degree of stability required, it is
bag-seal which can be retracted after take-off. not surprising that stability in ground effect has
They call this vehicle a Hoverwing. A hybrid been intensively studied. This is briefly reviewed
hovercraft-Ekranoplane results. In terms of per- in #32, and a more detailed asymptotic analysis
formance, paper #37 shows higher thrust weight is presented by Kiril Rozhdestvesky of St. Pe-
ratios for WIGS (Ekranoplanes), at between 4 tersburg in paper #22. Fortunately, a high de-
and 5, then Aframeev in #29, between 3 and 4, gree of longitudinal and laterial stability seems
and Fisher projects a value of about 7 for his obtainable through the application of known re-
Hoverwing-80, of which a model is already fly- sults; indeed, this makes the Ekranoplane fea-
ing. This value would be substantially in excess sible. It is particularly fascinating that a par-
T-6
ticular airfoil shape (S shaped camber) allows senger and car transport in ground effect will be
stability without a tail surface; this was men- very much brightened.
tioned by both Rozhdestvensky and Fisher but
is not documented in the written papers. Fi- 4 Other Interface Vehicles and
nally, the relative role of the upper and lower
surfaces in providing lift and stability are most Problems
important and interesting. In paper #30, the
A variety of isolated papers on SES vehi-
wing investigated at the NLR in the Nether-
cles, hydrofoils, ship control, missile water exit,
lands shows a thick foil (thickess/chord about
and special problems in interface hydrodynam-
20%) with maximum thickness far forward and
ics were presented. These contributed to the
with a flat bottom and trailing edge flap, Fig. 7;
breadth, balance, and interest of the Sympo-
this foil develops substantial negative pressures
sium, without allowing for evaluations of the
on its upper surface over the "bump" there, and
subjects involved. A paper, #15, on SES accel-
it seems remarkable that the upper surface pres-
erations (the cobblestone effect) due to dynamic
sure distribution is almost unchanged in and out
compression of the air cushion by waves illus-
of ground effect. On the other hand, "when a
trated again the problems of comfort associated
wing flies very close to an underlying surface,
with operation in a seaway. A number of the
the prevailing contribution to its aerodynamics
papers presented interesting hydrodynamic phe-
comes from the channel flow underneath." This,
nomena and ideas, #35 and #36, and #33 pre-
of course, simplifies design and analysis.
sented a review of an important emerging tech-
Waves inevitably play a large role in determin- nique for the simulation of certain high speed
ing the design and inevitably degrading the per- non-linear features at the interface.
formance of marine interface vehicles; for exam-
ple, main shortcomings of the Ekranoplane (pa-
per #29, section 3.2) include "the complexity of 5 Concluding Remarks
take off, particularly at rough sea, the necessity
The Symposium was highly successful in that
for qualified pilot team and their constant train-
it gathered together experts from both aero-
ing." Other authors also mentioned the high
dynamics and hydrodynamics, in almost equal
mean vertical accelerations during flight over
number, to discuss interface operations.
waves. It was therefore welcome to see data
Helicopter-ship mating poses severe problems
on wing performance as effected by flight over
requiring accurate simulation. Further research
waves, paper #21 from Kiev, including extensive
on the aerodynamics around and behind ship su-
data on smooth sea performance, too. The most
perstructures is necessary, both theoretical, nu-
important aspect of these data are the fluctua-
merical, and experimental, as are validation of
tions in verticial force due to flight over waves,
numerical results. Systematic wind tunnel data
from which the seaway induced vertical accel-
bases are desirable. Further fundamental re-
erations may be deduced. Since human toler-
search on unsteady three dimensional separated
ance to sustained mean accerlations is limited
flows and its application would be helpful.
to about 0.10 g, limitations on performance in a
The participation of a substantial Russian del-
seaway are implied. This has always been a se-
egation made possible a good review of the sta-
rious problem for hydrofoil craft and led to the
tus of the Ekranoplane. Technical performance
adoption in the US of fully submersible control-
exceeding that of aircraft has not so far been
lable foils. A monograph (in Russian) exists on
the dynamic control of Ekranoplanes, including achieved. Emphasis was put on equal devel-
a brief history which was published in St. Pe- opment of both hull borne and wing borne op-
eration (joint mode) for special applications as
tersburg in 1994; the author is A.V. Nebylov.
global marine disaster response.
Finally, a very upbeat presentation was made Other subjects such as SES, hydrofoils, missile
by H. Fisher, paper #37, describing the devel- water exit were briefly treated; further attention
opment of Ekranoplane vehicles, utilizing hov- to interface vehicles and their technical problems
ercraft ideas for take-off, as mentioned earlier. by this Panel would certainly seem warrented.
High performance was promised for a 100 kt ve-
hicle for the transport of 80 passengers. Should
this development succeed, the prospect for pas-
1-1
This paper presents an overview of recent developments in an effort to predict transient aeroelastic
rotor response during shipboard engage and disengage sequences. The blade is modeled as an elastic
beam undergoing deflections in flap, lag, extension and torsion. The blade equations of motion are
formulated using Hamilton's principle and they are spatially discretized using the finite element
method. The discretized blade equations of motion are integrated for a specified rotor speed run-up
or run-down profile. Blade element theory is used to calculate quasi-steady or unsteady aerodynamic
loads in linear and nonlinear regimes. The analysis is capable of simulating both articulated,
hingeless, and gimballed rotor systems. Validation of the rotor code is discussed, including
correlation with droop stop impact tests and wind tunnel experiments. Predictions of safe
engagement and disengagement envelopes, limited by excessive blade tip deflections or hub moments,
are presented. Future directions of study are also discussed.
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
1-2
elastic flap code was modified to model The assumptions in the original analysis in
articulated rotor systems in which blade motion Refs. 7-10 included that: the total rotor thrust is
is constrained from excessive upward and close to zero which precludes the need for inflow
downward flapping motion by mechanical flap and ground effect models; gravity forces are
and droop stops. Recent studies were conducted necessary since the blade operates at low rotor
to validate the rotor analysis in which a model speeds; ship motion effects on blade inertial
rotor system with rigid, teetering blades was loads are negligible due to low ship motion
placed aboard a scaled ship deck and tested in a frequencies relative to the rotor frequency; and
wind tunnel. The influence of the ship's airwake lead-lag degrees of freedom are neglected
was proven to be very important to the blade's because the portion of the engage/disengage
behavior. sequence in which the majority of tunnel strikes
occur is where Q.(t) is less than 20% of £20 . In
Overview of Previous Research this region Coriolis forces are small. However,
RESEARCHERS at Penn State University, in axial and lag degrees of freedom have since been
conjunction with the US Navy, have also added to the analysis and are detailed in Ref. 17.
been developing simulation tools to predict Quasi-steady airloads, including
transient aeroelastic rotor response during noncirculatory effects and either linear or
shipboard engage/disengage operations7' . This nonlinear separation using a Kirchoff model can
research differs from the efforts in Refs. 1-6 in be predicted. In addition, unsteady aerodynamic
several ways. The rotor model included both loads, including attached flow, unsteady trailing
flap and torsion elastic blade motions with edge separation, and dynamic stall effects can be
Bernoulli-Euler beam bending theory. Using the included.
finite element method, radial variations in elastic, The following sections will give a brief
inertial, and aerodynamic properties are easily introduction to the analytical development of the
accommodated; and different hub types are helicopter structural and aerodynamic model,
modeled with greater ease. Either a quasi-steady examples of validation of the model with
or a time domain unsteady aerodynamic model experimental data, and examples of the
based on indicial response is used to calculate parametric studies conducted to date. For
airloads in the attached flow, nonlinear example, the effects of collective and cyclic
separation and dynamic stall regimes. control settings, droop stop angle, and ship
Experimentally measured rotor run-up and run- motion on blade response were investigated " .
down rotor speed time histories were employed A preliminary study of the effectiveness of a flap
in the prediction of the transient blade response. damper in reducing tip deflections was also
As an example, the measured rotor speed run-up conducted9'10. The importance of torsion and
profile for the H-46 Sea Knight can be seen in unsteady aerodynamics to the simulation of the
Figure 1. blade response was also determined 9,10
Recall that ship motion effects on blade where KlM is the vertical wind component and x is
inertia forces were assumed negligible; however, the nondimensional distance along the blade.
ship motion effects on blade airloads are In the uniform gust distribution the
included. The ship roll motion aerodynamic individual components are defined as
model consists of the transformed relative WOD
Vr =V„dcos(aw)cos(¥wJ (10a)
conditions due to the ship roll angle and a wind
component at the hub height due to the angular Vr' = -V^ cos(or„, )sin(^ira,) (10b)
roll velocity. Ship roll motion is governed by Vzg""=-Vwoa.swla (10c)
\ w /)
sinusoidal variation of $s, the ship roll angle, and
can be expressed as where aw is the angle of the wind relative to the
r horizontal.
2ny^ The quasi-steady aerodynamics development
0,=0«xSin (6)
P follows an unsteady thin airfoil theory
development by Johnson". This development
where <j>imx is the maximum roll angle, y = £l0t includes the airloads due to virtual mass
is the nondimensional time, and P is the (noncirculatory) effects and allows for either
nondimensional roll period. The roll angular linear or nonlinear predictions using Kirchhoff's
velocity causes a lateral wind velocity at the hub nonlinear separation model. The unsteady
height given by aerodynamic was developed by Leishman and
Beddoes12'14. This time domain approach includes
In Ixf attached flow, unsteady trailing edge separation,
■-</> max eg
cos (7)
and dynamic stall effects. The attached flow,
unsteady model predicts both circulatory and
where hcg is the height of the rotor above the eg
noncirculatory airloads. The unsteady trailing
of the ship. edge separation model accounts for trailing edge
Two gust models that simulate the airflow
separation, leading edge pressure lags, and
over a ship's flight deck subjected to a crosswind
unsteady boundary layer effects. The dynamic
are explored in this analysis. The first model, the
stall model calculates the additional lift and large
linear gust developed in Ref. 2, is adopted to
downward pitching moment created by the
explore the effects of local wind variations over
shedding vortex.
the rotor disk on the blade response. This gust
consists of an upward wind component on the
III. Discretized Blade Equations of Motion
windward half of the rotor disk transitioning
The generalized Hamilton's principle, Eqn.
linearly to a downward wind component on the
1, is used to formulate the blade equations of
leeward half of the disk. The second model, the
motion and the finite element method is used to
uniform gust detailed in Ref. 7, is used to
discretize them. Each flexible element has seven
determine the effects of uniform vertical winds
degrees of freedom and is cubic in flap bending
on the blade response. This gust consists of a
and quadratic in torsion. The energy expressions
uniform downward velocity across the rotor disk.
are spatially discretized by substituting shape
In addition, each model features a lateral wind
functions into the elemental virtual energy
component in the direction of the relative WOD
expressions and integrated in space using a six
velocity. The wind velocity due to the either gust
point Gaussian quadrature method. The
model can be expressed as
elemental mass, damping, and stiffness matrices
Y«»« =ys""i +v* "j„+vrKH (8) and load vector are then assembled to form the
global mass matrix M, the damping matrix C,
where the subscript H denotes the hub fixed and the stiffness matrix K7. The global degrees
coordinates system. In the linear gust of freedom are q. The global discretized
distribution the individual components are equation of motion becomes
defined as
Mq+Cq+Kq=F (ID
Vr'=K^„^<¥,J (9a)
IV. Analysis
vr'=-V^„„M¥ml) (9b) The blade transient response is calculated by
Vzsu* = V-*«-Kcosfor -ye* yr^ ) (9c) integrating the discretized blade equations of
1-5
motion. Recall that the rotor speed and blade was undertaken in Ref. 15-16. The boundary
azimuth are functions of time and are used in the conditions of the blade change dramatically
calculations of M, C, K, and F which must be during such an event. When the blade is not in
recalculated at every time step. The time contact with the droop stop, the flap hinge is free
integration is performed by a 4th order Runge- to rotate and the blade acts as a hinged beam.
Kutta scheme. The blade static deflection is used When the blade is in contact with the droop stop,
as the initial conditions for rotor run-up the flap hinge is restrained from further
solutions. The steady state response at the downward rotation and the blade acts as a
operational rotor speed is used as initial cantilevered beam. Recall that a very stiff
conditions for run-down solutions. Blade rotational spring is added to K to freeze the
response can be computed in either the full finite hinge angle during blade contact with either the
element space or modal space, but modal space is droop or flap stop.
recommended for computational efficiency For computational efficiency, it is common
to transform the global equations of motion into a
Model Validation modal coordinate system
I. Blade Fan Diagram
so(*M0 (12)
The H-46 Sea Knight is a three-bladed
tandem rotor helicopter, but the present research where <D is the mode shape and a is the modal
only models a single rotor blade of the aft rotor amplitude. Substitution of Eqn. 12 into Eqn. 11
system. Proper stiffness and mass distributions and multiplication by Or yields
are verified by comparing the predicted static tip
deflection to experimental data. The predicted Mä + Cä + Ka = F (13)
static tip deflection of 15.3 inches is within 2.6
Two different stiffness matrices are used in
percent of the Boeing test data average. A fan
this transient analysis - one corresponding to the
diagram is also developed to determine how well
cantilevered condition of the beam and one
the finite element approach models the H-46 flap
corresponding to the hinged condition of the
and torsion rotating natural frequencies. Figure 4
beam. Therefore, two independent sets of natural
illustrates the first four coupled flap-torsion
frequencies and mode shapes are generated
frequencies overlaid with test data for the second
through the fourth flap modes from Ref. 7. The (0 = a>„ and if ß>ßns
flap-torsion elastic analysis is in satisfactory (15)
agreement with the blade data.
co = cor and if ß^ß,,
where co is the natural frequency, ß is the flap
hinge angle, and ßDS is the droop stop angle. The
equations of motion can be solved using three
different techniques. The first and simplest
technique is a direct solution of Eqn. 11 in full
finite element space, called a "Physical Space"
integration. The second technique is a solution
of Eqn. 13 using only the hinged modes of the
beam regardless of blade-droop stop contact,
called a "Modal Swapping Off" integration. The
third and most complicated technique is a
0.4 0.6 0.S 1 solution of Eqn. 13 using either the hinged or
Nondlmenslimal Rotational Speed
cantilevered modes depending on droop stop
Figure 4: H-46 Flap-Torsion Fan Diagram contact, called a "Modal Swapping On"
integration. In this method, the modal
II. Droop Stop Impacts amplitudes must be adjusted to maintain the
An important feature of articulated rotor physical shape of the beam when the mode
blades is the droop stop mechanism. During the shapes are changed or discontinuities can arise in
engagement and disengagement of the rotor the solution. The deflection of the beam can be
system, excessive flapping can cause the rotor approximated by either set of modal matrices and
blade to repeatedly impact the droop stop. An modal amplitudes
investigation to the behavior of a model blade
1-6
Droop Stop
The transient response of the tip deflection is each drop test. However, both the accuracy and
shown in Figure 6b. Much like the upward the computational efficiency of each solution
movement of the flap hinge angle, the cause of method are of importance. Each drop test
the upward motion of the blade tip at t = 0.04 simulates 0.5 seconds of real time. The Physical
seconds is due to the flexing of the blade while Space solution method requires the most CPU
falling. Both the experimental data and time, a total of 70 seconds, because 41 degrees of
analytical predictions capture this phenomenon. freedom are used. The Modal Swapping Off
The maximum tip deflection, wüp/L, of -0.17 at t solution is the next most efficient model,
= 0.25 seconds for the analytic methods is in requiring 24 seconds of CPU time. Instead of
excellent agreement with the experimental data. integrating 41 coupled equations of motion, only
0.15 . i . i i i . i I
i»*X ! 6 normal modes are required. The Modal
——Physical Space
0.1 ;0w*S Swapping On solution is the most efficient of the
5^ 0.05
\j ——~ Modal Swapping On
a Experiment
/jpf»*
three analytic methods, requiring only 13 seconds
:
ffn
u
; of CPU time because only four normal modes
"i? 0 \\ U
Ss were required for the solution.
\ \ i J ;
£ -0.05 The differences in CPU time to generate
CD : Static Droop
a
.e- -0.1 even the longest drop test solution are not very
\T "M substantial; however, the rotor blade and droop
-0.15
p§3» stop impact event is a very small part of a larger
■ . . . . i . . . . i
-0.2 picture - the complete rotor engagement and
0.1 0.2 0.3 0.4 0.5
Time (s)
disengagement simulation. Many
computationally intensive features such as blade
Figure 6b: Tip Deflection Time History
element unsteady aerodynamics which result in
motion dependent terms and large numbers of
The transient response of the strain gage at
degrees of freedom including flap and torsion in
x/L = 0.20 is shown in Figure 6c. There is
the aeroelastic rotor analysis make computational
excellent agreement in the peaks and valleys of
efficiency essential. A dramatic difference in
the measured and predicted strain. Note that the
CPU time can result using the three integration
strain becomes negative at t = 0.05 seconds,
techniques in the aeroelastic analysis in Refs. 6-
indicating the beam is actually curved upwards,
9. The Physical Space engagement solution is
and is indicative of the blade flexing as it falls.
again the longest, requiring 160 hours of CPU
The maximum strain of 1400 |i£ is captured well
time on an RS-6000 workstation. If the
captured by the experimental data and analytic engagement simulation is performed using the
predictions.
Modal Swapping Off integration, the size of the
1500
—— Physical Space problem is reduced to 6 normal modes. This
■—— Modal Swapping On
g\ 4 1 E D Experiment
reduces the length of the simulation to 10 hours,
1000 which is still fairly substantial. Finally, if the
fan 1 f
Modal Swapping On integration is used, 30
L In: Wö
minutes is required for the engagement
— 500 - Static -| ßpf -|
cd
\ Droop 1 / p 1 simulation. Clearly, when the issue of
55 computational efficiency is considered along
with accuracy, the Modal Swapping On
\ff\
U i i w i00 «V -
1 'pffi integration is the superior method for use in the
. , , , i .... i .... i , , , , i , . . . rotor engagement/disengagement simulation.
-500
0 0.1 0.2 0.3 0.4 0.5
Time (s) III. Wind Tunnel Validation Studies
Figure 6c: Strain Gage Time History Experimental data was gathered at the
University of Southampton on the engagement
As seen in the previous sections, each of the and disengagement of a model helicopter on
analytic techniques presented - the Physical board a scale ship deck5'6. First, airwake
Space Integration, the Modal Swapping Off measurements at five different positions were
integration, and the Modal Swapping On taken without the helicopter model. For each
integration provide almost identical solutions for rotor hub location, winds were measured along
1-8
the direction of wind flow. The variation over The experimental data gathered in Refs. 5-6
the rotor disk is shown for each of the five hub was then used to validate the transient dynamic
locations (A-E) in Figure 8a. The results tend to response predicted by the Penn State aeroelastic
indicate a very distinct division between laminar analysis. Correlation of the Penn State analysis
and turbulent flow regions. Figure 8b illustrates results with the single-bladed scale model rotor
the variation of flow inclination to the horizontal results was reported in Ref. 9.
over the rotor disk for the five hub locations. The comparisons between the
Note the large change in flow inclination for the experimentally measured rotor response and the
three leeward hub locations which is indicative of predictions are for deck positions A and E are
reverse flow. shown in Figures 9a-b and lOa-b, respectively.
For each figure, the experimental response
measured in Refs. 5-6 is shown on top and the
Deck Position A
theoretical predictions from the Penn State
analysis are shown on bottom.
Figures 9a-b show the comparison for deck
position A. Note that the experimental data
shows very little random behavior as might be
expected in a laminar airwake. In this case, the
theoretical predictions show good agreement
with experimental results.
-0.25 0 0.25 UK Wind Tunnel Data
Radial Station (r/R)
30 i r 2000
3D
180- ^^- Tv Deck Position C K 1500 S
150- —-""^^ o
"3120- Deck Position E
3 90'
Deck Position A 1
1 «o-
1 0
i -30 ^ \ Deck Position D| \ueck Position B 1 \
£ -60
| -90'
J
-120 ■
10 20 30 40 50 60
-150
.1X0 1 1 -1 1 V 1 1 Elapsed Time (sec)
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1
Radial Station (r/R)
PSU/NAWCAD Analysis
Figures 7a-b: Model Ship Airwake
Measurements from Refs. 5-6 2000
0 10 20 30 40 50 60
Elapsed Time (sec)
Figures 9a-b: Deck Position A
The conclusion, in agreement with Refs. 5-6, was maximum downward tip deflection occurs at
that different large scale airwake characteristics approximately 7 seconds into the engagement.
(i.e. different deck positions) are very important During rotor disengagement, the blade
to the blade sailing phenomenon deflections steadily grow. Note that when the
UK Wind Tunnel Data rotor brake is applied, the tip deflections grow
more rapidly and the maximum deflection occurs
within the last two seconds.
10 20 30 40 50 60
Elapsed Time (sec)
PSU/NAWC Analysis
0 10 20 30 40 50 60
Elapsed Time (sec)
Figures lOa-b: Deck Position E Figures lla-b: Rotor Engagement and
Disengagement
Parametric Studies
I. Typical Rotor Engagement and Disengagement II. Engage Envelope Study
After the validation was performed, a series The engage envelope analysis determines the
of parametric studies for the H-46 helicopter was blade-to-fuselage clearance for multiple wind-
conducted. First, a typical rotor engagement and over-deck conditions using the maximum
disengagement were examined. During all rotor downward tip deflection calculated from the
engage/disengage sequences, the collective is 3°, blade transient response. The safe engage region
the longitudinal cyclic is 2.5°, and the lateral is composed of all wind-over-deck conditions
cyclic is 0.07°. These control inputs are termed with blade-to-fuselage clearances greater than 8
the standard control inputs and are derived from inches.
the "auto" cyclic trim settings used in actual H- An H-46 baseline engagement envelope
46 operations. The standard control inputs are developed using standard control inputs and a
used for all parametric studies unless otherwise uniformly distributed gust with a;=150 is shown
noted. The experimentally measured rotor run- in Figure 12. Note that the H-46 is limited to
up profile in Figure 1 and the run-down profile wind speeds of 30 knots for port WOD
with rotor brake application at 55% of full rotor conditions and is limited to 35 knots for most
speed are used. A linear gust model, with Vwnd = starboard WOD conditions. The predicted
40 knots, Kvcn - 25%, and fKml = 90° was used. envelope shows safe engagements for winds
The unsteady aerodynamic model and dynamic speeds up to 40 knots for WOD directions of 45°
stall effects were used. Representative blade to 75°. Results of sensitivity studies performed
responses using are shown in Figures lla-b. in Ref. 7 showed that control inputs have a
Note for rotor engagement, the tip deflections moderate effect on maximum downward tip
decrease as the rotor speed increases, and the deflections. The two following studies show the
1-10
effect of 2° variations in collective and cyclic collective and cyclic inputs can moderately effect
pitch controls from the standard control inputs. the safe engage region.
330 ooo 030 330 000 030
"Safe Ehgagifl&sgiöh/y
210 180 150 210 180 150
Figure 12: Baseline Safe Engage Envelope Figure 14: Safe Engage Envelope for 2°
Lateral Cyclic Increase
The predicted H-46 engage envelope for a
collective pitch decrease of 2° from standard III. Flap Damping Study
control inputs is shown in Figure 13. A In an effort to reduce the frequency of tunnel
comparison with the baseline engagement strike occurrences, a preliminary investigation
envelope shows a 5 knot reduction in the safe was conducted into the effectiveness of adding a
engagement region for WOD directions of 45° to rotational damper at the flap hinge of the H-46 to
75° and 120° to 285°. dissipate energy in the flapping degree of
330 freedom. Preliminary results for the current H-
46 flap and droop stop settings, of -.54° and 1.5°
respectively, indicated that the flap damper,
regardless of its strength, was not effective in
reducing the blade deflection because the angular
range of motion was too small. At very large
damper strengths, the flap damper simply freezes
the flap hinge angle because the energy necessary
to rotate the highly damped hinge is much higher
than the energy necessary to deform the blade.
However, raising the flap stop angle permits the
damper to act through a larger angular range of
motion and dissipate more energy. Combinations
210 180 150 of flap stop settings from 1.5° to 10° and flap
Figure 13: Safe Engage Envelope for 2° damper strengths ranging from zero to five times
Collective Decrease the H-46's lead/lag damper strength, or 17500 ft-
lb/(rad/s), were investigated as means of
The predicted H-46 engage envelope for a reducing the blade deflection.
lateral cyclic pitch increase of 2° from standard Tip deflection time histories for rotor
control inputs is shown in Figure 14. Unlike the engagement are presented in Figure 15. Two
small decrease in wind speed over large portions flap stop settings and flap damper combinations
of the envelope for the collective pitch change, are examined. The flapping behavior of the rotor
note the substantial reduction in engagement blade becomes excessive at both 2 and 6 seconds
capability for port winds. WOD speeds are into the engagement because it is both rotating
reduced by 10 to 15 knots for WOD directions into the WOD and experiencing a strong upflow
from 120° to 150° and from 225° to 330°. These from the specified gust distribution. The
results illustrate that during engagements in maximum negative tip deflection occurred 7
uniformly distributed winds, small changes in seconds after engagement in both configurations.
1-11
For a flap damper four times the strength of the also result in high angles of attack at very low
lead/lag damper and a flap stop setting of 10°, rotor speeds.
enough energy is dissipated to prevent a tunnel An analytical model for steady-state
strike. aeroelastic response and stability of gimballed
tiltrotors was developed in Ref. 18. The rotor
i .-» : : : i
60
%% model included coupled flap-lag bending modes
•' '. l\
40 and torsion degrees of freedom. The gimbal tilt
/f\ \ I \
20 was described by using the gimbal degrees of
0
/' ' \ ''' III freedom, fiGC and fias , which are the pitch and
-20 ■Z' \ V\ /\;.A 7 1 U roll of the rotor disk in nonrotating frame. They
-40
Tip Deflection *
are determined by the trim equations in steady-
state analysis. For this type of analysis, the
H-46 Tunnel i
0 2 3 4 5 6
steady periodic response of the rotor can be
7 8
Time (s) obtained by considering the response of an
C = OxLag Damper, Flap Stop = 1.5 deg equivalent single blade. For transient engage and
disengage analyses, the motion of blades is non-
Figure 15: Flap Damping Study periodic. In addition, gimbal restraint impacts
make the individual blades respond differently
Current Research and the inflow distribution on all blades
MUCH of the current research at Penn State contributes to gimbal motion. The entire multi-
has focused upon the engagement and bladed gimballed rotor has to be modeled to
disengagement operations for tiltrotors17. Present analyze the transient response during engage and
tiltrotor configurations, such as the V-22 Osprey, disengage operations. The overall goal of this
feature the use of a gimballed rotor system. research was to systematically investigate the
Unique rotor characteristics include a gimbal transient response of gimballed tiltrotors during
spring and a gimbal restraint as shown in Figure shipboard engage/disengage operations.
16. In addition, gimballed rotor blades are
typically short and highly twisted. Engagement and Disengagement Analysis
PR0PR0T0R HUBv I. Structural Model Formulation
A fixed rotor support was considered in the
present research (i.e. rigid wing/pylon). The
multi-bladed gimballed rotor is modeled as
several slender elastic beams attached to a hub
and undergoing flap bending, lag bending, elastic
twist, and axial deflection. The blades are
discretized into a number of beam elements using
HUB UNDEFIECTED HUB DEFECTED
the finite element method. Each beam element
Figure 16: Gimballed tiltrotor consists of fifteen degrees of freedom18. The
motion of gimbal is expressed by using two
A gimballed rotor has three or more blades degrees of freedom in the rotating system.
attached to the hub without flap and lag hinges. Hamilton's Principle, Eqn 1, is again used to
The hub is attached to the rotor shaft by a develop the equations of motion. However, the
universal joint, or gimbal. Compared to contributions to these energy expressions must be
articulated or hingeless rotor systems, the summed over each blade
gimballed rotor has some unique structural and
aerodynamic characteristics related to shipboard SU = (^SU;) + SUCS +SUGR + «ft/,t (21a)
engagement and disengagement operations. The »1=1
where A^ is the number of blades, B refers to the shipboard aerodynamic environment are given in
blade, GS to the gimbal spring, GR to the gimbal Ref7.
restraint, and PL to the pitch link.
The variation of gimbal restraint strain III. Discretized Blade Equations of Motion
energy is given by An illustration of the blade finite element
discretization is shown in Figure 18. Each of the
6UGS = Kßcc ßGCSßGC + Kfm ßGS SßGS (22) flexible elements consists of fifteen degrees of
freedom. These degrees of freedom are
The degrees of freedom in the rotating system are
distributed over five element nodes, two external
used to express the motion of gimbal. The
and three internal, which describe the elemental
transformation matrix is given by
flap deflection (w,w'), lag deflection (v,v'),
ßoc -siny
COS\{f ß (23)
twist {ft), and axial deflection (u).
Pilch Bearing
ßas siny cosy Coning Flexure
blade 2#
blade 1#
where y is the azimuth angle of rotating frame,
and ß and ()> are the gimbal tilt degrees of
freedom in the rotating system. The gimbal
degree of freedom in nonrotating and rotating
coordinate systems are shown in Fig. 17.
blade 3# —*■*!
»2
• • • v
2
'1 ''2
»j *.i "4
"2
li w'2
(4
rotating system by
ySm„ = (/?cosy-0siny)cosym„ + ^Element©
(25)
(/?sin^ + 0cos^)sin^m„
Jemenf® |
where \pmax is the azimuth where the maximum
Gimbal Degrees of
tilt angle located and given by ^4 Freedom, pgmd <|>
The energy expressions are spatially where M, C, K, and F are for the entire rotor.
discretized by substituting the shape functions They include the linear structural, inertial and
into the elemental virtual energy expressions. aerodynamic contributions. The force vector
The virtual energy expression for multi-bladed includes the constant structural, inertial and
tiltrotor in terms of the elemental matrices and aerodynamic contributions as well as the
load vector becomes nonlinear structural and aerodynamic
contributions.
+ C"q7 + Kpq,"-F,")a
(27) IV. Hub Moments
During gimbal restraint impacts, it is
important to calculate blade and hub loads for
design limits. The flap bending moment, lag
bending moment, and twisting moment in the
where [kG] is the stiffness of matrix due to deformed coordinate system are defined by"
gimbal restraint and is expressed as
M; = £7 (v"sin(0 + <£)- w'cos(e + <ß))
(33a)
-k^cosj/sin/r -EC,f
k^colys+k^sidys,
+^A co^sir^
k]- -kk co^sir^ (28) M: =-EAe(u' + — +
2 2
)-EB Yä'9'
* (33b)
kßac&iriy+kß coly/ + EIt (v'cos(i9 + <j>)- w"sm(e + (*))
+L cosy/sir^
M- = GJip' + EAK't{e + <f)\u+—+—
where N is the number of finite elements for each
+ EBß'Y - EBß'(y'cosd + w'sin ö) (33c)
blade. For a gimballed tiltrotor, the blade is
fixed to the hub at the blade root. The hub is - [EC,f + £C,(w'cos6> - v'sin ö)j
attached to the rotor shaft by a gimbal. The
motion of the hub is expressed using two degrees where 9 is the pitch angle and defined as
of freedom in this analysis, ß and <|>, which rotate 0 = OQ+ 0ic cos \jfm + eu cos \ffm
with the rotor. The root slope, w[m and twist, (34)
+ etw{r-0.15)-Kpw[m
(j)"', of each blade are not restrained and can be
expressed as a function of two gimbal degrees of Rotor hub moments are obtained by transforming
freedom in the rotating frame, ß and 0 . the blade root moments to the non-rotating hub-
fixed coordinate system and summing over each
w'1m=ßcosy/m+</>smy/m (29) blade. The roll moment, pitch moment and torque
are expressed as
<p"' = -ß sin ym + (j> cos iym (30)
M" = 2^(M'cosy/-M' sinyf-M'ßr cosy/) (35a)
where the subscript 1 indicates the left-hand node N,
of the first element blade and \j)m is the azimuth M" =^i,M'siny/ + M'icosy/-M'ßrsiny/) (35b)
of mth blade in rotating frame and can be
expressed as Ml = 2JM'Jr +JW") (35c)
where M'", M™ and Mf are the flap bending
wm = \m-\) (31)
moment, lag bending moment and twist moment
at the root of mth blade in the undeformed
Eqns. (29) and (30) are then substituted into Eqn. coordinate system, and are given by
(28). The stiffness matrix, mass matrix, damping M" = M" cos 0 - M" sin <f> (36a)
matrix and force vector are assembled to form
w''
their global counterparts. The discretized M; =AT>'+(1 )(M;sin0+M;cos0) (36b)
equation of rotor motion is obtained as
M" =M"(1 )-M'.Wsin<j>-M'.w'cos<p (36c)
Mq+Cq+Kq-F=0 (32)
1-14
Model Validation
I. WRATS Bending Moments in Hover
Experimental data for blade bending
moments and twisting moment in hover condition
of a 1/5"1 scale model of V-22 WRATS gimballed
rotor was used to validate the analysis. The test
was conducted for hover condition with varying
collective pitch (zero cyclic pitch) at a rotor
speed of 875 RPM. Eighteen rotor modes are
used in the transient analysis, including three
flap, two lag and one torsion. The flap bending
moment, lag bending moment, and twisting
moment at six blade sections was measured. A
comparison of these experimental results with the
predictions of the present analysis is shown in
Predicted
Figures 20a-b and 21a-b. The flap and lag a Experimental Data
Azimuth (deg)
16 28 42 60 81 105 132 800
-JJ
600
400
200
I A
I
0 '-■—■ t
-200
-400
-600
-800 II Flap
■1000 Lag
■1200
0.0 0.1 0.2 0.3 0.4 0.5 0.7 0.8 0.9 1.0
Time (sec)
0.0 0.1 0.2 0.3 0.4 0.5 0.7 0.8 0.9 1.0
Time (sec) Figure 24: Bending Moment Response on
Figure 22: Transient Response for Gimballed Blade 1 at the Coning Flexure
Tiltrotor
Figure 25 shows the peak roll moment, peak
The first impact event is illustrated in Fig. pitch moment, and peak vector sum of roll and
23. The t0, th and t2 refer to the time of pitch moment during an engage operation at a
beginning rotating, first connection between rotor higher WOD speed of \i = 0.086. The hub
and gimbal restraint, and disconnection, moment vector sum is 18 times static hub load
respectively. The rotor first impacts gimbal for 4° gimbal tilt angle operations and 6 times
restraint when blade 1 rotates 16°. The azimuth static hub loads for 11" gimbal tilt operations,
where the gimbal impacts the hub is located at which is the maximum amount of gimbal tilt.
101°. The rotor keeps connection with gimbal
restraint while the blade 1 rotates to 20°, while
the azimuth of connection moves counter
clockwise from 101° to 118° during this period.
| 4° Tilt Angle |
| 11°TiltAngle| \
l\
Figure 25: Hub Loads During Run-up
Operation
Ongoing Work
Several areas of investigation are currently
underway. For articulated rotors, control of the
tunnel strike phenomenon is being investigated.
t^O.33 sec
One manner of reducing the tip deflections is to
%lad81=16° increase the flapwise stiffness of the blade by
^iadei=°° t,=0.28sec increasing the collective pitch setting. The total
tp=0 sec
flapwise stiffness is equal to
Figure 23: The First Impact Event
EIflap=EIyy cos2 6 +El„ sin2 6 (37)
II. Hub Moments
The flap and lag bending moment time This would only be performed in the very
histories at the inboard section of the coning low rotor speed region of a rotor engagement or
flexure are shown in Fig 24. It can be seen that disengagement where Q. < 20%NR. Figure 26
impacts between rotor and gimbal restraint cause shows the increase in flapwise stiffness achieved
substantial increases in blade bending moments. for various collective pitch settings. Two radial
locations are shown. The first location (r/R =
0.10) is just outboard of the pitch bearing and the
second location (r/R = 0.25) is the beginning of
the blade airfoil section. Note that the flapwise
1-16
stiffness can be increased over three times its 2.Newman, S.J., "A Theoretical Model for the
value at the standard collective pitch setting at Predicting the Blade Sailing Behaviour of a
r/R = 0.25. However, increasing the collective Semi-Rigid Rotor Helicopter," Vertica, Vol. 14,
pitch may also increase blade deflections due to No. 4, 1990, pp. 531-544.
the increased angle of attack. In addition, large
chordwise blade bending moments may be a 3.Newman, S.J., "The Application of a
concern. Theoretical Blade Sailing Model to Predict the
Behaviour of Articulated Rotor Helicopters,"
4 , i i . • ■ '
i 1 . i i > 1 i ' ' '
1 Aeronautical Journal of the Royal Aeronautical
r/R =0.10 Society, Vol. 96, No. 956, pp. 233-239.
U ——r/R * 0.25
..•
c
3
4.Newman, S.J., "The Problems of Rotor
Q.
(0 /; Engagement and Disengagement of a Shipborne
Stands rdSetting -'
2 ... Helicopter," Journal of Naval Sciences, Vol. 20,
c
i No. 1,1994, pp. 56-64.
Q.
a _—[
EE
.-j» --^T -.—i—r-—. I .... . . . 5.Newman, S.J., "The Verification of a
5 10 15 20
Theoretical Helicopter Rotor Blade Sailing
Collective Pitch (deg)
Method by Means of Windtunnel Testing,"
Figure 26: Variation of Flapwise Stiffness with Aeronautical Journal of the Royal Aeronautical
Collective Pitch Setting Society, Vol. 99, No. 982, 1995, pp. 41-51.
The other way to increase flapwise stiffness 6.Newman, S.J., "An Investigation into the
of the rotor is through active stiffening of the Phenomenon of Helicopter Blade Sailing," Ph.D.
rotor blade by means of piezoelectric actuators. Dissertation, Dept. of Aeronautics and
For gimballed rotors, pylon degrees of freedom Astronautics, University of Southampton,
are being added to the simulation to model the Southhampton, United Kingdom, Mar. 1995.
finite stiffness of the wing and pylon.
Incorporation of the airwake results from CFD 7.Geyer, W.P., "Aeroelastic Analysis of
analyses of simple ship configurations, such as in Transient Blade Dynamics During Shipboard
Ref. 20, is also underway. Engage/Disengage Operations," M.S. Thesis,
Dept. of Aerospace Engineering, Pennsylvania
Acknowledgments State University, University Park, PA, Aug.
This research was carried out with support 1995.
from the Naval Air Warfare Center Aircraft
Division at Patuxent River, Maryland (technical 8.Geyer, W.P. and Smith, E.C., "Aeroelastic
monitors: LCDR Mark Whittle, Mr. Kurt Long, Analysis of Transient Blade Dynamics During
Mr. Larry Trick, and Mr. Bill Geyer); the Naval Shipboard Engage/Disengage Operations,"
Aviation Depot at Cherry Point, North Carolina Proceedings of the 2nd International AHS
(technical monitor: MAJ Robert Hellar); the Aeromechanics Specialists' Meeting (Fairfield
NASA/Army National Rotorcraft Technology County, CT), Alexandria, VA, 1995, pp. 8-91-8-
Center; and the Office of Naval Research 114.
(technical monitor Dr. Pat Purtell). The authors
also want to thank Mr. Ben Settle of Bell 9.Geyer, W.P., Smith E.C., and Keller, J.,
Helicopters and Dr. Mark Nixon from US Army "Validation and Application of a Transient
at NASA LaRC, for valuable technical support Aeroelastic Analysis for Shipboard
and rotor data for tiltrotor configurations. Engage/Disengage Operations," Proceedings of
the 52nd Forum of the American Helicopter
References Society (Washington, D.C.), American
1.Hurst, D.W. and Newman, S.J., "Wind Tunnel Helicopter Society, Alexandria, VA, pp. 152-
Measurements of Ship Induced Turbulence and 167.
the Prediction of Helicopter Rotor Blade
Response," Vertica, Vol. 12, No. 3, 1988, pp. lO.Geyer Jr., W.P., Smith, E.C., and Keller, J.A.,
267-278. "Aeroelastic Analysis of Transient Blade
Dynamics During Shipboard Engage/Disengage
1-17
Operations," AIAA Journal of Aircraft, Vol. 35, 19.Hodges D.H., and Dowell E.H., "Nonlinear
No. 3, pp. 445-453. Equations of Motion for the Elastic Bending and
Torsion of Twisted Nonuniform Rotor Blades,"
11.Johnson, W., Helicopter Theory, Dover, New NASA TND-7818, 1994.
York, 1994.
20.Long, L.N., and Liu, J., "High Order Accurate
12.Leishman, J.G., and Beddoes, T.S., "A Ship Airwake Predictions for the Helicopter/Ship
Generalised Model for Airfoil Unsteady Interface Problem," Proceedings of the 54th
Aerodynamic Behaviour and Dynamic Stall Forum of the American Helicopter Society
Using the Indicial Method," Proceedings of the (Washington, DC), American Helicopter Society,
42nd Forum of the American Helicopter Society, Alexandria, VA.
Washington, D.C., June 1986.
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
2-2
the computational domain of Figure 3.1. The figure constant. There are many features of this flow that
indicates that the rotor thrust is exerted within the conflict with established understanding of both rotors
hatched areas. In these regions the Navier-Stokes and actuator discs. Figure 4.1 shows that the fluid flows
equations are modified by the addition of the vector B as upwards throughout most of the domain, including the
shown in Equation 3.1. The force applied to the fluid outer region of the rotor. Quantitative study yields that
acts vertically downwards. momentum in the downwash does not equate to the
thrust applied at the rotor.
1
y v1 y y V V y y1* V x X X x X X X X X X X X X X X X
1
X x w * y x * x * * X X X X y x X X X )< X X X X X X VSVN\\\\\\\\\\\,
The air flow around a hovering helicopter is In order to instigate the given relationship the
fundamentally different because there is no free stream computational domain must be constructed with two
velocity and all fluid flow is induced by the helicopter boundaries, an inner and an outer. At the outer
rotor itself. For this reason, applying any of the boundary, the extremity of the domain, the pressure is
available boundary conditions such as an imposed defined as zero. This is achieved using a 'pressure
velocity or pressure is unjustifiable and gives physically boundary', a boundary condition readily accessible
incorrect flow solutions. An example is the flow within most CFD solvers.
solution in Figure 4.1, generated by allowing the velocity
to vary around the boundary whilst the pressure remains
2-3
Outer Inner
Boundary Boundary
Computational
Domain
> >
P=0 1/2.Q.V
Forces Applied to
Fluid Here.
angle of 0°. The vectors represent speed and the incompressible and isothermal. Reference [1] was used
contours pressure variations of 25Pa. to determine the turbulent kinetic energy imposed at the
windward edge of the domain. Reference [2] was used
to determine a realistic mixing length for the initial
turbulence dissipation constant, epsilon.
Vo=30knots
14.00m
6. Ship Airwake
In Position 1, shown in Figure 7.1.1, there are vortices The only areas of recirculation are present at the
present on both sides of the rotor. These cause intersection of the ship's sides and the sea surface.
significant recirculation, which in turn increases the These are small in comparison to the vortices generated
power requirements to a value greater than the simple with the rotor in the other positions.
hover case, since the rotor is effectively in climb. The
flow through the rotor itself is approximately vertical. Position 1 2 3
%PH 126 99 74
Figure 7.1.3 displays the flow solution for the rotor over 7.2. Wind 30 Knots. Port to Starboard.
the ship's centreline. The rotor has no recirculation at
the tips. There is a high pressure region directly beneath These three flow solutions were determined with a 30
the rotor and on the deck. The favourable ground effect knot free steam velocity acting from port to starboard,
the ship generates, gives a power requirement which appears left to right in the figures. The pressure
significantly less than the hover. contours are spaced at 40Pa.
_——— i
zzzzz^~L~~
^^=
I^^I^^^ —
^^^^ter^rrrr-™"
>> ==
"'^^2^>t^r
ij.C*^-.. —
-,. --^
>~- "1
./) \\
f \
ship. This is reflected in the power exerted, as shown in noticeable feature of the flow solution is the lateral
Table 7.2.1. There is however a large vertical velocity velocity the rotor experiences, 10.9m/s or 21 knots. This
gradient across the rotor. The fluid is passing upwards is more than double the other two positions caused by
through the rotor on the port side and downwards at the only a small change in position of the rotor. As with the
starboard side, This represents increased control other two positions the fluid is flowing upwards through
requirements. the rotor on the port side and downwards on the
starboard side.
Position 1 2 3
%PH 0 -1 0
The rotor over the ship's centreline is shown Figure Figure 7.3.1 shows that the flow around the rotor in
7.2.3. Similar to the other positions the induced power Position 1 is largely influenced by the free stream and
is small due to the upflow across the rotor. The most the presence of the ship's helideck. The greatest
2-8
pressure gradients occur at the windward edge of the wind speed across the rotor is 65 knots, this is greater
deck, not at the rotor. The rotor is operating in a strong than the free stream velocity. The vertical flow of fluid
upflow resulting in a negative power requirement, as through the rotor varies by 13.3m/s between the port and
shown in Table 7.3.1. The lateral wind speed across the starboard reference points.
rotor is 23.8m/s which is less than the undisturbed free
stream velocity. Position 1 2 3
7.4.1. The power exerted at the rotor is only a fraction to flow upwards through the rotor at the windward end.
of the power exerted in the simple hover case. The flow At the leeward end of the rotor the air is flowing
across the rotor is a 8m/s, this is less than the free stream downwards. The net effect of the up and down flow is a
velocity. The variation between the vertical velocity zero power requirement. The lateral wind speed across
component at the port and starboard end of the rotors is the rotor, shown in Table 7.4.1, is 10.9m/s, this is less
large, 16.8m/s. than the free stream velocity.
Position 1 2 3
%PH 13 10 0
This study indicates that the control requirements of the WP = vertical velocity component at 5m port of rotor
helicopter are more likely to limit safe operations than centreline.
the power limitations. The vertical velocity variations
evident in the flow solutions are not found in any other Ws = vertical velocity component at 5m starboard of
normal operations, such as hover or forward flight. rotor centreline.
9. Nomenclature
A = area of rotor.
Ap = thrust/area at rotor.
p = pressure.
V = velocity scalar.
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
3-2
flow that effects this region. On the LHA, helicopters The need for numerical simulations comes from the
can land on many different locations on the deck, and high cost of determining the safe operating envelopes for
each of these can experience quite different flow fields. helicopters in a ship environment (and the huge test-
The LHA's can carry 9 CH-53D Sea Stallion or 12 CH- ing backlog). It would be very useful to have numerical
46D Sea Knight helicopters, and 6 AV-8B Harriers. The methods that could accurately simulate ship airwakes.
forward portion of the deck is primarily influenced by There have been other attempts at numerically simulat-
the separated flow off the deck edge. Very strong vortex ing ship airwakes. The airwake about a DD-963 ship
sheets emanate from these edges. One of the authors configuration was simulated using a steady-state flow
(Long) spent three days on an LHA (U.S.S. Saipan) and solver based on the 3D multi-zone, thin-layer Navier-
helped Kurt Long measure ship airwakes. We found in Stokes method.20 A US navy destroyer, DDG51 was
some cases the flow velocity ranged from 40 knots 12 feet chosen to validate an unsteady inviscid solver with an un-
off the deck to zero velocity 3 feet off the deck. In the structured grid and low-order method.12'13 No method
mid-section of the ship the very large island has a strong to-date has been entirely satisfactory for predicting these
effect on the flow and tunnels the flow tangential to the flow fields.
island.
Flow Nature of Ship Airwake
Simulation
From previous studies, it has been shown that the
key features of ship airwakes are (1) a low Mach num-
ber (about 0.05), (2) inherently unsteady flow, and (3)
large regions of separated flow. The large separated re-
gions from superstructure sharp edges are quite difficult
to capture accurately. In addition, the wind conditions
over rough seas have to be considered, such as, the at-
mospheric turbulent boundary layer and the effect of
the wind/ship speed ratio on the turbulence intensity.
When this ratio is increased, the turbulence intensity
will decrease and its spectrum will shift to a high value
in the streamwise direction. The wind direction can vary
a great deal, since the air flow can impact the ship at any
yaw angle (even 180 degrees). The complex ship geome-
Figure 1: Oliver Hazard Perry Class Guided Missile try makes unstructured grid solvers and parallel comput-
Frigate (length=445 feet, beam=45 feet.) ers very attractive. In this paper, preliminary attempts
at high order accurate ship airwake predictions have been
made by solving a steady flow field with a well-developed
CFD method (CFL3D19) and a perturbation field with
a high-order method. The result is high-order-accurate
3D simulations. We also have some preliminary results
for steady flow predictions from an unstructured solver
(PUMA).
dgH. _j
dF i
8G u
dH —
dR dS , dE (1)
w
Figure 2: Tarawa Class LHA (length 820 feet, beam dt 8x 8y dz dx dy dz
= 132 feet). where F, G, and H are the inviscid terms and R, S,
E are the viscous terms. The results presented here will
3-3
all be inviscid. The flow field is then split into a mean p0v' + p'vo + p'v'
and a fluctuating part:
pov0u' + p0u0v' + p'u0v0
q = q0 + q' (2) +p'v'u0 + p'u'v0 + (p0 + p') u'v'
where
p'v20 + 2p0v0v' + p'
G'=( (8)
+2pv'v0 + (p + p') v'v'
P
pu
p0v0w' + p0w0v' + p'v0w0
9 = S pv (3)
pw +p'w'v0 + p'v'w0 + (p0 + p') v'w'
e
v'{e0+p0) + +(v0 + v')(e'+p')
and
p0w' + p'wo + p'w'
I rto+T
. w'(e0+p0) + (w0+w')(e'+p')
p0u' + p'u0 + p'v!
p0v' + p'v0 + p'v' (6) The mean flow source term Q is time independent:
p0w' + p'w0 + p'w'
e' (dF0 dG0 ^dH0\ dR0 dS0 8E0
Q +
On the left hand side of the NLDE are terms related --{~dx- -dy- + -dz-)+-dx- + -^ + -dz- (10)
Methodology | r>T
C hallenge
73
I- I 3MSPZ
ä$U
j
:
60
\ t 2 3 4 5
accurate manner by taking advantage of computational PSU PSU PSU MHPCC NPACI
aeroacoustics (CAA) methods. The spatial flux deriva- Computer Site
tives are calculated using seven point stencils of the
fourth order optimized Dispersion Relation Preserving Figure 3: Timings for a ship air wake case on several
(DRP) scheme of Tam and Webb.21 The time integra- parallel computers.
tion is a fourth order accurate Runge-Kutta method.
Efficient computing performance is achieved by using
a three dimensional domain decomposition strategy. The
code is written in Fortran 77 plus Message Passing In-
terface (MPI)16 and is scalable in three dimensions. As
mentioned early, the ship geometry is very complicated, Timing of NLÖE code at SP2 at NPACI
(1.86;million grid points)
even for a generic frigate test model. This makes multi-
block grid simulations and domain decomposition very
difficult. In order to make the code scalable and flexible,
a three dimensional single-block grid is used. The whole
computational domain is divided into many three di-
mensional zones. The grid points are evenly distributed
across each processor.
The NLDE solver is implemented portably on paral-
lel computers, such as, the IBM SP2 (e.g. Penn State, 30 40 50
Npaci, MHPCC), SGI Power Challenge and Pentium Number of processor
II Cluster. A comparison of code performance for the
ship airwake run on various machines is shown in Fig. 3. Figure 4: Timings for a ship air wake case on SP2.
While a 24-processor IBM SP2 is 8.4 times faster than
eight 266 MHz Pentium IPs networked together, the SP2
costs roughly 28 times more than the PC cluster. Fig. 4 spots on the deck around the control tower. The air-
gives the wall clock time for a ship air wake case with wake influences on the helicopter are quite different in
1.86 million grid points using various number of proces- these two cases. For frigates the flow separation area
sors. A 64-processor SP2 is roughly 2.6 time faster than behind the hangar cube has a strong effect on a land-
a 16-processor SP2 (when problem size is kept fixed). ing helicopter, while on aircraft carriers and LHA's the
deck leading edge vortex and separation are the key flow
Results and Discussions phenomena.
In this paper some preliminary simulations have been
In the helicopter/ship interface problem, the most im- done for a generic ship shape (TTCP ship). In fact,
portant data for coupling the airwake solution to the this is a generic frigate model and it is shown in Fig. 5
dynamics analysis of the main rotor blades of helicopter in a computational mesh. It is 240 feet long, 45 feet
are the mean flow field, the intensity of flow perturba- wide and 55 feet high. It was chosen because there are
tions, and its dominant frequencies. Such an approach some experimental investigations using the same config-
is presented in this paper. So far we have been concen- uration. It is acknowledged that the ship superstructure
trating on two types of ships: (1) frigates with helicopter does not resemble a typical frigate superstructure in de-
landing pads on the deck behind the hangar and (2) air- tail. However, from an aerodynamic point of view the
craft carriers and LHA's with several helicopter landing airwake should still be representative of that for an ac-
3-5
tual frigate since this study is concerned with the macro- Of importance to the landing operation is the flow
scopic flow properties and large scale phenomena in the condition over the flight deck. Fig. 8 depicts the contour
hangar wake. of velocity magnitude at the ship's center plane. It is
Special attention is given to the helicopter landing shown that the large region of recirculating flow extends
area which is the square, aft section of the ship. There is over the flight deck and rises higher than the hangar.
a 20 feet drop down to the landing deck from the hangar This flow region is in the landing path.
structure, which will lead to vortex shedding over the Fig. 9 shows velocity vectors in two horizontal planes
deck causing landing approach hazards. 4.75 feet and 8.75 feet above the flight deck. Our numer-
The computational grid for this problem is 201 x 109 ical results are compared with a flow pattern obtained
x 85 which results in a grid resolution of two feet or from an experimental study18 in fig. 10. It shows the
less in each direction around the ship, grid stretching flow pattern from experiments, where four distinct flow
was used to enlarge the domain. So far both mean flow regions are behind the hangar. This three dimensional
based on the structured grids and NLDE simulations vortex and reverse flow has very low speed but generally
were based on the same grid in order to avoid three di- is very unsteady and yaw-dependent. Comparing to the
mensional interpolation. In fact, the NLDE needs much experiment, the physical flow features are well captured
fewer grid points than CFL3D. by the simulation. In the numerical plots, the vortex
pair in the higher plane is much close to the center line
Mean flow simulation based on and the hangar. This indicates that there is a horse shoe
structured grids vortex as shown by the topological drawing in fig. 10.
NASA Langley and Ames research centers have de- In fig. 11 the velocity vector on the flight deck floor is
voted significant resources in the past decades to devel- compared with flow visualization results for the TTCP
oping modern CFD technology. The CFL3D 5.0 pack- ship.24 The figure (a) is the results from CFL3D us-
age from NASA Langley is used here to simulate the ing structured grids and the figure (c) is from another
mean flow which will be given as a background flow solver (PUMA) using unstructured grids, which will be
to the unsteady flow computation of NLDE. The code discussed in the next section. The photograph in the
is a Reynolds-Averaged thin-layer Navier-Stokes flow middle is from oil flow visualization.24 There are dif-
solver for structured grids. A finite volume algorithm ferences in the attachment line and the position of the
with a spatial-factored diagonalised, implicit scheme is vortex center. This is probably due to our inviscid ap-
used in discretization of the partial differential equations. proach.
The upwind-biased-differencing using the flux-difference-
splitting technique is employed.
From the experimental results, it is known that the
flow is mostly separated, with free vortices originating
from the sharp corners. There are two types of separa-
tion: one due to viscosity and the other due to sharp
corners of the blocked structures. The former is heavily
influenced by the Reynolds number. The latter is purely
an inviscid phenomenon, independent of Reynolds num-
ber. The air wake is greatly influenced by both of them.
In this mean flow simulation, we are concerned primarily
with the inviscid phenomenon and used the Euler solver
of CFL3D. The TTCP ship computational domain is di-
yaw ingle wind (m/s)
vided into 10 blocks.
The Mach number chosen for the simulation is a high
wind case. The incoming flow speed is 41 knots. The wa-
ter surface is assumed to be a hard wall boundary. Fig. 6
and fig. 7 show the contour plots of velocity magnitude Figure 6: Flow speed contours on the surface of TTCP
on the surface of the TTCP ship for the zero and 40 ship with 0 degree yaw angle wind (CFL3D).
degree yaw angle wind cases from CFL3D results. The
symmetry property of zero yaw angle flow is captured A case with 40 degree yaw angle wind was also sim-
very well. The flow is accelerated around the sharp cor- ulated. The numerical results on the front and middle
ners and there are several reverse flow regions near the bridge-deck, and flight-deck are compared with oil vi-
walls close to each corner. After the blocked structures sualization photographs in fig. 12, fig. 13 and fig. 14.
there is massive flow separation; the separation line is The three dimensional separation lines are clearly shown.
clearly shown after each block. Comparing fig. 12 (a) and (b) the flow patterns on the
3-6
100 400
Figure 5: Configuration of TTCP ship and computational mesh (1.86 million grid points).
front part of bridge deck are very similar and close to For the unstructured grid flow field predictions we are
each other since this part of the flow is less influenced using the PUMA code1 from Dr. Chris Bruner (NAWC).
by the viscosity and our nonviscous simulation is more PUMA (Parallel Unstructured Maritime Aerodynamics)
capable of accurately capturing the flow. Comparing is a computer program for the analysis of internal and ex-
fig. 13 (a) and (b) and fig. 14 (a) and (b), the flow na- ternal non-reacting compressible flows over arbitrary ge-
ture is partially captured since the viscous effects become ometries. PUMA is written entirely in ANSI C and uses
stronger downstream. In the next section, the same com- MPI (Message Passing Interface) to ensure high porta-
parisons are made for a 30 degree yaw angle case using bility and good performance.
the unstructured grid solver PUMA. PUMA is based on FVM (Finite Volume Method) and
Finally in fig. 15 the performance of CFL3D (a serial supports mixed- topology unstructured grids composed
code) is shown. We use 1.86 million grid points and of tetrahedra, wedges, pyramids and hexahedra. The
for a typical run it needs about 10,000 time iterations. code may be run so as to preserve time accuracy for un-
For this TTCP ship configuration it takes days to get steady problems, or may be run using a pseudo-unsteady
the steady state results using an SGI Power Challenge formulation to enhance convergence to steady-state. Ei-
(Single Processor). ther explicit or implicit time integration may be used.
Primitive flow quantities (density, velocity, and pressure)
Mean flow simulation based on the are computed at the cell centers and saved on exit.
unstructured grids Because PUMA uses DMA (Dynamic Memory Alloca-
While we have simulated relatively simple geometries tion) , problem size is limited only by the amount of mem-
here, more complex geometries (such as the 'island' or ory available on the machine. Since PUMA is targeted
control tower on the LHA) are important. Preliminary for distributed-memory parallel computers, which usu-
results using unstructured grids are discussed in this sec- ally have an abundance of memory, little effort has been
tion.
3-7
im
A
•^
n:
I 9-9 -
■ 46
^TVfl ^^■■feAftk.. I 3.3
^BHI
24 m/s
24 m/s
::', v/y//^////y/W/V///J/ 7/
}j/oo</oooy/////f///i'/ii it
■V <>•>•/"/■>/■ v (■' y // </ Vj // ij i
(a)
Flow Direction
(b)
ry^'- /
(«0
Figure 11: Predicted surface flow velocities compared to Figure 12: Predicted surface flow velocities on the front
surface oil flow images on the flight deck.24 (a) CFL3D, part of bridge-deck compared to surface oil flow images
(b) oil flow visualization, (c) PUMA with 40 degree yaw angle wind.24 (a) CFL3D, (b) oil
flow visualization
«— 24 m/s _ 24 m/s
^ \ \ \\\\
I \ \\\'V
11 \
'/'/'/'/'/'/'l'l'l l X \
1
s f 's -s v v •!/ '!■' v ■•/ // //
(a)
Flow Direction
Flow Direction
j BK«j»:*-...-i.'.j-.;
Figure 13: Predicted surface flow velocities on the middle Figure 14: Predicted surface flow velocities on the flight-
part of bridge-deck compared to surface oil flow images deck compared to surface oil flow images with 40 degree
with 40 degree yaw angle wind.24 (a) CFL3D, (b) oil yaw angle wind.24 (a) CFL3D, (b) oil flow visualization
flow visualization
z
140
Tlmtnn fnr TFP SP -
a. 120
132
~:
«a>
J= 100 IBMSP2 -
5 SGI
9Z Power Challenge
3 »«
SI fiSt B2 > -
ii 60 67
E S
111 S [
£ 40 o_ ~
3 ■ox ;
E HE» S
Q.
= |
U 20
E
a II c 0.^ ;
Computer Name
** \ v \ •- v \
\ \ V N. >. ».
\ \ \ S.
•* \ \ N
(a)
\%mk
Figure 17: Predicted surface flow velocities on the middle Figure 18: Predicted surface flow velocities on the flight-
part of bridge-deck compared to surface oil flow images deck compared to surface oil flow images with 30 degree
with 30 degree yaw angle wind.24 (a) PUMA, (b) oil yaw angle wind.24 (a) PUMA, (b) oil flow visualization
flow visualization
[18] Rhoades, M. M., and Healey, J. Val., "Flight Deck [24] Zan, S.J., Unpublished data from the Aerodynamics
Aerodynamics of a Nonaviation Ship," Journal of Laboratory, Institute for Aerospace Research, Na-
Aircraft, Vol. 29, No.4, July-August 1992, pp. 619- tional Research Council of Canada, 1998.
626.
[19] Krist, S.L., R.T. Biedron, and C.L. Rumsey,
"CFL3D User's Manual (Version 5.0)," NASA Lan-
gley, Nov., 1996.
[20] Tai, T.C. and Carico, D., "Simulation of DD-963
Ship Airwake by Navier-Stokes Method," AIAA 93-
3002.
3-12
u'(m/s)
6
u'(m/s) 4.8
3.6
2.4
1.2
0
-1.2
-2.4
-3 6
•. - -4.8
JBiillllii -6
>"*"' VWfgTVSF?
6
v' (m/s) 4.8
3.6
2.4
1.2
0
-1.2
.-2.4
-3.6
*». j^ -4.8
6
w' (m/s) 4.8
1 3.6
2.4
1.2
0
-1.2
■I
-2.4
-3.6
-4.8
-6
Tsze C. Tai*
Naval Surface Warfare Center, Carderock Division
Code 5300, Bldg 4E Carderock Div.
9500 McArthur Boulevard
West Bethesda, MD 20817-5700, USA
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
4-2
geometry representation and obtained reasonably The NASA Ames 3DGRAPE code [8] is used for
useful steady-state flow results. basic grid generations. A cylindrical grid topology is
adopted for its capability to treat a body with a sharp
The present effort is an extension of previous nose. The topology is basically an H-0 mixed type,
work on the DD-963 and LPD ships. The same with H-type in the longitudinal plane, and O-type in
multi-zone, RANS scheme used in Refs. 4 and 6 is the crossflow plane. The outer cylindrical surface is
employed with an enhanced grid in the simulation. set at 2.5 ship lengths from the ship centerline. The
most forward plane is set at 1.0 ship length from the
3. SIMULATION METHOD bow of the ship, and 3.0 ship lengths for the wake.
The simulation method, or the computational An overall coarse grid is generated first by
method employed has been described previously. For 3DGRAPE using a multi-block procedure. The
completeness, the method is briefly described here. radial distances are then clustered near the surface
It contains two main elements: 1) the grid generation and stretched in the outer region for shear layer
and 2) the flow solving. development. The complete grid has a total of 187 x
123 x 57 points with 57 points in the radial
3.1 Grid Generation and Grid Topology direction. This number is larger than those used for
analysis of aircraft because of the very low speed
A structured, curvilinear, body-conforming grid freestream involved in the present work. Previous
is generated. The numerical model for the surface work [9] indicates that when using a compressible
grid is derived by using the Tai simplification flow code at low Mach numbers, an increased mesh
scheme called the area-volume rule for representing density is needed. The details of the grid generation
the surface components of a ship [8]. Figures 1 and and advantage of the grid topology adapted are
2 show the actual ship and resulting numerical presented in Ref. 7. The grid resolution was
models. In the latter, the cargo door and elevators enhanced by increasing the number of points in all
are closed and the antennas are neglected. three directions.
model has been regarded as the best turbulence All the computations were performed on the
model for flowfield dominated by vortical flow. Cray facilities at the DoD High Performance
Computing facility at the Naval Oceanographic
3.3 Boundary Conditions Center. Converged solutions were obtained in
about 8,000 to 9,000 iterations (time steps)
The boundary conditions for the Navier-Stokes requiring approximately 8 to 9 hours of Cray C-90
flow solver are: 1) atmospheric boundary layer flow CPU time. The large number of iterations is due to
upstream, 2) atmospheric pressure recovery the very low freestream Mach number used in the
downstream, 3) characteristic form of inflow-outflow compressible flow solver. The required CPU time is
at the cylindrical outer boundary, and 4) viscous reasonable and affordable in today's environment of
nonslip flow at the surface of the ship. The computer capability and resources.
atmospheric boundary layer is approximated by a
power-law profile: 4.1 Particle Trace
Fig. 5 - Particle trace over LHD ship at even keel at 12.87m/s (25 knots)
and zero wind angle. Astern view.
4-5
Fig. 6 - Particle trace over LHD ship at even keel at 12.87 m/s (25 knots)
and wind angle of 10 degrees. Perspective view.
TOUCH-DOWN
POINT
FLIGHT DECK
Fig. 7 - Particle trace over LHD ship at even keel at 12.87 m/s (25 knots)
and wind angle of 10 degrees. Front view.
4-6
Fig. 8 - Particle trace over LHD ship at even keel at 12.87 m/s (25 knots)
and wind angle of 20 degrees. Perspective view.
Fig. 9 - Particle trace over LHD ship at even keel at 15.44 m/s (25 knots)
and wind angle of 20 degrees. Astern view.
4-7
Fig. 12 - Longitudinal velocity distribution above Fig. 14 - Vertical velocity distribution above
touch-down; VM = 12.87 m/s (25 knots), ß = 0. touch-down; V«, = 12.87 m/s (25 knots), ß = 0.
The velocity in the transverse direction, V, has Larger variations in velocity components are
slightly negative values (from starboard to port) at observed along the line about one-ship length aft of
low altitude, but gradually becomes positive above the touch-down point. Completely out of the
100 feet from touch-down. It then diminishes to narrowed flow path and almost totally in the wake,
zero as the altitude increases, see Fig. 13. Again, the longitudinal velocity U exhibits large
these values are outside the boundary layer. The deficiencies at low altitude. It recovers with a steep
variation is rather small because of the zero wind ascent between 50 to 150 feet, and then gradually
angle. The vertical velocity, W, starts at a slightly approaches the freestream value as altitude
positive value (upward) and then becomes negative increases, see Fig. 15. The reason for changes in the
(downward) in the entire altitude range before it velocity slope above 150 feet level is believed to be
vanishes, see Fig. 14. due to the height of the superstructure involved.
4-9
o
"3 1
>
Fig. 15 ■ Longitudinal velocity distribution in the Fig. 17 - Vertical velocity distribution in the
plane one-ship length aft touch-down; plane one-ship length aft touch-down;
VM = 12.87 m/s (25 knots), ß = 0. V«, = 12.87 m/s (25 knots), ß = 0.
sy
P.Tattersall
C.M.Albone
M.M.SoIiman
Aero/Structures Department, DERA,
Farnborough, GU14 OLX, UK
C.B.Allen
Department of Aerospace Engineering,
University of Bristol,
Bristol, BS8 1TR, UK
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
5-2
predicted with good accuracy using the (inviscid) Euler successively embedded, such that smaller grid cells
equations. Accurate modelling of the downstream occur near a solid surface. In order to apply solid
development of the vortices, however, requires the surface boundary conditions, a particular type of
inclusion of viscous effects. There are a number of curvilinear mesh is associated with each "feature" of
arguments in favour of neglecting viscous effects, as the geometry. For example, an O-type mesh is
was chosen to do here, at least as a first step. The most associated with an aircraft wing section, whereas a
important is that, since the area of biggest concern for cylindrical polar mesh is associated with the
flow prediction is in the immediate vicinity of the intersection line between a wing and a fuselage. These
flight deck and separation from sharp edges can be curvilinear meshes overlap each other and the
captured by an inviscid method, it is reasonable to Cartesian blocks. They also control the embedding of
investigate how accurately the resulting flow field can the blocks. Communication between the various
be modelled without viscous effects included. In this meshes is governed by a set of hierarchical rules.
context, the information required from a flow
prediction is important. If the complete, accurate For application to ship air wake prediction, the whole
interaction and downstream decay of vortices is curvilinear mesh philosophy was removed from
needed, then an inviscid model is unlikely to be FAME, partly because generation of these grids, for
adequate. If, on the other hand, a general indication of bodies as complex as ships, is very time-consuming.
flow severity, such as maximum downwash in the Control of the embedding of the Cartesian blocks,
plane of the helicopter rotor, is sufficient, an inviscid originally carried out by the curvilinear meshes, now
model may well prove effective. takes place with a very small set of user-specified
parameters. The finest grid which can be used to obtain
It was expected that the influence of the helicopter a flow solution, on an SG Indigo 2 workstation with 32
rotor on the air wake would be significant and the MW of RAM, takes less than ten minutes to generate.
capability to model this by the actuator disc approach This includes all the flagging of the various types of
was included. Methods for improving the rotor grid point (internal to body, normal field, actuator disc,
downwash representation have been investigated interpolation required, etc.), for any geometry.
recently and are described later. It was intended that a
CFD tool should be produced which could be used for A computationally convenient device is currently used
routine design purposes within the next few years. This to describe the ship geometry. The ship is defined by a
means that a time-accurate, viscous method is unlikely collection of hexahedral elements, because each
to be a viable candidate. Of the two aspects, it was hexahedron can be divided into six tetrahedra and there
expected that it would be more important to model the exists a very simple test to find whether a point lies
variation with time. The evidence to date suggests that within a tetrahedron. This is useful because, during the
this is a valid assumption and this enhancement is grid generation process, special points, which lie inside
currently being developed. Initial results are presented the body but have neighbours outside, must be
later' in this paper. identified, since these are required for the
implementation of solid-surface boundary conditions.
2. MESH GENERATION More general numerically defined geometry can be
handled by routines existing within the FAME suite.
In light of the considerations described in the However, this has not yet proved to be necessary. A
introduction, it was decided that a "non-aligned grid" section through a typical grid, for an AOR (Auxiliary
approach should be adopted. This makes the generation Oiler Refueller), plus a simplified helicopter body, is
of a computational grid very easy, but necessitates shown in fig.l, highlighting the embedding of the
some special procedure within the flow solution code blocks near the ship and the non-aligned nature of the
in order to apply the solid surface boundary condition grid.
accurately. It should be emphasised, at this point, that
the approach taken was not to develop a state-of-the- 3. FLOW SOLUTION
art, non-aligned grid capability [3]. This would have
entailed starting "from scratch," requiring more Two alternative numerical techniques have been used
development time than was available. Instead, an to obtain the flow solutions presented here. In both
existing CFD method [4], known as FAME (Feature cases, a compressible flow solution code already
Associated Mesh Embedding) was used as the starting existing within the FAME system was modified to
point. solve the incompressible Euler equations, using the
technique of artificial compressibility [5].
FAME was developed, originally, for civil and military
aircraft applications. It solves the problem of 3.1 Governing equations
generating surface-aligned grids for highly complex
geometry by using a series of overlapping and The approximation of incompressible flow is a valid
embedded, Cartesian and curvilinear meshes. The one for the prediction of ship air wakes. The equations
whole region of interest (away from solid surfaces) is can be solved by conventional CFD methods by adding
covered by a set of regular, Cartesian blocks, which do a fictitious time derivative of pressure term to the
not, in general, align with the geometry. They are continuity equation. This terms vanishes in the steady
5-3
used to reduce turnaround times, since the main point deck will be separations from sharp edges, which can
of this exercise was to check that the CFD/CRFM be predicted by numerical solution of the inviscid
interaction process was feasible. Since much of the equations. It is essential to include some representation
detailed change in disc loading is concentrated in of helicopter rotor effects on the airflow. This can be
relatively small areas, particularly near the outer edge, done, as a first step, by a constant actuator disc
it may be that a finer CFD grid is necessary to capture approach. Improved modelling can be achieved by use
the resulting changes in the flow field satisfactorily. of a variable actuator disc. One way to decide on the
Fig.5b shows how the sequence of CFD/CRFM variation required across the disc is to use initial CFD
calculations produces a predicted change in collective results as boundary conditions within a helicopter
pitch and cyclic control settings and that the iterative performance code. Such a code can produce a new
sequence achieves convergence. It is hoped that the variation across the disc, which is then fed into an
veracity of the predicted changes, for this or a similar updated CFD calculation, before returning to the
case, can be checked against experimental data in the performance code, etc. This iterative process has two
near future. potentially important effects. The first is to produce a
more accurate CFD solution than that obtained with a
The air wake over a real ship superstructure is highly constant actuator disc model. The second is to provide
unsteady. Indeed, it is not obvious that a steady CFD predictions of the effect of the air wake on the
solution would be of any use whatsoever in predicting helicopter, hence aiding the understanding of the
this situation. The general conclusion from the testing handling problems suffered by the pilot during
carried out to date is that steady solutions can be of shipborne operations.
use, although more evidence is still needed to back this
up. However, it is undoubtedly the case that the effects The highly unsteady nature of a real ship air wake will
of variation of the flow field with time must be undoubtedly be an important consideration.
investigated. The route chosen towards time-accurate Modification of the existing code to include time-
CFD solutions was described earlier. An initial result varying effects has taken place, using an efficient
has been obtained for the geometry tested as part of the implicit scheme. Initial results are encouraging,
TTCP collaboration described in ref.14. A relatively although much more testing and potential development
coarse spatial grid was used to facilitate a more rapid is necessary.
turnaround of the solution. This particular test case
involves a port/starboard symmetric geometry and a Even with steady, inviscid flow solutions, much
yaw angle of zero degrees (wind from dead ahead). It is progress has been made on the prediction of realistic
to be expected that the real flow will vary with time, ship air wakes. Indications are that a useful predictive
but that a long time-averaged view of the flow pattern tool has been developed during this project.
will look symmetrical. Fig.6 shows two views of the
computed flow field, one corresponding to a particular
instant in time and one to an average of the solution
over a large number of real time steps. It is clear that REFERENCES
the instantaneous solution is highly asymmetric,
whereas the averaged view is almost perfectly 1. Breuer M, Lakehal D and Rodi W. Flow around a
symmetrical. This average solution agrees reasonably surface-mounted cuboidal obstacle: comparison of
well with that observed in the experiment [14], LES and RANS - results. In Notes on Numerical
although it was found, in the context of steady flow Fluid Mechanics, vol.53, Vieweg.
solutions, that a finer grid is necessary to really obtain 2. Albone C M, Catherall D C, Hall, M G, Joyce
a good representation of this situation. Thus, for a very Gaynor. An improved numerical methodfor
preliminary investigation, this result can be regarded as solving the transonic small-perturbation equation
highly encouraging. for the flow past a lifting aerofoil. RAE TR74056.
3. Powell K G. Solution of the Euler and
6. CONCLUSIONS magnetohydrodynamic equations on solution-
adaptive Cartesian grids. VKI Lecture series
A method has been presented for modelling the air 1996-06.
wake over a conventional naval ship, with particular 4. Albone C M. An approach to geometric andflow
attention paid to the region of an aft-located helicopter complexity usingfeature-associated mesh
landing deck. Grid generation is reduced to a very embedding (FAME): strategy andfirst results. In
simple problem, by use of the "non-aligned mesh" Numerical methods for fluid dynamics III, Oxford
technique, in which no attempt is made to align any University Press, 1988.
grid lines with solid geometrical surfaces. Accuracy is 5. Chorin A J. A numerical method ofsolving
enhanced in the region of solid surfaces by the use of incompressible viscous flow problems. Jnl. Comp.
grid embedding. Phys., Vol.2, pp. 12-26, 1967.
6. Albone CM. A second-order accurate scheme for
The airflow is modelled by solution of the Euler the Euler equations by deferred correction of a
equations. This relies on the assumption that the first-order upwind algorithm. RAE TR88061.
dominant flow features over an aft-located helicopter
5-7
7. Onslow S H and Peshkin D A. A TVDflow 13. Hansford R E. The development of the Coupled
algorithm for FAME. Proceedings 5th ICFD Rotor-Fuselage Model. 48* AHS Forum, 1992.
Conference, Oxford, 1995. 14. Wilkinson C H, Zan S J, Gilbert N E, Funk J D.
8. Allen C B, Onslow S H and Tattersall P. Parallel Modelling and simulation ofship air wakes for
implementation of an unsteady Euler solver using helicopter operations. A collaborative venture.
overlapping and embedded meshes. 4 Paper No. 9, AGARD Symposium, Amsterdam,
ECCOMAS CFD Conference, Athens, 1998. 1998.
9. Jameson A. Time dependent calculations using
multigrid, with applications to unsteady flows past ACKNOWLEDGEMENTS
airfoils and wings. AIAA paper 91-1596.
10. Allen C B. The reduction of numerical entropy This research is funded by the UK MOD under ARP package
generated by unsteady shock waves. Aero. Jnl., 15f, assignment 2, research objective 1.
Vol.101, No.1001, 1997,pp7-16.
11. Allen C B. Grid adaptation for unsteady flow ©British Crown Copyright 1998/DERA
computations. IMechE Journal of Aero Eng., Vol
211, 1997, pp.237-250. Published with the permission of the Controller ofHer
12. Payne P R. Helicopter dynamics and Britannic Majesty's Stationery Office.
aerodynamics. Pub. Pitman, London, 1959.
ILLUSTRATIONS
Figure 2a. Air wake over AOR + helicopter with no rotor effects-velocity vectors
Figure 2b. Air wake over AOR + helicopter with no rotor effects-stream ribbons
5-9
—9—9—f~P—F—$-~P--
Figure 3. Air wake over AOR + helicopter with constant actuator disc
Figure 4. Air wake over AOR + helicopter with variable actuator disc
5-10
Figure 5a. Results from sequence of iterations between CFD and helicopter performance code
' 5
Iteration No
Figure 5b. Results from sequence of iterations between CFD and helicopter performance code
5-11
/ S S s»
->—>—>-
t r / s -> s^
\ \ V \ \ \ \ t t / ^ _~ _
^*
\ \ \ \ \ \ \ \ \ i , s* ^
V \ \ \ \ \ \ \ \ \ I / s
V \ \ \ \ \ t / / ^
V N \ \ \\\\\ \ \ 1 r /
/ *"* v N \ \ \ \ \ \ 1 ! t / / /
^ V \ \\\ f / ///// /
S ' I \ N. ^M
-> 5> J
\ \
/ /
1
— •*.
/ N \ \ \ 1 / y
I
v \ / s s*
1. ABSTRACT 2. INTRODUCTION
This paper describes a French activity on helicopter Flying a helicopter above or around a frigate deck,
ship landing operations simulation improvement. landing on it or taking off from, are considered by
ONERA, under SPAe funding (Service des pilots as highly risked operations. Indeed, not only the
Programmes Aeronautiques), has developed an frigate moves, but also in the neighbourhood of the
aerodynamic disturbance model around the landing deck, the helicopter has to face with the changing
deck of a frigate La Fayette, and has tested its effects aerodynamic wake of the ship superstructure. This
on helicopter flight dynamics. unsteady flow provides high mean speed gradients to
which aerodynamic turbulence (fluctuations) is added.
Wind tunnel tests were carried out in ONERA-IMFL These conditions have important effects on helicopter
with a 1/50* scaled model of the La Fayette. Three global performance, and behaviour.
dimensional measurements of the mean wind speed and
velocity fluctuations were performed around the Under SPAe funding, ONERA has realised wind tunnel
landing deck area. tests on a 1/50* scaled La Fayette frigate model, and
has developed an aerodynamic wake model of the ship
The La Fayette aerodynamic wake model includes a landing area. This model was then used in order to
mean wake model [1] to which a turbulence model adds study the effects on flight mechanics.
velocity fluctuations. The turbulence model is realised
using power spectral densities of velocity fluctuations The main following topics are presented in this paper :
measurements. 1- Wind tunnel tests,
2- Analysis of the test results obtained,
The model was connected to the Eurocopter Helicopter 3- Definition and development of a mean air-wake
Overall Simulation Tool (HOST). Simulations of model,
flights above the deck with this model demonstrated 4- Definition and development of a turbulence model,
important effects of the ship air-wake on the helicopter
5- Implementation in HOST (Helicopter Overall
flight dynamics.
Simulation Tool),
6- Demonstration of the frigate wake effects (mean
This paper describes the test results obtained and the
velocities + fluctuations) on helicopter loads and
work performed in modelling the La Fayette wake and
flight dynamics.
its influence on the helicopter flight dynamics
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
6-2
Figures 9 and 10 are 3D flow visualisations of around the median axis (longitudinal), is to be noticed.
measurements at 50 kts with respectively 15° and 180° This asymmetry is due to higher deck's aerial elements.
(rearward) side-slip conditions. The effect of the lateral
hangar wall can be seen for a side-slip angle of 15°. b- Wind velocity = 50 kts. Side-slip ß = 15°
Comparisons of mean velocities evolution for 2 wind In order to study the side-slip effect on air-wake
speed conditions (50 kts and 25 kts) without side-slip velocity fluctuations, measurements have been done at
(ß = 0°) and 3 side-slip conditions at 50 kts (ß = 0°, 50 knots with a 15 degree side-slip.
15°, 180°) [1] provide the following conclusions on the
mean air-wake model usage domain : Two points have been selected, in order to illustrate a
comparison between the 2 tested side-slip
• A detailed database is available at 50 kts of frontal configurations (figure 16):
wind (ß=0°); 1- point A, which remains in the frigate superstructure
main wake for both cases,
• A detailed database is also available at 50 kts with 2- point B, which is out of this main wake at 15° side-
15° of side-slip. The mean flow changes slip.
considerably compared to the case of 0° side-slip.
However, it has been decided that interpolation Figures 17 and 18 are comparisons, respectively on A
between 15° and 0° of side-slip at 50 kts will be and B, of the 3 velocity components PSDs along the
"tolerated", in order to extend the database to vertical axis. Large differences can be seen between the
intermediate side-slips; 2 cases.
A partial database has been generated at 25 kts, As it was already concluded for the side-slip effect on
without side-slip. This investigation shows that mean velocities [1], these results show that the side-slip
except above the deck, the mean wind normalised angle is a quite important factor acting on turbulence
components have similar evolution. Above the PSDs and it is not possible to extrapolate these results
deck, these components will have to be interpolated for side-slip configurations higher than 15°. Even an
or extrapolated using the results at 25 and 50 kts; interpolation between 0 and 15° is critical. However, it
was decided to extend, by interpolation, the La Fayette
A 50 kt rear wind database is also available. aerodynamic turbulence model to intermediate side-
slips (0 to 15°) at 50 knots of wind speed.
Figures 14 and 15 illustrate the vertical velocity (w) Additional investigations have shown that the wind
PSD evolution respectively along the median axis (on speed effect on PSD amplitudes is much reduced when
the deck and one deck length behind) at a height of 2.4 the point considered is far from the deck.
m (fig. 14) and along the lateral axis crossing the centre
point A (from one deck width left side to one deck Concerning the frequency of PSDs maximum (fmax), a
width right side) at the same height (fig. 15). These relationship between fmax and the wind speed
figures show again that turbulence is much higher (Strouhal number similarity) can be expected:
above the deck than beside it. A small asymmetry
6-4
• A partial database has been generated at 25 kts, The approach consists in locating the helicopter centre
without side-slip. This investigation shows that of gravity in the test area elementary parallelepiped.
except above the deck, the normalised velocity Fluctuations PSDs on this point are defined using a
components fluctuations have similar evolutions linear combination of measured PSDs on the
with however, the respect of Strouhal number elementary parallelepiped tops, in respect with their
similarity. Above the deck, PSDs will have to be distance to the considered point.
interpolated or extrapolated using the results at 25
and 50 kts; Fluctuations are then processed using a signal
generation method ensuring the similarity between
• A 50 kt rear wind database is also available. PSDs of measurements and of the generated signal.
c- Velocity fluctuations computation from the parameters (helicopter attitudes, ground speed
identified PSD, using the following method components, ...) and trajectory coordinates. Figure 29
[4]: shows the aerodynamic loads on different helicopter
elements (main rotor, fuselage and horizontal
Example of u generation: stabiliser). Blades flapping and the local turbulence
terms (DUF-FGW, DVF-FGW, DWF-FGW) are also
u(x,y,z,t)=2ZyjS(x,y,z,fi)-^.co^2jfit+(fi(x,y,z)] presented on this figure.
1=1
N\ Number of samples in S These plots show a short deceleration at the beginning
ff.Temporal frequency for f sample of the simulation and a quick fall in altitude with a
large lateral deviation (20 m to the left).
(p>: Random phasebetweenOand TjZMithaunform
probability density It is interesting to compare these results obtained with
turbulence and mean aerodynamic velocities, with those
coming from the same simulation but in mean velocities
effect only.
Figure 25 is an example of vertical turbulence
generation on the deck centre (height = 2.4m) at 50
knots of wind speed. Figures 30 and 31 illustrate this case. Figure 30 shows
that the longitudinal component of the air-wake (UMF-
FGW) decreases during the approach because the
6. EFFECTS ON HELICOPTER LOADS AND helicopter comes closer to the deck. The vertical air-
FLIGHT DYNAMICS wake (WMF-FGW) is all the time negative (down
The frigate air-wake model has been connected to the wash) and changes not only with height but also with
Eurocopter simulation code HOST (Helicopter Overall helicopter X and Y positions above the deck.
Simulation Tool) [5]. The connection was done The combined effects of these two air speed
assuming the helicopter as a mass point. components result in a general decrease of the main
rotor lift (FZG-RP), and consequently a loss of altitude.
The air-wake model uses some general data coming The helicopter pitches down and accelerates because of
from HOST (wind velocity and direction), and specific the pitch moment decrease (MYG-RP). Simulation
ship state data (velocity and heading). The ship stops after 8 seconds when helicopter altitude is zero
trajectory is then calculated in the air-wake model, and attitudes have changed by -7° in pitch, -11° in roll
assuming a constant speed. Figure 26 illustrates this and -25° in heading.
implementation.
The comparison of these 2 trajectories shows that for
the first simulation (mean air-wake + turbulence), the
The air-wake model fulfils the following tasks: altitude loss has been delayed because of an initial lift
raise, this last being caused by 'favourable' initial
• generating the ship trajectory.
turbulence conditions. Other simulations have
• locating the helicopter in the test area grid.
demonstrated the opposite effect, with a higher altitude
• calculating the 3 velocity components at the
loss compared to the simulation with the only mean air-
helicopter CG.
wake. In these cases, 'non favourable' initial turbulence
generates a lift loss and a quicker altitude fall.
In order to demonstrate the frigate aerodynamic wake
effects on helicopter flight dynamics, off time open
It comes out from these open loop simulations that
loop simulations of flights above the deck were carried
turbulence can have a positive or a negative influence.
out with HOST.
However, these turbulence effects occur around a
general tendency given by the mean velocities and will
Typical results are presented on figures 27 to 29 for a
certainly have an effect on piloted simulations.
descending flight on the deck at a speed of 10 kts and a
slope angle of -1°. The frigate is still and faces with a
50 kt forward wind (ß=0°). The helicopter, a 3.5 t
7. CONCLUSION
Dauphin, is trimmed with the local mean air-wake
This paper presents an ONERA activity on helicopter
conditions at the starting point located on the
ship landing operations simulation improvement.
longitudinal symmetry axis (y-0), 9 m above the deck
and 23 m behind its rear extremity (X=25 m in the
The first phase of this activity started with wind tunnel
frigate axes).
tests in ONERA-IMFL, on a l/50th model of the French
frigate La Fayette. A detailed database was provided at
The simulation being realised in open loop, the
50 kts for 3 wind side-slip configurations (0°,
helicopter controls keep their trim values (figure 27).
15°,180°). A partial database was also generated at 25
Figure 28 illustrates the local wind components (UF-
kts of wind with 0° side-slip. The data analysis showed
FGW, VF-FGW, WF-FGW), the helicopter flight
6-6
REFERENCES
FREGATE LEGERE
«Gs=£T- version crotale
ED
FT
F^
m
\mtt-G*- i ^ I "H
Test Configurations
I 1
WHDPipKTWrlS
Planes height to the deck
FuelS"
Rearo* I
I
*""F
2,4 ml 4,4 m/6,4 m/ 8,4 ml10,9 m/13,4 m/15,9 ml 18,4 m
n.
g0.5-
-20 -10 10 20 30
| 0.5
Z>
-20 -10 20 30
0.2
-0.2
-20 -10 10 20 30
0.2
-0.2
-20 -10 0 10 20 30
Xf(m)
Figure 6 : Mean velocities evolution along the longitudinal axis on the deck
centre line (height=4.4m)
Wind = 50 kts, ß = 0°
LATERAL AXIS
0.5-- ON THE DECK
CENTRE
-20 10 20 30 SYMETRY
LONGITUDINAL
AXIS
§0.5-
3
-20 10 20 30
-0.2 30
-20 -10 10 20
0.2
-0.2 30
-10 0 10 20
Yf(m)
Figure 7 : Mean velocities evolution along the lateral axis on the deck centre
(pt A, height=4.4m)
Wind = 50 kts, ß = 0°
height = 6.4 m
Figure 9 : 3D visualisation of tests at 50 kts ; ß = 15°
frtrr n
^
nm
ww
height = 2.4 m
TURBULENCES NORMEES
■wtn*v^#
1 SO EOO 2SO 300 350 400
M O
1 50 200
WMpNfli2SO 300 350 400
5 0 0 0
D 3 5 0 5 3 5
0.01 0.01 V/.U 1 0.01
0 0 0
3 5 3 5 0 5 3
0.01 0.01 ""' 0.01
0
^ 0
,
3 5 ) 5 0 5 0 5
Z(m) 0.01 0.01 U.UI
0
/v_ 0
A^_ 0
^
E c 5 0 5 C J
f o.oiC 0.01 0.01
03
Q.
to 0.005 .005 -1.005 .005
r%
E o n 'V—.... o^- 0
~-—-
0 5 0 5 0 5 "0 5
Hz Hz Hz Hz
Figure 13 : u, v, w PSDs evolution above the deck centre
Wind = 50kts;ß = 0°
freq(Hz) X(m)
Figure 14 : Vertical velocity PSD evolution along the deck median axis
(height 2.4m) - Wind = 50 kts ; ß = 0°
Figure 15 : Vertical velocity PSD evolution along the lateral axis on the
deck centre (height 2.4m) - Wind = 50 kts ; ß = 0°
6-11
0 5 0 5 0 5 0 5
0 5 0 5 0 5 u 0 5
0 5 0 5 0 5 0 5
0 0
0.01
) 5 0 5
0.01
) 5
0.01
) 5
0 0 0
0.01
J 5
0.01
) 5
0.01
) 5
0.01
) 5
!
5 0.005 1.005 1.005 ,\ 1.005
f<—
0
f^_
0
r~~±^ 0
,—.,
t 0.01
3 5
0.01
) 5
0.01
) 5
0.01
J 5
"Q.
tS 0.005 1.005 1.005 ■ .
1.005
S o
'V.
(1
^^_.. 0 /4^ 0
~_..„..
~P=0P
pv=f(z) pw=f(z)
1 °'01
£ 0.005
CO
5 o
*- 0.01
(
o
5 0
0.01 1 0.01
S 0.005
0
I
0.01
5 0.005
0
0.01
(
5 0.005
0
0.01
( 0.01
5 0.005 1.005
0.01
( 0.01 0.01
5 o 0
^
Figure 20 : Comparison of u, v, w PSDs evolution in altitude on the deck centre
(ptA)
wind = 50 kts and 25 kts ; side-slip 0°
puw=f(z)
c 0.01 0.01 0.01
5 0.005 1.005
ot
0
0.01 0.01
5 0.005 1.005
0
0.01
( 0.01
5 0.005 1.005
0
C 0
0.01 0.01 r
0.01 0.01
5 0.005 1.005
0.01 0.01
E 0.005 1.005
8 0.01
S 0.005 1.005
0 5 0 5 0 5 0 5
Hz Hz Hz Hz
Hz Hz Hz Hz
Figure 21: Comparison of u, v, w PSDs evolution in altitude on point C Figure 22 : Comparison of u, v, w PSDs evolution in altitude on point D
wind = 50 kts and 25 kts ; side-slip 0° wind = 50 kts and 25 kts ; side-slip 0°
6-13
i Objective:
j Avoiding velocity discontinuity when entering the test area
i Method:
1 - Creation of a transition area.
2- Interpolation of air-wake mean velocity between the test area
and infinite free stream.
30 m 25 m
n
littiillli * —t-4-j— 30 m A
IT' 1 - - ■r/A
/f
lt- !*„> ----Z--
1 K=n
f> ,i
; i
E»i
u n xf k
1
(1
..
Spectre 1
h2
-T 1 r-
0 1 2 3 4 5 6
Figure 25 : Example of vertical velocity turbulence generation on
the deck centre
6-14
Wind
Frigate Speed Frigate Wake Aerodynamic velocities in the ship air-wake
Helicopter
Model
Frigate Heading
t Helicopter positions
• 5B-5
',,„!,„, ■■ Ml, ,,,,!,„,
TEMPS S
LK.'J
: "^-L :
X ■
YFH-F3H H . ZFH-FGW H
\ '
f i •.... i.... 1 1
CONDITIONS: equil2
DOF-R3W H/S.f DVF-rcw M/S., DWF-FGW M/S
:::::::::;:::.:::::::;:iSsj^:/:::;
TEMPS S
FY-CEL DAN
0
-20 ZZ\ZZ"^".'.'.'."'r'. \Z\Z'{Z'S^it
. . . . ; ,,,. 1 ....:... 11
„FY-EM? DAS
1. SUMMARY and decreased mean wind velocities in the lee of the ship
The operation of helicopters from a ship-based platform is a superstructure, and large-amplitude ship motion. Many of
challenging procedure from a pilot's perspective. In addition these limitations are due, in whole or in part, to the air wake,
to manoeuvring near a moving platform, the pilot must the region of complex flow around the ship superstructure.
respond to the complex, unsteady flow field which exists over Air wakes, such as that in the lee of a ship superstructure are
the flight deck. The flow over the ship superstructure characterized as complex, unsteady flow fields containing
generates an air wake which contains significant spatial recirculation regions, "flapping" shear layers, strong vortical
gradients in the mean wind velocities as well as increased flows and turbulence with length scales of the order of
turbulence levels; both may significantly alter the helicopter rotor diameter.
performance of an immersed helicopter rotor. To address this
issue completely, one must deal with a fully-coupled problem Inadequate yaw control will occur when the flow through the
including both the helicopter rotor and the ship air wake. tail rotor is reduced because of the air wake. Torque
However, as a first step, the ship air wake can be mapped out limitations can become important when the pilot is attempting
to estimate what inflow the rotor might experience. to track large vertical deck motions in high sea-states and as
the rotor translates toward the flight deck on the final descent.
A program is under way in the National Research Council of In the latter case the rotor is immersed in the reduced-
Canada Aerodynamics Laboratory (NRC/AL) to analyze the momentum flow of the air wake thereby causing a decrease in
air wake formed behind the Halifax-Class Patrol Frigate thrust and requiring the pilot to initiate a corresponding
(CPF). It contains an experimental component and a increase in rotor collective. A reduction in the usable cue
computational one. Using hot-film anemometers and the AL environment can occur because sea spray is obscuring the
5m Vertical Wind Tunnel, the experimental element maps out pilot's vision, or because the existing windshields on the
the flow field (14x13x16 matrix) in the vicinity of the flight helicopter do not permit adequate viewing of the flight deck
deck of a 1:50 scale CPF model at 0° yaw and 12° yaw. The during the final stages of landing. In the latter case, the pilot
region measured included the boundaries of the recirculation may choose to pitch, bank or yaw the helicopter in order to
zone that exists behind the hangar. An atmospheric boundary improve the visual cueing. Such changes in helicopter
layer profile was created by placing a set of horizontal rods attitude can result in changes to the stability and control
upstream of the model, which generated a properly scaled characteristics of the helicopter when compared to the same
variation of mean velocity with height. attitude away from a ship because of the non-uniformity of the
air wake. Increased turbulence arising from separated and/or
To complement this work, the flow field around a modified vortical flow in the lee of the superstructure will effect time-
CPF was computed using a Navier-Stokes flow solver. This varying changes in rotor output and be sensed by the pilot's
modification exists for these experiments only. No such inertial cueing system. These motion cues are likely to occur
modifications exist on the real ships. In order to facilitate grid over a frequency bandwidth wherein the pilot will attempt
generation, simplifications were made to the bridge and corrective action (McRuer, 1994). The air wake turbulence,
several of the smaller structures around the flight deck were characterized by intensities and length scales, will correlate
removed. A structured, multi-block, pressure-based Navier- with the pilot's "aerodynamic workload." That is, as
Stokes flow solver was used to compute the steady-state flow turbulence increases, the effort required to maintain trim at a
field. The atmospheric boundary layer was included in the fixed location over the flight deck will be increased. Large-
numerical simulation. This modified ship geometry was also amplitude ship motion, which gives rise to increased difficulty
placed in the wind tunnel for validation of the numerical in tracking the flight deck prior to landing, is caused by the
solutions. response of the ship to excitation from the wave spectrum. It
is evident that of the above-discussed factors leading to
The results of the wind tunnel experiments show the reduced SHOLs, many are due to the influence of the ship air
recirculation zone behind the hangar as well as the wake. In the case of smaller non-aviation ships wherein the
asymmetries in the flow field caused by the 20mm Phalanx helicopter must land in close proximity to the ship
CIWS found on the starboard roof of the hangar. The superstructure, the air wake will assume a greater importance.
numerical results agree favourably with the modified CPF
experimental results and give further details of the air wake. In the Canadian context, all at-sea helicopter landings are
made on small ships. Over the past five years the
2. INTRODUCTION Aerodynamics Laboratory of the Institute for Aerospace
The launching of helicopters from and recovery of helicopters Research (IAR) has developed a collaborative program with
onto small, so-called "non-aviation" ships presents the Air Vehicle Research Section of the Canadian Department
considerable challenges to a pilot. The operational envelope of National Defence (DND) to investigate the air wake and
for such manoeuvres (known as ship-helicopter operating rotor loads in and around the Canadian Patrol Frigate (CPF).
limits or SHOLs) is almost always reduced when compared to This paper deals with the analysis and characterization of the
the normal operating envelope for the helicopter. Reductions air wakes and comments briefly on the effects on rotor loads.
to the operational envelope arise for a number of reasons,
some of which are: inadequate yaw control, engine-torque Determination of air wake characteristics is not a trivial task.
limitations, a reduction in usable cue environment thus The current Canadian efforts are focussed on obtaining wind
requiring changes in helicopter attitude, increased turbulence tunnel and computational fluid dynamics (CFD) data. There
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
7-2
exists some full-scale airwake data acquired from a CPF wind represented wind velocity relative to the ship at a height of
over deck survey (Kowal, 1991) and in this paper the current 40.15m above sea level (ASL). This represents a wind speed
wind tunnel data is compared to those measurements where of 20.6m/s(40 kts) relative to the ship at a height of 10 meters
possible. This comparison is intended to "validate" the wind ASL. This height is consistent with the standard
tunnel as a tool for airwake measurements. When acquiring meteorological practice for a reference wind speed. However,
air wake data, wind tunnels have the advantages of reduced for the purpose of this paper, the wind speed as measured at
cost and time, and a controlled environment when compared the ship's anemometer height of 24.5 meters ASL will be used
to at-sea trials. CFD offers an attractive option for producing as the reference wind speed. In practice, a ship at sea has only
air wake data, especially when simulation is the end-product an anemometer with which to measure the relative wind
of the air wake data. CFD results must also be validated. In speed. Thus, this experiment simulated a 46-knot wind,
the present context, CFD calculations were made on a measured at 24.5 meters ASL, encountering the ship steaming
"Modified" CPF configuration, since it is possible to include at 10 knots into wind. The moving ship in this case would
only limited structural details in a numerical model. To show 56 knots on the anemometer reading.
overcome a possible limitation with respect to validation of
the CFD data, a second experimental configuration was wind The Reynolds number for the full scale CPF is approximately
tunnel tested, which had the identical geometry to the model 30xl06, based on the ship beam of 16.4 meters. The Reynolds
used in the CFD calculations. As a by-product of this test number for the experiment was 107xl03, based on the ship
plan, the air wake differences between the wind tunnel results beam of 0.328 meters. Although these are not equal, the
for the two configurations can be examined to determine the model had sharp edges that reduced the sensitivity to
sensitivity of the air wake to superstructure configuration Reynolds number effects. In addition, the Reynolds number is
changes. well in excess of the minimum 1 lxlO3 recommended for the
wind tunnel modelling of ships (Healey, 1992).
3. EXPERIMENT DETAILS
A 1:50 scale model of the above-water form of the CPF was Wind tunnel measurements were made of the longitudinal,
constructed from applicable drawings (see Figure lb). lateral and vertical winds over the CPF helicopter flight deck.
Several of the smaller structures on the ship and ahead of the Approach wind directions of 0° (head wind) and 12° from
helicopter hangar such as missile launch installations, wire starboard were investigated. The measurements were made
antennas, handrails, a small lattice radar-mast and 57mm with a TSI cross hot-film anemometer. A cross hot-film
cannon were not included in the model. From an anemometer is capable of measuring flow in the nominal
aerodynamic point of view however, the air wake should still streamwise direction and in an orthogonal direction that is in
be highly representative of a fully-detailed CPF since the plane of the films. By rolling the probe axis 90°, the films
individual wake signatures of small structures will blend into are then sensitive to the second orthogonal direction. In
the flow field as one moves aft. Exhaust flows from the ship's addition, the data set that provides the second orthogonal
funnel and mechanical units were also not included in the direction also provides a redundant set of data for the
model. streamwise flow component. For the film calibrations, the
relationship between cooling velocity and output voltage was
In the second part of the experiment, modifications were described by means of a least-squares fit to a quartic function.
made to the CPF wind tunnel model in order to match While the hot-film probe has good resolution and fast
precisely the "Modified" CPF model (MCPF) used in the response, it cannot differentiate between forward and
CFD analysis of the same problem. The structures reversing flows. Thus, under conditions of reversing flow,
modified/removed from the CPF model were: the 20mm such as in the near wake of the hangar, the longitudinal flow
Vulcan Phalanx CIWS from the starboard rear of the hangar directions determined from the mean velocity components
roof; the aft fire control radar from the front of the hangar will be in error. In highly turbulent flows, the results yielded
roof; the landing officer's shelter from the flight deck; and the by the probe will show a lower than actual turbulence
large lattice radar-mast from atop the bridge. A minor intensity and a higher than actual velocity component (Tutu
modification was also effected to the bridge. and Chevray, 1975). Specifically, the errors associated with
turbulence intensity values are small, being about 2% for
The CPF model was installed in the NRC-IAR Vertical Wind turbulence intensity less than about 50% and reaching 12% at
Tunnel. The 15' circular test section of the tunnel was a turbulence intensity of 70% (Keffer, 1986).
modified with a test section insert to obtain a rectangular 10'
x 7' open jet section with one solid wall representing the sea Each measurement consisted of 2048 data samples acquired at
surface. The solid wall provided a flat, vertical surface on a rate of 100 Hertz for each of the two hot-film sensors. The
which to mount the ship model. A turbulent, graded sea measured wind speeds were normalized by the reference wind
boundary layer profile was generated by a grid structure speed, Vane,, as determined from the wind tunnel freestream.
inserted at the test section entrance. The boundary layer The turbulence intensity components are expressed as the root
profile represented wind flow over rough water encountering mean square values of the wind gusts divided by the local
the ship steaming into the wind. This boundary layer profile streamwise wind component. More explicitly, the data were
is representative only as, in reality, the wind profiles would reduced to the following non-dimensional quantities:
vary according to the varying sea states and ship speeds.
U(x,v,z) V(x.v.z) W(x.v.z)
For an experiment such as this one that is insensitive to Vv anem
Reynolds number effects and conducted in incompressible
flow, the experimenter is free to choose any two of length, u'fx.v.z) v'fx.v.z) w'fx.v.z)
time and velocity scalings. With the length scaling chosen to U(x,y,z) U(x,y,z) U(x,y,z)
be 1:50 as previously mentioned, and the time scaling chosen
to be 10:1, the velocity scaling defaults to 1:5. where U(x,y,z) is the local mean streamwise velocity
u'(x,y,z) is the local rms streamwise velocity
The wind tunnel velocity was held constant with a freestream V(x,y,z) is the local mean lateral velocity
velocity of 5 m/s. The wind tunnel freestream velocity v'(x,y,z) is the local rms lateral velocity
7-3
W(x,y,z) is the local mean vertical velocity profiles were applied to the upstream, side and top boundaries
w'(x,y,z) is the local rms vertical velocity while the downstream boundary was treated as an "outlet" or
and Van,™ is the velocity of the relative wind at the constant pressure boundary. The upstream boundary was
anemometer height, 24.5 meters ASL. placed 3.9 ship lengths (L) ahead of the bow and the domain
was 14.3 L long, 7.5 L wide and 3.7 L high.
The hot-film anemometer was installed on a traversing
mechanism that was free to move in the vertical direction 5. RESULTS AND DISCUSSION
relative to the ship. In order to survey the air wake in three
dimensions, the CPF model.was moved laterally or 5.1 0° Yaw Wind Tunnel Data
longitudinally in the wind tunnel. The 14X x 13Y x 16Z The variation in streamwise velocity component in the
matrix of surveyed points defines a full-scale rectangular vicinity of the flight deck as a function of location is presented
volume measuring 41.3m x 41.9m x 14.5m, extending from in Figure 4. The origin of the coordinate system is the bull's
3.9m aft of the hangar and from 2.0m above the surface of the eye on the flight deck as shown in Figure lb. The figure
helideck. depicts a classical bluff-body wake in that the momentum
deficit is lessened as one moves downstream and the wake
4. CFD DETAILS widens as it progresses downstream. For the slices closest to
A CFD simulation will provide a more detailed picture of the the hangar (x < 0), a cut-out is indicated reflecting a region of
flow around the frigate. However, to get such an answer recirculating flow. Stationary hot-film data cannot be
using a reasonable amount of computing resources, the flow considered reliable here, and as demonstrated in Syms and
must be simplified by either simplifying the equations that are Zan (1995), valid data can be obtained with a flying probe.
solved or the geometry being analysed. The choice made here The flying probe was not used here. Streamwise turbulence
was to simplify the CPF shape. As mentioned previously, a intensities of thirty-five percent or more are used to define the
"Modified" CPF was designed in which smaller pieces were boundary between valid and suspect data. The asymmetry in
removed and the bridge shape was modified to introduce the air wake due to the presence of the Phalanx gun located on
more 90° angles. The details of the geometry are found in the aft starboard corner of the hangar is evident. This
Figure lc. asymmetry is lessened as x becomes more positive. In this
and sub-sequent similar plots, the x-direction has been
The full-scale MCPF was studied using the commercially stretched to aid in visualization of the data.
available software package CFD-ACE (CFDRC, 1998). The
code is a finite volume, structured, multi-block Navier-Stokes The projection of the flow vectors onto planes normal to the
flow solver using body-fitted coordinates. It is a pressure- free stream (Figure 4) depict the air wake directed inward
based solver employing a SIMPLEC pressure-correction over most of the flight deck, again reflecting the streamline
algorithm. The convective terms in the governing equations deformation toward the region of low pressure behind the
were discretised using a second-order accurate upwind hangar. Once more, asymmetry about the ship centreline is
scheme with a blending of first-order upwinding to maintain evident. The starboard side exhibits far greater lateral
solver stability. velocities than does the port side. In fact, it is as though the
downward vertical component of the port side is largely
CFD-ACE offers a range of turbulence models. For this replaced by an inward lateral component on the starboard
work, the k-e model based on Launder and Spalding (Launder side. This said, the lateral component of velocity reportedly
and Spalding, 1974) was used with wall functions. Previous has the least effect on helicopter operations for the 0° wind
work (Syms 1997) has shown that the k-e model with wall direction.
functions will produce answers similar to those of the k-u
model integrated to the wall on simplified frigate shapes. A similar plot, but for the vertical velocity is presented in
Thus with a fair degree of confidence, the advantages of using Figure 5. It is clear that the air wake is directed downward
wall functions can be realised. Grid spacing near solid over most of the flight deck, reflecting the fact that the
boundaries was kept so that y+ values ranged between 50 and streamlines are bending toward the region of low pressure
150. behind the hangar face. Again the asymmetry due to the
Phalanx gun is evident. The fact that the air wake is angled
The multi-block grid created for this geometry was "H"- downward suggests that as the rotor enters the air wake, a
topology based (Figure 2). Thirty-five blocks containing nose-down pitching moment will be generated and this is
approximately 903,000 grid points were sufficient to model corroborated by pilot testimony. The results suggest flow
the MCPF. This could be considered a medium-size grid angles of up to six degrees in the absence of rotor downwash.
considering the complexity of the MCPF model. Further
work is being carried out into refinement of this grid to Significant gradients exist in the time-averaged value of the
address grid sensitivity issues. streamwise and vertical air wake velocity components. As
these gradients are manifested over distances of the order of a
The ship was considered to be stationary in a steady rotor diameter, it could be expected that the non-uniform
atmospheric boundary layer giving rise to a steady-state inflow will have an effect on the helicopter trim.
solution. Profiles in streamwise velocity and length scale
(Figure 3) were based on a logarithmic profile (Simiu and The turbulence intensities for the streamwise and vertical
Scanlan, 1986). The surface condition was taken to be an velocities are shown in Figures 6 and 7 respectively. The
ocean with a wind velocity of 20m/s at 10 meters off the levels of turbulence increase monotonically as one approaches
surface. These conditions generate a surface skin friction the flight deck. Flow recirculation exists when turbulence
coefficient of 2.53x10 3 and a roughness length of 3.51x10 3 intensities exceed about 35%. Even in the region outside the
meters. The turbulent kinetic energy was set slightly higher recirculation zone, the intensities are still a factor of two to
(6.51 m2/s2) than that in the referred to analysis (3.26 m2/s2) to three greater than that found in the atmosphere. Since the
compensate for the lack of turbulence sustaining mechanisms length scales are of the order of a rotor radius, the increased
in the CFD model which exist in the real atmosphere. These turbulence will no doubt contribute to pilot workload.
Vertical turbulence intensities of up to 20% suggest local flow
7-4
angles approaching 30°. For a blade tip speed of 250 m/s in a Figure 8 compares the normalized streamwise velocity
20 m/s wind, this suggests local blade angle of attack components of the wind tunnel and at-sea data. There is some
fluctuations of about 2.3°, enough to invoke significant scatter in the sea-trial data, likely due to the short measuring
changes in rotor thrust, pitching moment and rolling moment. times. The wind tunnel data falls within this scatter range for
every location at the 5.5 meter height. The wind tunnel
5.2 Wind Tunnel and Sea-Trial Data Comparison underestimated the streamwise components at three of the five
For validation purposes, these wind tunnel data are compared 9 meter heights, but only by about 0.1 XJ/Vmem.
to full-scale at-sea data taken on a CPF (Kowal, 1991)
acquired using a standard three-component vane-type The lateral and vertical velocities, Figures 9 and 10, are
anemometer. The sea-trial data is more limited in that considerably smaller and more difficult to measure in both
measurements were acquired at fewer locations. The sea-trial cases than the streamwise component. Again, the wind tunnel
data were acquired for a period of 30 seconds, which is a data are in reasonable agreement. For these comparisons, the
practical limit in the situation, but probably an insufficient sea-trial data will be compromised by small changes in
length of time to ensure the results are statistically significant approach wind angle, while in the wind tunnel no such
for a stationary process. By way of contrast the wind tunnel concerns arise. The consistent positive values for the
data were averaged for a full-scale equivalent of 204.8 normalized sea-trial lateral velocity data at 20 knots is almost
seconds (3 min 25 sec). certainly due to a "red" wind. This likelihood is further
increased when one considers that the sea trial data was taken
In order to facilitate comparison of wind tunnel data with the under conditions of a prevailing 20-knot wind. Therefore, one
sea-trial data, the wind tunnel data were interpolated using a surmises, the 20-knot sea-trial data was achieved with a
bi-linear interpolation algorithm to arrive at values for stationary ship facing into wind. Without forward water
locations identical to those for the sea-trial data. The latter speed, the ship would be unable to maintain a heading for
data were acquired at two heights above the flight deck, 5.5 extended periods, being pushed off heading by both the wind
and 9 meters, for the following (x,y) locations: and the waves. The 30-knot data would thus be achieved by
steaming the ship at 10 knots into the prevailing wind, and
allowing much greater control of alignment with the
Table 1:
prevailing wind. Therefore, it is believed that the sea-trial 30-
Locations of Measurements for Sea-Trial Data
knot data would be more accurate than the 20-knot data.
Location X Y
(arbitrary (full-scale in (full scale in For the vertical velocity component, there is again
number) meters) meters) considerable scatter in the sea-trial data, reflecting the short
8 -6.5/-6.0 0.0/0.0 averaging times. The wind tunnel data are consistent with the
12 -1.2/-0.6 -5.5/-5.5 trends in the sea-trial data.
13 -1.2/-0.6 0.0/0.0
14 -1.2/-0.6 5.2/5.2 The turbulence intensities, Figure 11, are plotted as the
18 5.2/5.8 0.0/0.0 horizontal resultant (i.e. UV) wind intensity normalized by the
ship anemometer velocity. The data are presented in this way,
In the above table, the numbers preceding the slashes refer to as opposed to separating out the streamwise and lateral
the (x,y) location of the measurement at 5.5 meters above components, since the latter are unavailable for the sea-trial
deck and the numbers following the slashes refer to the data. Every wind tunnel data point matches, or falls within
location of the measurement at 9 meters above deck. The 1% turbulence intensity, of one of the sea-trial data points.
offsets (.61m) are due to specific anemometer positions on the
towers on which the anemometers were mounted. On a point of clarification, the wind tunnel data for the 5.5
meter height at Location 8 was omitted in the air wake plots,
The sea-trial results present data at two wind velocities, 20 Figures 4 and 5, due to high turbulence levels recorded in the
knots and 30 knots, although no Reynolds number effects hot-film measurement. However, the value is nevertheless
would be expected for this flow. The fact that different values presented here for comparison purposes. In non-reversing low
were obtained for each of the velocities is probably indicative turbulence flow, u'/U would be some value (<30%) and the
of the statistical uncertainty in the measurement and not of statistical skew of the data measured by the cross-film
different air wake characteristics. anemometer would be near zero. The actual measured u'/U
value at Location 8 is 47% and the statistical skew of the
In addition to the difficulties associated with obtaining at-sea U/Van,™ time history is 0.8. These values suggest a region that
measurements, it is also accepted that the wind tunnel cannot experiences intermittent flow reversal (Zan and Garry, 1994).
duplicate faithfully all details of the actual CPF. In particular, Thus data at location 8 are marginal in terms of acceptance
effects of stack gas efflux are not modelled in the wind tunnel based on previously defined criteria. As previously explained,
test, though this is not a difficult aspect to incorporate in an the result of such a flow condition on the interpreted data is to
experiment. increase the mean velocity and decrease, by small amount, the
turbulence intensity. Accepting the presence of error, the
Despite these limitations to the comparison of data sets, there wind tunnel value should be used with caution. It can be
is generally good agreement between the wind tunnel and sea- noted, however, that by reducing the U/Vanem value at this
trial data. Eighty percent of the data points match at the 5.5 location in compensation for the error, an even more
meter height while seventy percent match at the 9 meter favourable agreement might be obtained with the sea-trial
height. Where the data is not an exact match, the wind tunnel data.
data features consistent trends with the sea-trial data. For
example, local U/Vancm velocities increase with height. There 5.3 MCPF and CPF Air Wakes at 0° Yaw Comparison
are also consistent trends in location; for example, the The CPF and MCPF configurations were both tested in the
velocities at Location 12 are higher than at the other locations wind tunnel. The MCPF, a symmetric model, generates a
for both heights. classic bluff-body air wake over the helideck. Streamlines
follow downward and inward trajectories as described below.
7-5
The CPF air wake, however, is somewhat different. While experiment obtaining more detailed information (in particular
the flow on the port side of the CPF centreline is akin to a about the size and shape of the recirculation zone) is required
classic bluff-body air wake, the flow on the starboard side of to further validate the CFD results.
centreline is decidedly not. The CIWS generates a
streamwise flow deficit, and associated turbulence, that 5.6 12° Yaw Wind Tunnel Data
extends 15m aft of the hangar until smoothing into the larger In order to assess the effects of a non-zero incident wind on
air wake structure. The flow on the starboard side possesses a the air wake, the CPF model was yawed 12° to port in the
greater lateral inward component than the port side. Further, wind tunnel to effect a "green 12" wind. (Because of
the vertical flow above the port side of the helideck is almost installation restrictions, the actual angle was 11.8°). The
completely deleted on the starboard side due to the presence variation in the time-averaged non-dimensional U-component
of the CIWS. The presence of the CIWS is thus seen to cause velocity is shown in the upper half of Figure 14. The U-
a substantially different air wake structure over the helideck component is defined with respect to the ship axes system and
than might otherwise be expected. Another significant not the wind axes. The V-component, or lateral velocities
difference between the two air wakes is generated by the CPF (not shown) can be characterized in much the same way as for
large lattice radar mast. This mast contributes to a streamwise the U-component velocities, although the U-component
flow deficit, and associated turbulence, at the height of the velocity magnitudes are appreciably greater. Thus, the
mast and extends well beyond the stern of the ship. Other horizontal velocity contours are approximated by the U-
differences between the ship models were not seen to have component data presented. Again data are "blanked" where
appreciable impact on the airwake. reversing flow is suspected.
5.4 0° Yaw CFD Results While the predominant wake deficit is aligned with the free
The results of a numerical simulation provide details of the stream at higher elevations, the separation zone immediately
flow around the complete MCPF geometry. Since changes aft of the hangar and closer in elevation to the flight deck is
were made to the bridge and all smaller add-ons were more closely aligned to the ship axis. This suggests the
removed, analysing the flow ahead of the hangar would horizontal component of the air wake contains a gradient in
provide only a small amount of information pertinent to the direction as a function of vertical distance from the deck. For
study of the air wake. (It must be remembered that the high wind speeds as the landing is effected, this type of flow
changes to the bridge were assumed to have minimal impact could be expected to affect the thrust generated by the tail
on the flow over the flight deck). So, the present discussion rotor and the yawing moment acting on the fuselage. The
focuses on the area behind the hangar. change in tail rotor output and fuselage moment will induce a
yaw rotation of the helicopter, thereby requiring corrective
Figure 12 shows the general character of the flow over the action by the pilot. It would be expected that larger wind
flight deck. The transparent surface marks the region of the angles would result in larger gradients and vortical flows
recirculation zone (i.e. zero streamwise velocity). This which would require a larger corrective input from the pilot.
surface shows that the flow re-attaches just aft of the centre of At some point a control limit would be reached, thereby
the flight deck. The black particle trace shows the outline and defining one part of the SHOL. This is borne out in practice.
foot of the vertical horseshoe vortex standing in the The connection between air wake directional gradients and the
recirculation zone behind the hangar. The remaining traces SHOL limit is plausible, but remains a conjecture in the
show the rather benign flow around and over the hangar onto absence of experimental and/or numerical analysis.
the flight deck. The topology of this air wake is similar to
wakes seen behind three-dimensional bluff bodies (Larousse, The magnitudes of vertical velocity for the 12° yawed wind
Montinuzzi and Tropea, 1991). are depicted in the lower half of Figure 14. It is evident that
the windward side of the deck experiences an upwash caused
5.5 Wind Tunnel and CFD Data Comparison by the onset flow rising over the flight deck and probably
The MCPF geometry was also examined in the wind tunnel rolling up into a tight vortex with a relatively shallow helix
for purposes of CFD validation. To facilitate comparison of angle. Conversely, data for the leeward side of the flight deck
the CFD to experimental data, the CFD results were indicates a downwash on this side as the streamlines curve
interpolated onto the experimental grid. Figure 13 shows a toward the low pressure region aft of and beside the ship
comparison of the two sets of data. Displayed are a set of above the sea surface. These changes in direction of vertical
crossflow planes which are coloured according to the flow with lateral position could lead to rapid changes in rotor-
nondimensional streamwise velocity with the vectors generated rolling moments as the pilot traverses from the lee
representing the in-plane velocity components. side of the ship over the flight deck. However, SHOLs
defined by cyclic-input limitations are uncommon, suggesting
The first point to observe is that both flow fields show the the rotor downwash dominates in this particular case.
same flow topology. There is a wake behind the funnel and
recirculation zones behind the hangar and the stern of the 6. CONCLUSIONS
ship. The major difference between the measured and This paper has examined the air wake of a CPF using several
computed velocities exists in the flow gradients; the CFD sources of information: wind tunnel measurements, full-scale
showing much higher gradients. In general, the experimental sea-trial data and computational fluid dynamic simulation.
results show a much more diffusive flow. This results in the From this investigation, the following conclusions can be
computed solution exhibiting higher values in each of the drawn:
three velocity components. In particular, the streamwise
component shows a larger value at the higher elevations. This • Good correlation between wind tunnel data and full-scale
is a consequence of the flow being energized by the front sea-trial data has been demonstrated. Either could be
edge of the funnel and not being dissipated by the time it used for CFD validation.
reaches the flight deck.
7. REFERENCES
CFDRC (1998) "CFD-ACE Theory Manual" CFD Research
Corporation, Huntsville, Alabama, 1998.
Kowal, H.J. (1991) " Wind Over Deck Survey & CH-
124A/CPF Flight Deck Qualification Trials", Flight
Dynamics Technical Memorandum FD TM 91/05, Aerospace
Engineering Test Establishment, Department of National
Defence, Canada, December 1991.
I::
BifS; s;i ,A, , ,ft
H i r
\\ it
c) Modified Canadian Patrol Frigate (MCPF) Wind Tunnel and CFD Geometry
500
x=2707m
CPF Helideck
y is from -20.96m to 20-96m
z is from 2.00m to 16.50m
U/Vanem
ÖÖÖ ÖÖOÖÖÖÖO©
I I I
x=27.07m
CPF Helideck
y is from -2096m to 20-
z is from 4-50m to 16-50]
x = 27.07m
CPF Helideck
x = 20.72m
y is from -20.96m to 20.96m
z is from 2.00m to 1650m
x = 8.02m
x=1.67m
x=-468m
x=-1103m
x = 27.07m
CPF Helideck
y is from -20-96m to 20.96m
z is from 450m to 1650m
x = 802m
x = -468m
x = -1103m
0.15 ♦ 0.15
0.10
0.10
■ ♦ <► <►
0.05 ▲ 0.05 ■
-
A 0.00 A
0.00
-0.05 ■ ft a -0.05 A A
■ iL
A
II
-0.10 ■■ -0.10 ■ - ♦ Sea-Trial: 20 Knots
-0.15 < -0.15 1 ■ ■ Sea-Trial: 30 Knots
A Wind Tunnel
-0.20 -0.20
10 14 10 14 18
Tower Position Tower Position
■
-0.11 -0.11 ♦ Sea-1 rial: I J Knots
-0.13 -0.13 . M Sea-Trial: 3 3 Knots
A WindTunni:1
-0.15 -I -0.15
10 14 10 14 18
Tower Position Tower Position
Figure 10: Comparison of Wind Tunnel and Full-Scale W/Vanl:m Velocities
7-12
♦ ♦ <►
15 15
£
A A :« it
>
♦ .... ■■ 10 f ♦ A
10
■ X
i ■■ 1
■ T
♦ Sea-Trial: 20 Knots
■ Sea-Trial: 30 Knots
A Wind Tunnel
10 14 18 10 14 18
Tower Position Tower Position
Wind Tunnel
U/Vanem
11-
1-j
09-j
084
0-74
064
O-4J
O.3J
O.2J
Oil
0-)
-Oil
-024
-0-3-3
CFD
Figure 13: "Modified" CPF Wind Tunnel Results VS. CFD Solution
7-14
c
•8
o
tt
a.o
o
a
c~i
■o
u
CO
o.
2
co
l
II
N
c
>
a
8-1
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
8-2
Modelling the ship air wake, and its effect on technical aspects to be discussed elsewhere in the
helicopter behaviour, is regarded as one of the most symposium. The paper demonstrates the applications
significant technical challenges for the dynamic of air wake prediction in ship-aircraft qualification,
interface (DI) simulation community. The irregular and the value of air wake simulation in research and
nature of a typical ship superstructure, the effect of training. The paper discusses the piloting problems
ship motion and the influence of the helicopter rotor associated with ship air wakes and the necessity for
combine to produce a highly complex and dynamic air wake modelling. Methods of air wake prediction
flow field dominated by separations, vortical flows are discussed and model validation issues are
and recirculation. Validation of an air wake addressed.
prediction is complicated by the difficulties
associated with gathering quantitative measurements 3 OPERATIONAL PROBLEMS OF AIR WAKES
of the flow field around a ship, at full-scale or in a 3.1 Effect on pilot workload and performance
wind tunnel. When departing or returning to the ship, helicopter
pilots need to consider the conditions that enable
2.3 TTCP collaboration and involvement them to safely land on a moving deck. Although the
The development of simulation for application to the size of the deck area varies considerably for different
helicopter-ship DI has been a collaborative activity classes of ship, for ships such as destroyers and
under the auspices of The Technical Co-operation frigates the area is frequently not much larger than
Programme (TTCP) since 1990. The working-level the helicopter. Pilots must often position the
participants consist of representatives from the UK's helicopter with sufficient accuracy to engage an
Defence Evaluation and Research Agency (DERA), aircraft-mounted probe in a ship-mounted capture
the National Research Council of Canada - Institute mechanism, which may be as small as 2 metres .
for Aerospace Research (NRC - IAR), the Naval Air Significant factors affecting such operations include
Warfare Center (NAWC) in the USA and Australia's visibility (day, night, or degraded through poor
Defence Science and Technology Organisation weather conditions or spray), ship motion, and
(DSTO). The goals of the work are to progress the aerodynamic interactions of the helicopter with the
application of simulation to maritime helicopters in turbulent air flow near the ship. These all contribute
requirements capture, design, and test and evaluation to pilot workload, which may not allow the pilot to
with increased pace and effectiveness through safely take-off or land. Other factors limiting launch
collaboration. Within the group, the behaviour of and recovery include exceeding torque and control
helicopters operating in the vicinity of a ship is limits; in certain wind conditions, the pilot may have
addressed under the topic, 'Ship air wake effects on insufficient control authority, in particular tail rotor
rotor behaviour'. Through this topic, all participants authority in a crosswind, to maintain position relative
have been actively involved with improving their air to the ship.
wake models, conducting full- and model-scale
testing to gain an improved understanding of the key 3.2 Ship helicopter operating limits (SHOLs)
features of an air wake topology and to develop a In order to assist pilots, maritime helicopter operators
validation database. A number of approaches are develop SHOLs to define safe operating parameters.
being pursued concurrently aimed at applying These are usually represented in diagrammatic form,
Computational Fluid Dynamics (CFD) and other, an example of which is at Figure 1. The wind speed
more empirical, analyses to achieve a predictive and direction must be within the safe operating
capability and to enhance piloted simulation. The envelope shown on the diagram for the helicopter to
high potential pay-off from shared learning has land or take-off safely; if they are not, the ship may
proved to be a strong motivation for the different be forced to change course and speed to achieve a
nations to align activities where possible and to safe wind condition. SHOLs may also include
present and discuss the different facets of their work structural limits of both ship and aircraft, as well as
openly with other TTCP members. on deck limits to prevent toppling and sliding, with
and without securing devices such as lashings or a
2.3 Aims of the paper ship-mounted capture system. However, these limits
This paper is intended to provide an overview of the are not considered in this paper.
air wake modelling activities within the TTCP
collaboration. The paper details the collaborative The proliferation of ship and helicopter type
efforts being pursued by the member nations, under variations means that each individual country
the co-ordination of the authors. It summarises the generally operates its own unique combination of
work being conducted by each of the nations and ship and helicopter. There are no agreed
draws together the various research aspects and international standards for SHOLs and each country
validation tests. The aim is to present a usually determines their own. The development of
comprehensive and co-ordinated approach to these SHOLs is an expensive and exacting task
modelling and simulation of the air wake problem at currently determined through intensive and costly
the DI, putting into an operational context the more First of Class Flying Trials (FOCFTs), which limit
8-3
the final operational envelopes to environmental simulator training and may, in extreme
conditions similar to those encountered during the circumstances, prove dangerous. The air wake has an
trials. The use of modelling and simulation to important part to play in deck landing training as it
support SHOL development offers substantial has a considerable impact on both workload and
benefits in terms of reduced costs, improved safety performance, which, in turn, will influence the
and potentially larger operating envelopes. Perhaps control strategy adopted by the pilot. If the simulated
the most important component of such a simulation is air wake is more benign than in reality, a student may
a high fidelity air wake model that generates gain a false sense of security in the simulator, while,
representative levels of workload in piloted if the air wake is too severe, there is a danger that the
simulation and allows aircraft control limits to be pilot will lose confidence before he has even started.
accurately predicted. An air wake model of appropriate fidelity is,
therefore, an essential element of a training
simulation.
Red Green
4 AIR WAKE MODELLING AND VALIDATION
4.1 Background
For the purposes of this paper, modelling refers to the
use of numerical techniques to extract quantitative air
wake characteristics and to provide numerical data
for input to a simulator. Validation data refer to
acquired air wake data, from either at-sea trials or in
wind tunnel tests, which are used to provide
benchmarks for the modelling.
stationary data from which to generate valid coefficients at approximately 200 locations on the
representations of an air wake. Unfortunately, the SFS is scheduled for September of 1998. The surface
nature of full-scale atmospheric characteristics is oil flow revealed a surprisingly complex flow
such that data can rarely be collected for long enough topology for yawed winds and served to reinforce the
times to satisfy the stationary criterion. One notable difficulties of obtaining a CFD solution for a realistic
exception was an Australian at-sea trial where wind ship configuration.
conditions were benign for some time and the air
wake was thus generated by the relative ship motion.
do occur on the flight deck. The UK is developing a appropriate random time history that represents the
time-accurate inviscid code and is working towards profile of the turbulent winds.
inclusion of viscous effects. Future developments
include incorporation of moving geometries and Time-dependent turbulence is generated in the strong
inclusion of stack gas effects, as well as incorporation shear layers that exist between the flow past the
of the helicopter rotor. Much of this work is directed superstructure and the low velocity regions
towards solving for the air wake around Royal Navy immediately downstream of the superstructure.
ships to aid in ship design. Initially the turbulence is in small eddies and these
rapidly amalgamate to form larger eddies of the order
In the United States, CFD efforts applied to the SFS of that of the superstructure. As the shear regions
have used two approaches. In one case a NASA- dissipate further downstream, fewer new turbulent
developed time-accurate Navier-Stokes code was eddies are formed and the larger eddies decay into
used which produced encouraging results, although smaller ones. The turbulence intensities are
time-averaged data were not available. In another calculated from two new modules that relate to the
effort [4], a fourth-order time- and space-accurate flow behind bluff bodies (Profile Drag Turbulence)
code was used. In the latter case, the steady flow and the flow behind the edge of a bluff body (Edge
field solution was developed using an Euler solver Turbulence). The geometric properties of the ship
from NASA, and the unsteady flow field followed that contribute to these effects are already held in the
using a viscous solver. Significant CFD efforts steady air wake modules for the basic flow elements.
towards air wake modelling on United States Navy Turbulence intensities are calculated independently
ships have been reported by Tai [5] and Landsberg et for the vertical, longitudinal and lateral axes as the
al [6]. values are typically uncorrelated. Also modelled is
the attenuation of the vertical component of
4.3 Empirical methods turbulence near to a horizontal surface (such as the
deck of the ship) owing to the boundary condition
4.3.1 UK Advanced Flight Simulator simulation
which will not let any flow penetrate the deck and
The UK has implemented an air wake model in the
Advanced Flight Simulator (AFS) at DERA Bedford hence at this point the intensity must be zero.
under a contract aimed at establishing the potential of
In order to make use of the model results in a real-
a simulator to support deck landing training. The
time simulation, a look-up table is generated of
model, described by Woodfield [7], can produce
sufficient size to capture all the features of the model
estimates of the air wake around a ship of any shape
output without requiring a prohibitive amount of
for which the model has been configured.
computer memory for storage. This requires a non-
Configurations currently exist for the Type 23 frigate
uniform calculation grid which packs a higher
and the Landing Platform Helicopter (LPH).
number of calculation points where the flow pattern
is expected to vary rapidly with distance, and less
The model is built on the underlying assumption that
the main features of the flow field in the lee of a ship closely-spaced points where the flow is expected to
be approximately constant. It is assumed that both
can be predicted by adding together the effects of
the pattern of steady winds and the turbulence
basic flow elements located at various parts of the
ship geometry. Before being added together the intensities will be independent of wind over deck
relative magnitudes of the flows caused by each basic (WOD) velocity and therefore a new table is required
for each wind azimuth case where all the results are
flow element are calculated according to the
normalised by the WOD velocity. This assumption is
dimensions of the geometric feature to which they are
attached. The relationships between the magnitude expected to hold well for moderate and strong winds
of a flow and a geometry of a particular type have but may become increasingly less appropriate for
cases where the wind is weak and the flow pattern
been derived from empirical data and have undergone
may have significantly changed.
several stages of refinement since the creation of the
first version of the model.
A large amount of data and pilot comment has been
collected from using the UK air wake model
Air wakes behind bluff bodies can be described by a
throughout a range of research programmes. The
combination of steady wind components that vary
model datasets have also been supplied to
spatially but are independent of time and unsteady or
government representatives in Australia and the USA
turbulent components that vary in space and time.
in support of their own studies into the simulation of
The UK model was originally intended to provide an
estimate of the steady flow field pattern only, as a the DI under the auspices of TTCP. The combination
of these programmes has built a level of confidence
function of the position in space. A more recent
in the model allowing it to be used to investigate
version has been upgraded to also include a
prediction of turbulence intensity (the magnitude of some of the operational problems associated with the
deck landing task.
the time varying wind component) as a function of
the position in space, which can be multiplied by an
8-6
4.4 Experimental methods for validation Figure 3; Typical results from a conical scan
The difficulties associated with quantitative air wake
measurements for both full-scale and model-scale A significant limitation of current full-scale
measurements have been discussed. All of the measurement methods is the inability to measure air
collaborating TTCP nations have made such flow vectors over a wide range of points in a short
measurements, however, and there are encouraging space of time. Furthermore, most data can only be
collapses of the air wake data when velocities are recorded directly over the ship flight deck. To
normalised by ship anemometer airspeed, and overcome these limitations, the Shipboard Aircraft
positions are normalised by ship hangar height. LIDAR (Light Detection and Ranging) Turbulence
However, in one case full-scale air wake Sensor (SALTS) is being developed by NAWC
measurements made by two nations on a virtually Lakehurst, with UK Ministry of Defence (MOD)
identical ship operated by those two nations produced funding support. By using three co-ordinated lasers
a factor-of-three difference in the turbulence levels. mounted on or near the flight deck (Figure 4) a
picture of the flow field can be generated by scanning
4.4.1 Full-scale measurements the lasers over the flight deck and along the aircraft
The UK has conducted a number of full-scale trials flight path. Analysis of SALTS data will increase
aimed at collecting sufficient data to aid in the understanding of the air wake characteristics
validation of air wake models. A mast-mounted occurring through the launch/recovery area, and
anemometer, borrowed from NAWC Patuxent River, hence can be used to improve future ship and aircraft
was used in 1994 to record air flow measurements designs, to validate CFD codes and to model air wake
over the flight deck of a Type 23 frigate. A single systems for simulators.
mast with ultrasonic anemometers in 3 orthogonal
directions at 3 heights was located at a series of up to Measurement
14 locations on the flight deck. Simultaneous volume
measurements were made of ship angular motion,
linear accelerations, speed, heading and standard
wind-over-deck direction and speed.
development of the trials was that wind conditions database of air wake measurements which can be
were benign for much of the tests. While this used throughout the TTCP community to validate
permitted easy generation of 0° wind data, it was a models and improve understanding of the flow
greater challenge to generate data for yawed winds. structure over a ship flight deck.
Even though an air wake was often generated by
relative ship motion in the calm conditions, To complement the Australian full-scale trials, wind-
operational constraints during the trial meant that the tunnel measurements were made on a 1/64-scale
duration of recording at each mast position was often model of an FFG-7 during the early 1990s [9]. For
inadequate (typically 90 seconds rather than a more the wind tunnel tests, only mean flow could be
appropriate duration of 5 minutes). Also, the number measured and the measurement probe was limited to
of grid measurement locations (13 positions at three local flow deviations of less than about 40°, a serious
heights) was inadequate (Figure 5). deficiency in the recirculation region behind the
hangar. Measurements were not corrected for
blockage or interference.
• Mast Positions
empirical approach of the UK air wake model, so that An algorithm to regenerate the unsteady
a look-up table is used to approximate recirculation characteristics (spatial and temporal) of the measured
and ground effect at any point around the ship. This flow field was developed, and GenHel was then
type of model would reside with the ship, but it "flown" through this numerical air wake, with a
seems likely that a higher fidelity representation of thrust-to-weight ratio of unity. The spatial
the helicopter-ship system will require integration of correlation in the generated turbulence could
the rotor code with a real-time simulated air wake. incorporate cross-spectra of either the streamwise or
Real-time calculations of this type cannot be vertical velocity. No modelling of the lateral velocity
expected in the foreseeable future, but off-line cross-spectra or orthogonal component cross-spectra
analysis is being conducted in the UK, US and was attempted. This approach, while unparalleled in
Canada through the use of CFD codes. terms of accurate air wake modelling, is not able to
model the coupling of the rotor downwash and air
4.6.1 UK research wake. The calculated rotor loads at a given point in
In the UK research [3], a simple actuator disc space, as output from GenHel, were then analysed. It
technique has been used to represent the effect of the was reported that inclusion of spatial correlation
helicopter on the air wake. It is proposed that the effects increased the level of the unsteady rotor
effect of the modified air wake on the aircraft could response. This is to be expected since the turbulence
be calculated by extracting the velocity field over the scales are of the same size as the rotor diameter. It
rotor disc from a CFD calculation and feeding it to a was also found that the inclusion of the unsteady
helicopter performance code. A modified loading vertical velocity into the air wake model induced
over the actuator disc would then be applied as an greater unsteady rotor loads than inclusion of an
input to the CFD calculation and an iterative process unsteady streamwise velocity.
used to produce a representation of the complete flow
field. As a precursor to this research, the UK has In order to obtain quantitative information about the
recently conducted a land-based trial in an attempt to degree of coupling between the rotor downwash and
determine the degree to which the presence of the the air wake as it affects the rotor, the IAR is
helicopter influences the air wake over a flight deck. conducting sub-scale wind tunnel experiments
Air flow measurements, using a single laser involving a frigate model and a sting-mounted rotor
anemometer, were taken around a helicopter hovering that is traversed into the air wake. The geometric
first behind a hangar and second in clear air. Both scaling of the experiment is 1:50 and the rotor is
sets of data were recorded in nil wind conditions. powered by a small DC motor mounted on a six-
The results were compared with a CFD analysis of component balance. In this case, correct scaling of
the same conditions and initial comparisons were the rotor aerodynamics, specifically advance ratio
very encouraging [12]. However, perhaps the and thrust coefficient, will ensure the flow field
clearest indication of the influence of the helicopter coupling is correct. The rotor has the capability for
and solid boundaries on the flow field is given by the arbitrary collective settings (up to stall); however, the
pilots who consistently remarked on the difficulties capability for cyclic inputs is not possible at this
of holding position behind the hangar in comparison scale. Thus the rotor is not trimmed, but can be
to the hover in clear air. operated at any fixed collective position for a given
traverse. Ground effect is also inherently included in
4.6.2 Canadian research this experimental modelling.
Canadian work incorporating a numerical helicopter
into an air wake was described by Syms and Zan These experiments will be followed by modification
[13]. In this work, the rotor module of the Sikorsky of the GenHel rotor to match the characteristics of the
General Helicopter Flight Dynamics Simulation code wind tunnel rotor and then this numerical rotor will
GenHel [14 & 15] was "flown" through a numerical be "flown" through the measured CPF air wake.
model of an air wake. The air wake in this case was With the numerical rotor validated against the wind
obtained from cross-hot-film anemometer tunnel rotor in uniform flow, changes in rotor loads
measurements made behind a model of a frigate in determined from the experiment and the numerical
the IAR 0.9 metre wind tunnel, wherein an model will indicate the importance of the downwash
atmospheric boundary layer simulation was and air wake coupling.
incorporated. Measurements were made of the three
orthogonal mean and unsteady flow components. Measurements of rotor loads in the wind tunnel
Simultaneous measurements from two anemometers experiment have been made for several onset wind
permitted accurate measurements of the turbulence speeds and directions as well as with changes in ship
correlation between two physical points in the air superstructure. It has been determined that for a
wake region. A "flying" hot-film probe was also fixed collective setting in a 35 knot wind, the time-
used in these tests to permit accurate measurements averaged rotor thrust coefficient decreases about 10%
of velocities in the recirculation zone. for a rotor positioned 6 metres above the deck (full-
scale) and over the centre of the flight deck as
-10
compared to a rotor at 10 metres above the flight synchronised ship and helicopter measurements for
deck and displaced laterally by one rotor diameter. an S-70B-2 Seahawk helicopter operating from an
These positions represent the transition from FFG-7 frigate during a FOCFT in March 1994 [16].
alongside the ship to the low-hover, as practised by A data acquisition system developed by the
the British and Canadian navies. While a pilot will Australians was fitted to the ship, and another was
not hold a fixed collective during this manoeuvre, the fitted to the aircraft. Each system allowed up to 56
results suggest the magnitude of quantitative channels to be used, mostly analogue inputs, though
corrective actions that a pilot must take during the synchro inputs were used for data recorded from
landing. The pilot will, of course, decrease collective normal ship instruments and some aircraft
to execute the landing manoeuvre but these data instruments. The ship was instrumented to record
suggest the rate of descent will be greater than if the ship motion and wind velocity, and the helicopter
same control strategy were implemented for a was instrumented to record flight control positions
conventional ground-based landing. These effects and flight dynamic parameters. Duplication of some
may be overshadowed by the unsteady rotor loads measurements from alternative sources allowed some
arising from turbulence, or simply by the effort redundancy, which was utilised when instrument
required to track the vertical and pitch deck motion in failures occurred.
heavy seas. Whether the wind tunnel rotor can be
used to extract dynamic loads arising from the During these trials, though mean relative wind was
unsteady components of the air wake remains to be measured using the mast-mounted ship and reference
determined. anemometers, the air flow distribution over the flight
deck region and around the helicopter could not be
4.6.3 US research measured. Hence, for relative wind headings where
One US approach to modelling the rotor and air wake the ship hangar will significantly affect the air flow
interaction has been reported by Landsberg et al [6]. into the helicopter, the information is incomplete in
The primary focus of the work was the modelling of attempting to duplicate the trials in a simulation using
exhaust gas concentrations in the vicinity of the flight recorded pilot controls to drive the simulation.
deck as opposed to investigations of rotor
aerodynamics and helicopter flight dynamics. An in- 5.2 Torque-limited SHOLs
house developed unsteady-flow solver was used to In the absence of shielding of the air flow by the
compute the unsteady air wake and the ship stack hangar, torque would be expected to decrease with
exhaust gas trajectories for a DDG51 destroyer. increased airspeed up to about 80 knots. However,
Incorporated into the code was an actuator disc inspection of Australian FOCFT measurements
model to represent a 5000 kg helicopter with a 16.5 showed that torque remained relatively constant as
metre diameter rotor. This approach will result in a relative wind speed increased for headings where the
modified air wake due to the actuator disc; however, ship hangar significantly affected the air flow. This
the modified air wake is not fed back into the indicated that the flow was being effectively blocked.
actuator disc model, which would represent the fully-
coupled case. The work included modelling of the The blocking effect could be quantified in an off-line
atmospheric boundary layer to enhance accuracy of simulation by using GenHel without any allowance
the boundary conditions. The CFD air wake for ship air wake to predict torque for a trimmed
prediction indicated that the predominant frequencies aircraft behind the ship hangar. The ratio between
in the unsteady air wake are in a bandwidth likely to the measured and predicted torque was obtained and
affect helicopter flight dynamics, and this echoes the plotted against relative wind speed for each 20 degree
experimental (wind tunnel) work reported by Syms range of relative wind directions. Within this range,
andZan[13]. it was assumed that the physical characteristics of the
flow were reasonably constant. In general, a good
4.7 Effect of the tail rotor linear relationship between the torque ratio and
While the discussion so far has focused on the relative wind speed was obtained for each region. An
interactions of the main rotor and air wake, it is attempt was then made to use this relationship to
recognised that the tail rotor cannot be ignored. predict torque-limited boundaries of the SHOLs.
SHOL boundaries for beam winds are generally However, the boundaries did not match well with
determined by limitations in tail rotor authority. observations from the FOCFTs, primarily because of
Thus high-fidelity simulators used for training sensitivities of the boundaries to small changes in
purposes or SHOL prediction must include suitably torque.
accurate modelling of the tail rotor aerodynamics.
5.3 Piloted simulation
5 PREDICTION OF SHOLS Piloted simulation offers a more promising method
for recreating torque and pedal limited areas of the
5.1 Synchronised ship and helicopter measurements SHOLs. Additionally, more substantial areas limited
As a first step in using analysis and simulation in by pilot workload can also be investigated. The main
determining SHOLs, the Australians obtained
8-11
obstacle to achieving a suitable model of the whole simulation of the air wake remains, however, a
system is in representing the air wake. The most significant challenge due to the complex and highly
promising approach in the short term is to incorporate dynamic nature of the flow over the flight deck and
the UK look-up tables produced using the air wake the difficulties of obtaining validation data, at both
model installed in the AFS. Initial trials at DERA full- and model-scale.
have demonstrated the feasibility of this method [17]
and Australia will shortly attempt to reproduce The paper has described how, under the auspices of
FOCFT test data using the look-up tables generated TTCP, several organisations in the UK, Canada,
by the UK model. It should be noted, though, that in Australia and the USA are working collaboratively to
this air wake model, as with look-up tables produced facilitate the development of air wake models with
by current CFD codes, no account is taken of the the aim of improving simulation of the DI. Each
change in the air wake as a result of ship motion. nation is using CFD techniques to calculate the air
wake around a common SFS for various wind
5.4 Pilot workload modelling directions and comparing results against each other
A critical factor in the development of a SHOL is the and against wind tunnel data generated by Canada. A
pilot workload. When the workload is too high due more empirical approach is being pursued by the UK
to the air wake effects, the pilot will be unable to in an air wake model currently installed in the AFS
consistently achieve safe and accurate landings and and successfully used in trials aimed at supporting
the SHOL boundary is reached. The UK is currently FOCFT with simulation. The model will be
developing a modelling capability to assess and evaluated independently by Australia and the US in
quantify the effects of the air wake on helicopter deck similar trials in the near future. The US has proposed
landing operations. The project is divided into two inverse simulation as a means of deriving air wake
phases. The first phase uses a constrained simulation data from aircraft position and pilot control activity.
tool to estimate the control actions required to fly a Each nation has been collecting air wake validation
helicopter model along a prescribed flight path that data through ship trials and wind tunnel experiments
describes the deck landing task. The simulation and distributing the results to other members of the
incorporates a model of a human pilot with group, forming a validation database.
parameters to allow the pilot strategy to be varied.
This prevents the prediction of unfeasibly large or The air wake models have already been applied to the
rapid control inputs. The second phase utilises prediction of operational limits using a variety of
Wavelet Analysis to decompose the time histories of techniques, including piloted simulation and pilot
pilot control activity into discrete events which are workload modelling. It is anticipated that the
then analysed to indicate the level of workload synergy of effort achieved under TTCP will make
required. considerable savings in the time and cost required to
develop a high fidelity DI simulation and will
The aim of the research is to enable the investigation significantly enhance the maritime operational
of the mechanisms by which a helicopter performing capability of participating nations.
a deck landing is influenced by the presence of an
unsteady air wake and by ship motion. A piloted 7 ACKNOWLEDGEMENTS
simulation trial is planned on the AFS at DERA The research described in this paper results from the
Bedford to collect data for the assessment of the pilot
culmination of effort of perhaps forty individuals in
model and to identify suitable modelling parameters
the four nations. The authors gratefully acknowledge
for all stages of the deck landing task. The project
the substantial contributions of their colleagues
will culminate with the production of predicted without whom this work could not have progressed
SHOLs which will be compared against equivalent
as reported here.
data derived from piloted simulation.
The Canadian wind tunnel work is a collaboration
6 CONCLUSIONS between the Institute for Aerospace Research and the
This paper has shown how the modelling and Air Vehicle Research Section of the Department of
simulation of ship air wakes offers substantial National Defence.
potential benefits to maritime helicopter operational
availability through improved ship design, the 8 GLOSSARY
development of new technologies and qualification
AFS Advanced Flight Simulator (DERA)
techniques and more effective pilot training.
AO Auxiliary Oiler
Simulation allows any wind condition and ship AOR Auxiliary Oiler Replenishment
motion to be investigated and enables exploration of
CFD Computational Fluid Dynamics
areas of the flight envelope which may be dangerous
DC Direct Current
in flight. By modelling the air wake, helicopter-ship DERA Defence Evaluation and Research Agency (UK)
interface issues can be explored and SHOLs
DI Dynamic Interface
estimated before a ship is built. High fidelity
8-12
L.A.Maslov
N.O.Valuev
A.V.Zharinov
Gos NITs TsAGI, Moscow 107005
17, Radio Street, Russia
SUMMARY based naval missile-carrier aviation and submarines,
armed with winged missiles. In spite of the fact, that
An opinion about some reasons of relatively late World War II showed leading role of aircraft-carriers
appearance of the first aircraft carrier in the Soviet Navy among other classes of combat surface ships, «swim-
is expressed. TsAGI's works for provision of aerody- ming aerodromes» had not found their place in USSR
namic compatibility of ship-based aircraft and carrier defense doctrine, which was not supposed conduct of
are described. The works were directed on develop- war actions on alien territory. This circumstance and
ment of methodology of estimation of wind flow distur- also economic difficulties of parallel creation of aircraft
bances over flight deck and behind carrier by model carriers and ASBR have lead to the fact, that prolonged
experiment in wind tunnel. Its aim was design and real- time the soviet leadership at forming of programs of
ization of measures of reduction of these disturbances military shipbuilding gave absolute priority to ASBR,
and its gradients up to level required for takeoff landing capable to decide the strategic tasks.
operation safety.
The development of marine strategic forces demanded
Numerous model tests of different ship architecture with time the decisions of the tasks of protection of the
variants allowed to study the mechanism of flow forma- positional areas of its deployment from anti-submarine
tion and to estimate the effects of motion ship kinemat- means of antagonistically side. For this aim construc-
ic parameters relative to wind and of some features of tion of heavy aircraft-carrier cruisers (HACC) of the pro-
ship architecture on the flow structure. jects 1143, 1143M and 11434, equipped with VTOL air-
craft Jak-38, was deployed. The particular features of
Some recommendations were developed and created these ships applied architecture were a consequence
on aircraft-carrier cruiser «Minsk» for deck flow level- of their functions.
ing. The range of relative wind angles was determined
for safe landing on carrier «Admiral Kuznetsov». An absence of experience of creation of such objects
set before industry the quite a number of problems.
INTRODUCTION Whereat one of principal tasks was found to be the pro-
vision of aerodynamic compatibility of ship-based air-
According to its geographical position Russia is the craft and carrier-ship. With account of specifics of this
continental power. Therefore on many centuries its problem, its decision was laid on Central
main defense task was to restrain an aggressor from Aerohydrodynamlc Institute (TsAGI).
west and south directions. The lack of oversee territo-
ries and restricted emergence to Baltic and to Black METHODIC AND RESULTS OF INVESTIGATIONS
seas set before Russian naval fleet generally the tactic
tasks of support of maritime flanks of overland armies As the first step there was the choice of methodology of
on insignificant small distance from its shores. Such experiment on receiving of physical picture of flow over
conception persisted up to 60th years of XX century. the ship. Freeboard and super structures of aircraft-
carrier from point of view of aerodynamics are typical
Appearance in the stuff of the naval forces of the poten- bluff body with separated zones and vortex flow, fixed
tial adversary of atomic submarines with ballistic rock- on sharp bends of form. Since in this case the inertia-
ets (ASBR) radically changed the views of soviet leader- type effects prevail over the effects of viscosity, the
ship on role and the place of navy, and principally - on resulting flow picture practically does not depend on
its enlisted stuff of ships. Reynolds number, and, consequently, on scale effect.
This consideration allowed to go along the way of tests
The boundaries of anti-submarine defense have moved of models of aircraft-carriers in wind tunnels, giving for
on the thousands kilometers from coast, what demand- full-scale experiment the role of controlling instrument.
ed not only increase of autonomity and seaworthiness Besides, this gave the possibility of realization of
of ships, but also creation of principally new for the advanced investigations of perspective configurations.
domestic navy ships with aircraft armament. Anti-sub-
marine cruisers of project 1123, type of « Moskva», The wind tunnel with open working part and nozzle
each equipped by 14 anti-submarine helicopters, put diameter 2.2 m, having low level of turbulence, equal to
beginning of the application of deck aircraft in Soviet 0.3%, was used. In 1971 here the experiment on study
Navy. of pressure distribution on topside of HACC's hull for
the project 1160 was made. The results did not give
For a long time Soviet Navy task forces have not in open possibility to receive enough complete insight into the
ocean the air cover, which Is very necessary for ensur- flow picture around ship.
ing of combat stability. These functions was laid on anti- The measurement of velocity fields around carrier-ship
aircraft missile armament of the ships, and counterac- models became basic method of investigation. The
tion to aircraft-carriers of enemy was laid on coast- five-tube air-pressure probes mounted on spatial coor-
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
9-2
Output data of information-measuring system of wind It is necessary to note that fixed flow separation zones
tunnel were the components of local disturbed veloci- are formed above surfaces of decks immediately
ty Vx, Vy, Vz, vector fields Vxz=f(x,z) and Vyz=f(y,z), behind its bow edges. Its heights are in the range 1 -2 m.
and also distributions of angles of horizontal and verti- They do not exert considerable influence on resulting
cal flow washes en=f (y,z) and sv=f (y,z), en=f (x,y) and velocity field in wake, but it Is necessary to take into
£v= f(x- y)- As longstanding practice has shown, such account them at analysis of aircraft takeoff conditions.
set of information permits to carry out exhaustive analy- Described above picture is principal structure of flow
sis of features of flow in wake behind ship, structure above flight decks and in wake behind aircraft carrier of
around aircraft-carrier and to determine quantitative the considered configuration. Both hull and island are
characteristics of flow. Certain generalized picture of flowed as thick wing of extremely small aspect ratio.
flow, received on base of tests of series of aircraft-car- The vortex system undergoes some changes quantita-
riers' models, forms of which correspond to classic tively in dependence on absolute value of angle (, pitch
configuration with displaced on the starboard side and roll angles, and also on features of ship architec-
island and angled flight deck, is shown on Figs. 1 and 2 ture, but qualitatively it remains practically unchanged.
as an example. For example intensities of vortexes increase practically
according to linear law at an increasing of absolute val-
Vortex system of flow around aircraft-carrier is the fol- ues ( up to investigated range -25°<ß<25°).
lowing. In front of carrier the undisturbed horizontal
wind flow acquires vertical upwash due to which the Aircraft carrier in real conditions is subjected to rolling,
flow separation occurs from windward edge of fore- pitching and heaving. However correct modeling of this
deck. The vortex sheet is curtained into large-scale vor- process in wind tunnel with the aim of estimation of flow
tex core, named further as «bow vortex», which propa- disturbances is impossible. The reason of this is in high-
gates over flight deck. Vortex of analogous nature is frequency oscillations of scale model, required accord-
formed on windward side edge of ship's sponson, on ing to kinematics criteria. The oscillations are close to
which angled deck is located. This vortex is found vibration, what does not correspond to surveyed phe-
downstream up to distances behind the stern equal to nomenon. Therefore the investigations were restricted
0.5-0.7L (where L- length of ship's hull). Bow vortex, as to estimation of influence of limiting static roll and pitch
consequence of loss of its connection with feeding vor- angles on characteristics of stationary flow fields
tex sheet, is eroded gradually or joins with more inten- around heeled or trimmed models. Such positions of
sive vortex, coming off from edge of sponson. ship are possible at presence of battle or operational
damages.
Third component of vortex system of aircraft carrier is
the vortex, generated by island at angles (, which are Model experiment showed that heel on windward side
different from zero. Intensity of this vortex is compara- on comparatively small angle 9=3°( is accompanied with
ble with intensity of vortex, formed on edge of deck. noticeable reduction of intensities of vortexes (approxi-
Namely it exerts a crucial effect on parameters of air- mately on 50-85%), coming from corresponding deck
flow in zone of landing glidepath. edges. Opposite effect is observed at heel on leeward
9-3
side (Fig. 3). requirements from one side and was technologically
Influence of trim is more essential. So, intensity of edge and tactically acceptable from another side. The
vortexes at \p=1.5° (trim by the stern) increases in 1.5-2 accepted composition of LD included three flat and one
times and correspondingly decreases at i|)=-1.5°. The curved plates, as well as fairing on front wall of sponson
dead fluid zone after stern grows exactly likewise at trim
by the bow, because of it Is observed more strong fall of The flat plates turned the vortex flow into contoured
path of vortexes, propagating in aerodynamic wake passage between blast fence and curved plate, deviat-
behind ship (Fig.4). ing it from flight deck. The fairing with symmetrical
semi-elliptical profile practically removed separation
Vortex, generated by island, practically does not zone over deck and decreased deceleration of flow to
depend on angles of heel and trim. However the change acceptable level. Whole complex LD reduced lateral,
of characteristics of another elements of vortex system vertical and longitudinal disturbed velocities and its gra-
leads to some change of resulting flow intensity picture dients over flight deck to required values, as it is shown
in wake due to interaction. on Figs. 5 and 6.
There were made attempts of search of such shape- Full-scale measurements on HACC «Minsk» showed
formations of island, at which vortex formation behind it good convergence of the results with model experiment
was minimal. However since a use of radical means of ones. The flight tests confirmed correctness of the
aerodynamic improvement, as example spoilers and accepted decisions. All ships of this type were
deflected flaps, in this case is impossible to create on equipped with developed LD complex, and there are no
practice, recommendations are reduced to require- any takeoff accidents in process of long-term flight
ment of decrease of aspect ratio and relative width of operation of aircraft Jak-38 due to external aerodynam-
island. ic conditions.
Displacement of island to starboard from longitudinal The developed methodology allows easily to estimate
plane and closer to the bow favorably effects on condi- the external aerodynamic condition on the first in USSR
tions of landing, since in these cases vortex, generated aircraft-carrier of classic type named now «Admiral
by it, removed from glide path. Kuznetsov» (Fig.7) from point of view of safety problem
of takeoff/landing operations on the design stage.
How already it was noted, besides of kinematics para- Here, just like for Jak-38, the determination of require-
meters, there is a noticeable effect of some features of ments to parameters of air flow on trajectories of take-
ship's architecture on quantitative characteristics of off and landing for ship-based fighters Su-33 and MIG-
flow around it. For example the rise of foredeck bow 29K was as a point of departure.
part of HACC of the project 1143, type of «Minsk»
effects on flow above deck like trim by the stern. Previous experience of satisfaction of demands to flow
parameters along Jak-38 takeoff path needed in its
Just for this cruiser the first TsAGI experience was evolution. There are high velocity of approach and lim-
obtained in purposeful action on flow characteristic ited deck section of touching by arrester hook. In com-
over its angled flight deck. That was in 1981 during the parison with ground-based aircraft the requirements to
course of development of horizontal short-run takeoff accuracy of nominal glidepath keeping are increased.
of VTOL aircraft Jak-38 from «Minsk». This involve tightening of requirements to air flow para-
meters too. Besides, purposeful action on vortex sys-
Aerodynamic tests have shown, that intensive vortex tem of wake becomes difficult due to an increase of
comes from windward side edge of risen foredeck. The scale of the phenomenon. Therefore one of the way of
vortex deflects by containers, located on its way just on increase of landing safety was search of range of
flight deck. Moreover vertical front wall of large spon- angles (, inside of which flow parameters would be con-
son leads to appearance of well-developed separated formed to the requirements.
zone, covering up to 20% of runway. Flow in this zone is
characterized with high deceleration of incoming flow Such range was found in process of model tests, results
(up to 0.65V»). of which were confirmed subsequently by measure-
ments on ship. As test pilots noted, simulation of land-
At the analysis of safety of above-mentioned takeoff the ing outside the recommended of angle (range accom-
special requirements to parameters of air flow on take- panied with vibration of aircraft, mushing and in isolated
off runway of given aircraft must be taken into account. cases - with rolling-off. These phenomena, although
These data became basic criteria of conformity takeoff are not catastrophically, presented certain hazard for
conditions to required ones for safety. Study of data, operational pilot's. The necessity of additional control-
received in wind tunnel has shown, that it is impossible ling actions increased high psychophysical load on the
to satisfy these criteria without modification of ships pilot (Fig. 8).
architecture, directed on air flow leveling. To decide
this problem, TsAGI jointly with industry have carried Measurements of non-stationary component of dis-
out extensive research work, in process of which turbed air flow with help of hot-wire anemometers
dozens of so-called leveling devices (LD) variants were showed, that frequencies and levels of fluctuations over
tested. bow starting ramp and forward part of angled deck can
lead to pumpage of compressor. Therefore TsAGI rec-
As a result the such variant, was selected which ommended to set profiled semi-elliptical fairing on bow
reduced parameters of flow to correspondence with the and side edges of ships ramp and forward part of spon-
9-4
son. This permitted to eliminate formation of extensive Almost decennial experience of trouble-free drill opera-
separated zones, unfavourable influencing on condi- tion of deck fighters Su-33 on aircraft-carrier «Admiral
tions of flow around wing of aircraft at takeoff and on Kuznetsov» confirmed, in particular, correctness of
operation of its powerplant in consequence of high recommendations of TsAGI, developed on base of
degree of non-uniformity of flow. model experiments.
island vortex
bow vortex
t t / / / / / / *- ■
/ / / //*""
c) / / / ///-"'
I / I I //*"
\
\
V
\ I I i ///•"
/ Fig.2 Velocity fields in the wake behind carrier
\ \ I I t i / / '
\ \ model at different angles of relative wind
\ \ \ \ a)ß=-7°
b)ß=0
—<V\V c)ß=7°
£•■£-'- -
9-5
,b) _
modified version
required level
containers LD complex
b)
\ \ \ \ \ \ \ t ,
^ * \ \ \ \ \ v
VX -})!/,
\V $ \ \ \ \ J::
\ \.V*• X
V V
.. ' / '
Fig.6 Flow field over flight deck of HACC «Minsk» model ß= -4°
a) original version b) modified version
9-6
SIfitf
Näss?»
■äHEp
1111Hife
P
8
h>
deg. 4
Fig.8 Distribution of horizontal flow washes Eh= f(X/L) along glide path measured
-10 behind model of HACC «Admiral Kuznetsov» at different angles of relative wind ß
10-1
Bernard LANGLOIS
Senior Test Engineer, Dept. 870 (UAV)
Bombardier Services
Abstract Mirabel, Quebec Canada
Helicopters operating from small ships are limited in Sommaire
the maritime environment by high winds and rough Le bilan de revolution d'une simulation du mouvement
seas. In addition, man-made obstacles, such as, hangar de navire et son application dans la simulation de
wall generated turbulence, ship stack hot gas motor l'interface dynamique aeronef-navire est id, präsente.
ingestion, inappropriate deck lighting and markings Les operations d'helicopteres ä bord de navires autres
limit helicopters. Dynamic Interface (DI) is defined as que les porte-avions sont limitees, en general, par des
the study of the relationship between an air vehicle and vents et de la houle. Ces deux facteurs sont la cause
a moving platform. It is performed to reduce risks and principale du mouvement irregulier de la coque,
maximize operational flexibility. Countries with a large mouvement ä six degres-de-liberte don't plusieurs
number of platforms conduct DI testing as a matter of peuvent etre couples. Sur le pont, il en resulte un
necessity. An analytic approach to helicopter/ship ecoulement extreme de Vair provoque par des
dynamic interface testing is presented. gradients de vitesse violents (rafales) et par la
turbulence atmospherique. Ces conditions climatiques
A brief synopsis of the theory and calculation of the maritimes sont ä l'origine des contraintes
ship motion simulation program is presented. The operationnelles et des facteurs de base des etudes de
Ship Motion Simulation (SMS) Model is derived from l'interface dynamique aeronef-navire. L'interface
the relationship between the wave and ship motion dynamique est definie comme etant les relations entre
spectrum. It incorporates seakeeping philosophy and un vehicule aerien et une plate-forme en mouvement.
applies various definitions of seaway spectral Les etudes de l'interface dynamique sont realisees afin
formulation. SMS defines a seaway, computes the de minimiser les risques operationnels en maximisant
hydrodynamic and hydrostatic forces imposed on a la flexibilite tactique d'un Systeme aeronef-navire.
ship (defined as the product of its transfer function and
the seaway) and calculates a resulting ship time Les facteurs pouvant limiter la disponibilite d'aeronef
history. The simulation is an extensive treatment of a sont le mouvement du navire, la turbulence sur le pont,
floating object's response to the dynamic loads on its l'effet de sillage provoque par l'ecoulement autour de la
structure. superstructure et les conditions physiques du navire,
soi les differents obstacles ou l'etat du pont. La reponse
The application of ship motion simulation as a du navire est definie par la resistance cree par la houle
developmental operational tool is introduced. The lors du passage du navire. La simulation du
primary application of the SMS is in operational mouvement du ;navire est basee sur les caracteristiques
simulation such as aircraft launch and recovery; deck spectrales du navire et de la mer. Le spectre de reponse
handling; and flight readiness or availability. The est le produit des fonctions de transfert et du spectre de
Aircraft/Ship Interface Simulation (DI) is a la mer definie dans le cadre du spectre rencontre. En
mathematical description of conditions limiting the simulation d'interface dynamique, une limite est definie
availability of an air vehicle. Factors affecting an air au moment oil un incident est identifie. L'incident est
vehicle on a moving platform are primarily ship defini par un renversement en roulis ou en tagage
motion; wind-over-deck; ship airwake turbulence; and del'appareil, un glissement du train d'atterrissage sur le
deck conditions (wet, dry, oily obstructed, etc). Sample pont ou Vindication d'un decollage non controle. Un
helicopter/ship interface operational limits or incident peut etre identifie au moment de l'atterrissage,
envelopes are discussed. Spin-off projects into other du decollage ou du transfert de l'helicoptere entre le
fields of growth, such as visual aids, are developed. hangar et l'exterieur. Des exemples du calcul des
Ferrier et Langlois sont diplom.es de l'Ecole enveloppes operationnelles sont presentes ainsique
Polytechnique de Montreal. d'autres applications telque les aides visuelles.
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
10-2
While analytics may seem less taxing to the DI study (RAO). The response amplitude operators define the
process, it cannot replace experimentation. Envelope dynamic ship responses for a specified load/operating
studies will always require physical verification. condition [7].
Ship Motion Simulation The ship response spectrum is created as the product of
An important DI analytical tool is the Ship Motion the RAO and die driving sea spectrum (figure 1) over the
Simulation (SMS) which was initially developed by entire range of frequencies. The response spectrum is
Peter JJr. O'Reilly between 1973 to 1984 for the United reduced to sets of harmonic components for each degree-
States Navy. The program methodology uses spectral of freedom. Synthetic time histories are created
probabilities in order to produce deterministic synthetic stochastically by summing the harmonic components
time histories. over a given time period. A typical time history
equation is given by:
SMS Theoretical Synopsis
The Ship Motion Simulation (SMS) Model is derived
from the relationship between the wave and ship motion
spectrum [5]. It incorporates seakeeping philosophy and
applies various definitions of seaway spectral n=l
formulation, such as, Bretschneider [6]. SMS defines a
seaway, computes the hydrodynamic and hydrostatic where
forces imposed on a ship (defined as the product of its
transfer function and the seaway) and calculates a Ar DOF amplitude
resulting ship time history. The simulation is an w: a circular frequency
extensive treatment of a floating object's response to the e: phase angle
dynamic loads on it's structure.
Time histories are produced by the sum of 48 synthetic
SMS is divided into two basic themes, spectral analysis functions (k=48). Figure 2 displays a typical time
and the calculation motion histories in the time domain. history trace. In summary, the Ship Motion Simulation
The SMS fundamental relationship is: creates deterministic measures of ship motion from a
probabilistic spectrum.
Sr = S w(w) • RAO • / (V, m) (1)
Aircraft/Ship Interface Simulation
where: Sn Ship response spectrum The primary application of the SMS is in operational
Sw(w): Seaway spectrum simulation such as aircraft launch and recovery; deck
RAO: Ship transfer functions handling; and flight readiness or availability. The
f(V,m): Frequency mapping Aircraft/Ship Interface Simulation (DI) is a mathematical
V: Velocity description of conditions limiting the availability of an
m: Relative wave angle air vehicle. Factors affecting an air vehicle on a moving
platform are primarily ship motion; Wind Over Deck;
SMS can apply various definitions for the seaway. One Ship Airwake Turbulence; and deck conditions (eg: wet,
of the most common is the definition called the dry, oily, obstructed).
Bretschneider, which is given by:
An example of DI analysis involves deck handling. In
DI the limitations can be defined as the point at which an
/-1994.5\ aircraft/ship incident occurs. Incident means an
occurrence of aircraft turnover, pitchback or on-deck slide
Sw(w)=iP4H.2
4 (2) at any point from touch-down to hangar stowage and
W5T0 back to launch. Deck handling studies determine
turnover limits, sliding freedom, tiedown forces,
where: TJJ: period (sec) traversing factors, and pitch back limitations.
w: wave frequency (rad/sec)
DI Theoretical Synopsis
Sw(w) :seaway spectrum (m^-sec)
Motion of an aircraft on the flight deck is calculated in
Hs: significant wave height (m) terms of ship motion as a function of the aircraft model.
The aircraft model is considered an extension of the ship.
The model is defined by its landing gear footprint; deck
The spectral characteristic of a vessel is defined in the location and orientation; aircraft weight and intertias,
SMS by experimental or computational developed center of gravity, lateral drag area and center of pressure.
transfer functions termed Response Amplitude Operators The aircraft experiences ship transferred forces and
10-4
moments which create rectilinear and angular In the longitudinal, lateral and vertical directions, these
accelerations on the air vehicle. The accelerations can be inertial loads become:
numerically integrated to determine the position and
attitude of the helicopter relative to the ship as function
of time, for various ship motions [8]. In essence, the X\ /TllTi2Ti3\/ Fix \
aircraft is displaced as the sum of all forces, to which it Y = T21T22T23 Fiy (5)
is exposed ZJ Vr3iT32T33AFiz+wJ
The inertia! loads at the helicopter center of gravity where: Tij = T (<{>,8,\|/) (transformation matrix from
induced by ship motion is given by; ship's axis system to horizontal level/vertical axis
system).
W*AXcg are!
WAYcg (4) 4. roll
9= pitch
W*AZcg yaw
I C3SR = SwxRAO
1
At each of "n" freq
! h
RESPONSE Ä"" „ I -4 *""' o I A"**" c I ""c . """ c » " " (P «e lag at each
SPECTRA
ule u)e We C0e (Oe (oe
I I I I I I O Harmonic Analysis
imo
N,^.., A^.o^ev, A„. »..e,, AB,.«,^, AYl.Ul.eY^ """components
HARMONICS?-'«, Av2.W2.ev2 Ae2.U2.ee2 A02
vz ,ü) 2 'e 0
vi2 AY2.<02.eY2'•'•': , ,
•■ ■
COMPONENTS: *• '•
"I.---". *«.*••■.
A
AVn. d>». EVn Aen. «">„, Een A0„. <V £0n Yn. »„, eYn
TODR -on
FRMGÜDSrüel
\
F*,-»*^) (6)
A(
Klnertie) = tan'-1 (11)
(K)(LBL)
where: Ay =Aircraft projected area normal The simulation model assumes a constant wind, therefore,
to the VWj« component A$(wind) is constant throughout the simulation run.
Fwy =Lateral force applied at the aircraft However, A<t>(inertie) is continuously changing with ship
motion. The total incremental change in the aircraft roll
center of pressure due to wind with respect to the slup is given by:
The axial forces on the main landing gear due to the wind
force Fwv is given by: Afytotal) = A<|)(vent)+A<t»(inertie) (12)
Deck conditions, eg: dry or with substances, such as,
P ) water or oil, is a variable in the program. This parameter
FRMGwpM - (^y > 7£L CJ ro affects aircraft stability by changing the coefficient of
friction between the aircraft landing gear and the deck.
where: Aircraft handling systems are handled much in the same
FRMG
WIND
= Main Gear ri nt
^ 8 ) "^ force way. A maximum value of the encountered force load or
Fwy = Wind lateral force component geometric ship position is preprogrammed. When either
force loading or ship angular position is greater than the
WLcp = Center of pressure wateriine manufacturer's design limits, an aircraft incident is
WLQ = Ground wateriine registered. The aircraft operational limit is produced
LßL = Left wheel butteline owing to the break-down of the aircraft handling system.
RßL = Right wheel butteline
Scenarios are programmed for the "worst case" condition.
The incremental aircraft roll due to the wind is given by: For the greatest landing gear deflection, nose gears are
modelled unlocked and castored for turnover. The model
is lined up with the ship centerline and is rotated on the
deck to find the least stable, but realistic, orientation
(6gure3).
where: K = spring constant Referring to figure 3, the 'worst case' hinge line on the
flight deck about which the aircraft will turnover are
Axial forces on the main landing gear due to aircraft defined by Rto and Lto (right turnover and left turnover).
inertJal forces in the plane of the main gear, is given by: Each line is computed from its main gear position to the
nose gear swivelled for turnover. The azimuth of these
two lines are then determined with respect to the ship's
FRMGdnertie) = Y [ LßL +"RßL J (9) longitudinal axis, AZrto and AZlto.
The distance from the aircraft center of gravity (CG) to
where: each line is computed as TODR and TODL (right and
WLQ =Center of gravity wateriine left). They define the distance that the CG should move
F
RM(3nnertje\ =Right main gear axial force due to the for a turnover to occur (right or left). These lines describe
lateral inertia! force Y defined an angle TOR (right) or TOL (left). They are expressed
in equation (5). as:
Forward
WATULINIOnO
STAW
Starboard
Y
BLW
Hl/ VM 15t/ V» 15
\L
V»
/»• /»•
*lt\
$m
*«•'*■—L_ l*-*l«
/ "• ii A.
«•^"^-J— —\-~*$n
H«\
«•*—1 _J—-^JM
Real-Time DI Applications, Visual Aids meaningful value. The index is modular in design with
Application of DI tools to the operational environment the capacity of incorporating other parameters (e.g.:
has produced numerous real-time improvements. One wind-over-deck module) to improve energy index
such improvement is the CL352 Landing Period significance and applicability. The Energy Index
Designator (LPD) helicopter landing aid. The LPD equation of LPD Mk III measures lateral, vertical
supplies real-time information about the motion of any velocities and accelerations as well as roll and pitch
vessel as a function of helicopter operational limits. The angular displacements and velocities weighted by
system furnishes this information about any kind of dynamic coefficients. The equation in the Mk m is the
aircraft in any sort of sea condition on any sea vessel. sum of the squares of the various parameters and terms
LPD is designed to reduce pilot workload in completing representing real-time ship/aircraft interface motion.
ship interface activities by improving operational security
in the reduction of helicopter hover time. EI= a1y2+a2y2+a3z2+a4z2+a5r+a6r-ra7q +agq (17)
(where a„ z^... are weighted dynamic coefficients)
The LPD may show improved recovery
opportunity from its ability to identify the onset of As indicated in equation 17, the index contains
quiescent ship motion periods. This ability is based on acceleration, velocity and displacement terms which
ship motion as a function of the mechanical and dynamic determine the motion of the ship in the near future. The
limits of the helicopter. These limits are expressed by a LPD code calculates the rate at which a vessel can
scalar empirical formulation, termed, the energy index. displace due to natural hydrodynamic forces against the
The index identifies ship quiescence using displacement, structural and dynamic operating limits of the matching
velocity and acceleration terms. In short, the index air vehicle. The energy index uses eight parameters roll
furnishes information of the motion a ship must travel and pitch, their rates, lateral and vertical velocities and
in the near-term future. This does not suggest that the accelerations. All of the parameters are weighted by
index is predictive (using historical information to dynamic coefficients which are weighted according to the
extrapolate into the future). Rather, it capitalizes on the individual degree-of-freedom, the coupled degrees of
rate at which a vessel can displace due to natural freedom and normalized according to aircraft
hydrodynamic forces as a function of the structural and characteristics. The remaining two degrees of freedom
dynamic characteristics of the approaching air vehicle. (yaw and surge) are monitored for motion within certain
limits and may be incorporated more actively later if
The energy index is an empirical formulation warranted. The degrees of freedom, viz: roll, pitch,
designed to convert ship motion characteristics, aircraft lateral, and vertical, are considered the most important for
structural dynamic limits, and user experience into a motion sensitive tasks (in particular launch and recovery
of air vehicles).
10-10
established between each of the following four degrees of PHI PHI THE THE
VT VT LT LT
PHIMX THEMX LTMX L1MX
VTMX VTMX
freedom and their derivatives. A relationship is derived
for roll angle and roll rate, pitch angle and pitch rate, COEFFICIENT MATRIX CALCULATION
lateral velocity and lateral acceleration, and vertical MATRIX A = A1 A11 A12 A13
A2 A21 A22 A23
velocity and vertical acceleration. These relationships are A3 A31 A32 A33
CHARACTERISTICS
ACnVTTY
CHARACTERISTICS
(l.t. AIRCRAFT)
AND FORCE DIRECTION
DEPENDENCE (i.e. ROLL.
YAW, VERTICAL)
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
11-2
^hip-y/VV^syrsteni
ship accBlaratl
\' noise
Figure 1: Proposed approach
error
(measurement/
model)
The second essential element is a spatial defini-
^&*n
tion of the bullseye which is used to guide the UAV to
its centre. This is contrary to the instanteous defini-
tion of the bullseys used by traditional systems. The
proposed system calculates the virtual position of the
bullseys which is independent of sea created ship pur- Figure 3: Data fusion method
turbations. This methodology is described below.
An important advantage to this approach is that
the guidance and the flight control modules are de-
coupled from the perturbations caused by ship motion. inertial navigation system. INS data used are essen-
It is reasonable to assume that this approach can be tially accelerometric and vehicle orientation converted
considered more robust than that of the classic recov- into the inertial reference system.
ery systems. This approach, however, will have an The problem is to determine the components,
important impact on the architecture of the system. based on these data, of the AM vector and its derivat-
The primary structural modification is the creation of ive located in the inertial reference system (see figure
various data streams coming from several data sources 1).
defining the deck attitude. Several methods may be used to track the vehicle.
A method based on pseudo- measurement techniques
was presented in [1], other techniques may, however, be
3 Tracking UAV used. The data handling methodology used is classic.
This is illustrated in figure 3.
The technique of data fusion used is illustrated in fig- When the ship averaged accelerations are con-
ure 2. The illustration is of an autoland architectural sidered as input (originating from the inertial naviga-
system using a proposed sensor arrangement necessary tion system), the helicopter-ship system responds. The
for continuous vehicle tracking. Two optical sensor re- output is the definition of the relative position of the
ceivers provide angular positioning. The vehicle may vehicle as function of the ship. The air vehicle position
be tracked using angular sensors. is measured by a device, such as, an optical or radar
Using only angular information in a similar way tracking system. The mathematical model which de-
that radar is used, it is possible to determine the po- scribes the system is composed of: model describing
sition of the UAV relative to the ship. This alone is the state of the vehicle and ship (state variable model)
insufficient. The definition must be converted to the and a model describing the tracking system. These cal-
quasi intertial reference system. It is for this reason culated measurements are compared with actual meas-
that the information originating in the ship's inertial urements from the tracking system. The equations de-
navigation system is combined with that from the UAV fining the position of the UAV are those of a classic
11-3
the system.
It is known that the data provided by the track-
ing system (optical, for example) are excellent in the
long term but are relatively noisy. On the other hand,
the information provided by accelorometers are more
accurate in the short term while drift is an important
factor in the long term. Excellent results are produced
when combining these two types of data collection sys-
tems. By using this technique, the reconstruction of
the state vector is just as easily calculated in the ship's
Figure 5: Kaiman filter results
reference frame as in the inertial coordinate system.
However, the latter is prefered since it allows decoup-
ling of the vehicle command and control system from
ship motions. Figure 4 is a flow diagram for the mod- choices to make: the recovery point is defined as the
elling of this system. terminal point of the guidance recovery trajectory. The
optimal beginning of the final recovery phase must like-
w (UAV location)
wise be defined.
ship /
Instantaneous / ship
location / average
trajectory : Y(UAV)- Y(sWp)
4.2 Virtual bull's eye position predic- Vertical plane Horizontal plane
tion
The calculation methodology begins with a five minute
ship motion time history in the x, y, z directions (surge,
sway, heave). The autoregressive ship model (see figure
6) is derived from classic least squares techniques.
ship
sea motion
Chin
disturbances
e
—
S#
& »-*■
Figure 7: Virtual bull's-eye trajectory
b(t) y A
white noise
(0,1) /
least square
-«£=
algorithmes The first term of the right side of equation 1 can-
not be calculated since at k instance, the random se-
quence b(k +1), ....b(k + m) is unknown. At the inverse
Figure 6: Identification scheme
the outputs y(k),y(k — 1),... have been already ob-
served. Thus, considering the situation at time k, the
This model is used to predict the location of the best prediction for the output y(k + m) is given by:
bullseye at the instant of recovery. The deck posi- y(k + m/k)=Y(q-1)y(k)
tion is calculated with a model defining each of the It is the term used to predict the location of the
three degrees-of-freedom mentioned earlier. Consider deck.
a discrete-time single-input-output plant governed by
In summary, it is necessary to pre-calculate all
the equation:
the polynomial Y(q~1). It is important to remark that
each values of Y(q_1) are only valid for a considered
*(*) = Ä^W time m. The calculation may be made off-line based
on ship motion characteristics.
where Figure 7 displays the performance of this ap-
y output sequence (ship motion) proach. The left portion represents motion in the ver-
b sequence of independent random variables tical plane while the right part indicates motion in the
k current time horizontal plane. The dotted graph displays the flight
q forward shift operator qky(m) = y(k + m) in- deck motion while the heavy lined graph represents the
troduced by Astrom. virtual bullseyes trajectory during the landing phase.
It is noted that the model A(q~1) is valid at any
given instance. This calculation must be completed
5 secondes prior to the beginning of the vehicle descent.
The position of the deck is given by k + m where 4.3 Selecting the recovery moment
m is the position prior to the instant of recovery. This
given by the expression: The second point consists of selecting the optimal in-
stant that guarantees a safe recovery. A calculation
is made using an energy index which determines the
y(k + m)= _n&(fc + m) period where ship motion is below a level which en-
A{q~i)
sures a safe deck. In brief the Landing Period Desig-
rewriting the equation nator (LPD) is a helicopter recovery aid which calcu-
lates ship motion as a function of helicopter operational
limits. The energy index collapses ship and helicopter
y(k + m)= Xfa"1 W* + m)+ Y^^k) (1) limits into a scalar value while retaining the physical
significance of the helicopter ship system. The energy
X(<7-1) and Y{q~x) correspond to the minimal index is constructed such that 4 energy states are de-
solution scribed. The lowest energy state (very safe) guarantees
with respect to X of the following polynomial the aircraft operator a quiescent flat deck for a delayed
equation: period of time consistent with basic principles of New-
tonian mechanics. For more details on the LPD, please
A(q-1)X(q-1)+q-mY(q-1) = l (2) refer to [2].
11-5
the prediction technique well within acceptable para- VTOL" AGARD CONFERENCE PROCEEDINGS
meters (none of the success criteria failed). In higher 591, pp 25/1-25/8, (1996).
sea states, the only failed recoveries occurred owing to
exceeded vertical velocities. This infers that a vertical [2] B. de FERRIER,and J. SEMENZA. "Development,
Simulation, and Testing of the Landing Period Des-
rate predictor should be included in the final system.
ignator (LPD) Helicopter Recovery Aid" Proceedings
of the American helicopter Society, May 1994, Wash-
ington
7 Conclusion
[3] B. de FERRIER, B. LANGLOIS, F. BERGERON,
The purpose of this article was to describe the design C. REBOULET, V. FUERTES, J-C. BARRAL.
of an automatic recovery system for a VTOL type Un- "The design and test using simulation techniques of
manned Air Vehicle. From simulation results serveral an automated UAV-VTOL shipboard recovery sys-
tracking systems are possible. Two systems were con- tem" Proceedings of the American helicopter Society,
sidered. One tracking system used two optical receiv- May 1998,Washington
ers while the other used a radar tracking system. The
results were comparably acceptable. Simulations res-
ults, however, favoured the optical tracking system, in-
spite of the known problems associated with its use in
the maritime environment. Data handling fusion tech-
niques from several optical receivers may adequately
compensate for the deficiencies noted. An architec-
ture using a redondant optical tracking system is an
acceptable tracking technique. Prom simulation res-
ults optical tracking system'failures may occur during
the most critical phase of the vehicle recovery. Us-
ing redondant systems, it is highly unlikely that both
tracking systems would fail at the same time. The
costs involved, however, must be equally considered.
In addition, unlike radar systems, optical tracking sys-
tems are not confronted with the problems associated
by mulitpath. Careful trade-off analysis is necessary
when deciding on the components of the autoland sys-
tem.
What ever the technology chosen, data fusions
techniques used in this study were achieved using Kai-
man filter applications. The Kaiman filter application
reduces the tracking error to under 20cm permiting the
system to be a feasible solution.
The deck prediction module is considered indis-
pensable to ensure safe recoveries in agitated sea con-
ditions. The module calculates the virtual location
of the recovery deck at the precise moment of touch-
down. Simulation results supported deck recoveries un-
der 50cm from the centre of bullseye. It is important
to recall that the primary advantage of the autoland
system proposed its robust character and its capabil-
ity to operate even in agitated seas beyond sea state
6. Preliminary analysis has shown the system to be
feasible. The task at hand validate the system is by
assembling and testing the system at-sea.
References
[1] B. de FERRIER,and C. REBOULET. "Conception
et developpement d'un Systeme d'appontage mari-
time automatique d'un vehicule generique de type
12-1
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
12-2
■.(in)
1 H fp\
i(
T
Q<fut) Q cushion
X^^S„ ^ = ^ plenum
tion of added resistance with wave height does not hold small motions are defined by a small translation vec-
for an SES. tor if = (771,772, f?3)T and a small rotation vector Ü —
T
(T/4,?75,7?6) . The six small displacements, 771,...,776
The main goal of the present PhD research project is (i.e. surge, sway, heave, roll, pitch and yaw displacement)
to investigate how one can accurately compute the mo- are the actual unknown motions that have to be solved.
tions and added resistance due to waves of SESs. This Next to these unknowns two additional unknowns occur:
paper presents results of a new computational method the excess pressure in the cushion plenum, pc, and the
for motions and added resistance due to waves of SESs. excess pressure in the stern seal plenum, ps. The excess
These results are compared to the results of experiments pressures are assumed to be constant in space, so acoustic
that were carried out by Kapsenberg[l 1] at MARIN. The effects of the air inside a plenum cannot be resolved.
computational results are in good agreement with the
MARIN experiments as far as motions and cushion pres-
sure are concerned. The large added resistance due to 2.1 The equations of motion
waves that was measured at MARIN does not however
follow from the computations. New experiments have The motions of the SES are assumed to be small. Then
been carried out at the Ship Hydromechanics Laboratory the equations of Euler for the motions of a rigid body
of Delft University of Technology. These experiments can be linearized to the following system of differential
give new insights into to the origin of the added resis- equations:
tance due to waves of SES.
Mjk- d2i}k
The next section presents a description of the new compu- E : dt 2 =F i = i,...,( (1)
tational method for motions and added resistance due to k=l
waves of SESs. Section 3 describes the new experiments where Mjk is the is the kth component of the yh row
that were carried out at Delft University of Technology. of the generalized mass matrix, Fi, F2, F3 are the forces
Section 4 presents and compares the computational re- which acts on the vessel in ^-direction, y-direction, and
sults, the experimental results of MARTN and some new z-direction, and .F4, F5, F6 are the moments which act on
experimental results. the vessel around the z-axis, y-axis, and z-axis. Each
force is split up into the following components: a gravi-
tational force, a propulsive force, a hydromechanic force,
2 Computational method a force due to the air cushion and a force due to the seals.
The propulsive force is balanced by the resistance of the
vessel. The hydromechanic force is computed by means
This section presents the computational method for mo-
of a three-dimensional panel method. The force due to
tions and added resistance due to waves of Surface Effect
the air cushion follows from integration of the cushion
Ships. Up to now the method is only suitable for head
excess pressure over the deck. The forces due to seals
and following waves as only heave and pitch motions are
follow from the appropriate seal models.
considered. The effect of surge motion is neglected. Nev-
ertheless most of the theory is presented for six degrees
of freedom. 2.2 The equations for pc and ps
Consider an SES sailing in waves at a forward speed U.
The vessel carries out small oscillatory motions which are Next to the equations of motion two additional equations
superimposed to the translatory forward motion. These for the additional unknowns pc and ps have to be formu-
12-3
lated. These equations follow from conservation of mass one-dimensional analysis leads to good results:
for the air in a plenum and the equation of state of the
air in a plenum. The equation of state is taken to be the out) 2Ap
isentropic gas law. Figure 1 presents the system of plena Q{ ciAi (3)
P
and the air volume fluxes of an SES. The equations forpc
and ps read: where c; is the leakage coefficient which depends on the
local geometry of the orifice, Ai is the leakage area, Ap
Vc dpc dVc is the pressure jump across the orifice and p is the density
K(Pc+Pa) dt dt of air. The leakage areas under the seals follow from the
(2) seal models. The leakage area under a seal is highly non-
Vs dps _ (n(in) n(0Ut)\ dVs linear in the local relative wave height. When the relative
■K(p.+Pa) dt Ws S
^ ' dt wave height at the seal is large, the seal will touch the
water surface so the leakage area equals zero. When the
where Vc is the air cushion volume, Vs is the stern seal relative wave height at the seal becomes smaller a leakage
volume, pa is the ambient pressure and K is the ratio of gap will occur suddenly, and afterwards grow linearly.
specific heats of air; K = cp/cv « 1.4. Furthermore This process cannot be captured by a linear formulation.
Q(in) = Q(in) + Q(m) whjch ^^ sum of ^g fl()w trough
the cushion fan and the leakage flow from the stern seal,
and Qiout) = Q^6ut) + Q&"0 which is the sum of the leak- 2.5 The stern seal
age flow under the bow seal and the leakage flow under
the stern seal. The cushion volume Vc depends on the This subsection presents a brief descriptions of the
heave and pitch displacement of the vessel, the geometry stern seal model. A more complete description can
of the seals and the wave height in the air cushion. The be found in reference [18]. Lee[19], Steen[5] and
stern seal volume Vs depends on stern seal geometry. Masset et al.[20],[14] presented similar seal models.
The following assumptions underlie the stern seal model.
The model is two-dimensional in a longitudinal plane.
2.3 The fan system This implies that the wave height is assumed to be con-
stant in transverse direction. Therefore the seal model is
The air volume flux through a fan into a plenum, Q(m) most suitable for head and following waves. The gravi-
is approximated by the linearized steady fan charac- tational and inertial forces that act on the seal canvas are
teristic. Durkin and Luehr[12], Sullivan et al.[13], neglected. Ulstein[6] developed a seal model which in-
Masset et al.[14] and Witt[15] show that fans respond cludes the inertial forces. He found that inertial effects
in a dynamic way to oscillating back pressure. Sulli- are important for high frequent motions. When the seal
van et al.[13] show that the dynamic behavior of the fan does not touch the water surface air will leak from the
has a damping effect on the heave motions of a hovering cushion under the seal. This air flow will result in a dy-
box. Moulijn [16] also showed that the fan has an impor- namic pressure distribution under the seal which will re-
tant effect on the overall motions of an SES. Nevertheless duce the air leakage gap. Lee[19] neglected this pressure
all computational methods for seakeeping of SES, includ- distribution. In the present model and in the models of
ing the present method, use static fan characteristics. This Steen[5] and Masset et al.[20],[14] a simple stationary
is due to the absence of information on dynamic charac- one-dimensional modeling of the leakage flow is used.
teristics of lift fans. The seal canvas is assumed to have no bending stiffness;
it only transmits tension.
For a given bag configuration the following parameters
2.4 Air leakage determine the bag geometry and the tension in the seal
canvas: heave displacement, pitch displacement, cushion
Many authors (see for instance Nakos et al.[7], excess pressure, seal excess pressure, mean wave height
McHenry et al.[17], Masset et al.[14]) consider air leak- at the seal and mean wave slope at the seal. When the bag
age to be important and highly non-linear. Nevertheless, geometry and the tension in the seal canvas are known,
air leakage is often linearized or even neglected. Steen[5] the seal volume and seal force can be computed easily.
showed the importance of air leakage for the cobblestone The seal may either touch the water surface or leave a
effect. Ulstein[6] carried out extensive air leakage com- leakage gap. In the bottom case the cushion pressure is
putations using a non-linear panel method. He found that larger than the seal pressure, which results in the concave
the following formula which follows from a stationary cushion facing part of the seal.
12-4
2.6 The bow seal form the frequency domain results of the panel method to
the time domain.
The modeling of the finger-type bow seal is much more
simple than the modeling of the bag-type stern seal.
When the local deck height at the bow is smaller than the 2.8 Added resistance due to waves
height of the seal, the lower part of the fingers is simply
bent backwards at the water surface, and no air will es- The added resistance due to waves of a ship is the time
cape under the seal. When the local deck height is larger averaged resistance of the ship when sailing in a seaway
than the height of the seal, the seal will leave a gap above minus the resistance of the ship when sailing in calm wa-
the water surface, and air will escape from the cushion ters at the same speed. For convenience it is henceforth
plenum. The bow seal model is also two-dimensional (in called added resistance. Most authors consider the added
a longitudinal plane). Again the mean wave height and resistance of SESs to be large. Three components which
wave slope at the seal are used. The seal is represented contribute to the added resistance are distinguished:
by a flat boundary which roughly coincides with the fore-
most part of the fingers which actually closes the cushion 1. the usual added resistance of the hulls,
plenum. The part of the seal that is bent backwards at the
water surface is neglected. The frictional forces that act 2. the added resistance due to sinkage,
on the bow seal are neglected.
3. the added resistance of the air cushion.
2.7 Hydromechanics The usual added resistance of the hulls is caused by the
diffraction of the incident waves and the radiation of the
waves due to the motion of the vessel. This added resis-
This subsection presents a brief description of the solu-
tance component is also experienced by normal ships. It
tion method for the hydromechanical problem of an SES
can be computed for instance by the method of Gerritsma
sailing in waves. A more elaborate description of the
and Beukelman[25].
method can be found in reference [21].
The added resistance due to sinkage is basically an in-
The hydromechanical problem is solved by means of a
crease of the steady resistance of the hulls. When an SES
three-dimensional Rankine panel method. The water flow
sails in a seaway, the amount of air leakage increases be-
is assumed to be incompressible and non-rotational, so
cause the ambient waves cause large leakage gaps under
potential flow theory can be used. The boundary condi-
the seals. This results in a decrease of the mean excess
tions on the free surface and on the hulls are linearized
pressure in the air cushion, so a larger part of the SES's
around the undisturbed flow. This leads to a Kelvin free
weight has to be carried by the buoyancy of the hulls. The
surface boundary condition which includes some extra
draft of the vessel will therefore increase, which leads to
terms due to the presence of the air cushion, and a Neu-
a greater resistance of the hulls. This greater resistance
mann hull boundary condition. The hulls and a part of
is partly caused by an increase of the frictional resistance
the free surface are paneled with flat quadrilateral pan-
due to the larger wetted area of the hulls. The rest of it is
els. Each panel has a constant source an dipole distribu-
caused by an increase of the wave making resistance.
tion. The tangential derivatives of the unknown poten-
tial which occur in the free surface boundary condition The resistance of the air cushion follows essentially from
follow from analytical differentiation of a bi-quadratic the following equation:
spline approximation of the potential. Nakos[22] also
uses this spline scheme, but he also uses a quadratic sin- Ä(Be)=Pc-(C6-C.)-5e (4)
gularity distribution on each panel. When the boundary
value problem is solved, the pressure and the wave height where pc is the cushion excess pressure, £(> is the wave
can be calculated. The hydrodynamic forces which act height at the bow seal, £s is the wave height at the stern
on the hulls follow from integration of the pressure over seal and Bc is the width of the air cushion. Some ex-
the wetted part of the hulls. tra terms which account for the momentum of leaking air
should also be included. Then the added resistance due
The unsteady hydromechanic problem is solved in the to the air cushion follows from:
frequency domain, thus avoiding a complicated time step-
ping algorithm and saving much computational time. The R("$ = R^'Hin waves) — i?(00>(calm water) (5)
motions have to be solved in the time domain because of
the non-linear cushion and seal dynamics. Therefore the where the over lining denotes that the time averaged value
theory of Cummins[23] and Ogilvie[24] is used to trans- should be used.
12-5
"T- 1 I —i— 1 1
kA = 0.01 + + 4-
kA = 0.05 E 400-
kA = 0.10
kA = 0.15
experiment
1.5- * t +
+
+ +
+
■K-
££±
+
+
? -200-
kA = 0.01
kA = 0.05
kA = 0.10 + +
-400- kA = 0.15
experiment +
1 i 1 1
0.4 0.6 0.8 1 0.4 0.6 0.8 1 1.4
wave frequency [rad/s] wave frequency [rad/s]
Figure 3: Pitch RAO of the HYDROSES target vessel Figure 5: Quadratic added resistance operator for the
HYDROSES target vessel
kA = 0.01
kA = 0.05
kA = 0.10
kA = 0.15 located in the high frequency range. This resonance is
exp.fore
exp. aft very sensitive to for instance the behavior of the fan, and
is therefore difficult to predict computationally (see ref-
erence [16]).
Figure 5 presents the computed and the measured added
resistance divided by the wave height squared as a func-
tion of the wave frequency. MARIN measured added re-
sistance values up to two times as large as the calm water
resistance. In the case of conventional ships the added
0 0.2 0.4 0.6 0.8 1 1.2 1.4 resistance is proportional to the incident wave height
wave frequency [rad/s] squared. The scattering of the experimental results in-
dicates that this relation does not hold for an SES. The
Figure 4: Cushion pressure RAO for the HYDROSES measured added resistance is negative in some cases. The
target vessel computational results are much smaller than the exper-
imental results. There even does not seem to be any
correlation between the computed and measured added
well, but in the lower frequency range the computational resistance. The computational results also indicate a
results are larger than the experiments. It is not clear non-quadratic relation of the added resistance with the
whether the computational method over predicts the pitch wave height, although the scattering is much less severe.
resonance, or whether the resonant frequency is shifted. The discrepancy between the computed and the measured
This discrepancy is probably caused by the relatively added resistance was the reason for the new experiments.
harsh linearization scheme of the hydrodynamic prob-
lem. The effect of the non-linear cushion dynamics on
the pitch RAO is negligible.
4.2 The DUTSES model
Figure 4 presents the computed and measured RAO for
the cushion excess pressure. The pressure was measured This section presents preliminary results of the new
at two locations: one at the fore side of the cushion and
model experiments. These results are also compared to
one at the aft side. The computations and experiments computational results. The experimental results have not
are in good agreement. The effect of the non-linear cush- been analyzed thoroughly yet. Nevertheless these results
ion dynamics on this RAO is larger, but still not really
give some interesting new insights into the topic of added
significant. In the high frequency range the experimental resistance of SESs. All results are for the model version
results are somewhat larger, while in the low frequency with plates, and are presented on model scale. The model
range the computational results are greater. The latter
is towed in head waves at a speed of 3.72 m/s.
discrepancy is caused by the large computed pitch mo-
tions which cause air gap modulations under the seals. Figure 6 presents the measured and computed RAO for
The Heimholte resonant frequency of the air cushion is the wave force in heave direction. It displays results for
12-8
experiments
calculations
2 3 4 5 6 3 4 5
wave frequency [rad/s]
wave frequency [rad/s]
Figure 6: Heave force RAO for the DUTSES model Figure 8: Quadratic added resistance operator for the
DUTSES model
exp. A - 0.02 +
exp. A - 0.04 x
2 3 4 5 6
wave frequency [rad/s] 2 3 4 5 6
wave frequency [rad/s]
2 -
+ t
_i i i i i_
234567 234567 234567 234567
wave frequency [rad/s]
Figure 10: Added resistance of the air cushion; the experiments follow from the total added resistance minus the added
resistance of the plates
of the vessel was about 27 N. The computed and mea- cushion pressure amplitude is much larger during a wave
sured added resistance of the air cushion are in reason- force experiment than during a comparable free sailing
able agreement. The magnitude is about the same, and experiment. Therefore one might expect to find a larger
the trends do also agree. added resistance during wave force experiments. This is
not however the case.
The added resistance due to the air cushion can also be
calculated from the horizontal seal connection forces. The MARIN model has realistic side hulls instead of
The resistance of the air cushion follows from: plates. If this is the reason for the large added resistance,
ac) the major part of the added resistance should be attributed
Ä( -Fxs - Fxb -pc-Ad- 775 (6) to the hulls. This is not very likely as the realistic hulls
where Fxs is the horizontal stern seal connection force are still very slender.
(positive in forward direction, hence the minus sign), Fx\,
The added resistance was measured in two quite differ-
is the horizontal bow seal connection force, pc is the
ent ways. The MARTN model was self-propulsing by
cushion excess pressure, Ad is the deck area and 775 is means of model water-jets. The added resistance was
the pitch displacement angle. The added resistance of the
determined from a propulsion increase of the water-jets,
air cushion follows again from Equation 5.
which follows from a measurement of the flow through
Figure 11 presents the added resistance of the air cushion, the water-jet. In some cases an extra towing force was
where the experimental results were calculated from the added because the model could not reach the desired
seal forces and the cushion pressure. Again the results speed. The DUTSES model is always towed. The added
are in reasonable agreement. The results of both meth- resistance follows from an increase of the towing force.
ods for calculating the experimental added resistance of This method is, according to the author, more simple and
the air cushion agree quite well (compare Figure 10 with more reliable than the method adopted at MARTN. The
Figure 10). MARTN results might be affected by a decrease of the
efficiency of the water-jet propulsors.
4.3 Discussion The large speed loss of SES might also be caused by a
decrease of the thrust of the water jets. This would ex-
plain the large added resistance measured at MARTN. It
This section presents a discussion on the topic of added
would also explain the statement of Ehrenberg [9] that an
resistance. There is an obvious difference between the
SES has much less speed loss in waves than a comparable
experimental results of MARIN and the new experimen- catamaran, as his experience follows from an SES which
tal results. There are however many differences between
is propulsed by means of surface piercing propellers.
both experiments.
The model was free in heave, pitch and surge during the
MARIN experiments, while the model was rigidly con-
nected to the carriage during the new experiments. The
12-10
to
i i i i l
Figure 11: Added resistance of the air cushion, the experiments follow from the seal forces and the cushion pressure
[10] G.K. Kapsenberg. Seakeeping behaviour of a ses [23] W. E. Cummins. The impulse-response function
in different wave directions. In Proc. Second In- and ship motions. Schiffstechnik, 9(47): 101-109,
ternational Conference on Fast Sea Transportation 1962.
(FAST 93), Yokohamajapan, 1993.
[24] T. F. Ogilvie and E. O. Tuck. A rational strip the-
[11] G.K. Kapsenberg and P. Blume. Model tests for ory for ship motions -part i. Technical Report 013,
a large surface effect ship at different scale ra- Department of Naval Architecture and Marine En-
tios. In Proc. Third International Conference gineering, University of Michigan, USA, 1969.
on Fast Sea Transportation (FAST'95), Liibeck-
Travemiinde, Germany, 1995. [25] J. Gerritsma and W. Beukelman. Analysis of the
resistance increase in waves of a fast cargo ship.
[12] J.M. Durkin and L. LUehr. Dynamic response Technical Report 169s, Netherlands Ship Research
of lift fans subject to varying backpressure. In Center TNO.
AIAA/SNAME Advanced Marine Vehicles Confer-
ence, AIAA Paper No. 78-756,1978. [26] G.K. Kapsenberg. Added mass and damping coef-
ficients for a large ses including an appreciation of
[13] P.A. Sullivan, F. Gosselin, and MJ. Hinchey. Dy- scale effects. In Proc. NAV'94, volume 1,1994.
namic responce of an air cushion lift fan. In HPMV
'92 Conference, Washington, 1992. [27] J.C. Moulijn. Scaling of aircushion dynamics. Tech-
nical Report 1151, Delft University of Technology,
[14] J.F. Masset, J.P. Morel, and G.K. Kapsenberg. Large Ship Hydromechanics Laboratory, 1998.
surface effect ship (ses) air cushion dynamics: An
innovative methodology for theoretical modelling
validation. In Proc. Third International Conference
on Fast Sea Transportation (FAST 95), Liibeck-
Travemiinde, Germany, 1995.
[15] K.C.Witt. Lift fan stability for ses. In Second In-
ternational Conference on Fast Sea Transportation
(FAST93), Yokohama, Japan, 1993.
[16] J.C. Moulijn. Non-linear motions of surface effect
ships. In International Conference on Air Cushion
Vehicles (ACVs), RINA, London, 1997.
[17] G. McHenry, P. Kaplan, F. Korbijn, and
A. Nestegärd. Hydrodynamic analysis of surface
effect ships: Experiences with a quasi-linear model.
In Fast Conference, Trondheim, 1991.
[18] J.C. Moulijn. A model for flexible bag stern seals.
Technical Report 1082-O, Delft University of Tech-
nology, Ship Hydromechanics Laboratory, 1997.
[19] G.J. Lee. On the motions of high speed surface-
effect-ships in waves. Journal of Hydrospace Tech-
nology, 1(2): 13-35,1995.
[20] J.F. Masset and J.P. Morel. A test rig for the anal-
yses of a large surface effect ship seals dynamics:
design, manufacturing and results. In NAV'94 pro-
ceedings, volume 1,5-7 October 1994.
[21] J.C. Moulijn. Motions of surface effect ships. Tech-
nical Report 1051-O, Delft University of Technol-
ogy, Ship Hydromechanics Laboratory, 1996.
[22] D. E. Nakos. Ship Wave Patterns and Motions by
a Three Dimensional Rankine Panel Method. PhD
thesis, M. I. T., 1990.
13-1
SUMMARY
Sa«) •
Wave induced vertical accelerations (cobblestone [(m/s2)2s]
oscillations) of a SES in small sea states is studied.
'
Resonant spatially uniform and nonuniform dynamic
cushion variations are then important. A nonlinear time
domain solution is used. The nonlinearities are mainly
due to the flexible stern seal bag behaviour. It is
demonstrated that both the spatially varying pressure
underneath the flexible stern seal bag and the impact
between the bag and the water are important for the
vertical accelerations of the vessel. The influence of
main parameters characterizing the stern seal bag is Uy , «Hz)
discussed. 10
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
13-2
(/5+A^)tj5+fi5*f|5+C^Ti5+A53fi3
that are part of a ride control system. The placement of (2)
the louver system is essential. For instance if the louver +55*VC53*l3 = FfryFf®
system is placed midships, it will have a negligible
effect on the acoustic resonance mentioned above. The t is the time variable and dot stands for time derivative.
reason is simply that the acoustic pressure component M is the structural mass and L is the structural mass
h
has a small amplitude midships, while it has its moment of inertia around the 7-axis of the SES. Ay ,
maximum absolute value at the ends of the cushion. Bjh and Cy are hydrodynamic infinite frequency
Fig. 3 shows a proposed ride control system by [7]. A added mass, damping and restoring coefficients of the
louver system consisting of two vent valves in the front two side-hulls. F/1 is the linear hydrodynamic
excitation force on the side-hulls and F™ is the
nonlinear force acting on the SES due to integrated
unsteady excess pressure in the air cushion in direction
i. The nonlinearity in F™ is mainly caused by
nonlinear effects related to the flexible stern seal bag.
The hydrodynamic loads on the hull are expressed by
strip theory ([8]). This is not an important part of the
cobblestone analysis.
pvW 2 2 (4)
approximated by 3 cable segments with constant radius
of curvature R. This implies spatially uniform static
v. * ' <
difference pressure over each segment. The tensions in
,d$ segments 1 and 2 are constant and equal. There are
fNT(^-V(x,t) ■ n)dY = 0
J dn equal static difference pressures across segments 2 and
3, i.e., AP02=AP03.
Here NT is the transposed of the trial function matrix
N. The first term in Eq. (4) is rewritten by Green's The equations of motions for the flexible bag are
theorem as derived by a linear perturbation of the static solution.
Compatibility gives the relationship between tension
fNT^^dVc=fNT^-dT-fWNT-V^dVc (5) and the motions in the transverse and the longitudinal
vc r " vc directions. Since the radius of curvature R is constant
The analyzed domain is broken down into Nel small for each cable segment, the equations of motions can be
1-D elements in the longitudinal direction. It is assumed set up separately for each segment. Afterwards these
that <$>(£„t)=N(E,){<b(t)} where {$(t)} is a vector cable segments are linked together with the boundary
containing the node point values of the velocity conditions. The equations of motions for a cable
potential of the air flow. A linear variation is used over segment are finally written as
each element. [9] verified the numerical method by
comparing with the Finite Difference Method and by d\ + (T0+EA)dr]tEA
using an analytical solution based on mode MJ\n=T0- n„+AP„ (6)
superposition. ds2 R ds R
2.2 Local Air Cushion Model at the Stern MJ\=EA *\_EA*\n (7)
It is now focused on the details at the stern bag seal. A ds2 R ds
2-D problem is solved in the longitudinal plane of the
SES. The pressure is assumed spatially constant inside Here nn and n, are the motions in the transverse and
the bag. This pressure can be related to the pressure in longitudinal directions. Mm is the structural mass per
the air cushion by using continuity of mass flow in and unit length of the cable. E is the elasticity modulus of
out of the bag. The air inside the bag is assumed the material. A is the cross dimensional area of unit
compressible and an adiabatic relationship is used width and T0 is the static tension in the cable segment.
between the pressure and the air density. The bag s is the longitudinal coordinate along the cable
structure is modelled as a cable in the cross-sectional segments (see Fig. 4) and APu is the unsteady
plane. In the static case the bag is not touching the difference pressure across the cable. The structural
water, but the leakage height underneath the bag is boundary value problem is solved by "dry" mode
assumed to be initially zero in the time domain solution superposition. We can write
procedure. The aerodynamic pressure force acting on
the bag is an order of magnitude larger than the gravity Tl„(5,r) = Sa.(0<j>» (8)
force. The dynamic contributions from gravity will
appear as restoring terms and can be neglected r]t(s,t) = Zfl.-WtiW (9)
compared to the tension terms. The bag structure is
therefore modelled as a weightless but not massless Here aß) is the principal coordinate of vibration mode
cable. The studied bag geometry is a typical 2-loop bag number i and ({>" and $' are the mode shape functions
configuration (see Fig. 4). The static bag geometry is in the transverse and longitudinal directions. A finite
number of modes is used. Eqs. (8) and (9) are
substituted into Eqs. (6) and (7). The two equations are
multiplied with <j>J and cf>j, respectively and integrated
over the length Lcs of all the cable segments. The two
coupled equations of motions are added and give
where
segment 2 A
kinematic free surface condition is used to generate a force defined by the last integral on the right hand side of
new free air surface (see below), and the procedure is Eq. (11), will not be affected by the detailed behaviour of
repeated until there is no updating of the free air surface the flow around the spray. The vertical global motions of
or that the error estimate obtained from the conservation the bag and the effect of the vertical motions due to
of momentum is below a chosen value (e.g. 1 %). One incident waves are considered. A right-handed local xy-
important point is how the kinematic free surface coordinate system that moves with the forward speed U
condition is rewritten into an iterative scheme to update of the vessel is used. The origin is fixed at the lowest
the free air surface. To obtain a stable and accurate point of the static bag configuration. The x-axis is
solution, the kinematic free surface condition defined in positive towards the upstream direction of the
Eq. (12) is rewritten into the following iterative scheme undisturbed water flow and the y-axis is positive
upwards. The undisturbed free stream velocity U is in the
Ji., , N., , N., negative x-direction relative to this coordinate system.
(15) Since potential flow is assumed, the separation point
4>,"-C/4>„" must be determined a priori. The lowest point of the bag
is chosen. The body boundary conditions are transferred
It is used that £ =- nj/n2, ^x=nl^n+n2^s and to a straight horizontal line that corresponds to the x-axis
(t,v=n2<l)«~ni<i)t • Heren=(n,,n ). The slope of the first (y=0) defined above (see Fig. 6). The total
2
element on the free air surface is set equal to the slope of
the last element on the lip, and is not updated by the
algorithm described above. The reason is the Kutta
condition, that ensures a smooth flow at the separation
HBVP
point of the bag. Now both the velocity potential and its Simplified HBVP
normal derivative are known on the whole enclosing (3/at- u a/ax) 4> =o
surface. The air pressure is found from Bernoulli's
$=0
equation. The method has been verified against analytical $=0 <)> = 0 $=0
solutions based on conformal mapping ([9]). Since a
more simplified method is beneficial in a ride control
system, 1-D approximations have been studied. It has
L -Ut 0 7 2c(t) 0 / 2c(l)
In order to illustrate possible deformation patterns of the cross-section of the air cushion. This can be expressed
stern seal bag, the static and the first four mode shapes of as
a 2-loop flexible stern seal bag are represented in Fig. 9. dVC dVc dp
_ *C
These mode shapes are the first four used (see Eqs. (8) u'e 'H(Bc
dt dp dt
and (9)) to describe the deformation of the flexible stern
,a
seal bag structure. Here pc = -pc0my|j, e ' is the dynamic pressure in the
air cushion at x'=L. dVJdpc can be estimated by a
, = 1« (H*[)|-v^ «P.JI-, * S 19 |Hi|) —^ static analysis of the bag. The solution of the
1 Slate si
:Nv^ i
Slat* snapo
eigenvalue problem is t|r, =f|r, cosrx' where wc=vgr
and
' ' ! 3 3
|— ■ Mo
*,u,aPs;; ,.,..™,| \'ZL Maae
»
hape4(<
■"5™K tanrL+^ ^L=0 (20)
HficL dPc
:X
^i_y - ^^^
Here r is the wave number of the standing pressure
waves in the air cushion. Eq. (20) is the dispersion
Fig. 9 Static configuration and first four mode relation of the acoustic pressure waves inside the air
shapes of flexible stem seal bag. Static cushion. When dVJdpc is zero, rL=co<,L/vs=nit,
difference pressures: AP01= 500 Pa. AP02 n=l,2,* • • . This is a well known result for standing
= 5500 Pa. Bag length Lfc=3.16m. Bag pressure waves in a long tube with rigid ends. Since
height #,=1.4m.
^ b £4=0.6-106 N.
dVJdpp-0 for a flexible stern, the resonance frequency
M =4.3 kg/m.
is reduced relative to rigid ends. The actual value
depends on L,Hc,Bc, the ratio of specific heat of air
Vertical accelerations at AP, CG and FP with a flexible
ys and the steady excess pressure pc0 in the cushion.
stern seal bag are presented in Fig. 10. The air leakage
underneath the seals is neglected. If Fig. 8 is compared
Second one notes that the spectral peaks of the vertical
with Fig. 10, some differences are observed. First one
accelerations at AP and FP in Fig. 10 are reduced with
notes that the main spectral peak is shifted from
approximately 40% and 20% relative to Fig. 8. This
approximately 6 Hz down to about 5 Hz. This is due to the
reduction is explained by the flexible behaviour of the
flexible behaviour of the stern seal bag. The mechanism
stern seal bag and the coupling between the bag
may be explained as follows; the flexible bag is deformed
pressure and the air cushion pressure at the bag. The
due to the unsteady air cushion pressure so that the
differences between the vertical accelerations at AP and
deflected volume is in phase with the unsteady air cushion
FP presented in Fig. 10, may be explained by the mode
pressure. This leads to an equivalent horizontal velocity at
shape function of the pressure waves. If the previous
the bag, due to the assumption of uniform pressure over
simplified solution of i|/. is used, then the variation in
the height of the air cushion. A simplified model can
the longitudinal direction is given by cosrx'. The ratio
illustrate this, phenomenon. Assuming harmonic time
of tyhsp between AP and FP is cosrL. Since rL is less
dependence e e and neglecting damping, the following
than 7i, the absolute value of the ratio is less than 1.
eigenvalue problem can be defined
Fig. 10 shows that the flexible bag increases the
accelerations in the region just above 10Hz. This
-^ 0 increase is mainly explained by the volume change of
bsB-\s t
e^bsp ™bspjcx
the air cushion due to the deformation of the flexible
at X =Q
1W = ° ' (19)
stern seal bag.
*,bspj' at x'=L
dpc Hßc ■FP
AP
CG
Here i is the complex unit and i|/fa .e"""' is the velocity
potential of the air flow in the air cushion due to the
motion of the bag. Hc and Bc are the cushion height and
beam. pc0 is the steady mass density of air in the air 8 10 12
cushion. dVJdpc is a measure of the change in air cushion
Fig. 10 Spectra of vertical accelerations. Air
volume Vcdue to bag deformation. It can be evaluated by leakage neglected. Flexible stern seal
noting that the right hand side of the boundary condition bag. Head sea. T =1.8 s. H1/3=0.15 m.
at x'=L is a longitudinal velocity u 'averaged over the
13-8
The effect of a constant leakage height underneath the When vertical accelerations presented in Fig. 12 are
stern seal bag can be studied by comparing Fig. 10 and 11. compared with Fig. 11, one observes an increase of
The vertical accelerations in Fig. 11 are decreased with approximately 85% in the vertical accelerations around
approximately 10% relative to Fig. 10 around the first the first spatial pressure resonance frequency. An
spatial pressure resonance frequency. The reason is important reason is reduced damping due to smaller
additional damping caused by the air leakage underneath leakage height underneath the flexible stern seal bag.
the stern seal bag. The spatially varying pressure underneath the bag
forces the flexible bag to reduce the leakage height.
There is also a change in the excitation of the air
cushion pressure caused by the flexible bag motion.
'. lHzl
To evaluate the effect of the hydrodynamic impact flexible stern seal bag is "following" the water surface.
between the flexible stern seal bag and the water, an A main reason is that the spatially varying pressures
analysis is made with the same model and data as caused by the air leakage reduce the leakage. This
described in Fig. 13, but without the effect of water behaviour is a characteristic feature of a flexible stern
impact. The results are presented in Fig. 14. The seal bag.
w/o bagdeformation
with bagdeformation
'. [Hz]
Fig. 14 Spectra of vertical accelerations. Variable Fig. 15 Airgap underneath lowest point of static
air leakage. Spatially varying pressure bag configuration as function of time.
underneath flexible stern seal bag. No Effect of dynamic bag deformations. Full
water impact. Head sea. 7>1.8 Tp-\.i s. simulation model, (see also Fig. 13).
#1,3=0.1 m.
The previous analysis has in particular focused on the
largest differences between Figs. 13 and 14 occur in the dynamics of the flexible stern seal bag and its effect on
frequency region around the first spatial pressure the vertical accelerations. An important parameter from
resonance of the air cushion (=5 Hz). Spectral values for a design point of view is the height to length ratio of the
vertical accelerations are increased about 20% relative to flexible stern seal bag. Fig. 16 shows vertical
the results in Fig. 13. One possible explanation is that the accelerations
motion of the stern seal bag induced by the water impact
affects the air cushion pressure similarly as a piston at the
end of a long tube and may in this way cancel a part of the
Location: AP
pressure variation in the air cushion. A probably more
H„ /L„ = 0.44
important reason is that the hydrodynamic impact punches H„ IU, = 0.60
the bag out of the water immediately after the initial H„ lit = 0.75
impact so that an air leakage is recovered. The air leakage
introduces damping and therefore reduces the vertical
accelerations.
ratio of the flexible stern seal bag reduces the effect of the 2. Steen, S., "Cobblestone Effect on SES", Dr.ing.
spatial pressure variation underneath the bag caused by Thesis, Dept. Marine Hydrodynamics, Norw. Univ.
the air leakage. This reduction may be explained as of Science and Technology, Trondheim, Norway,
follows. The effect of the spatial pressure variation 1993.
underneath the bag is mainly important in the region where
the ratio between the leakage height and the height at the 3. Steen, S., Faltinsen, O.M., "Cobblestone
actual position is most influenced by the motion of the Oscillations of a SES with Flexible Bag aft Seal",
bag, that is the region near the lowest point of the bag. In J. Ship Res., Vol. 39, No.l, March 1995, pp. 25-
the present case where the height to length ratio is 41.
increased, this implies that the length of this region is
reduced and the effect of the spatial pressure variation as 4. S0rensen, A.J., Steen, S., Faltinsen, O.M., "SES
a result is reduced. Dynamics in the Vertical Plane", Schiffstecknik,
Bd 40,1993, pp. 71-84.
4. CONCLUSIONS
Cobblestone oscillations are theoretically studied in the 5. Faltinsen, O.M., "Sea Loads on Ships and
time domain. Focus is on the influence of flexible stern Offshore Structures", Cambridge University Press,
seal bags and how the associated aero-hydroelastic 1990.
problem can be simplified. It is shown by a case study
that: 6. Faltinsen, O.M., "Hydrodynamics of High Speed
Vehicles", Ch. 3 in Advances in Marine
The flexible stern seal bag reduces the lowest spatial Hydrodynamics (Ed. M. Ohkusu), Computational
resonance frequency of the air cushion. The elastic Mechanics Publications, Southampton, Boston,
bag deformations are important for cobblestone 1996.
oscillations.
The spatially varying pressure caused by air leakage 7. S0rensen, A.J., "Modelling and Control of SES
underneath the bag reduces the air leakage. This Dynamics in the Vertical Plane", Dr.ing. Thesis,
effect is important and increases the cobblestone Dept. of Engineering Cybernetics, Norw. Univ. of
oscillations. Science and Techn., Trondheim, Norway, 1993.
Hydrodynamic impact on the bag matters and reduces
the vertical accelerations. 8. Salvesen, N, Tuck, E.O., Faltinsen, O.M., "Ship
Hydrodynamic wave induced loads on the side hulls Motions and Sea Loads", Trans. SNAME, Vol. 78,
are not important. 1970, pp. 250-287.
Increased height to length ratio of the bag reduces the
vertical accelerations. 9. Ulstein, T., "Nonlinear Effects of a Flexible Stern
Vertical accelerations have a nonlinear dependence on Seal Bag on Cobblestone Oscillations of a SES",
significant wave height. Dr.ing. Thesis, Dept. Marine Hydrodynamics,
Norw. Univ. of Science and Techn., Trondheim,
The analysis assumes that the incident waves pass Norway, 1995.
undisturbed through the air cushion. The influence of the
change in air cushion volume and leakage area due to 10. Ulstein, T., Faltinsen, O.M., "Hydroelastic
diffraction of the incoming wave system by the presence Analysis of a Flexible Bag-Structure", 20th Symp.
of the side hulls, cushion pressure and bow seal should be on Naval Hydrodynamics, University of California,
studied. This includes interaction with the steady flow Santa Barbara, USA, 1994.
field. A quasi-steady analysis of the fans is used. [12]
reports that dynamic fan effects may significantly reduce 11. Ulstein, T., Faltinsen. O.M., "Two-Dimensional
the spatially uniform pressure response in the air cushion. Unsteady Planing", J. Ship Res., Vol. 40, No. 3,
Sept. 1996, pp. 200-210.
Since the cobblestone oscillations are difficult to study in
model scale [1], the theory needs to be validated by full 12. Sullivan, P.A., Gosselin, F., Hinchey, M.J.,
scale results. "Dynamic Response of an Air Cushion Lift Fan",
Proc. HPMV'92, Conference and Exhibits, ASME,
References Flagship Section, Arlington, VA, USA, 1992, pp.
1. Kaplan, P., Bentson, J., Davis, S., "Dynamics and ACV39-ACV47.
Hydrodynamics of Surface-Effect Ships", Trans.
SNAME, Vol. 89, 1981, pp. 211-247.
14-1
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
14-2
To achieve heel requirements, without degrading the order to decrease the course changing duration while
stabilization fins dynamic efficiency, a specific heel control carrying aircraft handling (see figure 2).
system was designed, which is the third mechanical
component of SATRAP. In the first design step, it was made of
Speed 25 knots
sea water ballasts and transfer pumps located in sponsons,
just under the flight deck, to get the maximum righting lever
arm. After an optimization stage, a better solution was chosen,
involving solid mass transfer.
- due to the larger flight deck beam, the wind heeling moment
• CdG-CVN with Cogite
is bigger on Charles de Gaulle than on former Foch CV, as m Foch-CV
shown by aerodynamical scale model testing, Jk CdG-CVN without Cogite
- also due to the larger flight deck beam and to the increase of 3 4 5 6
shipboard aircraft weight (11 tons Super Etendard versus Steady heel f)
20 tons Rafale), aircraft handling create higher heeling Figure 2
moment (x 1.8) than on former French CV.
The associated criteria are :
Heel compensation
- roll less than 3 degrees (max. value), while course keeping
Wind or course changing,
- heel (i.e. low frequency part of the roll) less than 1 degree in
straight courses.
Stabilisation fins
Figure 3
To ease control of the sub-systems, all of them acting on This SATRAP subsystem is equipped with twelve
physically bound phenomena, SATRAP is a very integrated independant 22 tons moving masses (trains), running athwart
system which fully performs three functions : ship under the flight deck, from side to side (figure 4).
- steadying horizontal ship motions Four speeds are available (0.2. 0.4, 0.6 and 0.8 m/s) : the last
one allows a complete transfer in less than one minute, the
- steering control (course changing and course keeping) lower speeds being suited for smoother transfers.
the Cogite subsystem, thanks to the following - the moving masses of the Cogite device devoted to the
characteristics : compensation of the whole low frequency motions. In fact.
Cogite response time makes it unsuited to oppose the roll
- two separate systems (two sub-groups of six trains) in two fast motions and. on the other hand, overheating and wear
separate rooms, to avoid loosing all the devices electro-mechanical constraints are not consistent with
contributing to a given function. ceaseless displacements of the masses.
- system designed to survive to the first failure, for any part of
It should be observed that a classical water-ballasting system
the system and associated ancillaries. i.e. for control units (3
still subsists on the CdG-CVN. and is available for manually
redundant processors), mechanical actuators, motor cooling, compensating the heel variations induced by different ship
electrical power ... The reconfigurations can happen even
load conditions.
during a transfer and are fully automatic. In case of damage
for one electro-mechanical element, the train is coupled to
The control scheme of figure 5.a is derived from satisfying
the adjacent one. and nominal performances are still
these constraints, and from the frequential complementarity of
available. the two actuators. The specific actions of the fins and Cogite
- high level of reliability and availability even in the fully are frequency uncoupled, because of the controllers structure
redundant state, i.e. insensitive to the first failure. (roll rate and acceleration feedback for the stabilization
system, and heel proportional-integral feedback for the
- very secure manual mode in case of several serious troubles. compensation system), and by means of appropriate filtering.
/ Figure 5.b
i
I Cogite righting
rudders ^ ship heeling moment ^ pseudo-derivation
deflection modeling (hydrodynamic displacement
ft -+- aerodynamic)
measured
wind
Figure 5.a
compensation ^/&\
controller ^££
moving-masses
lowpass speed SHIP
stabilization
controller fins deflection
bandpass
measured
roll
Figure 5
14-5
- the modeling naturally disregards the high frequency v correspond to the longitudinal and lateral components of
disturbances induced by the sea ; so, the control orders the ship speed, (p is a the roll, y the heading. GM the athwart
produced through it are "clean" ones, and do not require stability modulus, and M the ship mass : Mu. Mv, l,p and I¥
the same kind of filtering as their feedback counterparts. represent the masses and inertia of their corresponding axes,
including added masses and inertia. The hydrodynamic part
- since we get a model with a ship state vector can be developed in the following form :
(synchronously estimated) at our disposal, it is easy to take
advantage of them to predict the ship behaviour by f
simulation, over a short receding horizon (some secondes) ;
so. by using past, present and future informations, it is C,„2«|" \+C^2+Cxr2r2+C„/v
possible to implement a non-causal derivator filtering to
produce a moving speed without any additional phase lag F}" =Ip5V2 Cn,v+Cl7/+Cv5sin8+Cvi.2v|
with respect to the desired righting displacements. After
synthesis from an ideal frequency shaping, this FIR filter is < =Ip5LV: Ch,v+C,bsm8+C/pp+C^ p/V
described by the (finite) series of its impulse response
coefficients, permitting to compute the feedforward control < =lp5LV2 Cmy+C„rr+C„ssm8
action by straightforward convolution with the sequence of
estimated displacements. I
by denoting L and S the ship length and reference surface, p
Regarding the drawbacks, introducing an explicit modeling
into the control scheme could prove to be detrimental to the the water density, V the speed modulus. 5 the rudders angle,
robustness. In the present case, this is not crucial since using and the adimensional states as :
a model is mainly motivated by the search for early starting or
stopping of Cogite (more generally, anticipating any speed *=P L
;—rL
v = =
level change) . In practice, this trend information proves to be V Tn P -Wi ' V
unsensitive to the modeling errors, as confirmed by many
simulations intending to check the robustness of the method. The coefficients involved in the aerodynamic forces are
In the extreme, even a simplistic modeling could be useful to computed as a function of the relative wind parameters (speed
predict how the ship is going to heel in terms of the steering and direction), which are measured on board by the wind
angle, and therefore to determine the way the moving masses sensors. During the prediction stages with the internal
have to start for opposing this effect. However, in view of the model, these parameters are calculated by using the latest
coupling existing between the different axes and of the estimation of the real wind and the course changes of the ship
nonlinearities induced by the physical limits of the masses trajectory.
motions, it would be irrelevant to leave the model internal
state free, without taking care of its fitting and consistency Following the structure of equations (1), several ways are
with the real ship. This question will be adressed in §5, but used to make sure of the relevance of the trends issued from
implies to go further into the structure of the internal model the feedforward component (figure 6) :
first.
- The thrust Fxp is expressed in terms of u and propeller
revolutions Th through the propeller advance ratio. As far
5. Consistency of the internal model state as the internal modeling is concerned, the propeller number
The internal model is of the traditional manoeuvrability of revolutions is not measured, but is used as a control
modeling type, the different components of which have been variable for keeping the model longitudinal speed u close
identified from sea tests on scale models (for the to the ship measured one. In other words, the working of
hydrodynamic and propelling parts), and from wind tunnel the propelling model is fictitious and not linked with the
tests (for the aerodynamic part). Later on, the hydrodynamic real one ; it is only used for balancing some of the modeling
coefficients will be improved from real sea tests to be errors and for ensuring the consistency between the ship
performed soon aboard the Charles de Gaulle. The modeling and model speed variations.
structure obey the following equations : - the AM,, term squares with an estimate of the aggregate
modeling error on the roll moment. It is evently and
H A
Mji-M\r = FX +FX +F!' asynchronously readjusted for taking into account the
nonlinear effects induced by the masses displacement limit
Mvv+Mur = F?+F;
(+ 20 m). The goal is to allow the model heeling when the
=P real Cogite device is saturated (beyond the righting
/:. ..m
I9p+MgGMsmq> +AAf,„
(1) capability), and to integrate this compensation shortage
when the control is on the point of unsaturating the system.
V = /7cos<p This adjustment is calculated in terms of the moving masses
v position (dcs). and possibly of the difference between the
model and ship heels.
where the superscripts H. A, P, C represent the forces (F) or
the moments (M). respectively of hydrodynamic. aerodynamic,
propelling type and induced by Cogite displacements, u and
14-6
INTERNAL MODEL
-<g>>- lowpass |^-
<&
P.I.D. law
Th
SHIP
AMip
MODELING
if
wind
compensation
feedforward
control
zr.-i
i
autopilot rudders
or servo-controls
helmsman
TT
wind sea
T
heading
set point
Figure 6
- the AMV term represents an estimation of the modeling 6. Problems induced by quantifying
error on the yaw moment. It is used to be sure that the model the control
trajectory stays in agreement with the ship one (especially The feedforward and feedback control components, mentioned
the yaw rate). To this end, a servomechanism involves the up to now, were continuous variables, added for obtaining
model and measured yaw rate with the variable AMV as a the final order to be sent to the compensation device.
control input. Since it is imperative not to introduce high However, this order has first to be quantified (following the
frequency disturbances into the modeling, this yaw rate has aforesaid 9 speed levels - see § 3) before being used by Cogite
to be filtered first, and therefore it is a delayed model automaton. It explains the appearance of a "supervisor" unit
(delay T corresponding to the filter propagation time) in the scheme of figure 6. Its function is to quantify the
which is servocontrolled. Then, a predictive simulation control output by means of a standard threshold logic, while
avoiding over-delayed decisions of speed level changes, and
permits to recover this delay and to get synchronous
obviously ill-timed startings which coud result from a too
informations about the modeling state.
strict quantifying. For achieving this, in addition to the
instantaneous orders delivered by the feedforward and
To achieve describing the working of the feedforward
feedback modules, the supervisor uses a very short-term
component, it can be stated that the rudders angle used
during the simulation stages is associated either to the prediction (4 secondes) of these two components, permitting
to smooth the decisions to be taken. The feedforward
helmsman set point for manual steering, or to a reconstruction
component prediction is easily obtained by model
of the autopilot orders from heading set points and internal
simulation, and the feedback control prediction is issued from
variations of the model states.
a closed loop simulation of an Auto-Regressive modeling,
build and periodically adjusted with recent signal samples.
14-7
1-581 0 70 140 210 2B0 350 420 430 560 630 700 770 640 310 380 1050 1120
>
0-730
t I me
0.000 . A A.«
nniAr
v Aft i
0
v
VlrV^| » ^^J^ß\h fJVA 350 420 490 ' K. 560 630 700 7Wf 840 910 980 \f"I 1050 1120
-0.730
-1.581
\
H0-34B
70.174
0.000
0 140 210 280 350 420 490 5E0 630 700 40 910 980 1050 1120
-70.174
-140.348
1.003
CO
U
CD
*o t i me
=>c-1- i p , i i i i
Ü 70 140 310 ■ ' 281 350 420 490 560 630 700 770 B40 910 SBO 1050 1120
5
10
CO
c_>
/t i me
i
70 140 210 280 ■j.50 tlOy—«o—' 560 630 700 770 ihuj 910 9B0 / 1050 1120
CO
-r\
C_)
>
,,.,1-r1
0.3
0.0
70
l-P
140 2J10 1 260l 350
,Pn ,r
420 490 560 630 700 1 770 840 Bid 980
-J tVrrie
1050 1120
J
-0-6
3.615
4.808 —
Figure 7
14-8
On the other hand, because of the dead zone around 0 induced - (u-nmin) in m/s denotes the variations of the longitudinal
by the lowest speed levels (± 0.2 m/s), a weak heel residue speed from its initial value of 20 knots.
could remain in steady state since the static gain of the - v in m/s. the lateral component of the speed.
feedback component is limited. To cancel it. under some - psi in degrees, the heading.
circumstances, the integrator of the closed loop must have the - psip in deg/s. the yaw rate.
capability of ordering a moving masses displacement, a - rudders in degrees, the rudders angle.
posteriori converted into a speed command during the -phi in degrees, the roll of the ship (in dotted lines, an
convenient lapse of time. This fonction is devoted to the estimate of the heel being superimposed in solid lines).
supervisor too. - dCS in meters, the displacement of the 12 moving masses of
Cogite,
- vCS in m/s. the displacement speed (according to the 9
7. Validation of the approach by simulation available levels).
-fins in degrees, the deflection of the stabilization fins.
Figure 7 displays the results obtained during hard The manoeuvres carried out during this simulation can be
manoeuvres, the ship moving forward at the initial speed of read as a series of course-keeping and course-changing
20 knots, with variable wind conditions (rear wind at stages, followed by emergency manoeuvring at the end of the
time t=0) between 30 knots (at the beginning) and 20 knots sequence. The robustness issue is also addressed by means of
(at the end of the simulation). The sea disturbances large hydrodynamic modeling errors, i.e. the ship model used
correspond to a sea state 5. and fluctuate during the test as a for the main simulator is different from the internal one used
function of the ship encounter angle. The notations used for for computing the feedforward control component.
distinguishing the different plots of figure 7 are defined as :
ZX&f^^l^
Figure 8
14-9
Figure 7. where both feedforward and feedback components developments for her next aircraft carrier, take a great interest
are active, give an idea of the performances which can be in this new concept of compensation which improves the
obtained in realistic conditions, as the requests rate of the safety and operational capability of the carrier-borne aircaft.
compensation system. Outside the zones where Cogite is Likewise, an adaptation to others types of vessels, or navy
saturated (+ 20 meters in displacement), the residual heel devices, could be considered without too many difficulties.
remains below one degree, and the control proves to be
weakly sensitive to high frequency sea disturbances. References
By comparison, figure 3 depicts the roll motion when the H.Quintin
stabilization and compensation systems are not available PA CdG : rapport de sürete de l'environnement navire
(plu'NC). as the roll (phiFC) and Cogite control order when Attitudes de plate-forme. Rapport DCN ING/AN/BS/
the feedback component is active alone. The latter curves PACDG/Sürete / RPS 2/19. vol. 1. 2 et 3 - Version 2.0
allow to quantify the contribution of the feedforward Juillet 1996
component : it is easy to see that, in both cases, the masses
displacements are identical in the aggregate, but that adding E.Jenn, V.Lalumia. D.Ferrand
the feedforward action is useful for anticipating the speed Etude definitive de sürete de fonctionnement du Systeme
demands for about thirty seconds. In the end, this is Cogite. Rapport Technicatome DI/SCE/ 9505250 - 1995
translated into about 1.5° gain on the transient heel, which
permits to fulfil the required performances. S. Kummer
PA CdG - SATRAP : Girations optimisees interruptibles
More generally, a number of simulations were performed to Specification technique de besoin - Rapport DCN ING/
check the performances in all operational conditions, AN/BS N° 364/96 - Octobre 1996
stressing the transient stages of heeling. The robustness has
been tested too, by resorting to many types of modeling S. Kummer
errors, as well hydrodynamic as aerodynamic, and by SATRAP V2 : Procedures de comparaison des lois de
simulating various unknown damage conditions on the giration - Rapport DCN ING/AN/BS/PACDG/SATRAP/
actuators. However, about this, it must be stated that the PG -n° 2 - Octobre 1996
system is provided with automatic reconfiguration facilities,
in many damaged situations of the actuators (moving masses S. Kummer
locked or coupled, steering trouble) or sensors (loss of SATRAP V2 : Synthese de la comparaison des lois de
informations from wind channel), as far as they are recognized giration - Rapport DCN ING/AN/BS/PACDG/SATRAP/
by the acquisition and processing system. PGn°3-Vl.O-Fevrier 1997
8. Conclusions P.Perdon
Manoeuvrabilite du PAN - girations stabilisees
From the operational requirements of an aircraft carrier, this Rapport du Bassin d'Essais des Carenes - Etude 2341 -
paper has shown how the design stage has ended in defining Piece n° 11 - Mars 1991
and realizing an innovating and high-performance heel
compensation device, the Cogite system. On the other hand, it J.P.Jung, G.Hardier, M.A.Grandclement, P.Riot
has been shown, too. why the control laws structure was Integrated control of warship's steering and stabilizing
mostly conditioned by the actuators peculiarities and the system - 10th Ship Control Systems Symposium,
desired performances. In the present case, the use of a Ottawa- 1993
feedforward component in addition to the classical feedback
control has settled the trade-off between transient
performances and insensitiveness to sea disturbances. C.Lambert, G.Hardier
Compensation de gtte du SATRAP : girations
With regard to side advantages, disposing of an internal optimisees et interruptibles - Rapport DCSD-T
modeling allows to provide the OOD with a prediction of the 1/7996.20-Mars 1997
future heel on a receding horizon (of about one minute), thus
balancing the drawbacks induced by the discontinuous C.Lambert, G.Hardier
behaviour of the ship heeling. Definition des lois de commande et du logiciel du
Systeme Cogite du Charles de Gaulle -Rapport DCSD-T
Moreover, it must be highlighted that the working of the 2/7997.24 - Fevrier 1998
feedforward and feedback control components is completely
uncoupled. So. in highly damaged situations where an
extended loss of sensors signals prevents from safely using A.Le Pourhiet, G.Hardier
the feedforward action, the Cogite system will continue to be Porte-avions Charles de Gaulle : identification de la
piloted by the feedback structure only, the ship keeping the maquette de Castillon - Rapport DCSD-T 1/7997.42
benefits of automatic heel compensation. Juin 1998
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
15-2
When we tried to develop the real ship we came to the con- With Techniques Avancees as well as with L 'Hydroptere
clusion that the material most widely used then, i.e. naval those performances were not obtained easily. At sea experi-
aluminium alloy, was not strong enough to build the most ence, theoretical interpretation of the problems encountered,
solicited parts of the ship, namely the foils roots. modifications to the boats and, then, new sea trials, a lot of
So we came out with an hybrid design — what is sometimes such loops were necessary to understand the physics
call a foiler — that is to say a boat with small foils to assist behind the observed phenomena.
in the transversal stability but with no possibility to come
foil borne. This boat is Paul-Ricard (photo ©). It finished Several aspects of hydrofoil sailing theory have been revis-
second, five minutes behind the winner, in the 1979 double ited and it is the goal of this paper to present two of them.
Atlantic crossing race non stop. After some modifications, it The first one to be presented is the dependence of lateral
succeeded in beating by more than two days the very old stability on the general shape of the main foils.
Eastward Atlantic crossing record of Charlie Barr with the The second to be discussed is the importance of main foil
well named schooner Atlantic. twist on longitudinal stability.
Starting in 1987 two new projects were independently Before to go in those details I will quickly explain the gen-
developed in France. eral stability of an hydrofoil sailing craft.
One is an Engineers School students project "Techniques
Avancees" (photo ©) whose ambition was to beat the 3. GENERAL STABILITY OF AN HYDROFOIL SAIL-
Crossbow II world record of about 36 knots. ING BOAT.
The second, the dream of a young sailor named Alain There are two main differences between an aeroplane and a
Thebault, was to finalise the idea behind Paul-Ricard of an sailing hydrofoil:
ocean going sailing hydrofoil racer. • one is the strong lateral force due to the action of the wind
on the sails. With fast sailing boats, apparent wind is gen-
Due to circumstances, they are now closely linked together erally " close hauled " and the resultant lifting force on the
and they benefit of the same scientific and technical team. sail has a lateral component greater, and sometimes a lot
Techniques Avancees (the name is coming from the Engi- greater, than the longitudinal, useful, one. To give an idea,
neering School named " Ecole Nationale Superieure des this force is in the order of the boat displacement.
Techniques Avancees"), has been sailing regularly since • propulsive forces and drag forces are not at all in the same
1991 and it has beaten in 1997 the world speed record for plane, the distance between the forces generates important
its size class with 42.6 knots. moments (up to eighty kdn.m for L 'Hydroptere) and these
Before the drawing of its ocean going vessel, Alain moments have to be countered either by classical righting
Thebault and our team made a lot of sea trials with a scale moment of an archimedian hull or by the lifting forces differ-
model about seven metres long (photo ©) and a lot of model ential on the foils.
basin and wind tunnel tests have been also undertaken; the
fast and very long towing tank of Centre d'Etudes Aeronau- To deal with those problems, two main types of hydrofoils
tiques de Toulouse (CEAT), belonging to DGA, has been boats have been designed:
particularly appreciated. The decision to go full scale has • first generation hydrofoils are designed with a specific
been taken in 1992 after the scale model had proven its geometry and V shaped surface piercing foils so as to give
ability to sail at more than thirty knots in waves of up to the ship self stability characteristics for flight altitude, pitch
nearly one metre high. and roll; the pilot has only to control the course with the
The full scale boat, I'Hydroptere, has finally been launched rudder. These ships have more or less the same reactions as
in 1994, it sailed during all the winter season 94-95. In normal boats. Most of the motor hydrofoils (which are quite
June 95, during a race with the French Navy frigate Surcouf, numerous and are mainly used as fast passengers boats) are
the starboard arm broke in pieces with no apparent reason at entering this category.
about thirty knots. It took us a long time to understand the • second generation hydrofoils have fully submerged lifting
real reasons of the accident. Finally we redesigned several foils. Such a geometry has not a self stability characteristic
important pieces of the boat, we made it quite more stronger and, so, needs appropriate attitude and altitude sensors and
and, building on our at sea experience, we made some more the associated control loops.
modifications in order to have an even faster boat. The Most sailing hydrofoils are of the first type, but Long Shot,
rebuilt ship has been launched at the end of September 97 for instance, is of the second type, the control loops being
(photo ©). I allowed it to go progressively faster; at the fully mechanical.
beginning of December it was allowed to sail at up to thirty
six knots when Alain Thebault rushed she up to thirty nine This lecture deals only with first generation hydrofoils.
knots by a graceful day. More interesting for the architects After experimentation with a lot of different foils configura-
and engineers, it encountered the same day really harsh seas tions, there is now unanimity to consider that the best foils
and, as two dozens of strain gauges were continuously disposition is with two main forward foils, as distant later-
recorded, this allowed us to verify most of our assumptions aly as possible, supporting most of the ship weight and an
and computations. Photo OO was also taken that day, the horizontal fully submerged stabilising aft foil.
speed at the time of the shot is about thirty seven knots.
15-3
On quiet waters this aft foil controls very precisely the lon- of the aerodynamic force and the displacement passes
gitudinal trim. through point M and, as in the first theory, the leeway and
Transverse equilibrium is far more complex to deal with and the flight altitude adjust themselves so that the hydrody-
it is the subject of the next paragraph. namic resultant is exactly opposite to the sum of the two
other forces.
4. TRANSVERSE EQUILIBRIUM It can be shown that this equilibrium is stable if M is lower
than P, that is to say under the aerodynamic center of effort,
As for all mechanical equilibria one must have zero resultant and the equilibrium is unstable if M is above P.
forces and zero resultant moments.
Mere we deal with four forces, the two hydrodynamic forces All classical sailing hydrofoils have been built on that
on the foils, the transverse component of the aerodynamic theory.
force on the sail and, finally, the weight. For a practical reason, we have been obliged to adapt a ver-
Classical hydrofoils tical winglet at the tip
boats have straight ■■■< of the main foil. This
foils, lift is perpen- reason is easy to un-
dicular to the wing derstand: with very
middle axis, so the wide multihulls hy-
two hydrodynamic : drofoils, the lateral
forces converge at a aerodynamic force
fixed point. I call ! component can be
this point M ! greater than the ship
because it has some hydrodynamic resultant
j; weight; as the dihe-
resemblance with ( dral angle of the main
the metacentre of a fj foil is always quite
classical hull. | near to 45°, in that
The simplest hydro- i condition the leeward
foils theory (see I foil generates a verti-
fig. I) stipulate that J cal force greater than
the two other forces I the displacement and
are also passing | so it " climbs " out of
through this point. 8 the water, the wetted
This is obviously | useful surface de-
true for the gravity p creases, hydrodynamic
force when there is f charge by unity of
no heel, but it is | surface increases and,
more difficult to S finally, through stall
defend for the aero- I phenomenon or an
dynamic force which Fig. 1 Lateral stability f other phenomenon
Simplified theory
depends on the size, f specific to hydrofoils,
the twist, the angle »■- -■.--■•■ T~ .- -.--.-■■■••- -.--.-—.-.--..--.- • r——:--.T-^---.—-■.--r-r-TT -.'~~-.~-.-J'. ventilation, the lift
of attack , etc. of the decreases brutally,
sail. Anyway, this theory is quite simple. It says that the there is a violent side slip, the transverse equilibrium is de-
lateral component of the aerodynamic force induces an angle stroyed and there is an immediate hard landing!
of leeway, so changing the angle of incidence on the main To avoid this very annoying phenomenon we have added a
foils, increasing the down wind one and decreasing the up vertical winglet; if there is a stall on the main foil surface,
wind one, until to obtain equilibrium direction, the ampli- the side force is maintained and the equilibrium is generally
tude being automatically adjusted by the change in the automatically recovered before landing!
flight altitude. With this configuration of forces there is no When taking into account the presence of this winglet into
roll effect. the equilibrium equations, one can see that this winglet
enhances considerably the stability of the equilibrium and
A more realistic theory is as follow (fig. 2): point M is still a the range of conditions in which the equilibrium is possi-
fixed point in the median plane of the ship, aerodynamic ble; practice confirms this finding.
force and the weight converge at the point P which change This point can be explain graphically (fig. 3).
with roll angle. Roll angle adjust itself so that the resultant
15-4
hydvdynamlc resultant
Wind
Effect of orbital Hem
at the surface
Surface current
fig.4
Supposing the speed of the boat is thirty knots, this repre- efficiency of the aft foil in that situation determines its sur-
sent a variation of the angle of attack on the main foils of face, but the risk will stay, real seas being quite complex.
about
_, Z5 The first idea to reduce this risk is to work with a great nom-
a = tan :9.5°
15 inal angle of attack on the main foils, but this is, alas, non
This is not the only component in the variation of the angle practicable. For a ten degrees hydrodynamic angle of inci-
of attack, the two others are the vertical speed of the boat dence and thirty knots, the lift coefficient is about 1 and the
itself when climbing up or down the waves and, finally, its lift per square metre on the foil would theoretically be:
pitch angle. F= 1/2 p V2= 11 kdN.m-2
far too much compared with the limitations imposed by ven-
The real problem with sea going hydrofoils is to limit as tilation and cavitation.
much as possible the risk of apparition of a negative hydro- So, to adopt a nominal angle of attack more than a few
dynamic angle of attack and, in consequence, of a negative degrees would increase significantly the drag and be con-
lift on the main foils. If a main foil is in the situation of nega- trary to the objective of hydrofoils which is to reduce drag
tive lift, the flight altitude decreases, the wetted surface of as far as possible. Experience has effectively shown that
the foil increases and the negative lift increases; a hard land- performance was very sensitive to the value of the nominal
ing is unavoidable. angle of attack, the best angle being 2° or less.
As any sailor knows, problems are quite different going
windward close hauled, against the sea, or reaching in fol- The proposed solution is then two folds:
lowing seas. The difference is even greater with fast hydro- • first, to twist significantly the foil between the nominal
foils boats. "water line" and the root of the foil,
Close hauled, against the sea, negative incidence angles on • second, to have a significantly tapered foil, giving thus a
the main foils are not really anticipated. Time of reaction can greater surface to the high angle of attack parts than to the
be a problem as period of encounter of the waves can be as nominal angle of attack ones.
small as two seconds. Risks are limited to violent vertical The general aspect of such a foil will resemble the sketch of
positive and negative accelerations and even to "take-off fig. 5.
behind the crest of the wave. The bow can, may be, enter the The goal is to recover as fast as possible an important global
water when just beginning the wave ascent. positive lift when the tip of the foil encounters a negative
The risk, reaching in following seas, is to lose lift or, even, hydrodynamic angle of attack (fig. 6).This solution has of
to pass in regions of negative lift on the main foils. If the course its own penalty: a momentary sharp increase in drag,
ship goes faster than the sea — and he will normally try to but a lot less than the one corresponding to a full hard
do so— he will ascend the back of the wave. He will landing.
encounter there negative vertical orbital speed of water par-
ticles, himself having a positive vertical speed: the two The only other solution is to abandon the first generation
effect converge to produce negative incidence angles and concept and to adopt the second generation concept with
lift. The aft foil will modify the pitch to recover a positive incidence controlled foils; this is what has been done for
incidence (which needs an absolute pitch angle greater than the military patrol hydrofoils. Mechanical control solutions
the wave slope), but there can be a delay because vertical exists for sailing hydrofoils; they have been validated for
orbital speed can be less aft of the ship than forward and inner waters boats, not yet for ocean-going ones, but I don't
also because of pitch rotation inertia of the ship. The needed foresee insurmountable problems to adapt the inner waters
15-6
Bibliography photographies
photo O
photo ©
15-9
15-10
photo ©
photo ©
15-11
photo ©
photo
^1P~
Sill
11
photo ©
15-13
photo ©
15-14
photo©
Photo ©0
16-1
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
16-2
In the general case with non-zero heave, surge and pitch mo- Wake evolution equations We consider the vorticity to
tions of the foil, the velocity V and the acceleration A of a be confined in a narrow wake W downstream the hydrofoil
point P e U in 11 f are which is modeled by a zero thickness vortex layer of local
strength
V{P,t) = -U(t)ei+ V(0,t) +Cl(t) x (P-O)
flu,]
/ p{s)vdl M
-L p{s)r x ua
(10)
dt
*„.{ (w - q) ■ Vil+S ■ -Vpa
Dt J acting on the hydrofoil during its motion, the pressure p(s,t)
dT(ij,t) has to be evaluated. The tangential projection
+ «• dt
(6)
da 1 da + fa-T + g-r (11)
In these equations v, r are the unit normal and tangent vec- +
p dr
~dt 2dr
tors to T and Jj is the derivative following the motion of the
point P(£, t) of the free surface T. The first equation states of the Euler equation in a body-fixed frame of reference Hb
that P moves with velocity w. the normal component w ■ v is then considered, where the apparent forces
is fixed by the corresponding component qv — q ■ v of the
fluid below, while the tangential one, wT, can be arbitrarily
_(_^ei ^x(P-0)
chosen. Anyway, regardless the actual value of wT, the tan- =
dt
+ A(0) +
dt
gential velocity component uT of the fluid evolves according (12)
to the second equation, which follows from the tangential + n(t) x n(t) x (P -o)
projection of the Euler equation.
16-3
per unit of mass are introduced and a = ar is the total fluid through a boundary integral equation approach. In the
velocity in this frame of reference. By integrating (11) along present paper the vorticity £ reduces to the wake vorticity
the body surface the pressure 7 while the more general case with vorticity distributed over
a finite area is discussed by Landrini et al. (1998) for the
p(s,t) - p(s0,t)
•f
■I So
■ Ji — -p a I\
adt
1
2
2 S
'so
(13)
case of a circular cylinder beneath incoming waves.
The solution of the kinematic problem with boundary data
uT on T uv on W. -yw on W
+P / (fap+9)-rde
■/'
J Sn provides all the information requested to step forward in time
is finally obtained. The reference pressure p(so, t), say the the solution according to the procedure summarized in fig-
pressure at the trailing edge, is undetermined. Clearly the ure 3. In particular, both the free surface evolution equations
loads (10) are not affected by its actual value, while the and the wake evolution equations are integrated in time by a
pressure distribution on the foil at different time instants are fourth order Runge-Kutta scheme.
comparable except for a constant time dependent shift. We
discuss this point later. Integral formulation of the velocity field In the mov-
ing frame of reference 11, q —¥ us + U as |a;| —>• oo and,
The initial value problem The mathematical statement for finite times, the perturbation velocity field u is asymp-
of the problem is completed by prescribing suitable initial totically zero far from the hydrofoil. Then, under quite weak
conditions. To the purpose we assume that the hydrofoil hypotheses [1], u can be represented by the Poincare For-
starts to move from rest in an initially quiescent fluid. Then mula
the problem will consist in computing the disturbance u with
respect to the uniform stream U(t).
Dealing with incoming waves, it is assumed that the foil sud-
«(F) = Vp
L u ■ v GdSp + Vp x
I uxy GdSp
denly appears beneath a regular wave system ranging from Va X -fw GdSp
xi = — oo to x\ = oo. Again, the presence of the moving /
JW
foil will cause a perturbation u and a change in the wave (15)
height T] with respect to the unperturbed r/s. Although the where the first two terms take into account the presence of
transient stage is non-physical, this procedure avoids the the free surface and the hydrofoil and the Biot-Savart inte-
need of a numerical wave maker to generate the incoming gral represents the effect of the thin rotational wake W.
waves and simplify the damping of outgoing signals (see The direct use of (15) to solve the kinematic problem leads
Landrini et al. (1998)). to some numerical difficulties. It can be shown, in fact, that
an eigensolution related to the circulation V around the hy-
2.2 The numerical procedure drofoil exists [1] and that a singular value decomposition
General description of the algorithm In the following technique has to be adopted in order to solve the discretized
boundary integral equations [5]. A different approach to the
we consider only those cases for which the motion of the hy-
drofoil is a priori known. Let us now assume that at a given problem is here developed. By introducing a fictitious ve-
time instant to the flow field is completely known. In this locity field
case, without solving the field equations, a Cauchy problem
can be formulated for the boundary geometry and some of V(P) = - Vp / v • v GdSp - Vp x / v x v GdSp
JH JH
the fluid dynamic variables that can be prolonged up to the (16)
time tg + At. In fact inside the body %, the hybrid velocity representation
• the body motion fixes in advance the normal compo-
nent UU\H of the fluid velocity (cfr. eq. (5)). u(P) = /„.„<
Vpp • u, *■ GdSp + Vp x I UX V GdSp
Discretized integral equations For the purpose of dis- the hydrofoil. In the computations, the wake is replaced by
cussion it suffices to recall that, by introducing iV> points a set of point vortices with circulation Tk fixed at the time
on the free surface and NH (second order) panels on the hy- of shedding according to (9). Hence, the Kelvin theorem (8)
drofoil, an algebraic system of the form implies that Tk stays constant during the following motions.
Consistently, the Biot-Savart integral in (17) is replaced by
T^T U-+T Uv the sum of the velocity TkVpG(P,Pk) x fe induced by
T (19) the point vortices embedded in the flow field. For long time
evolution, to prevent irregular motions and instabilities dis-
■v- cussed by Krasny (see [12] and the literature cited therein),
A the kernel is desingularized by convolution with a cut-off
is obtained for the unknown normal velocity component and function which, in practical terms, implies a finite velocity
the source strength. In (19) the matrix coefficient A follows induced on near vortices and a zero self-induced velocity.
by assembling the self-influence and the cross-influence of An additional difficulty in modeling the wake appears when
the free surface and the hydrofoil, while the right hand side dealing with large deformations of the wake. In particular
the distance can increase or decrease too much. In these cir-
T -> T cumstances, at the expense of an approximation, vortices are
r = {uT } + splitted or merged, respectively, by enforcing the conserva-
tion of the total circulation and of the center of vorticity.
Finally, the computational effort becomes larger and larger
as time passes. A significant reduction has been obtained by
{ 7W } using multipoles expansion with fast summation techniques
W^H
(20) both for computing the self induced velocity and evaluating
takes into account the known free surface contribution due the influence of the vorticity field on the other fluid bound-
to the tangential velocity component uT, the influence of the aries.
body circulation T and the wake vorticity 7w, respectively.
Clearly, for not deforming bodies, the self-influence matrix 3. Discussion
[H -»• H] can be computed once and for all, as well as the After the transient stage, the flow field can attain a quasi-
self induced velocities. The solution of (19) is rapidly eval- steady regime: the wave profile is constant in a body-fixed
uated by means of an iterative solver while for the influence frame of reference and the hydrodynamic loads reach con-
of wake vortices a fast summation algorithm based on mul- stant values. In these circumstances it is easier to compare
tipole expansion is employed. An enhanced solver for (19) the numerical results with the experimental data available in
based on fast summation of multipoles expansion coupled literature. For clarity, we firstly discuss the flow around hy-
to an iterative solver was developed by Graziani & Landrini drofoil at zero geometrical incidence and then the case with
(1998). A deeper discussion of the numerical method can non zero angle of attack.
be found there and in Landrini et al. (1998) for the specific
problem at hand. Hydrofoil at zero incidence In [17], [18] Salvesen gave
the very first attempt to analyze the nonlinearities in wavy
Numerical treatment of the free surface A truncated flows generated by a submerged foil. He performed a wide
computational domain has to be used and, in general, both set of experiments, [17], by using a mathematically gener-
the velocity components uT and uv and the wave height r\ ated profile. Later, [18], he developed a second order theory
do not vanish at the edges, at least for long enough evo- in which i) the thickness and the chord are comparable and
lution time. Therefore, unphysical reflections and numeri- ii) the free surface is far enough to assume the disturbance
cal errors due to unbalanced edge singularities may occur. on T one order lower in the perturbation expansion scheme.
This drawback is avoided by artificial damping layers which Here we reproduce some of the Salvesen experimental con-
force the tangential velocity component and the wave dis- ditions. We start from figure 4 where, for the larger immer-
turbances to vanish when approaching the outer portions of sion d/c = 1.37 shown on the left-plots, the wave forms
T. In particular, modified free surface evolution equations are reported for increasing forward speed (from top down
can be written by introducing a damping function p(x), zero Fn = U/y/gc = 0.422,0.590,0.759,0.928, respectively).
everywhere except within the damping regions, which mul- The foil is placed at zero incidence with the trailing edge at
tiplies uT and 77 in the dynamic and kinematic free surface x/c = 0, y/c = -1.37. The non dimensional horizontal co-
equations, respectively. The length of the damping layers ordinate x/c is scaled by Fn2 so that, at the first order, the
and the maximum damping coefficient have been heuristi- same number of crests are plotted for each case. Although
cally determined. The accuracy of the computation is even- the physical dimensions of the adopted computational do-
tually maintained under control by comparing different so- mains are different, the discrete parameters for each case are
lutions for increasing domain lengths. refined enough to have results of comparable accuracy. Fi-
Finally, in order to handle the mean horizontal motion U(t) nally, if not stated otherwise, a smooth acceleration up to
in the frame of reference H of the free surface points, a fixed the final forward speed is used to reduce the oscillatory be-
computational window is considered and those points drop- haviour characteristic of the transient stage.
ping out downstream are thrown away and new markers are Regular wave systems are obtained for all cases and, at first
inserted upstream. sight, the measured wave height denoted by symbols com-
The sawtooth instability, that usually sets in for long enough pares well with the numerical profile (solid lines). Increas-
evolutions, are removed by high order filtering procedures. ing the forward velocity, the troughs are eventually under-
estimated. A wave profile shift is detected for the case
Numerical treatment of the wake The wake is a vor- Fn = 0.759.
tex sheet continuously emanated from the trailing edge of The results reported in the right-plots of figure 4 for the
16-5
smaller immersion d/c = 1.14 allow a clearer analysis of portant for the formation of the free surface as around the
nonlinearities. Because of the smaller depth of submergence, free surface itself where numerical diffusion could quickly
amplitudes are larger with respect to the previous cases but damp the generated waves. To rule out the uncertainty due to
the overall agreement is still satisfactory. Actually greater the graphic manipulation, the Lagrangian free surface mark-
discrepancies in hollows appear as the Froude number in- ers are simply connected with straight segments. For all
creases. The wavelength is rather well predicted also in the the cases shown, quite reasonable wave patterns can be ob-
worst case Fn = 0.928. Exactly for this speed (third right- served. For the deepest submergence (upper plot) only the
plot in fig. 4) we compare our results with Salvesen sec- present solution is contrasted with the Duncan data: the first
ond order theory (dashed line). The improvement is appar- trough just above the profile is well predicted as well as the
ent both in capturing the correct wave height and predicting two following peaks.
the wavelength. Salvesen himself, in fact, developed a (non- For d/c = 1.1626 some differences appear. In particular, the
consistent) third order correction (dotted line) which nicely trough induced above the suction side is under-predicted by
reproduces the experimental data and agrees better with the all methods. Steady BEM, o, and RANSE, •, solutions give
present numerical results. roughly the same values while a deeper depression is pre-
In spite of the good prediction of wave forms, the computed dicted by the present unsteady solver which is placed half-
loads are no more than acceptable. The wave drag deduced way from the actual data. This feature is repeated for hol-
by the Salvesen formula D = \pgH2 is no more than a lows more downstream. The two inviscid rotational models
global measure of the wave pattern because, as Salvesen dis- capture the first and second experimental wave humps well,
cussed, the measured drag is significantly affected by vis- while the third crest is rather underestimated. Also a sort
cous effects which varies with the immersion for the geo- of 'shift' of the wave forms can be detected and it may re-
metric scale here adopted. In figure 5, the wave drag de- quire an analysis of the discrete dispersion relation. As the
duced by the measured wave height H is compared with the RANSE code prediction is concerned, wave height is signif-
numerical results obtained by direct pressure integration (de- icantly under predicted although the wavelength is correctly
noted by •) and using the Salvesen formula. In some cases, captured. The reason for this disagreement has to be investi-
Salvesen used the first wave peak which is apparently af- gated, but recent results in [20] indicate that the assumption
fected by local effects due to the hydrofoil. Therefore we of turbulent flow over the whole foil may be one contributing
used both the first peak, A, and the wave amplitude far from factor.
the hydrofoil, y, to determine the wave-drag. For both the This behaviour is, in some sense, magnified for the shallow-
immersions but more clearly for the lowest one, we can ob- est immersion (bottom plot in fig. 7). In particular the free
serve an increasing under prediction as the Froude number surface deformation above the hydrofoil is by far more pro-
increases which is the 'signature' of the increasing difficul- nounced in the experiments. Following the Duncan sugges-
ties in capturing the measured wave-height. For the smaller tion, in the first computations performed the bottom of the
d/c (bottom plot in figure 5) we also report the theoretical basin was not taken into account because of the deep wa-
second order result (dashed line), in good agreement for all ter regime of the generated waves. On the other hand, the
cases but for Fn = 0.759. 'ground-effect' on the hydrofoil should increase the actual
body circulation which might explain the reduced depression
predicted. Anyway, although RANSE computation takes the
Hydrofoil with an angle of attack In this case the foil is
bottom effect, the same disagreement is observed. For the
placed beneath the free surface with a geometrical incidence sake of completeness, we eventually included the bottom ef-
a different from zero and, regardless the law U(t) adopted
fects by the method of images and the steady regime results
to reach the steady-regime forward speed, a more vigorous
obtained are practically identical.
vortex shedding fTE is expected in the initial stage. Con-
The last point to mention is the experimental uncertainty.
sistently, a circulation V around the body grows and together
No error bar is reported in Duncan paper and he generally
with the lift force component (see figure 6).
claimed 'distances were measured to an accuracy of about
Eventually steady state is reached and the predicted wave ±0.3 cm', which can explain the differences between the ex-
pattern is here compared with data measured by Duncan
periments and the reported numerical results only partly.
(1983). A NACA0012 profile with chord c at an angle of
attack a = 5° is considered. In this case, the immersion d/c
is measured from the mid-chord profile to the undisturbed Pressure distribution on the foil We like to discuss in
water level and results for the three values d/c = 1.2857, more detail the pressure computation. Once again, we can
1.1626 and 1.0345 are reported in figure 7, from top down use experimental data to validate the method. In [16], the
respectively. flow around a NACA 0012 with a = 5° placed at a depth
More precisely, the present results (solid lines) are compared d/c = 0.8 has been considered and for the case Fn = 0.35
both with the measured wave profile, o, and the numerical the authors reported the measured pressure coefficients on
predictions obtained by other numerical methods. Namely, the foil surface. Consistently with the Duncan observations,
a steady solver [19] based on a boundary integral formu- in spite of the small d/c, the reduced speed allows for ob-
lation for potential flow with lifting effects, o, and a finite taining a non-breaking solution which is reported in figure
volume code [15] to solve the Reynolds averaged Navier- 8 for tU/c = 30. As discussed, the pressure is obtained by
Stokes equations, •. the equation (13) and to compare with experiments the in-
For all the computational methods considered, tests are per- stantaneous reference pressurepo(s, t) has to be determined.
formed to check the invariance with respect to further refine- Fortunately, the contribution —pj adl eventually disap-
ments of the discrete solution. In the RANSE code, the part pears in steady regime and the reference pressure is simply
of the geometry which embeds the free surface is systemat- — |P(OQ — U2). Once this is taken into account, the pressure
ically refined using three different grids, the numerical grid coefficient Cp(s, t) = (p(s, t) — Pa)/\pU2 (dashed line) on
around the hydrofoil is kept unchanged. The numerical tests the foil for tU/c = 30 is compared with the measured val-
show that the grid refinement around the profile is not so im- ues (symbols) in figure 9. Although the method is relatively
16-6
crude, the overall agreement is quite satisfactory and it ap- highly nonlinear free surface flows.
pears suitable to discuss, at least from a qualitative point of In spite of the simplicity of the model, the quantitative agree-
view, the pressure behaviour in more complex flow condi- ment appears good enough to consider the present model a
tions. valuable theoretical tool for analyzing more general condi-
tions. In fact, arbitrary body motions and incoming waves
are, in principle included in the present model, are actually
Formation of breaking waves As the immersion de-
discussed in [14].
creases the amplitude of the generated wave system in- Further extensions are presently under analysis. If massive
creases, and for a small enough value of d/c the appearance separation occurs, the method can be generalized following
of breaking waves is expected. Duncan found [8] that a non Landrini et al. (1998) in which an operator splitting tech-
breaking wave profile can be still obtained for a NACA 0012 nique to solve the Navier-Stokes equations is coupled to the
placed at d/c = 0.9507 but, by perturbing the free surface, unsteady nonlinear free surface model. Also, as discussed
he observed the formation of a spilling breaker at the first in Bulgarelli et al. (1997), the model can be modified to
crest past the foil. Here, we consider the sudden start of deal with the flow generated by an aerofoil moving above the
the foil to obtain the results of figure 10. Symbols refer to air-water interface to analyze the unsteady flow of WISES-
the non-breaking experimental wave profile and nicely agree
concept vehicles.
with the present solution. In the enlarged view the emission
of a tiny jet at the first crest is clearly detected and, even-
Acknowledgments
tually, the simulation stops. We are not able to prolonge
The research was supported by Italian Ministem del
further in time the present computations: the coupling with
Trasporto e della Navigazione through INSEAN Research
the model for steady breakers proposed by Cointe & Tulin
Program 1997-99.
(1994) may be is the possible way to overcome this diffi-
culty.
References
Finally, we like to show some refined computations of the
growth of a larger plunging breaker past a NACA 0012 [1] P. BASSANINI, CM. CASCIOLA, M.R. LANCIA, R.
PIVA, 1991. A boundary integral formulation for ki-
slightly immersed, d/c = 0.25, with an angle of attack
netic field in aerodynamics. Part I: Mathematical anal-
a = 10°. The left plots of figure 11 show the wave mo-
tion due to the sudden start of the foil, as well as the starting ysis. Part II: Applications to unsteady 2D flows. Eur. J.
vortex. Time passes from top down and it is clearly seen Mech. B/Fluids. 10, 6 and 11, 1.
that, in spite of the large breaker just behind the foil, the free [2] BERTRAM, V., MUZAFERIJA, S.; PERIC, THIART,
surface just above it becomes soon stationary. The pressure G., 1998. A comparison of free surface viscous and in-
on H at the same time is plotted on the right. It is interesting viscid methods to compute the flow about a submerged
to observe that at the very beginning of the phenomenon the hydrofoil. SACAM '98, Capetown.
Cp distribution still resembles that one typical of a foil with
positive incidence and, actually, during the first stage the lift [3] U. B. BULGARELLI, M. GRECO, M. LANDRINI,
(not reported) sharply increases. As time passes the free sur- AND C. LUGNI, 1997. A simple model for the aero-
face becomes highly deformed and the pressure on the up- hydrodynamics ofEkranoplans. AGARD Workshop on
per side decreases. Unlikely, the pressure on the lower side "High Speed body motion in water", Kiev.
seems quite unaffected and, eventually, the lift after an initial
[4] C. M. CASCIOLA AND M. LANDRINI, 1991. Invis-
peak decreases. The simulation stops a bit after and we are
cid flows about a slightly submerged oscillating hy-
not able to give a steady regime value of the lift. Anyway,
it is well known from experimental data that the reducing of drofoil. First European Fluid Mechanics Conference,
d/c lowers the lift, that can become negative even for posi- Cambridge.
tive angles of attack. The late details of the plunging breaker [5] CM. CASCIOLA AND R. PIVA, 1990. A boundary in-
are shown in a natural scale in the figure 12. tegral approach in primitive variables for free surface
flows. Proc. 18.th Symp. on Naval Hydrod., Ann Arbor.
4. Conclusions and perspectives
An unsteady model for the inviscid flow generated by a hy- [6] R. COINTE & M.P TULIN (1994). A theory of steady
drofoil moving beneath a free surface has been presented. breakers. J. Fluid Mech., 276.
Rotational effects are modeled by a thin wake embedded in [7] R.M. COLEMAN, 1986. Nonlinear calculation of
a otherwise irrotational flow and shed from the trailing edge. breaking and non-breaking waves behind a two-
An efficient and accurate numerical procedure is developed. dimensional hydrofoil 16ift Symp. on Naval Hydrody-
Several examples to discuss the validity of the physical namics, Berkeley.
model and the performances of the developed algorithm have
been presented. Specifically, we have considered those flows [8] J. DUNCAN, 1983. The breaking and non-braking
characterized by a non-breaking steady regime. Resulting wave resistance of a two dimensional hydrofoil. J. Fluid
wave forms are compared with experimental results given Mech. 126.
by Salvesen [17] for a foil with zero geometrical incidence
and by Duncan [8] for a NACA 0012 at an angle of attack. [9] L. K. FORBES, 1985 A numerical method for non-
The agreement in all cases is satisfactory. linear flow about a submerged hydrofoil. J. Eng. Math.
For the same cases, drag predictions computed by pressure 19
integration fit well the experimental wave resistance. The [10] K.O. FRIEDRICHS, 1966. Special Topics in Fluid Me-
pressure acting on the foil has been compared with that one chanics New York, Gordon and Breach.
measured in [16].
Finally, refined computations in breaking cases are presented [11] J.P. GIESING & A.M.O. SMITH, 1967. Potential flow
to show the ability of the numerical technique in handling about two-dimensional hydrofoils. J. Fluid Mech. 28.
16-7
[12] R. KRASNY, 1991. Vortex sheet computations: roll- [17] N. SALVESEN, 1966. Second order wave theory for
up, wakes, separation. Lectures in Applied Mathemat- submerged two-dimensional bodies in Proc. of 6
ics, Vol. 28. Symp. on Naval Hydrodynamics, Washington.
[13] M. LANDRINI, M. RANUCCI, C. M. CASCIOLA AND
G. GRAZIANI, 1998. Viscous effects in -wave-body in- [18] N. SALVESEN, 1969. On second order wave theory for
teraction . Int. J. Off. and Polar Eng.. submerged two-dimensional bodies J. Fluid Mech. 38.
[14] M. LANDRINI, C. LUGNI, M. GRECO, 1998. Un- [19] THIART, G.; BERTRAM, V. (1998), Staggered-grid
steady flow around an oscillating hydrofoil beneath panel method for hydrofoils with fully nonlinear free-
waves. Submitted. surface effect, submitted to International Shipbuilding
[15] MuzAFERUA, S.; PERIC (1997), Interface-tracking Progress.
and interface-capturing methods for viscous free-
surface flows, 5th Symp. on Nonlinear and Free- [20] TZABIRAS, G.D. (1997), A numerical investigation of
Surface Flows, Hiroshima 2d, steady free surface flows, Int. J. Num. Methods in
Fluids 25, pp. 567-598
[16] K. H. MORI AND M. S. SHIN, (1988). Sub-breaking
wave: its characteristics, appearing conditions and
numerical simulation. Proc. of \lih Symp. on Naval [21] G.B. WHITHAM, 1974. Linear and nonlinear waves,
Hydrodynamics. 1974. John Wiley & Sons.
.01
F, = 0.422
7}/c
\
-.01 T
UlllMll , , , , 1 , , , , 1 , , , , 1 , , , , 1 , , , ,'
.04 r F. = 0.590
:
r,/c ; >^
-.04 - ^, V,,, 1,,,, 1,,,, 1,,, V,,1,
.
f\» A A A Ax
.1 Fn = 0.759
r
i)/c
-.1 L \ J \J \J \J \y
MMLM.IMM
_
.1 F, = 0.928
r
77/c
j- V , 1 , ... 1 .... 1
..ili.i.l
x/cF, 12 16 20 24 28
x/cFn2
Figure 4: Hydrofoil at zero incidence. The Froude number Fn = U/^/gc increases from top down. Present numerical
results (solid lines) are compared with experimental data, •, from [17]. Left: The trailing edge is located at (x =
0.,j/ = -d/c = -1.37). Right: The trailing edge of the foil is located at (a; = 0.,y = -d/c = -1.14). For this
depth of submergence and Fn - 0.759, second and third order results (dashed and dotted lines respectively in the third
right-plot) are taken from [18].
cD
.014 d/c=1.37
.012 % Pressure integral
Measured data
Measured data
1 1 1 1 I l 1 l
l Fn
Figure5: Hydrofoil at zero incidence. Drag coefficient CD = D/\pU2cm steady regime for the same cases of figure
(top plot: d/c = 1.37, bottom plot: d/c = 1.14). Wave resistance from measured wave height is taken from [17]. The
numerical CD is obtained both by direct pressure integration, •, and by applying the Salvesen formula to predicted
wave height A, V- Second order theory (dashed line in the upper plot) is taken from [18].
16-9
6 ^_ 1037TE/UC /^ cL
5
4
3
2 r/uc
1
Z—\—I^-M^F^H i i i i I i i IT I I —+—1 1 1 1—J \ 1 1 1 1 1 1
10 20 30 40 50
t U/c
Figure 6: Histories of the lift coefficient CL, body circulation Y and vortex shedding ^TE for a NACA0012 starting
from rest (a = 5°, d/c = 1.28, Fn = U/^gc = 0.5669).
d/c= 1.2857
i i i i I i i i i I i i i i I i i i i I i i i i t i i i i I i i i i I i i i i I i i i i I i i i i
i i i i I i i i i rf i i i I i i i i 1 i i i i I i i i i I i i i i I i i i I I i i i i 1 i i i i
■1 x/c l
Figure 7: Hydrofoil at incidence. Wave height r)/c past a NACA 0012 (Fn = U/y/gc = 0.567, a = 5°). Sub-
mergence d/c measured from the mid-chord point decreases from top down. Solid lines: present unsteady results, o
measurements, • viscous results from [2], o steady inviscid results from [2].
16-10
Figure 8: Wave profile past a NACA0012 foil (a - 5°, d/c - 0.8, Fn = 0.35). Discrete parameters: N/X = 10,
Nb = 120, At = ±c/U
1
.5
-a lower side
J L i i i i J I I L J L J I I L
Figure 9: Comparison of measured and predicted pressure coefficient. Symbols: measured data from [16]. Dashed line
present numerical results. Flow conditions: NACA0012 profile, a = 5°. d/c = 0.8, Fn = 0.35. Discrete parameters:
N/X = 10, Nb = 120, At = ±c/U
.2
d/c = 0.9507
.16 E-
.12
.08
.04 — •
h
f \m
\\»•
\«\ ■ /•
\■
7]/c \\•• s
•/■7
■7 \»
-.04 \»
•/%/ \«
\•
-.08 •/•/
-.12
- i i i i 1 , , , , 1 1 1 1 1 I , , i 1 i i i i 1 1 1 1 11111111111111
-1
</c
Figure 10: Formation of a spilling breaker past a NACA 0012, a = 5°, Fn = 0.5669, d/c = 0.9507, tU/c = 11.5279
after the sudden start. Experimental data, •, from [8]. Top left: detail in natural scale of the spilling crest.
16-11
7)/c
7)/c
.... i
i -
v/c c. ^ _«^ —X
I . ■ .
x/c 1.2 1.6 2 2.4 -.8 -.6 x/c
Figure 11: Evolution toward breaking of the first crest past a NACA 0012, a = 10°, Fn = 0.5669, d/c = 0.25.
Left: wave profile from tU/c = 0.1133 up to tU/c = 1.7007 time step AtU/c = 0.1133 Right: unsteady pressure
coefficient at the same time instants.
Figure 12: Details of the plunging breaker past a NACA 0012, a = 10°, Fn = 0.5669, d/c = 0.25, tU/c = 1.7007,
1.7544,1.7856,1.8125.
17-1
V.T.Savchenko
Inst. of Math. Mach. & Systems Problems
UNAS, 42, Glushkova Prospect
252187 Kiev, Ukraine
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
17-2
flat solid wall in (he hydrodynamic tunnel (HDT) Airflow rate Q for gas supply into the near wall
(Fig.l). layer was Q, = 1.2, Q2 = 0.6, Q3 =0 (liter/sec). This
corresponded to the three flow regimes:
The HDT has a 2000 ram long working part with l.The regime with the air layer between the foil and
the open free surface and rectangular 340x330 mm tire body. The cavity is closed behind the foil.
cross-section. The contour of the model was chosen 2. Gil forms between the foil and the body. The
that its main parts correspond to the profiles of cavity is closed in front of the foil.
vehicles being usually used (Fig.2). The head part 3. The regime of continuous flow about the foil.
of it was shaped as a parabola while the main part
was as a extended cylinder. The total length of a The coefficient of airflow rate was calculated
model was 387 mm, the radius of the cylinder part according to the following formula CQ = Q / nR"V.
was 30mm. The foil was inserted into a special Air concentration in Gil / was determined
clamp 30 mm away from the model's stern. It could depending on the airflow rate Q and a thickness of
be rotated and fixed at the required angles to the the laver Z
flow.
/ = "/ =T^
,N2 U)
There were two ring slits at the front of the cylinder
,+ ■1
part of the model. The first one was intended to
separate the fluid stream directly in front of it. The
V I
Parameters for regime 1 were: Zi = 3+4 mm. a
gas for forming either a developed cavitation flow
around the model or a two-phase near wall layer of = 0.015+0.02.
Parameters for regime 2 were: Z2 = 5mm, / =
liquid is supplied into the stream through the second
0.095, a = 0.04+0.045.
ring slit. The 5 orifices with diameter 2 mm were
made in the slit particularly for this purpose. There
The components of hydrodynamic forces Px and Py,
was an orifice to measure the pressure in the cavity
acting on the foil, were measured during the
Pk in the middle part of the model.
experiment. Px force acts along the up-running
stream ( a drag force), while Py acts across a plane
3. BASIC PARAMETERS
of a body's symmetry ( a lift force of the foil).
The experiments were carried out with a series of
The tests were carried out in order to estimate die
rectangular and triangular in plane foils with the
value and the type of changes of hydrodynamic
symmetric wedge profiles. The 12 types of rudders
forces acting on the rudders which operate near the
have been tested as a whole. Aspect ratio of this
body of revolution in each of the three flow regimes:
series was being altered within the limits of 0.9 < A
continuous flow, Gil between the hull and the
= S/c2 < 4 ( S is the area of a foil, c is the foil's
rudder, and the air layer between the rudder and the
chord). There was a series of rudders with the equal
hull.
lateral surface. It was convenient for evaluating
their hydrodynamic characteristics.
4. TEST RESULTS
The angle a between the plane of foil's symmetry The results of the tests are represented as a
and the plane of the semi-model's symmetry was dependence of coefficients of hydrodynamic forces
being altered within the limits of I a I =15° with the Cx and Cy on the angles of attack at flow regimes
step |Aal =3°. The tests were carried out at the described above. The typical regimes of flow about
velocity of free stream V = const = 9m/s. For the the model are displayed in figure 3 (photo a,b,c,d.e).
described model this corresponded to the non-
dimensional parameters as follows (Ref. 12): If the experiment is carried out in conditions
Reynolds number Re = VL/v = 3.2-10s; v is the described above, stable dependencies Cx(ot) and
kinematic coefficient of viscosity, Cy(a) take place in presence of a developed cavity
Froude number Fr = VA'gL = 4.6; g is the (regime 1) or in presence of the Gil (regime 2). To
acceleration due to gravity, show it better, figures 4, 5 display the measurement
Weber number We = cpV2/$=1.5-103; % is the results at deviations of rudders once in positive and
coefficient of surface tensions, p is the mass density, once in negative directions.
R is the radius of the cylinder part of the model.
The cavitation number depending on the flow The characteristics of foils Cx(rx) and Cy(oc) in the
regime was changed within the limits of 0.015 < c flow regime 1 and in continuous flow regime 3 arc
= 2(P-Pc)/pV2< 0.045. shown in figures 6, 7. The jumps discontinuity (the
1st type) appear in the third regime. They usually
17-3
shown in figures 6, 7. The jumps discontinuity (the estimation of the free surface influence (Ref. 14)
1st type) appear in the third regime. They usually shows:
occur when air is bursting from the main cavity
towards the foil. This situation are determined for
^yk ^yFr=
each case by conditions of the experiment. 0.1 (2)
^yk
Fig. 8 displays polars of the same foil for flow
regime 3 once with and once without flow The lift force value reduces almost 10% [14].
separation. Formation of a cavity behind a foil In the experiment we have observed the value of lift
reduces the hydrodynamic forces almost twice. We force reduction almost twice if an air layer formed
found even smaller forces in the first flow regime there between the hull and the foil. This
(fig. 9). Such decrease of the hydrodynamic forces, phenomenon can be explained only by taking into
which takes places when the continuous flow account all losses: the interference lose, the free
regime changes into the developed cavitation surface influence, and the partial loss of the area of
regime, is due to the loss of interference. the foil AS = c-Z. To unload the weight of the aft
As seen from the experiments the Gil with air part of the cavitating object we have suggested a
concentration f < 0.1 (Z=5mm) weakly influences special shape of the foil in plan. This gave a
on the Cx(oc) and Cy(a) dependencies and does not possibility to reduce the loss of hydrodynamic
reduce the interference between a foil and a body. quality down to the value of 0.12. The loss of
quality for rectangular foils was of the order of 0.4.
Presence of rudders sometimes influences on the
character of flow around a model. It is seen in CONCLUSIONS
photo (Fig. 3) that in practically all the cases of a
local cavity is formed nearby a rudder. It does not 1. Conducting the investigations on flow about the
disappear when a rudder is in its neutral position a series of protruding elements (of foils, rudders)
= 0 and it increases if a increases. Dimensions of nearby the bodies of revolution has given us an
such cavity are comparable in regimes 1 and 2 opportunity to find the significant, more than twice,
when the rudder has a angle of attack a > acr (acr = reduction of hydrodynamic forces on rudders if an
6° for this considered case). When this angle a< air layer between a hull of a body and a rudder was
otcr the length of a local cavity in the second present. This phenomenon is explained by taking
regime is in 10 times smaller than in the first into account the three factors: the complete loss of
regime. We should note also that a rudder creates a foil-hull interference, the partial loss of the effective
support to a cavity on a body and when a > aor its area of a foil and influence of the free surface of the
dimension decreases significantly. air layer (cavity).
V//////////////////////////////////////////////////////////////////////////M
fI
2
A —r 02 015
: 1 I ! o'
in
'X
\^ y=2x2/3
< ! ii
i
5 30
190
307
-« ►■
*c0^ Q
x
ü
:>;
op
:il
4= -a
lr C
£ (0
ü -r-
|8
2E
ES>
>• o
CD
x
O
>;
|z
a> gj
O T3
4) 2
O ^
OfM
o ro
c.E
■ge
d>
17-8
Ü«
to *
1\
£03 -i
P C
E «
(0 0)
"O ■
:>.:
c <
O i
<n i
a.
17-9
cy 'H50
V
0.2 -
%y_
0.1 - y4/
0
/o° k0.05 0.1
i i
0.15
i
0.2 a,
SUMMARY
H b- wave height of a wavy screen,
Some results of experimental studies c - wave velocity,
conducted at the Institute of Hydromechanics a - apparent frequency of wing's contact with a
of Ukrainian National Academy of Sciences wave,
applicable to the problem of ekranoplane ?77" . h
creation are presented in this paper. The aim of P= Strouhal number,
these research works was to determine the K
influence of proximity of the flat and wavy y - deviation angle of waves crests.
screen upon aerodynamics of the
ekranoplane's wing. Experimental installation
and procedure of the experiment are described. 1.INTRODUCTION.
Instantaneous values of lift, drag and position
of centre of pressure for a wing moving above
the flat and wavy screen have been defined. It In many countries of the world the attempts
has been determined the influence of the angle were taken during last decades to develop new
of attack, the aspect ratio, the distance to a flat means of transportation on water
and wavy screen as well as the wave length ekranoplanes. Ekranoplanes, as it is known,
and wave height of a wavy screen on the wing are vehicles intended for the flight in the
lift. The measurement results of forces and immediate proximity to a water surface. They
moments on the wing moving above the are based on using so- called "screen effect".
oblique waves are given. The essence of this effect lies in the sharp
increase of a wing lift and its lift/drag ratio on
flying at the low altitudes above the sea
LIST OF SYMBOLS surface. Lately, numerous experimental
models of ekranoplanes have been created in
different countries. However, only a few of
v o - wing velocity, them produced encouraging results. The most
H- wing span, successful achievements in this field were
b -wing chord, gained in the former Soviet Union. Substantial
scientific and manufacturing personnel were
S- wing area, enlisted for the development of ekranoplanes.
X = C/b- aspect ratio of a wing, As a result of a common efforts, under the
9° - angle of attack of a wing, guidance of the prominent designer
h - non-dimensional distance from a rear edge Mr.P.E.AIexeev a number of especially large
of wing to a screen, ekranoplanes with a great takeoff weight and a
H- non-dimensional distance from a rear edge high flying speed had been designed,
of hydrofoil to a free surface of a liquid, produced and field tested [1].
Cy, Cx, Cmz, Cmx, Cmy - coefficients of lift, drag, However, it should be admitted that,
pitching moment, rolling moment and yawing regardless of some successes, the creation of
moment, ekranoplanes is still the subject for
K. = Cy/Cx - coefficient of lift/drag ratio, investigations for the most countries of the
world. Many scientific and technical papers and
xc - centre of pressure coefficient, articles have been dedicated to the creation of
Xb - wave length of a wavy screen, ekranoplanes.
Paper presented at the RTO AWT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
18-2
As an example, we would like to make water the accuracy of the results obtained was
reference to survey works on this problem [2, less accurate. This was caused by the following
3, 4]. Nevertheless, it still remains many factors: high speeds of towing of comparatively
indistinct and insufficiently studied problems massive models; small absolute values of
in this field. One of the problems is the effect of aerodynamic forces to be measured; relatively
a wavy sea surface on the ekranoplane. low natural frequency of the measuring
In this paper, some experimental results on installation. In order to overcome these
this problem obtained at the Institute of difficulties and to increase the accuracy of the
Hydromechanics of Ukrainian National measurements, the hydrodynamic method for
Academy of Sciences are given. In the Institute the research into aerodynamics' problems was
a special experimental installation was created applied. This method was substantiated and
to research into aerohydrodynamics of tested in TsAGI [11, 5]. It amounts to the fact
ekranoplanes in late 70-es. Great volume of that a wing in the experiment was moving in a
works pertaining to the research into the water medium instead of air. In so doing, a
movement of ekranoplanes" models and their liquid flat or wavy screen was replaced with a
wing systems in the near to a flat and wavy solid flat or wavy screen. Changing from the air
screen had been carried out in this installation. medium to the water medium is quite possible
On fulfilling these studies, only one as, under conditions of the given experiment,
experimental work connected with the the compressibility effects can be disregarded.
movement of wing above a wavy screen was Such replacement is expedient as it allowes: to
known to us [5]. obtain relatively high forces to be subjected to
In this paper, however, it was considered a measurement; to attain fairly big Reynolds
one specific case of a wing's motion above the numbers (up to 2 • 106) at relatively low speeds
waves of only one certain length and height. of towing (up to 5-6 m/sec). In this particular
Our studies were conducted in compliance with case, the replacement of a liquid screen with
a larger-scale programme. This allowed to the solid one is possible because the tests of
produce data on the influence of a sufficiently wings were carried on at high Froude numbers
wide spectrum of characteristics of the wavy Jz = 4. As this takes place, the disturbances of
screen upon hydrodynamic characteristics of flow, caused by a wing movement, reach the
the wing moving above it. The results of our screen surface at a sufficiently great distance
studies were partially published in the articles beyond the wing and do not affect the results of
[6,7,8,9,10]. Tests results of separate wings on the experiment. The error out of such
their movement above a flat and wavy screen replacement does not exceed 1% [12]. It should
are presented below. also be pointed out that on replacing of a real
moving wavy surface with a stationary wavy
2. EXPERIMENTAL INSTALLATION AND screen, the speeds of air, caused by the
PROCEDURE. movement of a water surface, are not taken into
consideration in the experiment. These speeds
can be determined from the following formula
Experimental installation represents a [13]
hydrodynamic channel with the following
-2-7T-
n*
dimensions: length - 140.0 m, width - 4.0 m,
depth -1.2 m. The channel is equipped with two v = A- • cos2;r( -) (1)
towing carriages: slow-moving with the towing
speed up to 6.0 m/sec and fast-moving up to 25
m/sec. The channel is also fitted with a wave
maker - a generator of surface waves. In where A- wave amplitude;
this installation, it is possible to carry on the Xb -wave length;
towing tests of ekranoplanes' models with the
xand z- longitudinal and vertical coordinates.
wing spans up to 2.0 m above a free water
However, compared to the velocity of a wing
surface as well as the tests of separate
movement, these velocities, in many cases,
airfoils and hydrofoils at Reynolds numbers
amount to the value (1-2 %) and they can be
R = 106. The tests of airfoils above the calm ignored. The inherent velocity of a real wave
surface of water made possible to receive the movement at its replacement in the experiment
results with sufficietly pinpoint accuracy. But with the solid wavy screen can be
on testing airfoils above a choppy surface of considered in the following way (discussed the
18-3
case, when velocities of a wing movement and were positioned in the channel at a sufficiently
a wave coincide by a direction or are directly high depth so that the influence of a free water
opposite). surface was eliminated.AII tested wings had a
Apparent frequency of the wing contact rectangular form in plan and the same profile
with a wave is equal to: "TsAGI SR-16-6M" with a thickness ratio 6 %
and the chord fc=160-r250mm. Wings of various
7JZ ,2TT 2JT , aspect ratios X = C/b from A=0.5 to ,\=5.0 were
<7=o-0±cr,=—-v0±—-c=— -(y0±c) (2)
tested, where £- wing span. The length of flat
sections of the screen was 24.6. Waves' length
of the wavy screen varied in the range from
where: a0 - frequency of the wing contact with
a stationary wave of the screen; Xh=20»b up to lb=b. Waves' hight of the screen
oi - wave frequency; at different wavy sections ranged from
c - wave speed. H ^=Xb/20 toH h=Xb. The overall length of an
Strouhal number, which characterizes the underwater screen was equal to 70.0 m. On a
process considered will have the following form: wing passing above the wavy screen, the
following range of Strouhal numbers was
covered:
Sh=a'b = 2*'b (l/o±c)
=jp
(t/ ±c)
° (3)
v v
^0 K o o
In -b
P= (4) P = ^l^ = o,201-5-6,28
,.A' „e"
1
—
««
1 1
—
Fig.1
3. TESTS OF WINGS ABOVE THE FLAT flat screen "h" and also to a free surface of
SCREEN. liquid H As seen from the graph, the lift force
of a wing increases on approaching to the
screen from above (the case of wing
Results pertaining to the movement of a movement above the screen) and decreases
wing above the flat screen are illustrated by the on approaching of a wing to a free surface of
following data given below. liquid from below ( the case of hydrofoil
Dependence of a lift coefficient Cy upon movement).
an angle of attack 6 for the wing with aspect At small angles of attack (0<1c) and at little
ratio X=2 is shown in Fig.2. distances to the screen (h<0.2) there comes a
phenomenon of negative influence of the
screen upon the lift force of a wing. In case of
a wing movement under a free surface of
liquid at //<0.2, the partial uncovering of an
upper surface of wing is observed. On further
"//' reduction there starts the transition of a
wing movement to a gliding regime. As this
takes place, Cy of a wing tends to the half of
its value on infinity.
Dependence of the drag coefficient of a
wing Cx upon an angle of attack 0 for the
various distances of "h" and "H' is presented
in Fig.3. Here it is important the fact that
screen influence on a wing drag is different at
Q' little and high angles of attack. At 0<3°, the
j- '■!■ ■ \" Y "]»■ i'' i," i" -l' ?'■ i
proximity of the screen somewhat reduces a
wing drag. At 0>3°, a wing drag increases on
approaching to the screen. On wing's
Fig.2 approaching to a free surface of liquid its drag
decreases in the whole range of studied
This dependence is given for various distances angles of attack. This result has been obtained
from the rear edge of a wing to the at Froude numbers 72r=3.8.
18-5
(6)
35
i = 0.02
"K
/ .0.03
30 / \ / MS
25 \/\ / M
\
Ji
^5
20 - Ill / 002 Dfii
i5 - jt/jy Fig. 5
influence of the wing aspect ratio on the lift one It is clear from this graph that optimal, from
can judge from Fig.6. the standpoint of the maximum lift/drag ratio of
a wing near the screen, is a wing with aspest
ratio of X=3.
Comparison of the test results of two
similar wings: airfoil and hydrofoil is
presented in Figure 8.
The airfoil was tested near undisturbed
surface of water (scheme "A" in Fig.1).
Fig.6
In this Figure, the test results of seven wings
with different aspect ratios from X=0.5 to ^=5.0
are presented. In the graph, it is given the ratio
of a coefficient of wing lift near the screen
"Cyh" to its value in the infinity " Cym"
depending on the wing aspect ratio "X" for
different values of "h". It is evident from this
graph, that the most growth of the wing lift has
Fig.8
been attained at the wing aspect ratios from
1=2 to X=Z. The hydrofoil was tested near the solid screen
Therefore, the most optimal, from the
(soheme "8" in Fig.1). The airfoil had the
viewpoint of a growth of the lift force near the
absolute dimensions 1500 X 750 mm and was
screen, is a wing with the aspect ratio in the
tested at the speed 18 m/sec, and the hydrofoil
range of ^=2+3. - 500 X 250 mm and 6 m/sec, respectively. Wing
Figure 7 shows the ratio of maximum
polars Cy=jj(Cj of these wings are given for
lift/drag ratio of a wing near the screen "Kmaxh"
different distances "h". The consideration of
to its value in the infinity "K.maieB" depending on
these dependences shows a complete
the wing aspect ratio "X" for different values of qualitative agreement between the results of
"h". both air and underwater tests. The minor
quantitative discrepancies are within the
accuracy of measurements. These results
indicate a practical equivalence of two
Zfi
discussed methods of wing tests near the
screen.
0,05
-XfS- ^r xr^
Fig.9
Along the axis of abscissae, the profile of a Cy {(p) = a()/2 + ~Yjik ■ coskip + ^jbk ■ sin£<p (8)
wavy screen is plotted (vertical scale is k=\ k=\
increased). Analysis of given curves
demonstrates that on moving above the where: cp - phase angle of the lift force curve;
sinusoidal screen, the values Cy, Cx, and xc do a0/2 - mean for the period value of the lift
not vary in compliance with the sinusoidal law. force coefficient;
ak and bk - coefficients of Fourier series.
The maxima of curves Cy and Cx and the The set of these data represents the
minima of curves x.c take place earlier than the material by which using the formula (8) it can
leading edge of a wing reaches a wave crest. be reproduced the curve of instantaneous
Curves are asymmetrical relative to the vertical values of the lift force coefficient for any
lines passing through the extremal dots of values of wing parameters and the wavy
these curves. In so doing the left branches of screen from the range considered in this
curves, which correspond to approaching of a paper.
wing to a wing crest, are more complete. Below, based on these data, the
influence of some parameters of a wing and
All curves of instantaneous values of the wavy screen on the curve of instantaneous
coefficients of a lift force obtained during the values of a lift force coefficient is discussed.
experiment, were subjected to the harmonic The following characteristics of this curve will
analysis. be considered:
18-8
Fig 13
Hi
Figure 14 illustrates the influence of a 0,05 O.i 0,2 0,3
wave length of the screen Xb on the lift of a
wing.
Fig 15
0,528 0,3«
6,28 m 0,622, 0.3 ^
Fig 18
Fig 16 The exception makes the wing lift above
waves crests. The wing lift, in the range of
Strouhal numbers of P=0.628-0.314, can exceed
to some extent the lift above the flat screen at
the level of waves crests.
In Figure 17, characteristics C^, Cy3, and
Cy22 are compared with the wing lift above the
flat screen at the level of a wave's center line of
the screen. Here, we also discuss the case ,
when a wing moves at a steady hight above
waves crests h-,. It is seen that C^ and Cy22
exceed Cy02. It is natural that the lift value
above waves' trough Cy3 is less than Cyo2.
Figure 18 presents the case when a wing,
which at first was moving above the flat screen,
Xos 50? ojlz oyk then starts to move above the wavy screen. In
the beginning, small waves start to appear
under a wing (big Strouhal numbers) and then
Fig.17 the dimensions of waves increase (the
In all three cases, it is considered the
movement to the side of small Strouhal
movement of a wing above the wavy screen, numbers). In so doing , the distance from a
which consists of standard waves of a different wing to the flat screen and to a center line of
length. Common to these Figures is the fact the wavy screen remains constant all the time.
that the amplitude of oscillation of the lift curve Such movement is possible until a wing starts
(Cy1 — cy3) increases on falling of Strouhal
to touch waves crests. For example, at h02=Q.S
number. The increment of a wing lift above the
such moment will occur at P=0.314 and at
wavy screen (Cy22 — Cy02) also rises as
/jO2=0.15 this moment will take place at
Strouhal number goes down that is evident
P=1.04. In the Figure, it is shown for these two
from Figures 16 and 17.
The difference of these three mentioned cases, how rises the amplitude of oscillation of
Figures lies in the following. In Figure 16, the wing lift (C^ — Cy3) with the increase of
characteristics Cy1, Cy3 and Cy22 are compared the screen's waves dimensions (while Strouhal
with the wing lift above the flat screen at the numbers going down). One can see that in the
first case, the amplitude of oscillation of the
level of waves crests C^.The case is
wing lift in a limit attains the value
considered, when a wing moves at a constant
approximately two times greater than the
height above waves crests h.,. It is obvious, that
wing lift above the flat screen at the level of
all these characteristics are less than C^.
18-11
wave's center line. In the second case this and the coefficient of a rolling moment C„
increase amounts to the value in the order of upon the deviation angle of waves crests "y".
60 %. The results, presented in this Figure, give
once again an idea of the scale that the wavy
screen influences on the ekranoplane's
dynamics.
The foregoing results pertaining to the
movement of a wing above wavy screen refer
to the case, when a wing is moving
perpendicularly to the direction of waves
crests. Below, it is also discussed the case of a
wing movement above oblique waves, when a
wing does not move perpendicularly to the
direction of waves crests. Figure 19 shows the
experimental installation for the research into
the movement of a wing above oblique waves.
Fig 20
CONCLUSION
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
19-2
tions of motion without (o la Irodov) and with (a la clearance and pitch angle, M is the vehicle's mass
Zhukov) account of perturbation of forward speed. in cruise, S is a wing's reference area, p is specific
Then orders of magnitude of the terms are made density of air and iz is a coefficient of the longi-
and assumed on the basis of a simplified nonlin- tudinal moment of inertia of mass, determined by
ear unsteady theory of extreme ground effect [6]. means of a relationship
Eventually, an asymptotic form of the equations of
motion is derived for h —*• 0 and small periods of (4)
MC2
time from the moment of perturbation. The cases
of larger time scales are discussed. where Iz is the longitudinal moment of inertia of
mass with respect to the center of gravity. To track
the relationship between dynamic stability of the
2. LINEAR EQUATIONS OF DYNAMICS
vehicle with its cruise speed it is sometimes conve-
nient to use the following obvious equation
Here, the linearized versions of equations of longi-
tudinal motion will be deduced, corresponding to Uo
what is known as "quartic" and "quintic" descrip- H = Fr2ccyo, Frc = (5)
tion of dynamics of wing-in-ground effect vehicles.
In other words, linearized equations of motion and where Frc and cVo are correspondingly the cruise
corresponding characteristic equations are written Froude number and cruise lift coefficient. With ac-
with and without account of perturbation of for- count of (1), (2) and (5) the (two) equations of
ward speed. Instead of representative kinematic pa- motion with respect to perturbations of pitch and
rameters a (angle of attack) and h (relative ground ground clearance can be written in an alternative
clearance) utilized by Kumar and Staufenbiel, pa- form as
rameters 9 (pitch angle ) and h (relative ground fh
Fr2ccyo^ = Cy (6)
clearance) will be employed. In what follows all dt
unsteady aerodynamic derivatives will be incorpo- d29 _
rated. Fr\cyo
dt2
The pair of parameters 9, h is more practical in Using representation of perturbed lift and moment
aerodynamics of ground effect vehicles than that of coefficients in terms of derivatives with respect to
a, h. Use of 9, h was first proposed by Treshkov principal kinematic parameters 9, h one can write
(1970) and since had been adopted in most of the
Russian developments on WIGs. cJö + ^Ä + cJÖ + c*Ä (8)
where ß is relative density of the vehicle defined by where the coefficients A,(i = 1,2,3,4) are
the formula
/. = -^- Ax = -(mj+*,cj), (13)
P y(3)
pSC0 '
In (27), (28) c~y,mz are perturbed lift and moment m
coefficients, h, 9 represent perturbed relative ground A2 = -4^i4
ft Iz
* - cH+M"»2+*'.<£)]. (14)
19-3
d2e
1
ttiz
dt2 = yt 2c
( *° ~ c^u ' + m
°ö+ Incorporating both viscous and induced drag
19-4
sian one with x-axis directed forward, y-axis di- Making use of the steady flow equation, one can ob-
rected upward and z-axis directed to the starboard. tain an alternative equation for the perturbed un-
Abscissas are measured from the trailing edge of steady velocity potential
the wing. Equation (26) should be solved with two
d .dtp0]
boundary conditions
Assume that the gap under the tips of the endplates It can be seen from equations (38)-(41) that
is constant chordwise 6ep = const. Corresponding both derivatives of perturbation velocity and poten-
equations for the components of the perturbed so- tial with respect to kinematic parameters are in-
lution can be derived in the form versely proportional to relative ground clearance in
- cruise, i.e. are of the order of 0(1/h). Derivatives
_ \h (x) ] + ——(x)— of perturbed pressure with respect to perturbation
dx in height h and rate of variation of height h can be
\ d \ dip0 ] . determined in the form
"■[M«)^r]=0. (33)
PH>) = "2^^ = -2«.W^r (42)
£[M.)£]-. (-. "**' d4 *
Imposing conditions at the leading edges, one ar- p (x) = — 21 v0(x)— \- <p ) (43)
rives at the following boundary conditions at x = 1
for the equations (33) and (34). 3.2.2. Pitch perturbation
i In the case of pitch motions around the center
<ph(l) = 0, <ph(l) = 0 (35) of gravity h(x,t) = 0(t)(x - xcg) the perturbation
potential of unsteady flow can be represented as
Applying Kutta-Zhukovsky condition, one has
-h 4. . ... <p(z,t) = ipSÖ + tpS§ (44)
fc ft
-£-Ä
dx
+ -j£-Ä-0
dx
Ä-0 JUo (36) _Derivatives
. f. ol. unsteady
f , . , ,
perturbed
poten-
. ..
velocity
,
,T , ,. jj-i- r Aii- ■ i ■ i. tial with respect to perturbation in pitch 6 and rate
Neglecting second derivative of perturbation in height ;.
in (36), which implicitly would imply that Strouhal of Pitch 9 are governed by the following equations
numbers are small, one arrives at the following bound- J ~» J
ary conditions at x = 0. A. \he(x)^-] +-j-\(x - xcg)^] -
dxY dx 1 dxl "' dx J
df_
* A,~J a suction force contribution. For the case of flat
h$
H dx «K
The constants c% and C4 can be calculated with help
(51) ground cXp can be written as
cs = -hh*(l,t)[l + v(l,t)}2 =
ch,h,i,e=
I!
/ pW.'(x)dx (54)
curvature of wing sections of the order of 0(h), one 4.2. Derivation of asymptotic form of the
has equations of motion for h —► 0
<*.,"»,. =0(1), cXio=0(h) (65)
As per previous analysis the derivatives of aerody- Employing the above estimates and conventions about
namic coefficients have the following order of mag- the orders of magnitude ( in terms of h) and ne-
nitude glecting terms of the order 0(h) and higher, one
can reduce the equations (19)-(21) to
(cy,m,)h>kM = 0(h (66)
t = 0(l/h2). It can be shown that on the scale [4] Staufenbiel, R., Kleineidam, G., "Longitu-
of "large time" variation of speed of the vehicle is dinal Motion of Low-Flying Vehicles in Nonlinear
mostly driven by perturbations in height and pitch, Flowfields",in Proc. of the Congress of the Inter-
whereas on the scale of "very large time" variation national Council of the Aeronautical Sciences, Mu-
of speed is determined by perturbation of speed nich, FRG, 1980, pp. 293-308.
proper. In the latter case perturbed equation for [5] Zhukov, V.l., "Pecularities of Aerodynam-
speed is completely uncoupled from those for height ics, Stability and Controllability of an Ekranoplan",
and pitch and has the form Izdatelskiy Otdel Tsentralnogo Aerogidrodinamich-
eskogo Instituta im. N.E. Zhukovskogo, Moscow,
-dU' _o
2cx) U' (79) 1997, 81 p.
[6] Rozhdestvensky, K.V., "An Effective Math-
2
where r? = h t is "squeezed" time variable, cf = ematical Model of the Flow past Ekranoplan with
cY /h and cx = cx/h are quantities of the order of Small Endplate Tip Clearances in Extreme Ground
unity. Effect", in Proc. of the International Workshop on
"Twenty-FirstCentury Flying Ships", The Univer-
sity of New South Wales, Sydney, Australia, 7-9
5. CONCLUSIONS November 1995, pp. 155-178.
[7] Rozhdestvensky, K.V., "Ekranoplans -The
In the present paper an attempt has been made to GEMs of Fast Water Transport", Trans. Institute
derive asymptotic form of the equations of motion of Marine Engineers, Vol. 109, Part 1, 1997, pp.
of a lifting configuration in extreme ground effect 47-74.
through formal estimate of orders of magnitude of
different terms when h —»■ 0. It has been shown that
on (nondimensional) time scale t = 0(1), which cor- 7. APPENDIX
responds to distances of the order of the chord from
the moment of perturbation, the equations of mo- Formulae for the coefficients J3,, i = 1... 5
tion correspond to "quartic" formulation of Irodov
[1], i.e. the speed of the vehicle remains almost
constant. As different from Irodov, the asymptotic Bx = 4-[(2c, -cf - ck)iz - m»] (Al)
form of the equation, valid for vanishing h, does
not depend explicitly on relative ground clearance,
B2 = -4^[(c? - 2cx)(izck +mi- ytci)+
but rather on reduced density ß — /j,h and ratios of ß %z
design pitch and and curvature of the lower sur-
iz(2cyc - fie*) + chym2 - ceymhz - fime2]
k
(A2)
face to h, i.e. number of parameters is less by
one. On larger time scales of the order of 1/h and
1/h2 variation of speed is first driven by height and
pitch perturbations and, later is determined by the
speed perturbation proper. The latter conclusion H(m9x + izchy) + ceymhz- c^mez]+
confirms results, derived by Zhukov [5].
2cy(4mh2 + chxiiiz - chxm$z)+
,
S55 =
= ^{(**-^)(cT-2c, + 2CyJ)+
fl lg \ Cy
20-1
V.l. Korolyov
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
20-2
The longitudinal static stability of the objects with two - l + Ca(l + r,)jJ^-JAa (6)
carrying wings separated by some distance, which is de-
termined by the lenght of hull,-" tandem" system (Fig-
c.-=<#/c- (7)
ure 1) is being investigated in the present work.
For the wings with the rectangular plan form
2. DERIVATION OF THE MAIN DEPENDENCE.
l Vl (8)
At the calculated conditions the object moves at a rate ~ % 2
V with the given fit - defined angles of the hydrofoil
attack c*i and their distance from the boundary of sep- Using (2) and (3) we shall obtain
aration hi. It is possible for wings to have different
chords &i, relative distances from the boundary of sep-
<?V,M2, l + fi
(9)
aration hi = hi/k, relative lengthenings A = Si/bi. Cy,A2£ -/i
Here at the analysis of the main dependences we con-
sider wings with the rectangular plan form, but this
Taking
method can be used for wings with arbitrary form, if (10)
there is information about the aerohydrodynamic char-
acteristics. At every change of the fit, i.e. position of we shall obtain
wings relativily to the boundary of separation, caused
by external forces, carrying forces on the wings change,
this leads to the appearance of the longitudinal mo-
L. /J1+fl (11)
ment, which can be both restoring and destabilizing.
Value and sign of this moment can be determined tak- Putting (11) into (6) we shall obtain resulting:
ing into account the following considerations. Change
of the carrying force on the back wing at the change H = ^-L = fx(i + /i)|(i idMh)l
of the attack angle by the value Aa (turn of the ekra- Aa Dl <*i
noplane about transverse axis that goes across centre
of mass) can be determined as follows /2/:
1 Aa
Aft, = -Dlx I + TW + G)'»» (5) c (14)
a2 ^~
(12 - 14) equations are obtained for ekranoplans. Anal-
//6a
ogous expressions are applicable for the hydrofoil shipB,
Moment, conditioned by the change of the carrying The equations differ only by signs, in equations (12)
forces on wings, is determined in the form: and (14) for the hydrofoil ships the positive sign is re-
placed by the negative. It is necessary to note, that
AM - Dlfi(\ + 6)/(l +CifiI»1)~ when the wing system deviates by the angle Aa, not
20-3
only a moment appears, but also an additional force, stabilizing moment appears when the attack angle in-
which is denned by the expression crement on the back wing is positive, this moment
is conditioned by approach of the wing to the screen
and by the attack angle increase. Sign of the moment
^ = j(l+ £)(! +d/iW^-- on the front wing is determined by that parameter,
which more influence the change of the lifting force -
l + 'i increase of the relative distance to the screen h2 or
-h l + C2{l + h)hi -ph <*2 J (15) of the attack angle a-i- Therefore, in order to obtain
the maximum value of H, the attack angle a2 must
or, in other words, the inclinations are non-equilibrium. be as big as possible. (This condition determines also
The change of the lifting force is determined by calcu- the negative value //'00)- The value hi = l/bi, which
lations and has the order of characterizes the relative lengthening of the object un-
ambiguously determines the main component of H -
(16) screen Hh, which value is proportional to ltr Next
D cti
main factor, that influence the value of H is the charac-
Therefore it is necessary to correct the fit of the object teristic C = Cy/Cy. Its increase leads to the increase
in order to preserve D value constant for the calcula- of the metacentric heigh. For calculation of d values
tion of the recoveral moments at essential deviations dependences of Cy on the lengthenings and other pa-
rameters were used. These dependences were obtained
In majority of cases it is necessary to know the ini- for the profile ILATH CP-16-6M [4,5]. C values, calcu-
tial values H for small deviations, which do not need lated according to these data in the lengthening range
correction. In particular, such characteristic is impor- A = 2 - 4, attack angles a = 2° - 6°, distances from the
tant for determination of parameters of aerodynamics screen h = 0.05 - 0.2 for the wings of rectangular form
scheme of apparatus, which ensure its stability. It is with the relative profile thickness 6 = 0.06 are given
particularly actual taking into account the absence of at Figure 2. C value weakly depends on lengthening
experience in construction of ekranoplans of the dis- A, greatly increases while approaching the screen and
cussed scheme. In the work /3/ an expression was ob- in the attack angle range 0° — 6° has approximately
tained for the determination of the own frequency of linear dependence on the attack angle. Analysis of the
the longitudinal angular oscillations of ekranoplan.In influence of the above mentioned factors is fulfilled by
the dimensional form it looks like: calculations.
H value is calculated according to formula (12) at their
2 - 9/2 J-,
-pV2 change in the following limits: A; =2-4, a° = 2° -6°,
W2 = hi = 0.05 — 0.2. Relative lengthening of the object
2mr2 .•=1 ' »=1
lbi = l/b\ is taken 5. Influence of this factor is unam-
biguous, its increase leads to the increase of H. Influ-
1/a (17) ence of centering (position of the mass centre along the
4mr 2(^rE^^)']} length) was determined by calculation of every variant
The second term in the formula (17) usually constitutes of scheme at four centerings. (And the static condition
small part of the first one (3-4%) and can be neglected. was observed by variation of 62 value.)
Taking into account this circumstance it is possible to Calculation results show, that the value of H is mainly
obtain the following dependence, which links the meta- dependent on attack angles a\ and »2, value hi and on
centric height with the own frequency of longitudinal the position of mass centre (l\). In every case increase
angular oscillations (pitching oscillations): of H is facilitated by the increase of oci/012 ratio. Re-
duction of this ratio sharply decreases stability and can
lead to its loss. At Figure 3 the generalized results are
-gi H (18) given for the absolutely symmetric scheme (Ax = A2,
W2
experiments confirmed received results. narrow limits, where influence of A on C, is small so, in
We do not consider the questions of compatibility of every case A2/Ai = 1.1 and relative lengthening of the
stability and aerodynamics requirements {K quality), object hi = 25 were taken. Attack angles (with account
of possibility for the practical realization of the small of the angles of zero lifting force) a = 2° - 4°, relative
values of h and other peculiarities, connected with im- dippings hi = 0.4 - 1.6. Calculations were provided
plementation of the obtained results in practice. These at several values of k for determination of the value /1
questions are to be solved when the aerodynamic sche- (centering), which corresponds to the maximum value
mes of the definite projects are developed, on the base of H. Condition of statics was observed automatically,
of definite information - aerodynamic characteristics for the value 62 was determined in the course of calcu-
of wing with the account of their mutual influence, in- lations.
fluence of fuselage and other elements, form of wing in Hydrodynamic characteristic of hydrofoils - parameter
plane and so on. The given data are of approximate d = Cy/Cy (Figure 6) was determined according to
character. Nontheless these data give definite enough the data from the book /6/ - values Cy = f{h,X),
view of the influence of main factors, permit to analyse obtained theoretically for the plane rectangular hydro-
the possibilities of different compositions. foil. Influence of thickness and curvature of the axial
are of the profile on Cy was accounted for by the change
4. STABILITY OF HYDROFOIL SHIPS of the angle of zero lifting force. Calculation results,
corresponding to the greatest values of H are given at
In the former Soviet Union a series of projects for the Figure 7. As it is seen from the table, correspondence
ships with the little sink underwater hydrofoils was de- of the influence of ratios a,- and hi of the calculation re-
veloped /l/. Hydrodynamic scheme of these ships is sults and practice is quite unambiguous. Dependences
analogous to the aerodynamic scheme of ekranoplan of H on mass centre position (h) for the variants of
with the "tandem" system of hydrofoils (Figure 1). calculations with the greatest values of H are given at
Stability problems of their movement were solved by Figure 8. In every case the value of /"1 changes in nar-
empiric method - the optimum combinations of hy- row enough limits (approximately 0.4 - 0.5), this also
drofoil system parameters, which ensure high hydro- corresponds to the results from practice. The carried
dynamic quality and stability were found as a result out comparison shows that the influence of the factors
of numerous experiments. So, all the constructed hy- a,-, hi, h on the value of H is determining; the devel-
drofoil ships have good stability and can serve as some oped method of H calculation quite objectively reflects
standard when new projects are estimated. Analysis of the link of metacentric height with these factors.
the main parameters of the carrying systems of these
ships shows, that the definite enough ratios are char- 5 SOME ESTIMATIONS AND DEDUCTIONS
acteristic for them:
- position of the mass centre along the length is deter- Ekranoplan must satisfy two main requirements - to
mined by the value h 2 0.48; have high aerodynamic efficiency and high level of safety,
- lengthening /&, = //&i — 25; particularly stability of movement. Taking into ac-
- chord ratio of the bow and stern hydrofoils 62/&1 = count the scale of damages due to inadequate efficiency
0.9; or stability, when conflicting requirements are to be
- ratio of the lengthening of hydrofoils was in the lim- observed, the problem must be solved by way of com-
its A2/Ai = 1.1 - 1.35, in majority of cases the attack promise, with the priority of requirements, that en-
angles of the stern hydrofoil were less, than that of the sure safety. One of the advantages of the two- support
bow hydrofoil (c*i < a^); scheme, that was considered above, is that the require-
- dipping of the bow hydrofoil in every case is more, ments for optimum choice of parameters for wings from
than that of the stern hydrofoil (h2 > hi). the point of view of quality and stability in some cases
Here are given the average values of parameters. Their coincide. Thus the main (back) wing, carrying 65 -
variability can be explained by the influence of numer- 70% of the load, must have, according to both require-
ous factors, which are not taken into account - hydro- ments, small attack angle a 1 = 2°-2.5° . Aerodynamic
foil deadrise and hydrofoil sweep, profile form, influence scheme of the stabilized ekranoplan is more efficient in
of form of wing in plane, position of the struts, hydro- principle, that the scheme with control, for it excludes
dynamic influence of the front hydrofoil on the back appearance on the control elements - stabilizers - of
one and other individual peculiarities of the projects. the control "sinking" forces.
However, general tendencies, that determine influence Stability on the vertical travel for ekranoplans and hy-
of principal factors, manifest themselves quite clearly. drofoil ships of "Tandem" scheme is ensured in every
In this connection comparison of these data with the case at the account of appearance of restoring force,
calculation results according to formula (12) are inter- which is defined by the value Cy.
esting. Calculations were fulfilled in the real range of Besides, for the scheme, that is asymmetric relating to
parameters, which was developed in the practice of ex- the mass centre (such scheme of "tandem" system is
ploitation of such ships. The ratio A2/Ai changed in the most stable), at vertical travel the restoring mo-
20-5
6. CONCLUSION
0,Z5 05 WS fO /?
0,25 05 0.75 /? i0
■c
20
A>
-H /
0,6- /
S
i
10 /
\/
\/
oM
Xz * 2 2 2 3 3 3 2 3 4
hz 0,f 0,1 mmoos
0,2 d\ z 4 G 2 * e 6 6 €
A/ / 2 3 4 5 6 7 8 9 to
Fig 2 Aerodynamical wings characteristics of ecra- FiS-4- Dependence of H on main parameters of ecra-
, noplan.
noplan.
Fig.3. Stability characteristics of ekranoplan with iden- Figf DePendes of H on mass center of
Positi™ ecra-
tical forward and after wing. noplan.
20-8
a1 = 2°
Fig.7. Dependence of H on main parameters of hydro-
foil ships.
21-1
A.V. Nebylov
State University of Aerospace Instrumentation,
67 Bolshaya Morskaya, St. Petersburg, 190000 Russia.
E.Th. Zhigalko
Faculty of Mathematics and Mechanics, St. Petersburg State University,
P.O.Box 154, St. Peterhof, 198904 Russia.
1
The work was executed under support of
Russian Foundation for Basic Researches
according to the project 96-01-00010.
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
21-2
There are numerous functions directly or ratio can with a drop to a basic surface
implicitly ensured by aerodynamic achieve two. This property together with
properties of LE. They are: a lift the effect of altitude self-stabilization (see
capability, a property of the self- Fig.1,
stabilization of flight (on a height, load Fy > G, the gravity force, when y<hk)
change, pitch and bank angles), distinguishes an ekranoplane from an
controllability (on a velocity, altitude, pitch, airplane and is important enough for
course, roll) providing with maneuvers making the concept of WIG plane
(take-off from water, landing) buoyancy attractive, even at its known basic
and stability in sailing, basing of ASP, joint shortages. The most important among
with ASP maneuvers (start, landing, these is the problem of an ekranoplane
emergency), alternative application. There flight stability, under which its ability is
are other functions apart from the understood to independent (without
transport ones affecting the realizability of interference of a pilot or system for
the project, for example, a safety, a automatic motion control, SAMC) return to
manufacturability and interaction with an the initial condition of driving after random
environment. perturbation.
When estimating the take-off weight of The solution of the problem may be
ASP is chosen as 600 tons, the landing provided with the proper choice of the
weight is 60 tons, a square of a wing is glider form and the mass center location.
600 m2. The booster velocity is 0.6 M that The importance of the question for
covers a range of 3000 km. Basing on ekranoplane results from the altitude of its
parameters of known machines and using flight being too small for a piloting error
extrapolation one can calculate that the correction.
ekranoplane valid for the booster function
own mass will reach 1500 tons while a Fy/F,yoo
chord and a square of the main wing are
60 m and 3000A772 correspondingly.
The scale increase of a flying apparatus 1.5 -
erects new problems on the design, for
example, conjugated with necessity of the
arrangement of a sufficient number of G/Fyoo
drives of a limited single power. The most 1.0
appropriate among existing at the moment
' hi/b
is the Russian drive NK-44, and the
Russian company AO NTK "Engines NK", 0 0.1 0.2 y/b
has declared readiness to produce it in Fig. 1. Typical lift dependency on flight
lots. altitude
( (^ „ \\ -i
G-R G-Ryx> V*
yco
v.M- [pya-G)y + hkRyx In -1 (1)
V F
yx> JJ Ryoo
21-5
Hl ■^s» <~ 1
f"
T*n
*-_£> \
.uf) oH>
a b
Fig.3 Arrangements of ekranoplanes
2 X
3* + X +
Rear wing 1* X
*
2 - - X ~ —
*
3 + X ++ + ++
ASP1) - + X - + - + +
2 Displacement of GC Hor. + + + X + + +
+ + +
Vert. X -
1
) In this line the sign "+" shows presence of adequate aerodynamics.
The suggested variant of the ekranoplane a booster-carrier for ASP horizontal start
tandem configuration - ASP carrier can be and landing and do the similar operations
compared to another variant of in other fields.
ekranoplane - catamaran (Fig.2) REFERENCES
described in [1]. It is closer to the airplane
configuration and allows to flight both in 1. Tomita, M., Ohkami.Y., NebylovA.V.,
the WIG-effect action zone and outside it. Sokolov V.V., "The concept of heavy
However, it is possible to consider the last ekranoplane use for aerospace plane
function which is superfluous for examined horizontal take off and landing", WIGs
ekranoplane's application. Without tandem
RINA International Conference,
arrangement, this variant could be
London, UK, 1997.
aerodynamically more favorable, but it has
2. Tomita, M., Tsurumaru D., Saotomi T.,
smaller margin of stability in comparison Ohkami.Y., NebylovA.V., Sokolov
with variant "c°", and with "Orlionok" that V.V., "Feasibility study of a rocked
forces to increase essentially the load to powered HLHLSSTOwith
aerodynamic control elements and to ekranoplane as a take off assist", 7th
increase the responsibility of automatic AIAA International Aerospace Planes
stabilization and damping systems up to and Hypersonic Technologies and
the top. The final conclusion about Systems Conference, Norfolk, Wa,
advantage of any considered variants can
USA, 1996.
be made only on the basis of detailed 3. Nebylov, A.V. "Measurement of
research and simulation in view of all set Parameters of Flight close to the Sea
of essential circumstances. Surface", Publication of SAAI, St.
7. CONCLUSION Petersburg, 1994, in Russian with
The proper choice of arrangement of an summary in English.
ekranoplane can ensure its high 4. Diomidov, V.B. "Automatic Control of
properties: the carrying capacity and flight of ekranoplanes", Elektropribor,
velocity. Properties are ensured of the SPb, Russia, 1996, in Russian.
self-stabilization of a flight altitude as well 5. Zhukov, V.l. "Peculiarities of
as of pitch and roll angles. Applied criteria Aerodynamics,Stability and Handling
have been proved by practice of use of of Ekranoplanes", RINA, International
large Russian ekranoplanes "Orlionok", Conference on WIGs. London, UK,
"Lun". The self-stabilization is combined 1997.
well in flight with a SAMC action which is
attribute of the control circuit for
ekranoplane. A heavy ekranoplane can be
22-1
E.A.Aframeev
Krylov Shipbuilding Research Institute
44, Moskovskoe Shosse
Saint Petersburg, 196158
Russia
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
22-2
which is limited with solitary examples of Aspiration for reaching the high lift-to-drag
large ekranoplans application in Russian ratio (aerodynamic efficiency) of ekranoplan
Navy and some small ones in other countries. wings owing to its movement near the
underlaying surface was, as it is well known,
Meanwhile the important qualitative a pioneer technical idea, starting the
conclusions can be made from the available appearance of the ekranoplans. It was
R&D and design works, experience of assumed that high aerodynamic efficiency
ekranoplans creation and operation. These would provide the advantage of the
conclusions will allow to outline the ekranoplans even over the air transport
conceptual base for ekranoplans building for means in respect of such features as thrust-
the present time and more remote future: the to-weight ratio and fuel consumption, i. e.
aims of the creation, technical appearance, eventually in respect of the economical
ways of practical introduction into the system efficiency.
of transport means.
In the majority of the countries developing
The complex works in the sphere of future the ekranoplans this idea is considered now
ekranoplans were carried out in Russia in so important that their arrangement, in
1970-1980 at Krylov Shipbuilding Research essence, provides only the aspiration for
Institute under the guidance of the author reaching as high aerodynamic efficience as
with participation of design and research possible to the detriment of providing some
organizations of shipbuilding and aircraft other necessary features. Such particular
industries. In the process of these works the approach can be considered as appropriate
author suggested and grounded the idea of one for the development and experimental
the transition to the creation of the second stages of research when the aims of
generation ekranoplans with high ship ekranoplans practical introduction into the
features. The proper author's conception (A- system of transport means are not so urgent.
conception) of two-mode ekranoplans was However on the stages of practical realization
developed. Later on the subsequent R&D of of the ekranoplans the more detailed
the second generation ekranoplans and the consideration of all diversity of the factors
transport system on their base were carried connected with both the creation of the
out [1]. ekranoplans (parameters of structure
materials, the engines for the power plant,
The fundamental theses of the above stated technology of building and so on) and their
author's conception are presented in this operation (tasks, conditions of sailing and
paper on the base of the analysis of real basing, economical factors of the operation
experience of creation and operation of the and so on) is necessary.
available ekranoplans and development of
perspective ones. The suggestions concerning Unfortunately, in the most of investigations
the development of ekranoplans and their carried out the impact of the real factors of
immediate practical application are stated. creation and operation of the ekranoplans
upon their technical appearance isn't taken
In accordance with the author's classification into account properly. The design is
the ekranoplans are divided into superlight unsufficiently attended with economical
(takeoff weight up to several tons), light (up research which moreover are excessively
to several tens of tons), middle (up to 100- general and rough.
500 tons), heavy (up to 800-1500 tons) and
superheavy (up to several thousands of tons). As regards the ideas connected with the
The principal theses of this paper are spheres of ekranoplans applications it is
dedicated to the middle and large necessary to say that they aren't very diverse.
ekranoplans, however in many respects they As a rule, the use as the passenger transport
are correct for the lighter ones too. mean in civilian sphere or for the troops and
armament transportation, in particular, the
2. THE IDEAS forces of quick deployment in military one
Now as before the ekranoplans building are considered. The lack of widespread view
needs the ideas determining the technical on the modern and future needs of mankind
appearance of the ekranoplans and the in marine means doesn't allow to determine
spheres of their application. the requirements upon the technical
22-3
3.5. The problems of weight perfection 3.6. The problems of the power plant
3.5.1. The level of weight perfection As a rule, the power plant of the ekranoplans
The level of weight perfection of the are completed with converted aircraft
ekranoplans is determined with their full engines, i. e. they aren't optimized for the
weight efficiency showing the share of permanent work at low altitudes. It is
possible payload (i. e. the sum of load and necessary to place the engines in the bow of
fuel) in the total weight of the vehicle. the ekranoplan with sufficient advancement
before the wing for providing the effective
In is advisable to compare the full weight blowing. Such arrangement leads to the
efficiency of the available first generation increase of hull dimensions, the appearance of
middle ekranoplans having the takeoff weight massive pilon carrying the engines, difficulties
up to 450 tons and the perspective second in vehicle centering, danger of engines
generation heavy ekranoplans having the flooding when sailing and taking off and so
weight of 700-800 tons with that of the heavy on. The special equipment for the deflection
aircraft. The full weight efficience of the first of the engines streams under the wing or for
generation ekranoplans is a several times turning the engines as a whole is also
smaller as compared with the aircraft and the required what complicates and increases the
competitive parameters are achieved only for weight of the power plant. The above stated
the second generation heavy ekranoplans has a negative effect on the parameters of the
[Fig4], aerodynamic and weight perfection of the
ekranoplans.
The analysis of the relative weight of the
main glider units for the ekranoplans and 3.6.1. Thrust-to-weight ratio
aircraft shows that their empennage weight The total thrust-to-weight ratio of the
are equal to each other and the weight of the ekranoplans representing the specific value of
hull and wings of the ekranoplan is required adjusting engines thrust is
considerably greater than that of the aircraft sufficiently similar to that of the heavy
[Fig5]. aircraft because of using the takeoff with
22-5
From the other side, it should be not 4.1. The need for the change of the
expected that the cost of the creation and fundamental conception
operation of the ekranoplans would be From the above stated it follows that the
considerably higher than that for aviation pioneer technical idea of achieving the
means. ekranoplans advantages owing to the high
value of aerodynamic efficiency can't be
Depending upon the specific conditions of determinative for their practical realization.
operation, length of the lines, value and The basic ideas for the perspective
constancy of passengers and cargo traffic, ekranoplans must be found not in the sphere
availability of the other transport means, of competition with aircraft or
possibilities of creation of the required hydroaeroplanes in respect of aerodynamic
infrastructure and service and so on, the use efficiency. The ekranoplans created up to
of the ekranoplans can be turn out profitable now essentially as the kind of aviation
or not competitive. engineering can't win in a competition with
22-6
The demand of sailing for a long time leads to probably, the single type of hight speed
corresponding arrangement of the hull, power transport means of so great weight using
plant, lifting surfaces and flight control above the water surface. The ekranoplans
systems which ought to be dry, not damaging will be appllied in the spheres requiring the
with the waves and floating objects, not quick transportation of passengers and cargo
exposing to corrosion and so on. along the water areas when it isn't advisable
or unprofitable to use the aircraft or
After fulfilling the conditions of sailing in the hydroaeroplanes, sailing for a long time
process of ekranoplans designing all (including the sailing in conditions of roungh
measures must be taken for increasing of sea), fulfillment of various works on the sea
their flight characteristics, including the surface; superheavy ekranoplans will be
aerodynamic efficiency, owing to the applied for transoceanic transportation of
"ground" effect. At the same time the passengers and cargo with high level of
requirements of efficient takeoff and landing comfort and safety, transportation of large-
must be considered. sized cargo and so on.
The above stated determines the solutions It is obvious that the greatest potentialities of
concerning the aerodynamic arrangement of the ekranoplans building are in the sphere of
the hull and lifting surfaces, pover plant and large vehicles in full measure realizing the
the takeoff and landing systems, in particular, possibilities of two-mode operation in
the transition to the "compound wing" conditions of rough sea and saving no rivals
arrangement of the lifting surfaces for the amond the transport means. Therefore in the
second generation ekranoplans. course of ekranoplans development it is
necessary to aim at the realization of the
The approving of the stated conception gives vehicles of as great as possible weight what is
the ekranoplans so specific qualities that they really possible at the present level of aircraft
in truth become the new marine transport and ship power engineering, construction
means, differing from the analogues with its materials, instrument - building, building
technical appearance, performance, areas of technology and so on.
application and so on. The ekranoplans safety
level will be incomparably greater than that of The next step in the ekranoplans building
other types of aircraft. development at the present level of
The conception of two-mode operation is engineering will be the creation of the heavy
suitable at one or another extent to all classes ekranoplans with the takeoff weight of 700-
of the ekranoplans, from superlight till 800 tons, what is practically can be realized
superheavy ones (in spite of the possibility of just now. As it is evident from the
the latter to have the advantages over the illustrations of the previous part, the basic
aircraft in respect of the aerodinamic technical characteristics of such ekranoplans
efficiency). already will be at the level of those of the
heavy aircraft.
4.3. The next step in the development of
the ekranoplans As the result of the above mentioned
It is obvious that all classes of the complex works carried out in 1970-1980
ekranoplans will be developed and improved. years the completely technically grounded
It is also obvious that they will not replace design of the 750 tons ekranoplan is available
the other types of high speed marine craft and in Krylov Shipbuilding Research Institute.
aircraft (hydroaeroplanes) even in the future, The share of the payload and hiel reserves for
but will be the addition to the available this ekranoplan is about 250-300 tons, what
transport means. Having the possibility of prowides the range of operation up to 3000 -
two-mode operation superlight ekranoplans 4000 km. The takeoff, landing and cruising
will occupy the technical recept between flight are possible at sea states not less than 5
small boats and small hydroaeroplanes, light balls (waves height of 3% ensuring is up to
and middl ones - between the other types of 3,5 m). The speed of the flight - 400-450
high speed craft and hydroaeroplanes up to km/h, the sailing speed - ab. 25 km/h. The
the largest ones, heavy ekranoplans will ekranoplan helly complies with the
represent the class of high speed marine requirements of two-mode operation
transport means which have no real fleet and providing both the sailing for a long time in
aviation analogues, superheavy ones will be, conditions of rough sea and the returning to
22-8
the port - shelter from any point of its the point of emergency. Experience has
operation area in the sailing mode driving proven that the use of aircraft or
with the diesel engines of low speed. hydroaeroplanes doesn't solve all problems
connected with the rescue and is often
Because of hight technical characteristics of unefficient in unfavorable
the heavy ekranoplans the sphere of these hydrometeorological conditions, because of
application is wide. For example, it may be some accident-related reasons or because of
the commercial passenger and cargo their own limited capabilities. The use of
transportation (including the use as a car - helicopters is limited to coastal areas. The
passanger ferry), patrol - guard duties ships and vehicles that happen to be near
(including the constant duty in the accident sites remain the main rescue means
economical zones), rescue operations, as before. However, most of them don't have
tourism and so on. It seems' unquestionable the special rescue equipment and is little
that with the advent of heavy two-mode suited for rescue operations.
ekranoplans another, not mentioned above
spheres of their application will be The use of above mentioned heavy two-mode
determined. second generation ekranoplans with weight
of 750 tons provides the basis for effective
The real creation of such ekranoplans will be marine rescue system organization. Such
the important event in the development of global international rescue systems should be
marine transport means and will allow the intended to rescue crews and passengers of
marking to solve some problems which were ships, submarines and aircraft, personnel of oil
difficult to solve before. drilling platforms and other technical facilities
at sea and technical facilities themselves and
4.4. Creation of global international sea render emergency aid in environmental
rescue system based on the heavy catastrophes at sea.
ekranoplans is the challenge for the
international community It should be emphasized that the effective
The challenge of the effective rescue of rescue may be carried out if two components
people and technical facilities in the case of are available: quick notification about the
disaster at sea is one of the burning world catastrophe and quick arriving of the rescue
problems. It connects with total considerable means to the point of distress.
intensification of mankind activity on the
world ocean areas, growth of passenger and Now the international community already use
cargo shipping traffics (in particular, the oil the COSPAS-SARSAT satellite
and chemical product), wide air communication and navigation system
transportation over the seas and oceans, created by Russia, the United States, Canada
fishing development, intensive developing of and France for rescue operations at sea. This
marine zones of minerals resources system is being intensively modernized for
extraction. providing the immediate reception of distress
signals and information about the emergency
As estimated, about 1200-1300 lives are lost object location, circumstances of accidents
aboard the sea-going ships and about 400 and weather conditions in the area of the
ships with total capacity of about 1,5 million accident.
register tons are lost annually. Financial
losses of ships-owners caused only with the The ekranoplans - rescuer, designed by
loss of the ships are equal to 1-2 billion Krylov Shipbuilding Research Institute on the
dollars, the losses of consignee are not less. base of above mentioned heavy ekranoplan
Disasters and losses of the ships transporting with weight of 750 tons, may be the second
the oil or oil-chemical products lead to hard component of the rescue system. This vehicle
ecological consequences. can quickly deliver the wide range of rescue
equipment, from rafts and motor boats to a
However in spite of the above stated some boom defense ship, a bathyscaph and a
kind of integral international rescue systems helicopter, to an accident site. Owing to the
which can render emergency aid in possibility of two-mode operation such
catastrophes at sea is absent till now. ekranoplan can sail and manoeuvre at sea for
Primarily it is connected with the lack of the a long time even in conditions of very rought
transport means which can quicly arrive to sea and rescue the people and render the aid
22-9
The author of this paper suggests to begin the However the creation of such ekranoplanes is
practical realization of the global international real only in the remote future.
rescue system just now.The efforts in the
development of heavy two-mode rescue The estimations and analysis of superheavy
ekranoplan, building and operation the series ekranoplan show that in principle, it is
of them and the system as a whole can be possible to create them on the base of the
joined within the framework of an available construction materials and
international consortium of the shipbuilding prospective conventional and nuclear power
and aircraft companies and with the financial plants. Therewith the use of nuclear power
support of the project by all countries plants is more preferable in respect or their
interested in rescue at the sea. weight and dimensions because during the
service life of the reactor's cores the
The creation of suggested system really may esentially unlimited endurance is provided
become the "design of century". with the same cargo-capacity. The demands
of environmental safety of operation of so
great hight-speed transport means are met if
the nuclear power plant is functioning
22-10
normally. The state-of-the-art scientific and transport means. Now it can be said with
technical level make it possible to start the assurance that the ekranoplans inevitably will
real development of such power plants iust be a part of the total system of marine
now, but its complication and high cost don't transport means and the technical feasibility
allow to put it on the agenda. of the ekranoplanes of all classes is
sufficiently clear.
It is possible that the further development of
superheavy ekranoplans may lead to the
encouraging results in respect of its real REFERENCES
creation. For example, the hopeful future of
creation of efficient and safe power plants for 1. Aframeev, E.A., "Will the WIG Craft
them is connected with using of hydrogen as Become Transport of the Future?", in "Navy
a fuel for gas-turbine engine. From technical and Shipbuilding Nowadays", Krylov
point of view it is possible to start the Shipbuilding Research Institute, International
development of such conception just now. Conference Proceedings, February 1996,
ppAl-22-l-A-l-22-6.
5. CONCLUSION
It is obvious that the stages of ekranoplans 2. Aframeev, E.A., "Heavy-class Ekranoplans
building development will be corresponded to Make up Global Rescue System at Sea", J.
with the total level of world engineering and Military Parade, #24(28), July-August 1998,
economy and will be determined by real pp93-95
needs of mankind in high-speed marine
22-11
• • •:
• •
00 H/b
0,10,25 0,75 1,25 2,0 3,5 h3%,m
i i i i i— i
L
%
100
90
80
70 h3%,m
2 3 5
■
h3%
m
4
Hydroaeroplanes, h3%
P+F/M
Planes
0,6
0,4
0,3
1st generation
WIGs
0,2
X X X M.tons
100 200 300 400 500 600 700
Mhl/M
Mw/M r
-0,4 WIGs ;
■0,3
* * Planes
0
•0,2 * WIGs
PlaneS
-0,1 fi $ % * *$
— 0 Mhl/M
Mh/M __ js Mw/M
Mv/M Planes
— • Mh/M
0,04
o Mv/M
0,03
WIGs
0,02
° 8
Planes
0,01 8
0,08
WIGs
0,07
0,06
Mw/M
0,19
0,18
0,17
0,16
0,15
0,14
0,13
0,12 Planes
0,11
0,10
0,09
X X X X X X
350 400 450 500 550 600 650 700 750 800M/Sw, KG/M
Fig 7. Influence of wing load upon weight of wings of WIGs and planes.
T/M
1st generation
0,32 - WIGs
0,30 -
C~ Two - mode
0,28 * heavy
0,26
0,24
0,22
0,20
0,18
X
3 100 200 300 400 500 600 700 M.tons
F/P-L, KG/t-km
1,00 -
0,90 - I st generation
0,80
0,70
WIGs
0,60
0,50
I-Two - mode
0,40
0,30
0,20
0,10
j i_
0 100 200 300 400 500 600 700 M.tons
Fig 9. Fuel efficiency of WIGs and planes
140 t
V,m/s
120 + 1,2
IE/M
C/3
fa
fa
o
z
o
o
fa
o
fa
o
z
o
l-H
fa
23-1
and
G. Fantacci
Intermarine S.p.A.
Via Alta
19038 Sarzana (La Spezia)
Italy
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
23-2
the present configuration is to carry 800 passengers and 100 cars and the Split Orifice Flow Model for the prediction of the lift-
at a cruise speed of 120 knots over a distance of 850 km. The to-drag ratio of WIG vehicles. Rozhdestvensky (Ref. [20])
emphasis in the present project is on hydrodynamics, aerody- presented one- and two-dimensional mathematical models for
namics and their interaction, on high-speed waterjet propulsion, the non-linear aerodynamics of lifting surfaces in ground effect.
on the optimization of thick composite structures and on obsta- Other methods apply the lifting line theory, matched asymptotic
cle detection systems. expansions and various implementations of the vortex lattice
As project co-ordinator, Intermarine is involved into all aspects method, see Hooker (Ref. [14]). A more recent example of the
of the SEABUS-HYDAER project. As a partner in the project, vortex lattice method approach can be found in Day and Doctors
NLR is involved in the aerodynamics of the wing of the craft. (Ref. [9]).
Activities in the aerodynamics field have been defined in the Also panel methods have been used for the calculation of the
analysis of the wing and its high lift system in ground ef- flow around three-dimensional wings near the water surface.
fect. However, since the performance of the complete vehicle is Goetz et al. (Ref. [12]) used the PANAIR panel method for cal-
strongly related to the stability and control of the complete ve- culations on a low aspect ratio rectangular wing with end plates
hicle, it is not meaningful to consider the wing isolated from the and trailing edge flap deflection. Chun and Park (Ref. [7]) per-
hydrodynamic components and the propulsion system. At the formed steady and unsteady panel method calculations; the latter
start of the project, such an integrated approach has not yet been showed that the ground effect for a wing above waves is lower
pursued. Therefore it was imperative to perform a preliminary than above flat ground. Kühmstedt and Milbradt (Ref. [17])
design study into the controllability of the complete configura- presented an application of several two- and three-dimensional
tion and postpone the detailed wing design until constraints on potential flow methods to the airfoil and wing design of a WIG
the wing design goals have been achieved from a preliminary craft. Chun et al. (Ref. [8]) presented results of panel method
design study. calculations not only for wing configurations, but also for a
complete WIG craft. Correspondence with experimental results
2 BACKGROUND for the wing configurations are shown to be good, except for
From the beginning of the 20th century it was noted that a wing situations where the wing is in close proximity to the ground.
flying in close proximity to the ground experiences an increase This conclusion has also been drawn in the other panel method
of the lift and a reduction of the induced drag. This phenomenon, references.
called ground effect, has been studied since then, because it com- With the continuing development and application of Euler and
plicated the take-off and landing of aircraft. Already in 1912 Navier-Stokes codes, these methods have started to be used
Betz sought to discover what lay behind this phenomenon, see also in the analysis of WIG configurations; Hirata and Kodama
Hooker (Ref. [14]). A historical overview of several projects (Ref. [13]) applied a Navier-Stokes solver for calculations on a
for Wing-In-Ground (WIG) effect craft until 1980 has been pre- three-dimensional wing with end-plates in ground effect.
sented by Ollila (Ref. [18]). As for cruising conditions, the ground effect plays also an im-
Wing-In-Ground (WIG) effect craft have been under develop- portant role at lower speeds in take-off and landing high lift
ment in various countries, such as China, Germany, Italy, Japan, conditions. Steinbach and Jacob (Ref. [23]) conducted a com-
Russia and the United States. From the 1960's Russia put large bined experimental and theoretical investigation into a high lift
efforts in the development of "ekranoplans" (Russian for Wing- airfoil configuration. They presented the opposite effects of
In-Ground effect vehicles). The central role was played here by ground influence on lift for two-dimensional airfoils with de-
Rostislav Alexeyev from the Hydrofoil Design and Construc- flected high lift systems (lift decrease due to the "forward-wash"
tion Bureau. They developed several large ekranoplans (take-off effect) and for low aspect ratio three-dimensional wings in high
weight from 120 to 550 tonnes). Nowadays there is an extensive lift condition (lift increase due to the image trailing vortices).
amount of open literature available on their research. Examples Filippone and Selig (Ref. [ 10]) investigated with a panel method
are: Chubikov et al. (Ref. [6]), Rozhdestvensky (Ref. [21]) and the ground effect on a thin and a thick airfoil. They showed op-
Kirillovikh (Ref. [16]). These papers show that many aspects, posite effects in extreme ground effect on the thin airfoil (lift
such as aerodynamics, propulsion, stability and control, have to increase) and on the thick airfoil (lift decrease). The reason for
be taken into account in close interrelation in order to obtain this is a different balance of the "forward-wash" effect and the
an efficient transport vehicle. From a United States point of "ram" (stagnation) effect. They also studied three-dimensional
view, the same conclusion was also put forward by Balow et al. effects of the flow about low aspect ratio wings by VSAERO
(Ref. [3]). panel method calculations; these effects are responsible for the
Also smaller scale WIG craft have been developed and built. much lower lift coefficients compared to the two-dimensional
An example is the work performed by Jörg in Germany since case.
the early 1960's on several small craft (Ref. [15]). Most methods for aerodynamic calculations on WIG configura-
In the future an increasingly important role for WIG craft is tions consider the water surface as not deformable. However,
envisaged, mainly for civil applications. This prophecy is ex- the non-uniform pressure field which is created by the wing,
pressed in many publications, see for example Rozhdestvensky alters the water surface and generates a specific wave pattern.
andSynitsin(Ref. [22]). Development of commercial WIG craft In this sense the aerodynamic flow-field and the hydrodynamic
is in progress in Russia, Australia, Germany, China and Korea. flow-field are strongly coupled. Bulgarelli et al. (Ref. [5])
presented a mathematical model to describe the unsteady flow
A wide variety of techniques have been employed to study the about a wing moving in the proximity of the air-water interface.
aerodynamics of WIG vehicles. Ando (Ref. [1]) gave a short As the air-water density ratio is very small, the usual assump-
summary of empirical methods, classical theoretical methods tion is to neglect completely the free surface deformation and to
23-3
consider the WIG craft in flat ground effect. The aerodynamic foils at the mid-chord location.
flow-field and the hydrodynamic flow-field are then assumed to
be decoupled. The take-off foils are vertically positioned in such a way that
As already indicated, stability and control of WIG craft is an im- they are fully submerged in the planing hull state in order to
portant aspect which needs careful consideration. Studies into generate maximum lift to decrease the take-off velocity as far as
this subject are not äs numerous as into the performance of these possible. Applying take-off hydrofoils will allow the take-off
vehicles. Gera (Ref. [11]) discussed the static and dynamic sta- at a lower speed than in case the configuration only features an
bility characteristics of wingships. aerodynamic wing to generate lift. After take-off the vehicle
rises with increasing speed due to the lift force generated by the
The present paper presents the status of the limited preliminary take-off hydrofoils and the wing, such that just after take-off
design of the complete craft and of the aerodynamic analysis of the hulls are already well above the surface surface. Also the
the wing in ground effect. The next section presents the layout V-shaped take-off foils become decreasingly submerged with
of the configuration, as it is defined at the start of the project by increasing speed. At the speed where the vehicle becomes fully
Intermarine. Section 3 deals with the preliminary design inves- airborne the take-off foils are no longer submerged, therefore
tigations. Section 4 shows first results of the wing aerodynamic not contributing to the hydrodynamic drag in the airborne speed
analysis. Finally, section 5 contains preliminary conclusions on regime. The control foils are located at the bottom of the struts
the results obtained up to now. and remain submerged over the entire speed range of the craft.
An example of an existing WIG craft with hydrofoils is the Lip-
3 THE SEABUS CONFIGURATION pischX-114H.
The initial configuration of the SEABUS has been developed Propulsion of the vehicle is provided by a waterjet, located just
and defined by Intermarine. It is shown in figure 1. under the wing at the end of the front hull. The waterjet nozzle
can be tilted from a horizontal position towards a 10 degrees
The wing features a low aspect ratio in order to take advan- downward position. In this way the thrust can also assist in the
tage of the ground effect. The outer parts of the wing can be generation of the required lift force during take-off. The water
folded to reduce the span for manoeuvring operations in har- inlet is located at the bottom of the front strut. The inlet duct
bours. The high lift system of the wing consists of a full-span runs through this strut upwards to the pump which is located
trailing edge single-slotted Fowler flap. Passengers and cargo in the front hull. The thrust is generated by a gas-turbine, also
are accommodated inside the inner part of the wing. The ab- placed inside this hull, which is connected by a gearbox to the
sence of a fuselage is favourable for achieving a high lift/drag pump.
ratio of the configuration. The main characteristics of the SEABUS configuration are:
Three hulls are connected in a trimaran configuration to the
lower surface of the wing; one hull in a forward position in
the plane of symmetry, the two others in an aft starboard and • weight:
port position. The main purpose of the three hulls is to provide - take-off: Wto = 500 ton
buoyancy during floating operations at low speed in harbours
and in take-off and re-entry operations. - payload: Wp = 266.5 ton
• aft strut: are not discussed in the present paper. It has to be realized that
these data are a first iteration. The hydrodynamic behaviour of
- 20 % thick, base vented, parabolic profile struts and foils at the speed range in which the SEABUS is in-
- span: b = 7.40 m tended to operate is an unknown area in which super-cavitation
phenomena play a dominant role. It is exactly one of the goals
- chord: c = 3.33 m
of the present project to obtain hydrodynamic data up to speeds
- sweep angle: A = 0° of 120 knots. The same holds for the propulsion; the flow in the
inlet at high velocities is also strongly infuenced by cavitation.
• front take-off hydrofoil:
This will also be investigated in the project.
- 6 % thick, base vented, curved, parabolic profile The aerodynamic characteristics of the wing and the high lift
system are calculated with well-established preliminary design
- straight-tapered, taper ratio: A = 0.5
methods, taken from Ref. [24]. It also includes corrections for
- aspect ratio: A = 8.26 the lift, drag and pitching moment due to the ground effect. The
- area: S = 40 m2 effect of the deformable water surface has not been taken into
account.
- V-shaped, dihedral angle: T = 30°
• front control hydrofoil: Beside the geometrical data already discussed, a number of
additional data has also to be defined to perform the static anal-
- 6 % thick, NACA 16006 profile ysis. The wing profile characteristics are defined in such a way
- airfeed at 50 % chord on the upper and lower that, at the cruise speed of 120 kts, the wing lift for zero angle
side, for lift variation for control purposes of attack, plus the component in lift direction of the pressure
(Ac(pro/„e=±0.15) related part of the force on the water inlet, equal the weight of
the craft. The inlet features an inclination angle of 30 degrees
- straight-tapered, taper ratio: X = 0.39
with respect to the vertical axis. The wing lift coefficient in
- aspect ratio: A = 5.36 cruise equals: CLwina = 0.39. The wing planform, as
- area: S = 10 m2 specified in the previous section, has been used. The flap/wing
chord ratio has been selected at 0.35. No attempts have been
made to optimize this ratio. The center of gravity and its range
The two aft take-off hydrofoils feature each only half of the are defined; the average location is at 30 % of the wing root
front take-off hydrofoil: at the starboard-side aft strut only the
chord from the leading edge.
starboard-side of the front foil, 5 = 20 m2, at the port-side aft
The height of the wing above the water surface and its angle of
strut only the port-side of the front foil, 5 = 20 m2. Each of
attack depend on the phase of the flight. From zero up to take-
the two aft control hydrofoils features half of the front control
off speed the craft is in the planing hull state where it is planing
hydrofoil area: 5 = 5 m2, at the same values of the other foil on the three hulls. The height above the surface is selected at
parameters. 2.4 meter. In the foilborne state from the take-off velocity up to
In summary we have: the fully airborne velocity the height increases with increasing
speed. The height is determined by the extent of submergence
• total take-off foil area: 40+20+20=80 m2 of the take-off hydrofoils which is required to generate, in com-
bination with the lift on the wing, the total lift on the vehicle.
• total control foil area: 10+5+5=20 m2 The total lift equals the weight of the configuration. Therefore
the height in the foilborne state results from the calculations. In
4 THE PRELIMINARY DESIGN CALCULATIONS the airborne state the height is defined at 9 meter, such that the
A limited preliminary design for the SEABUS configuration take-off foils are above the water surface. This implies that the
has been performed. The aim was to determine the performance craft is flying at a relative flight height of 18.5 % of the mean
of the vehicle and the amount of propulsive power required for aerodynamic chord length. A wing with an aspect ratio between
obtaining static equilibrium with regard to lift, drag and pitching 2 and 3, flying at a relative flight height of 20 % of the mean aero-
moment. Also it has to provide the design lift coefficients for dynamic chord length is considered acceptable for an efficient
the wing, the wing pitching moment limitations in both cruise transport vehicle, see Kirillovikh (Ref. [16]). Rozhdestvensky
and high lift configurations and the angle of attack limitations, (Ref. [20]) states that the relative ground clearance should be in
which will be used in the detailed design of the wing and the high the range of 5-10 % of the mean chord for obtaining efficient
lift system of the wing. This section presents several aspects of WIG craft from the viewpoint of aerodynamics. The seastate
the preliminary design and results of calculations. (waveheight) comes also into the picture, requiring a larger per-
centage.
4.1 Aspects of the preliminary design The minimum take-off velocity is determined by the condition
The lift and drag data over the entire speed range for the hydrody-
that the take-off hydrofoils and the wing carry the vehicle (the
namic components of the SEABUS configuration are provided
front and two aft control foils together are assumed to generate
to NLR by the partners in the SEABUS-HYDAER consortium
no net lift); the value of this velocity is therefore a result from
which are responsible for the design, testing and validation of
the calculations. The minimum airborne velocity is determined
the hulls, struts, take-off foils and control foils. Data on the
by the condition that the wing fully carries the vehicle; the value
propulsion (water inlet and water jet) has also been supplied to
of this velocity is also a result from the calculations.
NLR by the partner responsible for this subject. These data itself
23-5
Finally, the angle of attack of the craft in the planing hull state maximum at the take-off speed due to the wave-making drag
and the foilbome state can only be chosen within strict limits. of the hulls which are no longer submerged in the foilborne
Restrictions are the requirements that the wing trailing edge does state. In the foilbome state the power required is nearly con-
not touch the water surface and that the water inlet and, at the stant, because the hydrodynamically wetted area decreases with
take-off velocity, the take-off foils remain properly submerged. in creasing speed due to the increased height of the vehicle
A limited angle is also a requirement to avoid shifting of the above the water surface. As shown in Table 4, the drag of the
cars in the cargo-space, since it will be in practice impossible configuration decreases in the foilbome state with a factor two,
to fasten all cars sufficiently to the floor. In the present prelimi- while the speed doubles from 35 to 70 knots. In the airborne
nary design calculations the angle of attack has been set at zero speed range, being at a constant height of 9 meter, the power
degrees. In the airborne state the angle of attack results from the required approximately increases up to the overall maximum of
analysis by the condition that at all speeds in this state the wing 90 MegaWatt at 120 kts. This is in contrast to conventional WIG
generates a lift force which equals the weight of the vehicle. craft without submerged components above the take-off speed
for which the overall maximum power required is determined
The hydrodynamic, propulsion, aerodynamic and additional by the wave-making drag at the take-off speed.
data are combined in a computer program in order to calcu- The high value of the maximum installed power required affects
late the performance over the entire velocity range, for mini- the operational efficiency in an adverse way. Ando (Ref. [2])
mum required power, of the complete vehicle based on static devised the non-dimensional "Modified Effective Lift-Drag Ra-
equilibrium of the lift, drag and pitching moment, i.e. only tio" parameter: (WV/P)MCT, where W is the weight, V the
starboard/port-side symmetric conditions are considered. All cruise velocity, P the installed power and Mcr the Mach num-
calculations are made for the take-off weight of the vehicle and ber at cruise. Ando presented plots for this parameter based on
for the average location of the center of gravity. the gross-weight Wg and on the payload Wp for a wide range of
transport vehicles. For the present SEABUS configuration the
4.2 Results of preliminary design calculations numbers are: Wg = 500 ton (gross-weight), Wp = 266.5 ton
Calculations have been performed in order to establish the inputs (payload), Vcr = 120 kts, Pin = ±90 MegaWatt and Mcr =
for the detailed aerodynamic design of the SEABUS configu- 0.18. These data result in two values of the efficiency parameter
ration. The results of the calculations are summarized for a (WV/P)Mcr:
number of velocities in the tables 3 and 4.
The take-off velocity is 30 knots; the presently calculated mini-
mum airborne velocity is 70 kts. At the latter speed the vehicle 1. one based on gross-weight; this value of the SEABUS
can change the foilborne state into the airborne state by simul- configuration equals the value for an efficient helicopter
taneously increasing the height above the water surface and in- in comparison with various other transport vehicles, as
creasing the angle of attack. After take-off the increased height presented by Ando.
above the water surface enables to deflect the wing trailing edge
2. one based on payload; this value of the SEABUS configu-
flap to 10 degrees, and above 45 knots to 15 degrees.
ration exceeds that of helicopters, and equals the value for
In the foilbome state the lift force on the take-off hydrofoils de-
a passenger car and, although operating in a higher speed
crease with increasing speed, while the wing increasingly takes
range, the value for less efficient jet aircraft.
over the function of lift generation device. This is also presented
in Fig. 2 in which the total lift coefficient of the vehicle for static
equilibrium is plotted; also shown are the lift coefficient of the 5 THE DETAILED DESIGN CALCULATIONS
take-off hydrofoils (front and two aft together) and the wing. Detailed wing design activities have recently been started. This
All coefficients are based on the dynamic pressure of the airflow section describes the work performed on the wing profile and
and the planform area of the wing (to allow addition). trailing edge flap profile design and on the analysis of the perfor-
Figure 3 presents the lift coefficient of the wing and the angle of mance of the three-dimensional wing and the high lift system.
attack over the speed range; also the trailing edge flap deflection
angles are indicated. In the planing hull and foilborne state the 5.1 Wing and flap profile design
angle of attack has to remain zero degrees. In the airborne state Recently, detailed analysis of the performance of the wing has
the angle of attack varies with speed in order to achieve at all been started. Based on the lift coefficients for the wing, ob-
speeds equilibrium of forces and pitching moment, while fluctu- tained from the preliminary design calculations, wing profile
ations of the flap deflection angle, which is at its maximum of 15 design can be performed. Initially, Intermarine has chosen the
degrees at 70 kts, are avoided, such that the flap can be retracted Göttingen 1020 profile for the SEABUS wing. A first itera-
smoothly with increasing speed until the high lift system is no tion in this profile design has been executed. Using a two-
longer required at 100 kts. dimensional panel method with ground effect capability, cou-
Table 4 shows the drag on the components and on the entire pled with a boundary layer method, see Ref. [19], a profile with
configuration. Clearly the hydrodynamic components account enhanced lift capability has been designed during an earlier de-
for the largest portion of the drag. The drag contribution of the sign iteration, allowing for larger angles of attack up to seven
wing is 15percentat 120 knots. This is also illustrated in Fig. 4 degrees. Modifications to the original profile relate to the upper
where the drag of the various components is given as percentage side near the nose and the adoption of a flat lower side, see Fig.
of the drag of the entire vehicle. 6, where the profiles are shown at airborne height.
Fig. 5 presents the installed power over the speed range. As for The trailing edge flap profile is based on the single-slotted flap
"conventional" WIG craft, the power required reaches a local of a three element airfoil. The flap/wing chordratio is 0.35. The
geometry of the flap has been modified in order to integrate it in
23-6
retracted position smoothly in the wing profile. The flap/wing h[m] Uoo[kts] «n */[°] cL CD L/D
chord ratio has been kept the same. Fig. 7 shows the flap profile la oo 50 0 15 0.8376 0.0300 28
in retracted and deflected position at a foilbome height. lb 6.15 50 0 15 1.2668 0.0156 81
Since wing analysis is performed with a non-viscous potential oo 85 2.1 5 0.5705 0.0267 21
2a
flow method, a modification to the wing profile with relation 2.1 5 0.7803 0.0165 47
2b 9 85
to the flap cove has to be made in case of calculations on the
3a CO 120 0.1 0 0.2780 0.0160 17
wing with deflected flap in order to prevent unrealistic results
3b 9 120 0.1 0 0.3939 0.0104 38
in the cove. To simulate flow separation in the potential flow
calculations, the cove has been faired smoothly from the profile
Table 1: Wing lift and wing drag coefficents at three conditions
lower side to the trailing edge, see Fig. 7.
from the preliminary design calculations
It is mentioned that the two-dimensional wing and flap pro-
file design, which has been performed up to now, is of limited
scope. Complete wing design activities would require several it-
/«[m] Uoo[kts] «n Sf[°] CL CD,_J L/Di
la CO 50 0 15 0.8357 0.0900 9
eration loops of preliminary design calculations, as presented in
the previous section, two-dimensional profile design and three- lb 6.15 50 0 15 1.4511 0.0670 22
dimensional detailed design, as is presented in the next subsec- 2a CO 85 2.1 5 0.5971 0.0460 13
tion. 2b 9 85 2.1 5 0.9055 0.0429 21
3a CO 120 0 0 0.3027 0.0119 25
5.2 Wing-In-Ground effect calculations
3b 9 120 0 0 0.3824 0.0093 41
A number of aerodynamic calculations on the SEABUS wing
have been performed. The calculational method is the PDAERO Table 2: Wing lift and wing induced drag coefficents at three
panel method, see Ref. [4]. This method is capable of the calcu- conditions from the PDAERO calculations
lation of the three-dimensional inviscid, subsonic potential flow
on arbitrary aeronautical configurations. The method includes
an option for (undeformable) ground effect calculations. effect calculation to 250 seconds for the calculation on the wing
The geometry of the wing is based on the planform which has with deflected flap including ground effect.
been defined by Intermarine and which is presented in section Pressure distributions over the wing and flap, for two spanwise
2. The wing and flap profile are discussed in section 4.1. The locations (rj=y/(b/2) = 0.2 and 0.7) and spanwise lift distribu-
flap/wing chord ratio is constant in spanwise direction. tions are presented in Fig. 12 through 19. At all conditions,
Calculations have been performed at one foilbome and two the ground effect increases the pressure at the lower side of the
airborne conditions (all conditions are starboard/port-side sym- wing, while the distribution at the upper side remains relatively
metric): unchanged. The ground effect is most prominent at the foilbome
condition where the wing is closest to the ground.
1. L/oo=50/fcts,a = 0°,<5/ = 15° The results for lift and drag are given in table 1 and 2. Table 1
contains the lift and total drag from the preliminary design cal-
2. C/oo = 85 kts.a = 2.1", 8f =5° culations, table 2 the lift and induced drag from the PDAERO
calculations. In both tables results without and with ground
3. Uoo = 120kts, a = 0°,6f=0°
effect are presented. Comparison of lift coefficients shows that
without ground effect the lift values are in fair correspondence.
The full-span, single-slotted trailing edge flap is of the Fowler With ground effect larger differences occur for the two high lift
type. The chord extension due to the translation of the flap in conditions.
backward direction is set at 30 % flap chord at 5° flap deflection Comparison of drag coefficients shows that the panel method
and at 50 % flap chord at 15° flap deflection. The extension values, consisting of induced drag only, are higher than the val-
percentages are constant in spanwise direction. For both deflec- ues resulting from the preliminary design calculations in the two
tion angles, the gap between wing and flap is 2 per cent of the high lift conditions, but lower at cruise condition.
wing chord without flap deflection, again constant in spanwise These wing analysis results should be regarded as results in
direction. progress for the SEABUS configuration. The generally fair cor-
The panel distribution is shown in Fig. 8 and 9 for the wing with respondence between the results on lift from the preliminary
retracted flap and in Fig. 10 for the wing with flap deflection. design calculations and from the wing analysis indicate that
Fig. 11 shows details of the wing and flap paneling, including the wing design can be developed further from the present lay-
their wakes, near the wing trailing edge. The wake of the main out. Design of the wing and the high lift system (flap gap and
wing, trailing off in downstream direction over the deflected overlap) is possible through modification of several parameters,
flap, is taken parallel to the upper surface of the flap up to the such as wing leading edge curvature, which define a complete,
trailing edge of the flap. three-dimensional wing.
At each condition calculations have been performed with and
without ground effect in order to investigate this phenomenon. 6 CONCLUSIONS
The number of panels for the clean wing geometry is: 2700 (in A description has been presented of a preliminary design in-
ground effect effectively 5400), for the wing with deflected flap: vestigation into the SEABUS WIG concept which features hy-
4500 (in ground effect effectively 9000). The calculations have drodynamic control surfaces and a water jet propulsion system.
been executed on the NEC SX-4 supercomputer of NLR. Run A computational tool has been developed allowing to calculate
times vary from 55 seconds for the clean wing without ground static equilibrium states with relation to lift, drag and pitching
23-7
moment in a rapid manner, taking severe operational limitations [11] Gera, J., "Stability and Control of Wing-In-Ground Effect
in terms of the angle of attack into account. The preliminary Vehicles or Wingships", AIAA 95-0339, 33rd Aerospace
design results indicate a drag level dominated by hydrodynamic Sciences Meeting and Exhibit, Reno, NV, Jan. 9-12,1995.
contributions and as a consequence a large installed power. The
[12] Goetz, A.R., Osbom, R.F. and Smith, M.L., "Wing-In-
resulting operational efficiency of the SEABUS, based on gross-
j Ground Effect Aerodynamic Predictions Using PANAIR",
weight, equals the value of efficient helicopters; the operational
AIAA-84-2429, AIAA Aircraft design systems and oper-
efficiency of the SEABUS, based on payload, equals the value of
ations meeting, San Diego, CA, Oct. 31 - Nov. 2,1984.
passenger cars and, although operating in a higher speed range,
the value of less efficient jet aircraft. These efficiencies indicate [13] Hirata, N. and Kodama,Y., "Flow Computation for Three-
the need for further design iterations to reduce the hydrodynamic Dimensional Wing in Ground Effect Using Multi-Block
drag level of the SEABUS concept, by reducing the size of the Technique", Journal of the Society of Naval Architects of
struts and the control foils, to increase the operational efficiency. Japan, Vol. 177, June 1995. pp. 49-57.
7 REFERENCES [14] Hooker, S.F, "A Review of Current Technical Knowledge
[1] Ando, S., "Note on Prediction of Aerodynamic Lift/Drag Necessary to Develop Large Scale Wing-In-Surface Ef-
Ratio of WIG (Wing-In-Ground) at Cruise", Proc. of fect Craft", Intersociety Advanced Marine Vehicles Conf.,
FAST'93, Yokohama, Japan, Dec. 13-16,1993, pp. 1561- Arlington, VA, June 5-7,1989, pp. 367-429.
1572.
[15] Jörg, G.W., "History and Development of the "Aerody-
[2] Ando, S., "Some Topics on WIG (Ekranoplan) Design", namic Ground Effect Craft" (AGEC) with tandem wings",
Trans, of the Japan Society for Aeronautical and Space Proc. of the Symposium "Ram Wing and Ground Effect
Sciences, Vol. 39, No. 124, Aug. 1996, pp. 184-198. Craft", The Royal Aeronautical Society, London, England,
May 19,1987, pp. 87-109.
[3] Balow, F.A., Guglielmo, J.G. and Sivier, K.R., "Design
and Evaluation of a Midsize Wing-in-Ground Effect Trans- [16] Kirillovikh, V.N., "Russian Ekranoplans",Proc. of aWork-
port", AIAA 93-3952, AIAA Aircraft Design, Systems and shop on Twenty-First Century Flying Ships, Sydney, Aus-
Operations Meeting, Monterey, CA, Aug. 11-13,1993. tralia, Nov. 7-8, 1995, pp. 71-117.
[17] Kiihmstedt, T. and Milbradt, G., "Aerodynamic Design of
[4] Beek, CM. van, Piers, W.J. andOskam, B., "Aerodynamic
Wing in Ground Effect Craft", Proc. of FAST'95, Liibeck-
Analysis of Slipstream/Wing/Nacelle Interference for Pre-
Travemiinde, Germany, Sept. 25-27,1995, Vol. 1, pp. 597-
liminary Design of Aircraft Configurations", AGARD-CP-
608.
498, Paper No. 6,1992.
[18] Ollila, R.G., "Historical Review of WIG vehicles", Journal
[5] Bulgarelli, U.P., Greco, M., Landrini, M. and Lugni, C, of Hydronautics, Vol. 14, No. 3, July 1980, pp.65-76.
"A Simple Model for the Aero-Hydrodynamics of Ekra-
noplans", Proc. of the AGARD Fluid Dynamics Panel [19] Oskam, B., Laan, D.J. and Volkers, D.F, "Recent Ad-
Workshop on High Speed Body Motion in Water, Kiev, vances in Computational Methods to Solve the High-
Ukraine, Sept. 1-3,1997, AGARD-R-827, Paper 16. Lift Multi-Component Airfoil Problem", AGARD-CP-
365, Paper No. 3,1984; alsoNLR MP 84042 U, 1984.
[6] Chubikov, V, Pashin, V., Treschchevsky, V. andMaskalik,
A., "Ekranoplan: A High-Speed Marine Vehicle of a New [20] Rozhdestvensky.K.V, "Nonlinear Aerodynamics of Ekra-
Type", Proc. of FAST'91, Vol. 1, June 1991, pp. 641-648. noplan in Strong Ground Effect", Proc. of FAST'95,
Lübeck-Travemünde, Germany, Sept. 25-27, 1995, Vol.
[7] Chun, H.H. and Park, I.R., "Analysis of Steady and Un- 1, pp. 621-630.
steady Performances for 3-D Airwings in the Vicinity of
the Free Surface", Proc. of a Workshop on Twenty-First [21] Rozhdestvensky, K.V., "Ekranoplans - Flying Ships of the
Century Flying Ships, Sydney, Australia, Nov. 7-8, 1995, Next Century", Proc. of a Workshop on Twenty-First Cen-
pp. 23-46. tury Flying Ships, Sydney, Australia, Nov. 7-8, 1995, pp.
47-70.
[8] Chun, H.H., Park, I.R., Chung, K.H. and Shin, M.S.,
[22] Rozhdestvensky, K.V. and Synitsyn, D.N., "State-of-the-
"Computational and Experimental Studies on Wings In
Art and Perspectives of Development of Ekranoplans in
Ground Effect and a WIG Effect Craft", Workshop Proc.
Russia", Proc. of FAST'93, Yokohama, Japan, Dec. 13-
of Ekranoplans and Very Fast Craft, Sydney, Australia,
16,1993 pp. 1567-1670.
Dec. 5-6, 1996, pp. 38-59.
[23] Steinbach, D. and Jacob, K, "Some Aerodynamic Aspects
[9] Day, A.H. and Doctors, L.J., "A Study on the Efficiency of
of Wings near Ground", Trans, of the Japan Society for
the WIG Concept", Proc. of a Workshop on Twenty-First
Aeronautical and Space Sciences, Vol. 34, No. 104, Aug.
Century Flying Ships, Sydney, Australia, Nov. 7-8, 1995,
1991, pp. 56-70.
pp. 1-22.
[24] Torenbeek, E., "Synthesis of Subsonic Airplane Design",
[10] Filippone, A. and Selig, M.S., "Low Aspect-Ratio Wings Delft University Press and Martinus Nijhoff Publishers,
for Wing-Ships", AIAA 98-0761, 36th Aerospace Sci- 1982.
ences Meeting and Exhibit, Reno.NV, Jan. 12-15,1998.
23-8
W = v ;hicle weight
SEABUS - lift force
h hulls take-off foils wing
£/«, a=0 lift force lift force lift force a Sf
[kts] [m] [ton] %w [ton] %w [ton] %W CL„ing [deg] [deg] state
water inlet
front hull
water jet nozzle
*-take-off'foils + wing
L take-off foils
*- wing
C
L 5.0-
4.0--
3.0--
2.0--
1.0- -
Uoolkts]-
o.o-
5 jfe id iö iö iö iö fö iö tö i\)o Ao iio
Figure 2: Lift coefficients for static equilibrium over the speed range
CL wing
3.0 idL 5^
4
ö,=25°
•£
2.5--
CL,a[°]
2.0 +
1.5
1.0
0.5 --
0.0--
-0.5--
ö iö iö Iö io~ ^Ö <Jö~ V 90 100 110 120
Figure 3: Wing lift coefficient and angle of attack for static equilibrium over the speed range
23-11
Drag percentages:
wing
-Q-- struts
-X- take-off foils
-A- control foils
tot
75--
^.*
-o-o.
50
l**x*->*
25--
^-Jt ' =
— xxxxxxxxxxx
fe ife iff ifc ife ^ <fc 70 to bo ibo Ao lh
Figure 4: Drag of various configuration components as percentage of the total drag at static equilibrium over the speed range
required
P[MW]7
75-
50-
25-
Vbotkts]—►:
0-
fe ife iö iö iö iö <fe % ifo tö ifeö ho do
Figure 5: Installed power required at static equilibrium (lift, drag and pitching moment) over the speed range
23-12
airborne height: 9 m
water surface
V_
Figure 6: The initial and modified wing profile (of earlier design iteration, allowing angles of attack up to seven degrees) in airborne
state
foilborne height: 6 m
Figure 7: The wing profile (of earlier design iteration) and the trailing edge flap profile in foilborne state
23-13
s V
\\K^N.
^ *Sv ^^S^.
^>x^\ W^^v
TTUMT^
Figure 11: Panel distribution over the wing with extendedflap and the wakes; detail near the flap
23-15
-1.5
Figure 12: Pressure distributions over the wing andflap at r] = 0.2forUco = 50kts,a = 0°,<5/ = 15°, foilborne state
Figure 13: Pressure distributions over the wing andflap at rj = 0.7forU^ = 50kts,a = 0°, 8f = 15°', foilborne state
23-16
Figure 14: Pressure distributions over the wing and flap at 77 = 0.2 for Uoo = %5kts,a = 2.1°, <$/ = 5°, airborne state
-1.0--
Figure 15: Pressure distributions over the wing andflap at r) = O.lforUco = &5kts,a = 2.1°, 5j = 5°, airborne state
23-17
no ground effect
with ground effect
1.50-r
1.00--
0.75--
Figure 16: Spanwise lift distribution over the wing for V'«> = 50kts,a = 0°,<5/ = 15°, foilborne state
no ground effect
with ground effect
Figure 17: Spanwise lift distribution over the wing for Uoo = S5kts,a =2.1°, Sf = 5°, airborne state
23-18
-1.0
Figure 18: Pressure distributions over the wing at n = 0.2 and rj = 0.7 for Uoo = \20kts, a = 0°, Sf = 0°, airborne state
no ground effect
with ground effect
1.50-r
1.25--
1.00--
0.75—
0.50--
0.25--
0.00
to ti tr «b 0I4 ts to öt? ts fa to
Figure 19: Spanwise lift distribution over the wing for Uoo = 120fcts, a = 0°, 5; = 0°, airborne state
24-1
The first commercial ekranoplan "Amphistar" and prospects for the development of
passenger ekranoplans
Dr. D.N. Sinitsyn
Dr. A.I. Maskalik
Technologies & Transport JSC
Sormovskoye Shosse 1-a
603603 Nizhny Novgorod, RUSSIA
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
24-2
It is based on a composite wing of a large extension efficiency the 2-nd generation ekranoplans are just
(A.=4-6) with functional subdividing of its area into as good as the modern subsonic passenger aircraft.
"lifting" and "blowing-under" parts. Such a wing
and maximum reduction of the stabilizing surfaces Thus, at the end of the 20,h century it became
area allowed to achieve the aerodynamic quality of possible for the mankind to implement a new type of
25 - 30, that is two times higher as compared to the perspective transport means that can be used in all
1-st generation ekranoplans; this also helped to seasons - ekranoplans that ensure highly efficient
substantially improve ekranoplan seaworthiness and safe transfer of passengers at speeds exceeding
and amphibian capabilities. 600 km/h, at low (screen) flight altitudes over water
(wave height up to 3m), ground, snow and ice;
The adoption of the new configuration implies a besides they possess amphibian capabilities and
range of problems to be resolved, for instance ability to come out directly to the shore, to move
provision of high precision and safe operation of and to be based on it.
automatic damping and stabilization systems. At the
beginning of the configuration implementation we Besides the "Amphistar" ekranoplan the
developed the first stage of the aerohydrodynamic "Technologies & Transport" JSC has developed the
configuration of the 2-nd generation ekranoplan, Marine Passenger Ekranoplan MPE-200 (Fig.ll).
which still retains stabilizers system, though with The aerohydrodynamic configuration of this
significantly reduced area (Fig.5). ekranoplan is given in Fig.12. It consists of a body
(1), a lifting wing with intermediate plates (2) and a
On the basis of the described configuration the twin-fin tail with high set horizontal tail plane (3).
"Technologies & Transport" Joint-Stock Company In the bow of the ekranoplan there is a starting
involved in development of passenger ekranoplans power unit (4) consisting of two high-bypass-ratio
and formerly constituting a part of the Central turbofan home produced engines NK-87 with
Hydrofoil Design Bureau worked out a wide range steering nozzles (5).
of passenger ekranoplans, see Table 2[1], the first
specimen of which (the boat-ekranoplan At the intersection of tail fins and horizontal tails
"Amphistar") was successfully built and certified there is a flight power unit (6) consisting of two
by the Russian Maritime Register of Shipping. turboprop engines NK-15 with coaxial propellers
AV-90.
In passenger ekranoplan design the company
"Technologies & Transport" made use of flexible To support the craft in amphibious modes and to
pneumo-cylinders as hydrodynamic elements of an reduce the structure workload in take off / landing
ekranoplan. As compared to the stiff analog used in modes the body and the intermediate end plate
the 1st generation ekranoplans the flexible pneumo- lower surfaces are equipped with pneumo-cylinders
cylinders show higher weight efficiency in providing acting as shock absorbers (7).
floating and gliding as well as give an opportunity
to substantially decrease overloads in take-off / The wing of the craft is mechanized with deflecting
landing modes and provide a wide range of flaps (8).
amphibian capabilities.
To control the ekranoplan the upper part of the
In figure 6 there is shown a range of tasks that wing panel is equipped with spoiler-ailerons (9), tail
include optimization of all ekranoplan elements fins - with jar rudders (10,11) and the horizontal
from the point of view of technical capabilities, as tail - with an elevator (12).
well as prospective commercial application.
General layout of the ekranoplan is given in Fig. 13.
On the basis of the defined design tasks we worked In the bow there are a starting power unit, starting
out a scale range of passenger ekranoplans of the equipment and switch gear compartment, towing
2-nd generation. Figure 7 shows their tactical and and anchor - handling gear, aerial post of the radar
technical data. This range consists of ekranoplans anti-collision station, wheel compartment and radio
with a displacement from 10 up to 500 tons, room.
passenger capacity from 20 to 500 persons,
seaworthiness - up to 3.5 m wave height and range - The wheel compartment has working places for a
up to 6000 km. commander, second pilot, flight engineer and
navigator.
The level of technical perfection of these
ekranoplans, their transport, fuel and weight In the bow passenger cabin section the seats are
efficiency in comparison to other vehicles is shown placed on two decks. The upper deck is equipped
in figures 8 - 10. On the conventional Gabrielli - with 81 airplane seats and toilets. The main deck
Kärmän diagram ekranoplans have their own bow section is divided into two cabins: the first one
application "niche", where they are unrivalled in with 32 passenger seats, a kitchen and a buffet and
transport efficiency. In terms of fuel and weight the second - with 43 seats.
24-3
The horizontal tail is equipped with the stern power Fig.4. The 2nd generation ekranoplan
unit. Fuel is held in four fuel tanks in the center aerohydrodynamic configuration.
wing section.
Fig.5. First stage 2nd generation ekranoplan
The undertaken research works testify that aerohydrodynamic configuration.
exploiting these craft for passenger conveyance is
effective and highly promising not just owing to Fig.6. The problems to be solved while design ing
their high speed, seaworthiness properties and ekranoplans.
economic efficiency, but also due to the safety
aspect. High passenger conveyance safety as Fig.7. The 2nd generation civilian ekranoplans.
compared to aviation is provided in an emergency
by the ability of an ekranoplan flying low over the Fig.8. Gabrielly -Karman diagram.
water to execute "soft" landing at any moment.
Fig.9. Ekranoplan fuel effectiveness.
The prospects for the future development of
passenger and cargo ekranoplans depend, on the Fig. 10. Ekranoplan weigh effectibeness.
one hand, upon general acceptance of the second
generation aerohydrodynamic configuration and on Fig.l 1. Marine passenger ekranoplan MPE-200.
the other hand, upon further development of its
potential to increase weigh efficiency, economic Fig.12. Aerohydrodynamic configuration of the
ratio, safety, seaworthiness, to improve overall MPE-200 ekranoplan.
strength and effectiveness of engines, to provide
control system reliability, to decrease the craft Fig.13. General layout of the MPE-200 ekranoplan.
weight etc.
References
Fig. 1
24-5
Tt^/r^VZL^i—J-
Fig. 2
24-6
SlSilÄSS
Fig. 3
24-7
Fig. 4
24-8
Fig. 5
24-9
Ekranoplan design
AHDC and Power unit an« External loads Automatic system Complex equipment fo:
general control system Strength of the Radio communication technical maintenance
configuration construction and navigation
equipment
max P (KW)
effectiveness of anti-
max Hwave (3%) operating in min y (vert, collision measures
sea overload)
stability and conditions providing autonomous simplicity and reliability
controllability min M navigation of maintenance
min Ce (moments)
amphibious Automatic Control
capacity (fuel min G System effectiveness
consump (constr.weight) and safety
load capacity tion rate)
min equipment weight
centre of gravity
location
The design task of commercial ekranoplans is to optimize all the elements from the point
of view of both technical potential and future commercial use
Fig. 6
24-10
Passenger capacity
18 150 250 340 460
Displacement (t)
10 90 210 300 470
MPE-10 MPE-100 MPE-200 MPE-300 MPE-400
Fig. 7
24-11
Transport effectiveness
displacement crafts
IgK hydrofoil, air cushion craft, catamarans
ekranoplans
3- MPE-200
helicopters
2-
sub-sonic aircraft
1— supersonic aircraft
asymptotic
evaluation NASA
lgV
Fig. 8
24-12
Technical comparison of
ekranoplans with conventional
vehicles
Fig. 9
24-13
Weight effectiveness
G useful. G hull.
G vert.tail= G hull.= G useful. = G comm. + G t
G total. G total.
Modern aeroplanes
IL-86
206 91.5 62.9 44.4 30.5
AN-22
225 105.9 70.0 47.1 36.0
DC-10
195.2 90.1 57.2 46.1 29.3
Boeing 707-320C
148 85.7 35.4 57.9 23.9
The second generation ekranoplans have relative hull weight on the level of
modern aeroplanes, hut they are inferior in weight efficiency due to the
installation of equipment for safety in sea operations
Fig. 10
24-14
Fig. 11
24-15
Fig. 12
24-16
GO
24-17
CA
U)
E
CD
c
1c CO CO CO CO CM in q co in
■c
o > d
CD
CO
d d co d
CD
CD I
W
o o o o o o o o o o
(power), t, speed
o o o o
km/h
Max thrust Max CO CO o in i^ o
CM CM CO CM CO co in CO
in10. m
CN o
O) o>° m o"o
d
X
d
X o q
o o o o X
00
x *- i^ o
■<»■
^< in
CM ^ CM CM CM CM S2- CM T-
(h,p)
■** CM II II II II II II II it II II
n
0.
II
0.
CN
II
a.
¥
Q.
co co
a. a.
CO CO m co m co II CO (0
a. o. a. a. a. a. z a a.
o O
o
o §8 o o 2 CO
Power unit
CD CO
a. ■ o o o a>
CM C» o co CO 05 CO CO CO CM
6Q 6 o
S=stern
o ■ I «— 1 1
B=bow
co D
co
1
o r- I l~-
cca.
1^
a. ™ CM
CC I x -i
^ *: 00 CN CM CO
X a: z> < CO CM Z
CM DC m co m co m cö CO GO 5 m cd
o S8 m co O
CM co
O) <0 ^ CM
c "-
CD Q.
A it
O £; IE o
to
d
O o o 00 CM
CN
O
co
1- CO
JD CO
£ » << CN
II ii
<< <<
CM
II
CN
II
<<
CM
II
CM
II
II
II
c
o
■</> O
c CO o> CD"
CD
I CO CD co a> CO
E X LO X o X
m X oo
TJ X
_i
X
q O)
d
X
o"
X
d
X
O)
d
&
cri
CN
■x
si in
v- co
^<
o
X
o
X X X
o
X
en
00
CO
X
q in X
CO X o
>
CD
d d en d ■«r
X
oo cd
O CM CN CM CM en co o> in
(take off weiqht), t
displacement
CO CM CO CO 00 O o in o
o CO
CM CO CD CO 03 in CM
d
Max
CD .22
CD ID
c c <" -9
ü CO <0 Q. S i i ■ ■
Q. Q.
o "> i i ■ ■
la
CO < £ E
iS
c +
JS
Q.
O
c
SS
©
c CO
o Ü
a.
CD o
2 >.
Q. 2
o c co
c CO
5
■ O).
0) Q.
w T3 O
O c c
c CN
Q.
co in
co o
CO ■
CM
i > ■ 9
XL 5 5 5 5
HI w w w CO CO CO
a> - co m oo o
n CN ■* CD i-~ 05
24-18
Seaworthiness in
landing modes
CO in O LO o
CO
Ö CO CO o
1
LO
CM CM
0)| o o o O o
o o o O o
•<* CO o LO o
CO co
ü 2
o
CO
Cruising
vo O o
O
T LO o
speed
CO CO
o LO
km/h
CO
o
Max thrust (power)
t(h.p) B=bow
4-*
Q.
25 55 ^ LO
cq co 55 co 55 co 55
si T- T—
S=stern
CO +*
o X Tf x ^< X X X X
CM v- CM CM •* IN CO CM
CM II II II II II II II II
II m W CO CO CQ W CQ in
z 0. Q. 0. 0. a. a. Q. 0.
N
c
CD
X>
CD l>- in P~ LO r- LO
'c Q. <0
o CO x- 00 T- op Y
CD
-o <*> .. *: ik
CD zz z z z z
I
O
<0
CD
Q
CM
CM
Q
X
CM
X
CM
X
■*
X
CM
X
CO
X
CM
0. 5 m cd cd in m' in CQ CO
Aspect
<^ 00 CO CO LO LO
Wing
Ö
ratio
CN
CO CM LO o
c X X
^< X
CM
X
o X O) X
o
= CO ffl lO
X
CO
CM CO CO
LO
ra c X
q X X
_l X- 00 co
Si
OS
x—
LO co I--
o o o o
o
t} o LO
C CM co1 LO
x—
JS CN
o
1
O o
o
Ö
Q. CO
CO
co
LO
CO P
O
c 5&
2
it —*
O .£ _ o
= ™ 2
0) 0)
B
CO "55 "in CO
a> co t3 o o o o
c c
0) >s
w <2 1— CO
f
5
CO JD
5 co o f* 3 M
o -^ M
O •=:
Ü 5 f*
o 5
w£
ffi c
Passen-
o o o o
gers on
•*r io CO
board
CO
CM CO 1-
O d)
'<? c
g TO
© "55
c © c i_
0) Q CO
Q.
ra O) o
o to
c !c <o £ co 8 S T
co a) ~ co UJ .§ SLU ~ co UJ
Q. ~ co UJ
8
J3 Ü 0)
£ 9- E
< 2 Q.5 2 Q.2 2 Q.S 1 i
2 Q. 2
re © (•) - CM CO ■* LO
25-1
1. INTRODUCTION
Development and construction of ekranoplans ne-
cessitated solution of challenging problems of aero-
dynamics of vehicles, advancing near interface of
two media : water (snow, ice, ground) and air.
During the period of development of practical
prototypes of ekranoplans "Orlyonok", "Lun", "Am-
phistar"( Ref. 1-5) in Russia there were created
basics of theory of motion of these vehicles as well
as experimental facilities and installations for the
purpose of solution of principal problems of aero-
dynamics and hydrodynamics of ekranoplans. At
present time, when the first Russian commercial
ekranoplan "Amphistar" ( principal designer D.N.
Sinitsyn ) has been built (Ref. 5 ) , the state of the
art in aero- and hydrodynamics of this high-speed
craft can be characterized in the following way
2. AERO-HYDRODYNAMIC CONFIG- Fig. 1 Aerodynamic configuration of ekranoplan
URATION (AHC) of the 1st generation
Aero-hydrodynamic configuration of ekranoplans Implementation of this AHC secured the follow-
of the first generation ( Figure 1 ), which was ing principal features of ekranoplan:
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
25-2
• Detachment from water ( snow, ice, ground ) In the eighties D.N. Sinitsyn proposed an AHC
with further acceleration into a cruise flight; of an ekranoplan of the first stage of the second gen-
eration (Ref. 1,4), see Figure 4. This AHC employs
• Stable ( without use of automatic control sys- a scheme of a "composite wing" with functional sub-
tem ) cruise flight in ground effect; division of wing area into a "lifting" and "power
augmented" parts. Introduction of such a concept
• Lift-to-drag ratio L/D = 15 — 17; enabled to considerably reduce the area of horizon-
tal stabilizer (tail) as compared to the ekranoplans
• Seaworthiness ( as capacity to take-off and of the first generation, increase lift-to-drag ratio up
land in conditions of rough seas) up to hw5% = to 20, improve seagoing and amphibious properties
2.5-3.0m; of the craft.
• Good characteristics of steerability and manou-
verability in all modes of motion;
K /
\ /
1 IV II IV III
1' 1
12 16 Fr
r
20 Fig. 3 Alekseev's flying wing configuration.
On the basis of the aforementionned AHC the
Joint-Stock Company "Technology and Transport"
developed a scale series of commercial ekranoplans
Fig. 2 Characteristic drag curve of ekranoplan of with technical parameters as presented in Table 1.
the 1st generation ( I - floating, II - planing, III - One can see from this table that the ekranoplan of
flaring in ground effect, IV - intermediate regimes ) the second generation MPE-400, comparable in its
main dimensions to the ekranoplan of the first gen-
eration "Lun" at a similar cruise speed of 500 km/h
Figure 2 presents characteristic drag curve of will have range about 6000 km ( against 2000 km
an ekranoplan of the first generation plotted versus of "Lun") and (take-off) seaworthiness of 2.5-3.0 m
displacement Froude number Fro ■ wave height (against 2 m of "Lun").
In the seventies R.E. Alekseev proposed an aero-
hydrodynamic configuration of ekranoplan of the
second generation (Ref. 1) of the "flying wing" 3. AERODYNAMICS
type, see Figure 3. Due to increase of the aspect
ratio up to A = 4 - 6 and use of 5-shaped wing sec- Developed as a result of large effort of Russian and
tions lift-to-drag ratio of such a configuration can foreign researchers (Ya. M. Serebriysky, R.E. Alek-
be augmented up to K = L/D = 25. Therewith the seev, Ya. I. Voitkunsky, G.A. Pavlovets, V.K. Tresh-
ekranoplan based on such a configuration can oper- kov, A.N. Panchenkov, V.N. Treschevsky, L.D. Vol-
ate far from the underlying surface ( out-of-ground kov, K.V. Rozhdestvensky, V.P. Shadrin, T. Nishi-
effect) still retaining sufficiently high lift-to-drag ra- yama, S. Widnall et al.) the theory of wing motion
tios of 10-12. During development of the "flying in ground proximity enables ( at conceptual and
wing" configuration there were encountered serious preliminary stages of design) to solve practical tasks
problems of hydrodynamics and take off (landing) of choice of the geometry and reciprocal location of
which still await their solutions. lifting elements of the AHC of ekranoplan as well as
25-3
1.6 \ ea
1.2
0.8
0.4
e Ac
0.0
I
0.0 0.2 0.4 0.6 0.8 1.0
The influence of the angle of attack a0 at large ßg ratio of momentum of gaz blown under the wing
distances from the ground has but a weak ef- to that in the jets under the wing.
fect upon derivatives c* and m*; for smaller
ground clearances augmentation of a0 results
in reduction of c* ; derivative mf increases
with increase of a0 for wings of large aspect
ratio and diminishes for wings of small aspect
ratio.
PAR lift
tilted thrust
4. HYDRODYNAMICS nozzle
PAR drag
On the basis of research, conducted by Russian re-
searchers (R.E. Alekseev, A.I. Maskalik, V.K. Dy-
achenko, E.A. Kramarev, E.I. Privalov et al.) there
was created theoretical and experimental knowledge
base relevant to modes of motion of ekranoplan in
contact with water, which can be assigned to "hy-
drodynamics" of wing-in-ground effect craft. As a
matter of fact, according to the concept of motion
of the developed ekranoplans, power augmentation Fig. 7 Scheme of generation of lift force in power
(blowing under the wing) is enabled before they augmentation mode.
start to move, i.e. at zero speed.
Presence of jets of the engines under the wing
changes the character of interaction of the ekra- Using the above formulae one can evaluate the
noplan with the underlying surface. Among the efficiency of power augmentation as a function of
most important results, obtained from theoretical the geometry of the wing (/, 6), angle of inclination
and experimental investigations of ekranoplan's mode of the wing with respect to the hull (<p), position
of motion in contact with water, one can single out of the engines (9e,ßg), location of the flap above
the following the ground (h), number of functioning engines n
Generation of considerable lift upon the wing and mode of their operation (T, u), pitch angle of
subject to power augmentation by means of blowing ekranoplan ip. Magnitude of real lift-to-drag ratio
the exhaust of the engines underneath (Figure 7) is in PAR mode of the developed ekranoplans is K =
largely due by stagnation of the engine jets, blown Y/X = 16 - 17.
between the wing and the ground.
Development of mathematical models of motion
Figure 8 shows dependence of nondimensional
of an ekranoplan in this mode requires knowledge
magnitude of average pressure under the wing p/pu2
of its hydrodynamic characteristics with account of
(u velocity of gaz at the engine nozzle exit) upon
blowing (H + B)
relative (equivalent) ground clearance
brh
heq = 0.42
lib
+ 1.89
""t+f] (3)
L-UHB
_ %YH+B
pV2S
_ 2XH+B
pV2S '
,„*
0.04
\A
0.02 X? •
0.00-
0.00 0.04 0.08 0.12 ^ 0.16
-0.4
-0.8 -
Fig. 8 Average pressure p/pu2 in PAR generated
cushion versus equivalent ground clearance heq
MZo
MZB = Tb (11)
are correspondingly nondimensional drag, lift and The methodology of research enables to devise
longitudinal moment due to power augmentation, an optimal power augmentation system for an ekra-
noplan in each particular case. The efficiency of
2T power augmentation is, to a considerable extent, de-
ct - 2
(12)
PV S termined by achievable minimum speed of detach-
ment of the ekranoplan from water. Reduction of
coefficient of thrust (T) of the engine.
the speed of detachment of the craft from water
Figure 8 shows nondimensional hydrodynamic
brings about reduction of maximum external loads
coefficients versus clearance. With increase of the
acting upon an ekranoplan, i.e. eventually the ve-
ground clearance h magnitudes of hydrodynamic
hicle's economic efficiency and seaworthiness.
forces vanish. The factors influencing cVH, cXH, mZH
are parameters h and >p which determine the wet-
ted area of the ekranoplan. Investigated in the Figure 10 presents dependence of detachment
wide range of variation were the influence of po- speed of ekranoplan on its take-off mass for differ-
sition of the nozzles of PAR engines, aspect ratio of ent efficiencies of power augmentation. It can be
the wing and loading of the power augmented area seen that more efficient blowing (at 8j = 10 — 15°)
of the wing upon efficiency of the blowing under the reduces magnitude of the speed of detachment from
wing, in particular, upon lift-to-drag ratio in PAR water.
25-7
5.2 Lateral motion of ekranoplan These equations can be used when solving prac-
tical problems of lateral dynamics of ekranoplan.
A distinguished feature of ekranoplan as compared Conditions of lateral stability of ekranoplan at small
to the airplane in lateral motion is considerable de- relative ground clearances have the form
pendence of lateral moment coefficient mx on ground
mßy <0 (26)
clearance h, see Figure 11.
condition of oscillatory course stability
+ i<°
m,
loading * (27)
mn — m- (28)
2/i
m~ <0 (29)
r
V m*0
loading * 6. EXPERIMENTAL METHODS
b) -i 1 1 r
math modelling
• • full-size trials
0
-» » » •—«~
Fig. 11 Principle of natural lateral stability of •5
ekranoplan in heel when moving near the under-
lying surface (a - without heel, b - with heel)
As a consequence, ekranoplan, differently from
the airplane, has natural self-stabilization capabil-
ity near the ground.
The following expression was obtained for the
derivative rrQ.
ml = A| (24)
Equations of lateral motion of ekranoplan have the
following form
m„
7 2/J7 Fig. 12 Transitional processes due to perturbation
of longitudinal motion of an ekranoplan of 1201.
P a Among experimental methods of investigation of
m£/? = 0
IT, stability those making use of radio controlled and
piloted self-propelled models as well as full-size tests
"--(f4y-e^- are viewed as the most reliable. Radio controlled
models provide an opportunity of collecting neces-
sary data without endangering the lives of test pi-
(25) lots. When making use of such models, controlled
25-9
Range (km) 400 1000 1000 1500 1800 3000 4500 6000
26-1
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
26-2
2.3 Impact and Planing study the effect of the aft lines and transom shape
Subsequently, Theodore Von Karman (1929) realized on the resistance of fast naval ships, for which high
that the flow about a nearly flat body impacting speed data (substantially beyond the wavemaking
a horizontal water surface at high speed was also peak) existed for David Taylor Model Basin Hull
essentially dipole-like in the near field, and expressed Series (64). They were able to predict high speed
the vertical loads, like Munk, in terms of added mass resistance very well, and showed that it was related
variation in time. This comprised the first water to a vortex wake produced in the shadow of the
impact theory. Wagner (1932) justified formally Von maximum beam of the ship. This wake is akin to
Karman's work and pointed out that the impact the wake which originates at the trailing edge of an
theory could be applied to slender planing surfaces extended delta wing, i.e. a slender wing on which
(a narrow delta, for example), thereby combining the trailing edges are swept back from the maximum
the ideas of Munk and Von Karman, in the process span, creating a backward facing extension to the
neglecting the generation of waves by the planing rear of the wing. This problem had been treated by
surface. This kind of planing theory was used aerodynamicists using slender wing theory in the
by flying boat hydrodynamists in the US during 40's and 50's. In their work, Tulin and Hsu satisfied
W.W.II, where seaplanes played a distinctive role. exactly the boundary conditions on the hull in each
Its feasibility depended on the very high forward transverse plane, but the free surface condition was
speed of the seaplane. only satisfied in the linear approximation on the
horizontal plane in each 2D section.
On a slender planing surface, part of the energy put
into the water (drag ■ forward speed) is due to lift
and is lost in the wake, and another part is lost 2.6 Slender Bodies Maneuvering
in the spray. Tulin (1957), used 2D + t theory for In all of the cases mentioned above, except impact,
slender planing surfaces and showed how to mini- the bodies are considered in a steady homogeneous
mize the energy lost to spray through camber of the flow and in a fixed orientation. But these restric-
surface, he also showed that for a flat or presumably tions are not always necessary, and by the early 60's
slender surface, the energy loss in the wake and hydrodynamicists and aerodynamicists interested
due to spray were always equal, but not in general in the stability and control of slender bodies like
when the body is cambered. The theory included planing surface, projectiles, missiles, submarines,
consideration of non-linear effects at the wet chine, and torpedoes, had applied the Munk theory to
where a singularity appears in Von Karman's model, the calculation of stability derivatives which arise
but which bifurcated into a stagnation line (inboard during turning, maneuvering, and oscillations.
of the chine), and a spray jet (outboard of the
chine) in Tulin's model. Similar approaches based 2.7 Ships in Waves; Strip Theory
on 2D + t theory, including the advantages of high In the early 50's, in a very fruitful application,
resolution numerical computation are still in use Korvin-Kroukovsky (1956) and Lewis (1955) of the
today: Takagi & Niimi (1990), Vorus (1996), Zhao Stevens Institute applied cross flow theory to the
et al (1997), Fontaine & Cointe (1997). Gravity is prediction of the load induced by the passage of
neglected. ships through waves. In this theory, the flow at
each transverse section is treated as 2D with the
2.4 Slender Wings Including Separation transverse velocity due to the wave providing the
The use of Munk's 2D + t model was also extended dynamic excitation. This has always been called
to slender wings, fully wetted, like deltas, following "strip" theory. The transverse forces calculated
W.W.II, including supersonic flows, by R.T. Jones on this basis include hydrostatic (Froude-Krylov),
(1946) of NACA Langley, and somewhat later, added mass and damping due to the generation of
Clinton E. Brown, also of Langley, introduced a 2D beam-wise waves. This 2D + t (strip) theory, ig-
model with a vortex above each of the two leading noring the interaction between different parts of the
edges of the delta, on the lee side, in order to ship, has been applied now for about fifty years, very
model the effect of the flow separation which was successfully, and has only lately been supplanted
commonly observed on the delta. In a related by more advanced analytic and numerical theories.
problem, Tulin (1959) used the 2D + t theory to Its success is extraordinary, considering that it is
approximate the effect of leading edge cavitation for also based on the linearizing assumption of small
slender delta wings operating at high speed in water, waves and small motion. In the conventional strip
a flow already studied experimentally by Reichardt theory, the forward motion of the ship enters the
at Göttingen. theory only through the kinematical effects created
by the ship's encounter with wave trains, where the
2.5 High Speed Displacement Ships relative motion of the ship and waves determines
In principle, the problem of displacement ships in the frequency of forcing, etc. The very visible and
the high speed limit (very high Froude number) important effects created locally at the bow of
can be treated by 2D + t theory, and this was the ship as waves are first encountered depend on
undertaken by Tulin and Hsu (1986) specifically to the widening of the hull; this effect is specifically
26-3
neglected in comparison to that due to waves and review of Tulin (1976). This means that there
wave induced motion of the ship. In this respect is a small region of the flow centered about the
strip theory is more approximate than 2D + t theory bow, within which the behavior of the flow is very
in its treatment of forward motion effects. different from the more regular flow surrounding
both it and the rest of the ship. This region grows
2.8 Bow Interaction with Waves; Deck Wet- with increasing speed. For a fine ship at low speed
ting; Non Linear 2D + t Theory. its size scales with the wave number g/U2, where
In practice, the visual motion of a ship in stormy U is the ship speed, Tulin & Wu (1996), and when
weather is comparable to the ship draft, and the its size approaches the ships draft, h, it more likely
effects of this motion combined with large waves scales as hU/^/gH, Fontaine & Cointe (1997).
requires non-linear theory for the wave response
near the hull. In the late 80's, H. Maruo, then at the 2.11 Splash and Divergent Waves.
Ocean Engineering Laboratory, UCSB, showed the The splash commonly seen at the bow of fine ships
applicability of a non-linear 2D + t theory involving is the central physical feature arising in the region
approaching waves at the bow, ship motion, and of flow non-uniformity. This splash would seem
appropriate non-linear boundary conditions in the to be an impact-like phenomena, already observed
free surface, including the effect of gravity, Maruo in the case of slender planing surface and treated
(1990). This method was subsequently implemented there by the 2D +1 theory in the absence of gravity.
successfully, Mauro & Song (1993; 1994) and Wu In the case of ship splash, the rising water is
(1997), utilizing high speed, high resolution numeri- eventually halted by the action of gravity and falls,
cal computation. in the process creating a divergent wave crest which
radiates out away from the body. This may then
2.9 Early Effects with Gravity be followed by at least one more divergent crest.
The work of Maruo et al, did not represent the The entire process has been on mapped out using
earliest attempts to include gravity effects in 2D +1 high resolution 2D + t numerical computations,
calculations. The flow around a wedge shaped bow Wu (1997), Tulin & Wu (1996). The theory seems
of a very small angle, was treated analytically using highly applicable both near the bow and for the
the 2D + t approach and linearizing the free surface entire divergent wave pattern, as the divergent
condition, Ogilvie (1972); and Chapman (1975) features calculated in this way match in great detail
again used the linearization to study a surface the same features as calculated by general ship wave
piercing plate oscillating in yaw and sway, with good codes, the notable difference being that the 2D + t
experimental agreement. He went on, Chapman calculations can follow the divergent waves through
(1976), to calculate the steady loads on a plate at to breaking, while breaking never occurs in the low
incidence, but using the full non-linear boundary resolution general codes, Tulin & Wu (1996).
conditions within the 2D + t approximation, again There were earlier pioneering attempts to use 2D +1
with good agreement with experiments. to study the splash on a hull. Fritts et al (1988)
and Calisal & Chang (1989) used a fully non-linear
2.10 Ship Waves numerical method; they partially succeeded, but did
The idea of applying 2D +1 theory to the prediction not have sufficient resolution to accurately predict
of wave generation by a shape (calm sea) was taken this splash.
up in the early 50's by W. Cummins of the David
Taylor Model Basin in his Ph.D. dissertation at 2.12 High Speed Ships and the Future
American University in Washington. He concluded As the ship speed increases, the entire inner flow
that only the divergent waves could in this way be is occupied by splash and its consequences, wave
simulated. This opened up the prospect to treat breaking, the appearance of a wake behind the
high Froude number wavemaking effects, dominated maximum beam, and the appearance of a rooster
by divergent waves, using 2D +1 theory. Even this, tail behind the ship, as mapped out by Wu (1997),
however, has taken a very long time to be realized. see Figure 4. The missing transverse waves become
For normal ships, the transverse waves (crests insignificant in the near field. At this stage the
normal to the ships direction) are responsible for use of 2D + t theory to predict the resistance and
most of the ship's wave resistance, and effective motions of high speed ships comes into its own,
non-linear methods exist for their computation. Faltinsen L Zhao (1991), Fontaine & Cordier (1997).
The divergent waves have been and still are much Its use for slamming loads, including hydroelastic
more poorly understood. Since they are waves of effects, has begun, Magee & Fontaine (1998), and
higher frequency, it is natural that they arise from would seem most appropriate.
regions of rapid change in the flow field, notably As the ship speed increases, the region of greatest
the region of the ship bow for fine ships. Here interest in the flow shrinks down to the immediate
non-linear complexity abounds. It has been known region of the hull and the wake beyond it. Very
for some time that the mathematical solution is high numerical resolution is required to map out
non-uniformly convergent near the bow, see the the processes in this region, see figures 4-10. In
26-4
these calculations the equivalent of 106 grid points in a rational way. To improve the situation, formal
were used. This is very far beyond the capability of small perturbation expansions came into use, and
today's 3D ship codes, which are therefore unable eventually the development of formal asymptotic
to accurately resolve splash and breaking. The methods as described by the CalTech school in the
situation worsens for 3D codes if the ship's motion early books of Cole (1968) and van Dyke (1975). An
is unsteady or into waves, whereas the 2D + t example of the successful use of small perturbation
calculation is not basically modified, Maruo (1990). expansions in hydrodynamics is in the case of
It seems increasingly obvious that the 2D + t supercavitating flows in 2D, Tulin (1964), where
method will become heavily utilized for the study engineering results of considerable utility were given
of a wide variety of problems connected with the through second order. Later, formal asymptotic
growing subject of high speed ships, and especially methods were used to great effect in the case of high
since the prediction of hydrodynamic loads grows in speed water entry by Cointe (1991).
importance. The generation of waves by ships involves consider-
able mathematical complexity and until now formal
3. THE FORMAL MATHEMATICAL asymptotic methods, although applied, have failed
APPROACH TO 2D + t to result in an innovative useful theory. Meanwhile
3.1 Formal Asymptotics useful ad-hoc theories abound: Guilloton, Neuman-
The history of approximation theories in aero- Kelvin, Dawson (Slow Ship), fully non-linear
hydrodynamics is marked by the successful applica- Computational. All of these essentially capture the
tion to engineering problems of ad hoc approximate mechanics of the generation and propagation of
theories at the hands of highly skilled innovators, the transverse waves in the normal' range of ship
for example: Michell, the Göttingen School, Munk, operation. They fail, however, to capture adequately
Von Karman, Wagner, and later Korvin-Kroskovsky, divergent wave phenomena.
Dawson, Maruo, and others. The validity of these As we have noted, the ad-hoc 2D + t theory does
theories has always rested on comparison with seem to capture the mechanics of the divergent
experiments, as in the entirety of science. It has waves, including splash, and not only at high
to be added that even when not suitably correct, speeds.
approximation theories have often led to enormous Very helpful asymptotic analysis of the situa-
physical and often mathematical insight into the tion, Fontaine (1996), Fontaine & Cointe (1997),
actual mechanics of the flow processes. Fontaine, Faltinsen & Cointe (not yet published),
There are at least two common difficulties with ad have recently been carried out. They justify the
hoc theories: (i) it is difficult sometimes to know 2D + t approximation using matched asymptotic
the best, most consistent form of theory; (ii) it is expansions. Using the slenderness of the body as
often very difficult to improve the approximation the perturbation parameter, these analyses lead to
26-5
4. EXAMPLES OF COMPUTATIONS
The near-field problem shown in fig.2 is solved
numerically using a Mixed Eulerian Lagrangian Figure 4: Free-surface elevation around a Wigley
method. Details of the numerical schemes can be hull, 6/L=0.05, h/L=0.1, FL=0.30, from Tulin &
found in Maruo & Song (1994), Fontaine (1996). Wu (1996).
The aim of this section is to demonstrate the
capability of the method to perform high resolution
simulations of nonlinear flows where phenomena results of Raven (1996) for the wave pattern gives
such as jet formation and wave breaking occur. confidence in the general pattern obtained by the
2D +1 theory.
4.1 Steady Flows Compared to a fully three-dimensional approach, a
4.1.1 Wave pattern big advantage of the 2D+t approach is that the mesh
Figure 4 shows a typical picture of the wave pattern size can be refined so that really high-resolution
obtained by 2D + t computation, corresponding to simulations are achieved. As can be seen from fig.4
the flow around a Wigley hull. The main features through 8, diverging waves may be very steep and
are the presence of diverging dispersive waves that may eventually break as confirmed by observations.
originate both in the bow and the stern regions. In Their evolution is therefore influenced by nonlinear
the latter case, they seem to be generated by the effects, which require high resolution simulation to
relaxation of the "rooster tail". be properly taken into account. Another advantage
A general discussion of the generation mechanism of the 2D + t approach is that highly nonlinear
and characteristics of these diverging waves is given phenomena such as jet formation and wave breaking
by Tulin & Wu (1996). Although the transverse can be simulated.
waves are not present in these simulations, com-
parison with the three-dimensional fully nonlinear
26-7
vx1 xV-V;^?;i'(V:?>'/J,''/';Y>':
MMNMRP U = 3 m/s
w
Figure 5: 2D + t simulated bow wave by frigate hull,
FL = 0.5, from Tulin & Wu (1996).
m/s
A/h=0.11
A/h=0.22
■3
I
A/h=0.33
\!
A/h=0.44
Maruo H. & Song W., 1994. Nonlinear Analysis of Wagner, H., 1932. Über Sto/3 und Gleitvorgänge an
Bow Wave Breaking and Deck Wetness of a High- der Oberfläche von Flüssigkeiten.", ZAMM, Vol.
Speed Ship by the Parabolic Approximation, 20th 12 ,pp. 193-215.
Symposium on Naval Hydrodynamics, University Zhao, R., Faltinsen, O.M. & Haslum, H. A., 1997.
of California, Santa Barbara. A simplified nonlinear analysis of a high-speed
Michell, 1898. The wave-resistance of a ship. Philo- planing craft in calm water. Fourth International
sophical Magazine, Serie 5, Vol. 45, 1968. Conference on Fast Sea Transportation, Sydney,
Munk, M.M. 1924. The aerodynamic forces on air- Australia.
ship hulls. NACA Kept. I84.
Ogilvie, T.F., 1967. Non-linear High-Froude-
Number Free-Surface Problems, Journal of En-
gineering Mathematics, volume 1, p.215-235.
Ogilvie, T.F., 1972. The wave generated by a fine
ship bow, Ninth Symposium Naval Hydrodynam-
ics, Vol. 2, pp. 1483-1525.
Raven, H.C., 1996. A solution method for the non-
linear ship wave resistance problem. PhD disser-
tation, Delft Technical University.
Song, W and Maruo, H., 1993. Bow impact and deck
wetness: simulations based on nonlinear slender
body theory. Proc. IS OPE conf, Singapore.
Takagi, K., Niimi, A., 1990. A Theoretical Approach
to Bow Deck Wetness of a High-Speed Ship, Jour-
nal of Ship Research, No. 3, Sept., pp. 163-171.
Tulin, M.P., 1957. The theory of slender surfaces
planing at high speeds, Schiffstechnik Bd. 4, Heft
21, pp 125-133.
Tulin, M.P., 1959. Supercavitating flow past slender
delta wings, Journal of Ship Res., vol. 3, no. 3,
pp. 17-22.
Tulin, M.P., 1964. Supercavitating flows, small per-
turbation theory, Journal of Ship Research, vol.
3, no. 3.
Tulin, M.P., 1976. Ship Wave Resistance, A Survey,
In Proceedings of the 8th U.S. National Congress
of Applied Mechanics, pp. 217-246.
Tulin, M.P., Hsu C.C., 1986. "Theory of high speed
displacement ships with transom sterns, J. Ship
Res.,30(3):186-193.
Tulin, M.P. Wu, M., 1996. Diverging Bow
Waves, 21t/l Symposium on Naval Hydrodynam-
ics, Trondheim, Norway.
Van Dyke, M., 1975. Perturbation Methods in Fluid
Mechanics, The Parabolic Press, Stanford, CA.
Von Karman, T., 1929. The impact on seaplane
floats during landing, NACA Tech. Note No.
321.
Vorus, W.S., 1996. A flat cylinder theory for vessel
impact and steady planing resistance. Journal of
Ship Research, no. 2, pp. 89-106.
Ward, G.N., 1949. Supersonic flow past slender
pointed bodies. Quart. J. Mech. Appl. Math.
2, Part 1, 75-97.
Wu, M., 1997. Prediction of deck wetness, divergent
bow waves on fine ships-nonlinear numerical stud-
ies. Ph.D. thesis, University of California, Santa
Barbara.
27-1
Dmytro I. Cherniy
Department of Cybernetics, Taras Shevchenko National University
64, Vladimirska St., 252601 Kiev, Ukraine.
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
27-2
where W0 =W((O0,t) is a velocity of filling of <D(z, 0 = O(co, /), V(z,t) = V + (&, t),
the moving boundary with new liquid particles in a
boundary point C00. where the singular integrals are regarded in terms of
principal value of Cauchy.
For the determination of the pressure function in any
point Z of D a presentation of a Cauchy- In view of analyticity of the z(t) , V(z,t) func-
Lagrange's integral may look like: tions, for an operator of an individual derivative it is
true that:
^ = _R>^+I^1I^_G(Z)) (1-3)
P 1 dl 2 J -V(z,t\k>\ (1.12)
where for a complex conjugate and complex velocity
dtk Z(
° D tk-
in a point z e D* a following presentation is justi- where
fied:
D_ (1.13)
8<X>(z.t) 1 ef(v(t),t)d&{t) (1.4) — + V(z,t)^: + V(z,t) —
J'(-.0:
& 2;r/ ■-«(/)
Dt dt dz dz
Since all the kinematic and dynamic characteristics
V(z,t) = V(z,t). (1.5) of the flow inside the domain are determined through
the boundary integrals, the problem (1.6)-( 1.11) is
Let a boundary L of a closed domain D is a set stated on the boundary L =dD , at that the pres-
of sectionally Lapunov's curves and in every point
Z = CO of the responding curve tangent sure P((0,t) is regarded as specified in all its
a{a) a(B) points and determines external effect on the bound-
T(ffl) = e' and normal «(©) = ze'
ary. A potential of mass forces is an analytical func-
basis vectors are defined, where Ot(co) is an incli- tion given in D .
tation of a tangent to the OX-axis in a point CO of
For the particle trajectory in the vicinity of its initial
the boundary L . position a Taylor series expansion is valid:
In this case an impenetrability of solid boundaries in
a point CO 0 may be presented as: *) = S £Ü^o)
k\
(,_,)4 { o)
(1.14)
AW
representations (2.1), (2.2). where density f is a jZ^zZT*^ ->N \ *;»*■,
■"^Wki? ^.L*T
sectionally continuous bounded function on a
D
\^ \LII -dtl0
w
\ Lite,]
X
smooth Z.then 3m and 3*^0 }m and "für,}/«
(coherent partitioning of the curve L ) that for
Vz G D \ Q the following estimations are 3. Consideration of kinematic effects
valid:
Since the maximal over module values of velocity
(3.3)
\Lk\< min - max \W\ / max . 1/max IK (co ) ^.-" r-'ü
k = \,..., m .
1 1 (3.4)
a « niaxmm< A, mm ma\\W\ / max
K(/)'2 M-
STEP ■ 13
'
A J\
■■■■ \ _J :-
:-^r--■■:;_■-■ ; -----^
STEF = 21
J
■S?\ %<y
I
1 iTLP • «
CO,. -co. cor -CO,+iU = l,.... (4.2)
>
are observed.
\ J References
1. M.A. JlaBpeHTbes, B.M. LIIaöaT
MeTonw TeopMM (|)yHKUHR KOMnjieKCHoro
nepeMennoro. M.: "HayKa", 1987, 688 c.
2. H.H. MycxejiHLUBHJiH
CHHryjiapubie HHrerpajibHbie ypaBHeHHH. M.:
THOMJ1. 1962.600 c.
3. 0.JX- TaxoB
KpaeBbie sanami. M.: "HayKa". 1977. 639 c.
4. A. H. HeKpacoB
CoSpaHHe coMHHeHHfi, T. 2, M: H3aaTejibCTBo AH
CCCP. 1962.708 c.
5. H. E. KOHKH
Co6panHe coHMHemiH. T. 2. M.: H3naTejrbCTB0 AH
CCCP. 1962.584 c
6. Jl. H. CeiiOB
njlOCKHC 3aiiaMM rHflpcaHHaMHKM H a3pOJIHHaMHKH.
M.. «HayKa». 1980,448 c.
7. r.B. JlOrBHHOBMM
TMiipOUHHaMHKaTeHeHHtl CO CBOÖOÜHblMH rpam-i-
uaMH. KHCB. HayKOBaayMKa, 1969. 216 c
8. G. Birkhoff. E. H. Zarantonello
Jets. Wakes and Cavities. Academic Press Inc. Pub-
lishers, New York, 1957.
9. B.H. MoHaxoB
KpaeBbie 3a.aaHn co CBOöOUHMMH rpammaMH jxnn
3jijinn™MecKMx CMCTeM ypaBHeHHfi. HOBOCH6HPCK:
"HayKa". 1977.424 c.
10. L. M. Milne-Thomson
Theoretical Hydromechanics. London Macmillan
Fig. 5-6 is the same comparison for a wedge (angle and Co. LTD. New York, St. Martin's Press, 1960.
30°). 11..I.L.Walsh
Interpolation and Approximation by Rational Func-
tions in the Complex Domain. AMS Colloquium
Publications. Second Edition. 1960.
27-6
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
28-2
premiere ; les seuls parametres resultant de la premiere phase sont de l'ordre de grandeur de (R0U0IL?. Dans le cas general
qui conditionnent la seconde phase sont la pression du gaz pD et les effets de pression imposee, de pesanteur et de penetration,
la vitesse du projectile U0 au debouchage. Differentes figurant explicitement dans l'equation de Bernoulli (2), doivent
hypotheses sont faites pour etablir un modele simple : les effets etre du meme ordre de grandeur:
de la viscosite et de la capillarite sont negliges; les pegL~peU2oR20/L2~\Pa-Po\ (4)
perturbations produites par l'ogive du projectile sont ignorees ; IJ s'en deduit que L est de l'ordre de grandeur de RoF . La
le projectile, cylindrique et elance, est assimile ä un cylindre longueur de reference permettant de normaliser les equations
semi-infini. La longueur de la bulle est supposee petite par etant precisement L, on pose :
rapport ä l'immersion de tir de sorte que l'effet de la surface
libre de la mer peut etre neglige. La pression p„ du gaz au §
R„FA -R0F2li r
-R0F2" (5)
debouchage est supposee voisine de la pression hydrostatique
au plancher P«, de sorte que la bulle soit elancee, conformement
aux observations faites lors de tirs reels. D a ete demontre de
facon theorique [1,2] et numerique [9] que les hypotheses de
"■i P*= . „ ^2/3
chaleur specifique cv est pVcvAT tandis que la quantite de 3.2 - Chasse avec un gaz incondensable
chaleur diffusee dans le liquide est de l'ordre de On considere la phase d'expulsion du projectile de la fosse. Le
XAAtATI{At%f5 avec % coefficient de diffusivite thermique de gaz est suppose homogene, de sorte que la seule variable est t.
l'eau et A surface de l'interface. La constante de temps des Soit qi le debit du gaz injecte dans la fosse, p la masse
echanges thermiques est done de l'ordre de pcv^sVIXA soit volumique du gaz et xc l'abscisse du culot comptee ä partir du
-0.02 |xs pour une bulle d'air de 1 mm de diametre ä la fond de la fosse. La conservation de la masse s'exprime par:
pression atmospherique dans de l'eau degazee. Pour un temps dp
S x
o c-^ + PUSo = Qi (13)
d'observation plus long revolution d'une petite bulle est par
consequent isotherme. La taille des cavites gazeuses qui nous En negligeant l'energie cinetique du gaz dans la fosse par
concernent etant de l'ordre de Im, les echanges par conduction rapport ä son energie interne, la conservation de l'energie
thermique dans la couche limite le long des parois ne peuvent s'ecrit, en l'absence d'echange de chaleur avec l'exterieur:
pas etre negliges. dp dxc
Dans le modele hydrodynamique la loi d'evolution du gaz est
supposee polytropique. D'apres l'observation des phenomenes
L'indice 1 correspond ä l'injection : M\ est le nombre de Mach
thermodynamiques, revolution n'est bien sür ni adiabatique ni
et Ti la temperature. L'equation du mouvement du projectile
isotherme ; la condensation se traduit par une loi d'evolution
s'ecrit en negligeant la trainee de forme devant la poussee au
qui n'est pas a priori polytropique et il n'existe pas de relation
culot:
directe entre la pression p et le volume V du gaz ä un instant t:
MtdU . ,
p depend de revolution thermodynamique depuis le debut de j-f-ä-ghpgfx. xc0)=p-P„ (15)
l'injection.
La duree des phases de lancement etant de l'ordre de la avec M masse du projectile et xco abscisse initiale. La
seconde, la variation de la masse du gaz par dissolution dans resolution des equations (13) ä (15) donne les valeurs de U(t),
l'eau est negligee. Par exemple la constante de diffusion de xc, p et p pour differents gaz injectes [4].
l'air dans l'eau est de 2 10"9 et la dissolution d'un petit volume
d'air necessite plusieurs minutes en statique [13]. 3.2 - Chasse avec un gaz condensable
L'ecoulement de l'air et du gaz accelere fortement la La condensation a pour effet de diminuer la masse du gaz.
dissolution, mais la constante de temps reste probablement Avec #Cp et qcm les debits de condensation ä la paroi et en
beaucoup plus grande que la seconde. Ce resultat est corrobore masse, la relation (13) devient:
par des tests de degazage d'eau realises dans la cuve
depressurisee decrite en 4. La constante de temps pour le xcS0^ + pUS0=q,-qcp-qcm (16)
degazage y est de plusieurs minutes en presence d'ecoulement La quantite de chaleur degagee par la condensation etant
et avec une interface air-eau de grande surface. supposee entierement transmise ä la paroi, l'energie du gaz est
L'hypothese principale du modele elementaire developpe ci- diminuee de l'energie interne csTqcp et du travail des forces
dessous repose sur la constatation que la chaleur latente de exterieures comblant le vide laisse par le gaz condense, soit
vaporisation L est grande par rapport ä la quantite de chaleur (y-l)cvTqcp [4]. A l'equilibre la temperature du gaz T est rsat,
necessaire ä un echauffement de quelques degres: temperature de la vapeur saturante. L'equation de l'energie
L=2,3 106J/Kg et Cv=l,5 103 J/Kg pour la vapeur d'eau. La (14) devient:
detente du gaz devrait etre principalement regie par la
dp dxc
condensation en masse lorsque la vapeur est saturee et les
echanges thermiques devraient etre regis par le transfert de la dt
(17)
chaleur latente de condensation ä la paroi. Les etudes sur les (r-i)cpTl(i+ 7-1 Mj)T-YcJsa,qcp 7-1
transferts de masse et de chaleur avec condensation dans une Z ö
o ä
o
couche limite [14] montrent que le probleme est difficile ; il Le tube etant suppose parfaitement conducteur et ä temperature
semble que pour une forte proportion de condensable, la vapeur constante Tp, l'ecran thermique qui limite la condensation est
reste sursaturee dans la couche limite et que la condensation le film d'eau condensee sur les parois. L'eau condense sous la
s'effectue uniquement sur la paroi. Les conductivites forme d'un film d'epaisseur e(x, t) qui depend de l'abscisse et
thermiques du tube, de l'eau et de la vapeur etant du temps. L'ecoulement du gaz et la pesanteur provoquent un
respectivement d'environ 40, 0.5 et 0.01 W/m/°C, la vapeur mouvement vertical dans ce film. Une etude des ordres de
peut etre considered en premiere approximation comme grandeur [4] et des estimations numeriques [6] montrent que
parfaitement isolante et le tube comme parfaitement l'ecoulement du condensat influe peu sur les echanges
conducteur. Le film d'eau condensee constitue un ecran thermiques; les effets de gravite compensent en partie
thermique qui regule la condensation ä la paroi. l'entrainement du au gaz. Le film de condensat etant mince,
Le modele thermodynamique est valide en comparant les (~0.3 mm), la variation de temperature suivant l'epaisseur est
previsions numeriques et des mesures ä grande echelle pendant presque lineaire. La chaleur latente de condensation est egale
l'expulsion du projectile. Ces resultats qui portent sur des au flux de chaleur transmis par le film, soit:
evolutions de la vitesse du projectile et de pressions dans le
tube et au culot du projectile ne sont pas reproduits ici. Ce (18)
e at
modele de la phase d'expulsion est mis en ceuvre en supposant
d'abord le gaz incondensable, puis en prenant en compte les X etant la conductivite du condensat. Si la variation de est
effets de la condensation. Le modele thermodynamique elabore negligeable, on en deduit:
est ensuite couple avec le modele hydrodynamique presente en
-Tp).(t- r) (19)
2 pour estimer les effets de la condensation sur la striction de
la bulle. Les effets thermodynamiques principaux ayant ete mis oü t est compte ä partir du demarrage du projectile et T est
en equation, il suffit ensuite de normaliser ces equations pour
l'instant de mise en contact avec le gaz. T est nul sauf sur la
mettre en evidence les parametres de similitude et determiner
surface laterale de la fosse de lancement.
des conditions d'essais ä petite echelle.
28-5
3.5- Approximation par une loi d'evolution polvtropique Pour simuler ä petite echelle la condensation en film il faudrait
Pour mieux comprendre l'effet de la condensation en masse, respecter l'invariance du parametre de similitude H qui varie
on propose un schema simplifie qui ne se limite pas ä la comme l'echelle geometrique ä la puissance 7/4. Cette
presentation de resultats numeriques. Si tout le gaz est similitude ne semble pas realisable: si dans une petite
condensable, l'equation de l'energie (26) peut etre reduite ä installation on utilisait un gaz condensable ä une temperature
une relation lineaire entre dT et dV ou dp et dV. En effet la elevee et dans des proportions importantes, la condensation
derivation de l'equation d'etat du gaz donne une relation aurait un effet preponderant par rapport ä la detente du gaz
lineaire entre dmv, dp, dV etdT; l'equation de Clapeyron (25) Hqcp"»E. Le film de condensät ä petite echelle est trop mince
relie dp ä dT. Par elimination de dp et dmv entre ces trois pour representer un ecran thermique comme ä l'echelle un.
equations, on a: Pour une simulation ä petite echelle, il faudrait une valeur de
AT tres petite pour respecter la valeur de H. E est preferable
dV J Mrn„
L— + dT\
V
-DC+/^-i)2 (27) d'utiliser un gaz incondensable ä la temperature ambiante avec
AT=0 de facon ä ce que gcp=0 plutot que de chauffer le melange
Comme L» RVT, si T et mv varient peu, cette relation peut gazeux en contact avec de l'eau. L'evolution du gaz pourrait
etre approximee par une evolution polvtropique ; d'apres (32) etre approximee par une loi polvtropique. A suffit
*L~*L-L-»*E. (28) pour la similitude de s'assurer que l'indice n moyen est bien
nV T RCT p respecte.
Dans les tirs ä l'echelle 1, l'indice polvtropique obtenu est La meme demarche de normalisation des equations est
d'environ 1,1. En presence d'un melange gazeux contenant des realisable pour les equations de la condensation en masse ; ceci
gaz incondensables, les deductions precedentes peuvent etre permet de mettre en evidence des parametres de similitude tels
generalises: il est montre en [6] que 1'evolution du gaz peut que (32). La chaleur latente etant grande par rapport ä la
encore etre approximee par une loi polvtropique tant que la quantite de chaleur necessaire pour echauffer le melange de gaz
masse de gaz incondensable n'est pas superieure ä environ 20 L»RVT, la condensation en masse induit d'apres le contenu de
fois celle de gaz condensable. Malgre la complexite des la section 3.5 une evolution presque polvtropique, quelles que
phenomenes thermodynamiques il semble qu'ils peuvent etre soient les conditions d'essais. En premiere approximation la
schematises par une loi pV = Cste avec un indice n determine similitude thermodynamique necessite done que l'indice n soit
lors d'essais ä l'echelle reelle. II convient lors des simulations voisin de la valeur 1,1 mesuree ä l'echelle un. Une similitude
ä une petite echelle de verifier que la valeur de ce parametre n plus complete de la condensation en masse necessiterait de
est bien reproduite. Pour s'en assurer il suffit de calculer n par tenir compte de la composition du melange gazeux ce qui est
regression sur les mesures pendant la phase d'expulsion du incompatible avec la similitude de la condensation sur les
projectile. parois.
position du Systeme permet de simuler une loi de mise en la maquette. L'effort transitoire du jet est obtenu en soustrayant
vitesse pre-programmee. de ce signal global des termes provenant des mesures de
Un element important de l'analyse experimentale etant la l'acceleration par un accelerometre installe dans la maquette.
visualisation de la striction de la bulle et des jets rentrants, la Les termes de pression et d'acceleration,. assimiles ä une
cuve est munie de quatre hublots en verre de dimension de surface apparente et ä une masse sont calcules ä chaque tir,
1,2x0,2m2. Pour eviter que la cuve ne se deforme trop aux pendant la phase d'expulsion du projectile, par correlation
pressions elevees, ce qui pourrait induire des contraintes double entre les signaux d'effort, d'acceleration et de pression.
dangereuses vis ä vis de la resistance des hublots, l'enveloppe Des contröles sont faits dans Fair pour s'assurer qu'en
metallique de la cuve a ete prevue tres rigide. l'absence de jet le traitement du signal fournit effectivement un
La course du verin est de 1,5 m avec une vitesse maximale de effort nul.
5 m/s et une acceleration maximale de 125 m/s . n est La pression initiale dans la fosse pi est mesuree par un capteur
susceptible d'entrainer une maquette de 10 kg. D est equipe statique eau-eau. Pour parvenir ä une precision d'environ 0.5
d'un capteur de position lineaire qui a ete etalonne en statique mbar (soit environ 0.05 sur D), on mesure la difference de
par rapport ä un cathetometre. L'entree du verin en bout de pression p- Px entre la fosse et le plancher. La pression de l'air
course dans la fosse de lancement est rendue possible grace ä dans la cuve est aussi mesuree, mais la precision necessaire
un jeu de 1 mm entre les rayons de la fosse et de la maquette. pour 1'interpretation de cette mesure est moindre. Toutes les
Apres la mise en place de la maquette la fosse est rendue mesures sont acquises pendant environ une demie seconde sur
etanche en gonflant un joint place pres du plancher. un Systeme d'enregistreurs de transitoire Lecroy 6810 ä une
La fosse est prevue demontable pour tester differentes volumes frequence d'echantillonnage choisie ä 10kHz. L'acquisition est
V0 conformement ä l'analyse de similitude. Elle est composee declenchee par une barriere optique detectant le passage de la
d'une partie haute cylindrique de 150 mm de long au diametre maquette.
de la maquette augmente des jeux mecaniques, et d'une partie Une tentative de mesure de la temperature dans la bulle a ete
basse interchangeable de 190 mm de diametre. Les infructueuse car l'inertie thermique des petits thermocouples
caracteristiques de l'installation sont prevues pour couvrir utilises est encore trop grande pour parvenir ä une constante de
largement le domaine des conditions de lancement temps de l'ordre de 1 ms.
envisageable en D, E et F. Des essais ont ete realises pour deux
valeurs du volume V0, dont l'une est environ six fois plus 4.3- Visualisations
grande que l'autre, en gardant constant/), E et F. En general, La duree de la striction etant de l'ordre de 50 ms, la
le gaz de chasse utilise est de l'air, mais, pour certains essais visualisation doit etre rapide. Une technique d'ombroscopie a
avec le petit volume, on a aussi utilise un gaz lourd dont la tout d'abord ete utilisee. Deux stroboscopes places l'un au
valeur de yest voisine de 1.09. Pindice n moyen de devolution dessus de l'autre fournissent une energie de l'ordre de 0,25
approximativement polytropique est calcule ä chaque essai par Joules chacun en environ 18 (is jusqu'ä une cadence de 1000
regression pendant la phase de mise en vitesse de la maquette, images par seconde. Une rafale d'un nombre reglable d'eclairs
pour verifier qu'il est proche de la valeur reelle. Deux est declenchee ä la cadence souhaitee par un signal emis par la
maquettes, geometriquement semblables, de rayons R'o et R"o station de travail pilote. Un ecran diffusant est place devant un
(R"o = 1,6 R'o) ont ete utilisees. La hauteur d'eau au dessus du hublot pour obtenir un eclairage uniforme. La striction de la
plancher ne peut pas toujours correspondre ä la profondeur bulle est enregistree soit sur un film de 35mm avec une camera
d'immersion en similitude geometrique, soit que la pression ä tambour tournant soit en memoire avec une camera video
ambiante devienne trop faible soit que la cuve ne soit pas rapide. Bien que les flashes soient tres courts, le mouvement de
assez haute. la camera ä tambour tournant induit un flou de bouger. La
Le circuit etanche d'alimentation de la fosse peut etre connecte video rapide ne presente pas cet inconvenient, mais sa
ä un reseau d'air comprime, ä une pompe ä vide ou une resolution est limitee par le nombre de pixels.
bouteille de gaz. La technique ombroscopique ne permettant pas de bien voir les
jets rentrants (figure 7a), des essais de visualisations ont ete
4.2 - Eauipement de mesure effectues en eclairage stroboscopiques indirect. Les resultats
Le culot de la maquette est equipe de capteurs de pression sont un peu meilleurs que precedemment mais insuffisants pour
Druck pdcr42, dont la bände passante estimee ä partir de caracteriser les jets (figure 7b). Ensuite une technique de
mesures faites au tube ä choc est de l'ordre de 32 kHz, ainsi visualisation par fluorescence induite par laser a ete
que d'une rondelle de force piezo-electrique Kistler 9104; la developpee. Un plan lumineux perpendiculaire ä la direction
fonction de transfert du capteur de force mesuree dans Fair et d'observation est obtenu avec un laser continu de puissance
dans l'eau indique une frequence de coupure d'environ 1kHz. 30W. De la rhodamine dissoute dans l'eau emet par
La contre-pression dans la maquette est celle de la surface fluorescence une lumiere de faible intensite dont la frequence
libre, ce qui permet de choisir des capteurs d'une etendue de est differente de celle de l'eclairage. Des filtres optiques
mesure de 0,35bar qui peuvent cependant subir une surpression permettent de dissocier la lumiere se reflechissant sur
de 3 bars sans endommagement. Les capteurs de pression ne 1'interface perturbee de la bulle de la lumiere emise par
pouvant detecter les jets rentrants que s'ils percutent leur partie fluorescence, ce qui fait apparaitre l'eau des jets rentrants sans
sensible, la configuration du culot avec un capteur de force est, etre ebloui par la lumiere reflechie. L'intensite lumineuse etant
parmi celles testees, la mieux adaptee ä la detection des jets tres faible, l'utilisation d'une camera video rapide intensifiee
rentrants. Un capteur de pression est place au centre du culot, est necessaire (figure 8).
tandis qu'un second capteur est place derriere le plateau de Le comportement de la striction de la bulle change en fonction
mesure des efforts pour fournir la pression dans la bulle meme de sa longueur, c'est ä dire principalement en fonction de la
pendant l'impact d'un jet. L'equipement du culot de la valeur de D [3]. Pour D>0 la bulle se ferme rapidement au
maquette est represente sur la figure 6. plancher. Lorsque D decroit, il apparait une fermeture non
En plus des efforts transitoires d'impact des jets, le capteur de ponctuelle le long de l'axe qui laisse une emulsion dans son
force est sensible ä l'acceleration de la maquette et ä la sillage. Pour D fortement negatif le gaz s'echappe lateralement
difference de pression entre la cavite gazeuse et l'interieur de le long du plancher tandis que la bulle se scinde ä une abscisse
28-8
intermediate entre le plancher et le culot. Si de plus E est petit conditions de lancement. Lorsque la bulle est tres longue les
(<3), la pression du gaz reste longtemps superieure ä P~, les caracteristiques des jets presentent une dispersion importante
bulles sont tres longues et leur rupture s'effectue parfois apres dont il est difficile de cerner l'origine. Sur le plan structural,
la sortie de l'eau du projectile. l'effort maximal du ä l'impact d'un jet n'est pas significatif, la
grandeur interessante est l'impulsion.
4.4 - Temps de rupture Les recoupements entre les visualisations et les mesures au
Des temps de rupture adimensionnes tT calcules et mesures culot de la maquette permettent de caracteriser les conditions
sont presented sur la figure 3. Bien que l'approximation de la d'apparition de jets rentrants, d'estimer les evolutions de leur
theorie asymptotique ne soit pas bien respectee pour des bulles vitesse et de leur section en fonction du temps. Pour de raisons
courtes avec D>0, les valeurs calculees sont proches de celles de confidentialite il n'est pas possible de donner ici l'ensemble
mesurees quel que soit E. Pour D>0 1'augmentation prevue de des resultats obtenus. La figure 13 presente un exemple
t*avec E decroissant est bien confirmee par l'experience, mais d'enregistrement de pression et d'effort avec des cliches au
on constate que les valeurs calculees sont generalement moment de l'apparition de jets rentrants. En reperant sur les
inferieures ä celles mesurees. Les valeurs de tr, relativement cliches l'instant de rupture de la bulle et sur les
peu precises pour la maquette de rayon R'o, ne semblent pas enregistrements de pression et d'effort l'instant de l'impact
dependre du diametre. d'un jet, on en deduit le temps mis par le jet pour atteindre le
Suivant les previsions du modele thermodynamique, il est culot et done une estimation de sa vitesse moyenne. Des essais
necessaire que revolution de la pression soit en moyenne dans des conditions bien maitrisees d'impact de jets dans l'axe
polytropique d'indice n voisin de 1,1. Lors des essais realises du capteur ont montre que le signal enregistre par le capteur de
avec de l'air et une fosse de grand volume, l'indice de la pression est de l'ordre de grandeur de la pression d'arret
detente du gaz est voisin de 1,4. Lors d'essais realises avec de stationnaire pU02/2, car la constante de temps des phenomenes
Fair mais avec un faible volume de fosse, la vapeur d'eau transitoires n'est que de l'ordre de 0,2 msec. Les mesures de
condense subitement juste avant le debouchage du projectile pression d'impact sont ainsi transposables ä l'echelle reelle et
(figure 9); la pression presente un palier au voisinage du elles fournissent une estimation de la vitesse des jets ä
debouchage qui est caracteristique d'un flash de condensation. l'impact. Ces vitesses des jets deduites des cliches, de
B en resulte une augmentation du temps de fermeture de la l'intensite de la pression d'impact et du temps ecoule entre la
bulle t* (figure 10). ; lorsque D<-\ les valeurs de t* obtenues fermeture de la bulle et l'impact sont toutes de l'ordre de 2U0
avec la fosse de petit volume V0 sont relativement plus elevees en valeur absolue. Cette vitesse augmente avec D pour D
que celles avec un grand volume. Pour une detente d'un gaz fortement negatif, car le jet principal est precede par une
incondensable avec y=\A, par suite du rapport des volumes, la projection de gouttelettes ä grande vitesse. La comparaison
temperature au moment du debouchage devrait etre de -38°C entre la pression et la force d'impact permet, en supposant
pour la petite fosse et 9°C pour la grande fosse. En presence l'ecoulement quasi-stationnaire, d'estimer la section du jet et
d'une faible proportion de gaz condensable, le gaz devient de la comparer ä celle visualisee.
sursature avant le debouchage dans le premier cas, ce qui La duree d'existence des jets ne se deduit des visualisations
declenche une variation de pression non representative du reel. car l'interface de la bulle se perturbe trop rapidement pour voir
Ce phenomene parasite a ete elimine en utilisant un gaz lourd, le jet jusqu'au moment de la sortie de l'eau; les estimations de
dont le rapport des chaleurs specifiques plus proche de 1 induit la duree du jet ä partir des enregistrements d'effort et de
des variations de temperature plus faibles. pression indiquent de fortes variations en fonction des
parametres de similitude.
4.5 - Jets rentrants A l'aide d'un traitement d'images on peut extraire des
Lorsque D>0, les bulles sont courtes et quelques visualisations le contour exterieur des bulles et en deduire leur
millisecondes apres la rupture, un jet fin et rapide est projete volume apparent. Si la pression dans la bulle evolue de fagon
sur le culot. D est suivi d'un jet de forme conique dont le polytropique, le volume de gaz V apres la rupture s'obtient ä
diametre atteint de l'ordre de 25% du diametre du projectile. l'aide de la mesure de p; la difference entre le volume
L'eau provenant du jet qui s'ecoule le long du culot rejoint apparent et V correspond au volume occupe par le jet rentrant.
l'interface de la bulle et la rend opaque. Dans certains cas cette On peut ainsi estimer le debit moyen du jet. La figure 14
perturbation de l'interface met en oscillation la bulle ä sa presente ce resultat, compare aux estimations obtenues
frequence propre, ce qui se traduit, dans le cas ou un jet directement de l'observation du jet (suppose de revolution).
percute le culot, par des oscillations importantes de l'effort
mesure (figure 11). Cette oscillation disparait lorsque l'eau 5 - CONCLUSION
s'ecoulant le long du culot est deviee avant d'atteindre Le modele hydrodynamique asymptotique de la striction de la
l'interface. Lorsque D<0 les caracteristiques des jets ne bulle a mis en evidence les deux parametres de similitude D et
semblent pas parfaitement reproductibles: ils sont parfois E qui correspondent ä la difference de pression entre le gaz et
dissymetriques et precedes d'une emulsion d'air et d'eau. Au l'eau au moment du debouchage et ä la derivee normalisee de
moment de la rupture les perturbations de ['interface la rendent cette grandeur. La resolution numerique montre que le
parfois trap opaque pour voir distinctement le jet. On a parametre influant le plus sur le temps de rupture de la bulle,
represente sur la figure 12 la valeur maximale de l'effort par exemple, est D, que l'influence du nombre de Froude est
d'impact mesure en fonction de D pour differentes valeurs des faible comme prevu et que la linearisation de la detente du gaz
nombres E et F : les essais ont ete realises ä P«, constant, et non n'est pas une approximation süffisante pour D<0. Le modele ne
ä p0 constant comme l'aurait exige le respect de la condition permet pas de depasser le point singulier de la rupture de la
E=Cste. Le capteurde force detecte presque toujours un impact bulle.
dont Pintensite est minimale au voisinage de D=0,5. Pour D La these soutenue est que l'effet thermodynamique principal
fortement negatif la bulle se scinde presque simultanement ä vis ä vis de la striction et l'apparition des jets rentrants est la
deux abscisses produisant deux jets rentrants; les efforts condensation de la vapeur en masse. La condensation sur les
maximaux mesures presentent alors une grande dispersion, parois du tube et sur l'interface accelere un peu la striction;
depassant largement les valeurs obtenues dans les autres pour que la condensation ne devienne pas preponderante ä
28-9
petite echelle, il ne faut pas chauffer le gaz et il faut que la 12 - Molin B., Dieval L., Marcer R., Arnaud M.,
pression partielle de vapeur soit faible. La loi d'evolution p(V) "Modelisation instationnaire de poches de cavitation par la
doit etre en similitude ä petite echelle ; il suffit, en premiere methode potentielle et par la methode VOF", 64mcs Journees
approximation, qu'elle soit de la forme pV =Cste avec le de l'hydrodynamique Nantes 1997
meme indice n qu'ä l'echelle un. 13 - Lefranc J.P. & al, "La cavitation", Presses Universitaires
La theorie et les calculs ne permettent pas de prevoir les jets de Grenoble 1995
rentrants et leurs effets lors d'impacts ; l'installation de 1TMF- 14 - Legay-Desesquelles F., Prunet-Foch B., " Heat and mass
Lille a ete adaptee pour simuler des jets axisymetriques dans transfer with condensation in laminar and turbulent
des conditions representatives du reel. Les visualisations et les boundary layer along a flat plane ", I.J.H.M.T,29, 95-105,
mesures simultanees de pression et d'effort au culot montrent 1986
qu'il existe toujours des jets rentrants, mais que dans certains 15 - Collier J.G., "Convective boiling and condensation ", Mc
cas, ils n'atteignent pas le culot, tout au moins avant que le Graw Hill 1981
projectile ne sorte de l'eau. L'extrapolation des mesures 16 - Huetz J., Petit J.P., "Techniques de 1'ingenieur. Notions de
d'efforts et de pression permet de dimensionner les structures transfert de chaleur", A1541, A1550
pour eviter tout risque de deterioration par des jets rentrants. 17 - May A., "Water entry and the cavity running behavior of
La prevision theorique ou numerique de ces jets rentrants reste missiles". NAVSEA AD-A020.429 report 75-2 1975
un probleme ouvert. 18 - Waugh J.G. & all, "Hydroballisties modeling", NUC
report TP447, AD-A007.529. 1975
6 - REMERCIEMENTS 19 - Paquet J.B., "Water jet impact on a rigid wall",Conference
on offshore Mechanics and arctic engineering. Lisbon 1998
Les travaux presentes dans ce document ont ete realises avec le
support financier de la Direction des Systemes de Force et de la
Prospective du Ministere de la Defense francais.
REFERENCES
1 - Dyment A., "Modele asymptotique de l'ecoulement produit
par l'impact d'un projectile dans un liquide", Comptes
rendus de l'Academie des Sciences t.310 Serie n, p.1589-
1594, 1990
2 - Dyment A, "Transient behavior of a gaseous cavity attached
to a projectile in a two phase flow", Symposium in honour
of Pr. Guiraud, Paris, Lecture notes in physics n°442 p205- /
220, Springer 1994 Jet
3 - Dyment A., Flodrops J.P., Paquet J.B., Dupuis D.,
Marchand D., "Gaseous cavity at the base of an underwater
projectile", Aerospace Science and Technology. Gauthier- 7— —rm
Villars ä paraitre en 1998
4 - Danis-Gobert A., "Etude du lancement en immersion d'un {po, Vo}
projectile par une chasse ä gaz incondensable et
condensable ", These. Universite de Lille 1994
5 - Gobert A, Flodrops J.P., Paquet J.B., "Resorption de la
cavite gazeuse derriere un projectile lance en immersion" , {p, V} D6bouchage Deuxidme
Association Technique Maritime et Aeronautique 1994 phase
6 - Paquet J.B., Gobert A, "Modele elementaire de la bulle de
culot d'un missile hydrobalistique", Rapport IMFL n°95/21 Figure 1 Lancement du projectile
1995
7 - Paquet J.B., Flodrops J.P., Dyment A, Gobert A.,
"Simulation sur modele reduit et elements de modelisation
de la bulle de culot au lancement d'un missile
hydrobalistique", Rapport IMFL n°93/46 1993
8 - Paquet J.B., Flodrops J.P., "Bulle de culot d'un missile
hydrobalistique : moyens pour caracteriser des jets rentrants
ä une petite echelle", Rapport IMFL n°97/16 1997
9 - Paquet J.B., "Cavite formee par un cylindre semi-infini
penetrant dans une nappe d'eau", Rapport IMFL n°90/22
1990
10 - Semenenko V.N., "Computer simulation of unsteady
0,50 --
supercavitating flows",AGARD R-827 High speed body
motion in water 1998
11 - Mahjoub F," Modelisation diphasique des evolutions 0,00
transitoires d'interfaces liquide-gaz ; application aux bulles
attachees ä des corps en mouvement", These. Institut
national poly technique de Grenoble 1995
Figure 2 Approximation de la detente du gaz
fosse de petit volume, F=8,y= 1,4
28-10
Capteurs de pression !
-12 -10
%> Capteur d'efforts !)
1,50
1,45
x(xR02IV0)
1,40
135 1,18
-•—0,94 Vapeurpure
1,30 -H-0,65
Figure 5 Vue generale de Installation experimental Figure 8 Clichg tomoscopique de la bulle et du jet
D=1,E = 6,F = 5
28-11
Flash de condensation
300
F (Newtons)
Avec oscillations
-50 ,100
i f(ms)
3,0
■ E- 1.75 F„a^(JtR02P„)
o MaqueitcR'o
lE- 2.6
DMaqueiteR"o
a 2,8 <E< 3.7
A Maquctlc R "o, bassc pression
2,5 D3.8 <E< 4.5
.4.7 <E< 5.5
0,20
oE- 6.2
2,0
A7< E<8
1,5
«»%t f -
1,0 0,10
03
"SVood
0,0 H 1 1 1 1 1 1 1 1 1 1
-e#o-
-12 -10 -8-6-4-2 0 2 4 6 10 -15 -10 10
0,30"
ql(KR ,2Vo)
0,25
•
0,20" •
• • D • D
0,15
• • • •
• •
. 0,10 •* •
• •
0,05" • •
D
1—ffcftft- H H 1 [ —1 1
-4-2 0 2 4 6 8
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
29-2
The main idea of undisturbed motion consists in The useful volume on scheme is shown by dot-
that the object should be moved not moving the ted line. This scheme is possible to be named by
medium apart but moving it inside self. Thus, the scheme "Body with channel" in the case of axisym-
object saves the former order of elements in wake metric execution. It is interesting that if we reflect
29-3
the half of this scheme relative to the solid wall, we Q is the flow rate through the propeller.
can obtain another scheme (Fig. 3 ) named "Body
in channel". The first component is the reaction of the propeller
jet, the second is the sink resistance i.e. the reac-
In Fig. 3, the updating of scheme "Body in chan- tion of fluid drawn in the propeller. We note that
nel" is shown in the case of viscous fluid flow outside the boundary layer
with formation of boundary layer with displace-
ment thickness 6*. Then, to reduce potential dis- V2y/1 + K2, = V0. (1)
turbances the corresponding restriction of channel
to value of boundary layer calculation thickness 6* Comparing the flow energy in front of the body
is nessesary. with energy behind one, we can calculate the power
transmitted to flow by the propeller:
The water-jet propeller is a special case of the pro-
peller in closed channel. The theory of such pro- N
= U [Vq2(l + K2q)-Vo2]V2dF+ (2)
peller is worked out in detail [14, 15, 18].
Here, F2 and Fe are the jet section areas and an- For numerical analysis of described schemes the
other part of the wake: uniform mathematical model was developed. It
V2 is the velocity in jet or wake past a body; permits to calculate axially symmetric configura-
P2 is the pressure in this place. tion including the central body, displacement walls
If in the propeller jet the velocity V2 and pressure of the channel, thin shielding surfaces and two pro-
P2 are constant then presented relations can be rep- pellers: nose and stern (Fig. 4). The dimensions of
resented in the form elements of this system and their mutual location
can be arbitrary. Action of the central body and
Pa = pQV2y/l + K2q-pQV0, thick channel walls on flow is taken into account by
2(P2-PQ)
using ring sources located on the body surface and
where K pV? > the channel walls. The action of shielding surfaces
29-4
— by action of system of ring vortices located on The length L is equal 1000 mm;
this surface. The velocity field created by propeller The diameter D is equal 212 mm;
is treated as a disk of sinks and a jet located behind The diameter of internal channel d is equal 80 mm;
it. The velocity in jet is defined according to the The electric motor power is 60 Wt (27 V);
propeller load [4]. This velocity value, radius and The mass is 14.25 kg;
propeller location were given. The buoyancy P& is 1.75 kg;
Some results of computations are shown in Figs. 5, The screw with four blades Ds is 75 mm.
6. The figures represent axial and radial velocity The experiment with described model was con-
distributions along the body axes at two distances ducted in the hydraulic laboratory of the Institute.
from it for conditions with shielding surface (curves The horizontal hydraulic channel with length 17 m
Vx and VT) and without it (curves VXQ and V^o), the was used. Water to the channel is carried by pump
former is being multiplied 10 times. station from underground basin of the laboratory
by closed cycle. The scheme of execution of exper-
The problem of flow according to the hydrody- iment is shown in Fig. 7. The hydraulic channel 1
namic scheme "Body with channel" (Fig. 2) was with width 0.6 m with glass walls with height 1.0
analyzed by Merculov V.l. [1—2]. He showed that m is equipped by devices for water rate regulation,
the disturbances of the external flow introduced by flow damper in entrance part of the channel and
the motion of such a body can be made small by backwater board in the channel end for regulation
selecting the length of the bounding planes so that of flow depth. The water rate value was determined
x — ±x0. The velocity and pressure disturbances by triangular weir gauge.
on the inner wall in the plane y — ±1 are equal to
The model 2 was installed in flow by described
[2]
■KCTdrP way with help flexible wires - tension members 4
Aux ; Ap =
2Sh??f 2Sh2?f' hooked up to plugs 3. The supplying wires 5 were
stacked on channel bottom in the water flow direc-
It follows from these equations that the required tion and removed through the free surface to the
length x of the bounding planes at a given wa- removal distance about 3 m from model. In this
ter displacement <7<jr can be computed is one as- way the influence of disturbances created by them
sumes the permissible velocity and pressure distur- was eliminated. The power supply of model was
bances. The practicality of this flow pattern with executed from control block 6. The electric motor
constraints on disturbances of the surrounding fluid revolutions were regulated and the parameters of
was demonstrated experimentally. consumed currents were measured due to it.
The model parameters are: The most convincing demonstration of wave drag
29-5
destruction is given by photos of experiments for 6. Dubrovsky, V.A., "The research of wave drag of
three regimes (Fig. 8 a) b) c)): catamaran", Sudostroenie, Leningrad, 1968, No 7.
7. Dubrovsky, V.A. "The main features of hy-
a) the self-propelled regime of internal displace- drodynamics and technical-operational qualities of
ment object (body) (Pe = F);
ships with small areas of waterline", Sudostroenie,
b) the tugging regime (Pe = 0) (internal object 1993, No 8-9, pp 3-6.
inside a body, scheme "Body in channel"); 8. Numata, E., "Predicting Hydrodynamic Behav-
c) the self-propelled regime (Pe = F) for scheme ior of Small-Waterline-Area Twin-Hull Ships", Ma-
"Body in channel". rine Technology, 1981, 1, vol. 18, N 1.
9. Solomentzev, O.I., "The comparison analysis
5. CONCLUSIONS of one-hull ships and catamarans", Sudostroenie,
In present work we define the undisturbed mo- Leningrad, 1990, No 5, pp 6-8.
tion of body as such motion, when the information 10. Dubrovsky, V.A., "The expansion of opera-
about this motion will absent outside some limited tional opportunities of high-speed SWATH ships",
area of space containing the moving body. Sudostroenie, Leningrad, 1990, No 1, pp 5-6.
11. Nilsen, J., "Aerodynamics of controlled shells",
To create such motion it is necessary to move Oborongiz, Moscow, 1962, 474 p.
through the internal cavity of body to its stern part 12. Busemann, A., "Aerodynammischer Auftrieb
the fluid particles located in front of the body and bei Uberschalgeschwinddigkeit", Luftfahrt-Forsch,
to restore their initial characteristics. 1935, 12, pp. 210-220.
13. Leibenberg, L., Bunt, E.A., "Wave drag coef-
The work confirms the basic opportunity of similar
ficient of a model 'Busemann biplane catamaran",
motion creation for the case of uniform rectilinear
Experiments in fluids, 9, 1990, pp. 352-354.
motion of the special shaped body in low-viscous
14. Kulikov, S.V., Hramkin, M.F., "Water-jet pro-
fluid. The feature of this shape is the internal chan-
pellers", Sudostroenie, Leningrad, 1970, 352 p.
nel availability inside the body. The analytical ex-
15. Basin, A.M., "Some problems of theory of
pressions for computations of the body shape and
water-jet propellers", Trudy Academii rechnogo
necessary differences of the velocity and pressure on
transporta, No II, 1953.
surfaces bounded the control volume are presented.
16. Ivchenko, V.M., "About most profitable pro-
peller complex" Trudy NTO Sudproma, 1963, No
The experiments with models executed according 45, pp 73-76.
to the two main theoretical schemes " Body with 17. Miln - Tompson, L.M., "The theoretical hydro-
channel " and " Body in channel " have shown dynamics". Mir, Moscow, 1964, 655 p.
that in regime of uniform motion Pe = F in fluid 18. Basin, A.M., Miniovich, I.Ya., "The the-
around the body the disturbed fields of the velocity ory and design of propeller screws", Sudpromgiz,
and pressure did not arise (excepting the turbulent Leningrad, 1963, 760 p.
wake area).
REFERENCES
1. Merculov, V.l., "About opportunity of undis-
turbed motion of body in fluid" Izv. SO AN SSSR.
Ser. tehn. nauk, No 2, 1981, pp 37-44.
2. Merculov, V.l., "The fluid motion control",
Nauka, Novosibirsk 1983, 174 p.
3. Savchenko, Y.N., "Some Special Cases of flows
over Bodies", Fluid Mechanics - Soviet Research,
vol. 21, No.l, Jan. 1992.
4. "The propeller screws. The modern methods of
calculation", Sudostroenue, Leningrad, 1983, 296
P-
5. Kostyukov, A.A., "The interaction of bodies
moving in fluids", Sudostroenie, Leningrad, 1972,
309 p.
29-6
i=t t't.+L-V
/ \ .. .
Zones
ADC /N-Nus\ ABC
■ A c t i o n s
/ \
Caplure S / Transfer Reconsiructions
Compression / & Delay \ % ejections
V
Fig. 1
Fig. 2
Fig. 3
29-7
/ 2 S _3
Fig. 4
CONTROL OP PLOW
VarCAP R- 1.5
g
>
0.05 -
-0.05
Fig.-5
CONTROL OF PLOW
VarCAP R-7.5
0.002
g
0.001
-0.0 01
Fig.6
29-8
29-9
■K5»*-'
Fig. 8 a) b) c)
30-1
Paper presented at the RTO AVT Symposium on "Fluid Dynamics Problems of Vehicles Operating near or in the
Air-Sea Interface", held in Amsterdam, The Netherlands, 5-8 October 1998, and published in RTO MP-15.
30-2
take-off transition, the vessel maintains 1:3.35 from the 80 Seater HW 80 and
stability, and no manual control is were tested 1997. Hoverwing 2-VT
necessary. completely fulfilled the performance and
As soon as the so-called „flare" mode is characteristics in accordance with the
taken after take-off, the inlet port behind given requirements in regard to
the propeller is closed, and full thrust is take-off, cruise, manoeuverability and
available for cruise. Closing the inlet landing, during tests on the Baltic Sea
port also deflates the bag-type skirt and on the Ijsselmeer in the
sealing at the end of the air cushion, and Netherlands. Fig.7 shows HW 2 Vt in
is then also folded. Operation
Before landing, the inlet port is opened
again, the rear sealing inflates 6. Conclusion.
immediately after throttling up when Creating a static air cushion by using
touching the water surface. The build parts of the propeller slip stream du-ring
up of the air cushion makes the front take-off, a reduction of hydrodynamic
finger skirt sealing swing down drag comparable to that of an SES is
automatically, so that the air cushion is achieved. During cruise the dynamic
fully working again during touch down. pressure of the airspeed replaces the
This makes the landing extremely soft static air cushion and automatically folds
and reduces structural loads. the sealing skirts to the lower side of the
One problem of this basically new catamaran body, which then creates
configuration was the catamaran float aerodynamic lift and makes the vessel
design. The best position for the step, become a ground effect craft.
the design ot the sealing skirts and the The catamaran body functioning as the
best trim for take-off, were optimized in main wing, offers, due to it size and
cooperation with the Ver-suchsanstalt airfoil shape, a thickness which allows
für Binnenschiflbau e.V (VBD) at the comfortable accomodation of
Duisburg, Germany. Using scale models passengers inside without the need
with extensive measurement equipment building a separate fuselage.
which transmitted the test data, the FF holds patent rights for the Hover-
complete speed range from wing Technology and could confirm the
displacement to flare mode was required performance and charac-
analysed in order to achieve reliable teristics as predicted by calculation and
information about the dynamic experiment with the manned test craft,
characteristics and performance of the the Hoverwing 2 VT.
new configuration. The achievable
reduction of drag, and the resulting
possible reduction of power to be
installed, is shown in Figure 6
Performance potential.
Q
<
O
_i
I
AIRFOILBOAT.
UJ
Q.
0,25
SEA FREIGHT
trr.^
I
Airfoilboats WIG
aerodynamic lift
-i Luftkissenfahrzeug'
I '" Selenrümp7" ""
Luftkissenfahrzeug'
SWAACS
Auftrieb
HYACS durch
aerostatische -
Kraft
Tragflächen-
boot
f
A Auftrieb durch Auftrieb durch
hydrodynamische Kraft Verdrängung
SWAACS = Smal Waterplane Area Cushion Ship HYSWAS = Hydrofoil Small Waterplaner Area Sh
SSACV = Semi Submerged Air Cushion Vehicle LAHNS = Large Hydrofoil Hybrid Ship
SWAT! I * Small Wateriane Area Tw.n Hull HYACS = Hydrofoil Air Cushion Ship
SWASH = Small Waterplane Single Hull
10
target
8
Hoverwing-80
Airfisch-3
WIG - ships
waterplane
2 -
Fig. 3
Helicopter <—>J
Aeronautic
Technology r* Aircraft
I
o
\ o
ACV <r—> WIG >
O)
C
■(0
CO
/
Displacement
Bfiäi
Nautic
Technology
U Hydrofoil
o
c
MO TYPE SORT
. A most«
:UGD
M I
r^4- AIR WLET CLOSED
>--.
— _r~—"
Flare - Mode
HOVERWING Fig. 5
PRINCIPLE
to
0
0 20 40 60 80 100 kts
Fig. 6 BMBF
Phase
Performance Potential
30-7
1^
o
o
CD
Gi
o
o
•o
CN
o
00 ö
OS
CN
u
lO
LU CO
.2
S| o
c
r^ O CO
Ö)
G>
GD-1
GENERAL DISCUSSION
Fluid Dynamics Problems of Vehicles
Operating Near or in the Air-Sea Interface
Amsterdam, The Netherlands
5-8 October 1998
The other fusion that took place here is between the scientists and engineers from Western
Europe, North America, Australia, Russia and Ukraine. I think that this was a very important
aspect of this meeting, and I will have a little more to say about that later. Of course, it takes
money to do these things, and I think I would like to say thank you from all of us for the
funding. I hope that our East-West cooperation will be long and fruitful and continue in the
future. It is very stimulating.
The meeting has, if you look at the title, had two aspects. I think that is the way it worked
out. One concerned the fluid dynamics connected with these vehicles, and there are people
here who are more interested in fluid dynamics and the aerodynamics or hydrodynamics of
the vehicles, than the vehicle engineering and performance itself, and then there are the
vehicles. That makes the meeting a little hard to discuss from a technical viewpoint, but I am
going to try. My remarks must of necessity be very informal since time hasn't permitted a
proper digestion of everything heard here - that will come later in a written report.
There was a continuum of vehicles represented here. Dr. Fischer in his last very fascinating
talk extended that continuum by suggesting a marriage here at the interface between
hydrofoils and Ekranoplanes. In the overall picture, there is a larger continuum starting at
one end with stationary platforms in ocean engineering, where there are huge developments
GD-2
and very wonderful engineering and science. Then ships, including the problem of helicopter
mating. I hadn't been aware ofthat problem, but it seems most serious. Beyond ships and at
higher speeds we have hydrofoils and SES vehicles and then at the high speed end, the
Ekranoplane. There were some members of this continuum which were not specifically
discussed here in the regime of fast ships, which are getting very accelerated development
now: planing boats, catamarans, and swath ships. Finally, at the low speed end I mention the
MOB, which means Mobile Offshore Base, and the US Navy is now conducting a research
program connected with studying the feasibility of building a floating base which would be at
least 5,000 feet long and at the present time is conceived to be 400 feet in breadth, which
would be able to land our largest aircraft, the C-130, and which has enormous storage
capacity inside these structures, which would detach into maybe 1,000 foot modules for self-
propulsion to sites around the world. This is, of course, motivated by the desire to have
independent bases offshore to conduct whatever operations are necessary, as in the Persian
Gulf, for instance. That leads to fascinating problems of various kinds. Finally, beyond the
Ekranoplane in the high speed end of the continuum we have seaplanes, which detach from
the surface. I think that Bernard Masure had hoped that seaplanes would get a greater
discussion than took place here; there was a paper on the subject which was canceled. Just
before I came up to speak, Mr. Swick from Lockheed Martin put in my hands an amazing
picture of a proposed C-130 float plane. I asked him if there is a requirement for that, and he
said we are working on it. So, there may be a future in seaplanes. Of course, we have
seaplanes for fire fighting, I see those often in Santa Barbara. That is a part of the continuum
not to be neglected, especially by all of you. So that is the continuum, and it would be nice
to settle the question whether Ekranoplanes have a future in it and what it might be, and the
same for seaplanes, too. I don't think in a meeting of this duration or representation it is
really possible to do that, but we had some indications for the future and some hopes
expressed by various people.
As far as fluids are concerned, there was not a continuum of fluids here. Particular problems
and areas were spotlighted which arose in connection with specific interface technologies; for
example, in connection with deck aerodynamics, which was the subject of the first set of
lectures here. I would say there are two fluid subjects which arose from that. One was
organized vortical flows. Here I am thinking particularly of the flows around carriers where
you have these large vortex structures. We heard about some very nice experiments from
TsAGI presented by Dr. Maslov, and we also saw some numerical calculations which we
shall refer to shortly. On the frigates, where you have the backward facing step that the deck
represents, the flow underway and due to wind is very complicated. It is a bluff body flow, a
3-D unsteady flow with separation, and it was very apparent that our means of calculating
these flows is developing but limited. This is an area which everybody in fluid dynamics
recognizes as a frontier, i.e. to develop the ability to go on and calculate 3-D complex flows
with separation. There are a number of fields in which these flows are important in ocean
engineering and building aerodynamics. It would be interesting to compare what they do in
those fields in comparison with what we heard here in connection with the frigate deck
problems, especially from the building field, where they have exactly those same flows and
where they are very important in practice.
I can identify some issues in connection with these flows and one is the issue of Euler
equations versus turbulence modeling. It was fascinating to see that people use Euler
equations without viscous or turbulent stresses, and nevertheless, they can get separated
GD-3
flows, vortical instabilities and so forth. As I learned from a British aerodynamicist here,
more is known about this in terms of the fundamentals than was evident in the lectures.
Apparently, flow separation can occur as an artifice of the numerical process. I understand
that it is a function of the corner radius of the body and of the grid size. Of course, if you get
a shear zone started, i.e. if you introduce some vorticity into the flow, even artificially
through numerical means, then that interface can be unstable in the Kelvin-Helmholtz sense
and a kind of turbulence ensues. So, you can see how you can get complicated unsteady
flows from Euler equations alone. However, whether you can accurately predict something
like the size of the separation region behind the deck, which is important, in connection with
the helicopter problem, without introducing real stresses, is very doubtful, I should think.
But these simulation phenomena are most interesting, and I think that the question of the
validity of Euler simulation is an important issue for people to deal with in future.
The other great issue is the balance between experiments and numerical modeling. I think a
lot of wisdom has to be exercised in connection with deciding how to go about some of these
things. When we don't have analytical means or means which are still in doubt, where there
are big questions, experiments then become extremely important. We saw here a number of
interesting experiments and experimental results. These things have to go hand in hand, but
it is obviously very important, especially in connection with questions about Euler equations,
that numerical modellers do systematic comparisons and careful comparisons with
experiments. In my experience in life, it doesn't pay to assume a theory is correct in fluid
dynamics; if you don't check it with experiments, you will usually pay seriously. So, there
are very important things to do. I think there are many research opportunities in these fluid
dynamic areas involving unstable and separated flows and their simulation. The right
balance between experiments and theory plus simulations will be required.
The second general area, where we can group things together, is ground effect aerodynamics.
Here we have the Ekranoplanes which use the Russian terminology. We have several
aerodynamic regimes here, the image regime, in accord with Kiril Rozhdestvensky's span
regime; the ram regime; and the chord regime. It is obvious from what we heard in
Rozhdestvensky's talk and others that a vast amount of useful aerodynamics has been done
that is directly useful for the analysis of stability near the ground. There are emerging
numerical methods which will also be useful. Remarkable applications have been made in
Russia to the actual design of vehicles. I am thinking particularly here of Alexeev of
Nizhny-Novgorod, and I will say more about him later; we have heard here about really
remarkable developments, that I hadn't known about in detail before. Although much has
been done, there obviously is still good aerodynamics to do and systematic validation. It
wasn't clear to me for example how much of the theory we heard about has actually been
compared with detailed experiments. Of course, it is more than force coefficients, if you are
talking about details of the flow. Eventually, if these vehicles evolve, improvements, even
marginal improvements, will be sought, and that means more aerodynamic testing. The very
successful and innovate developments of Fischer in Germany, which we heard about only
this morning should also be mentioned, including his technique for tethered testing indoors.
would carry very heavy equipment, including a tug. Following this suggestion, it occurred to
me that a subject for future meetings might be "marine vehicles in rapid disaster response."
If you think about such marine vehicles, you go from lifeboats to cutters to air/sea rescue,
and Ekranoplanes; it seems an interesting and important subject involving the interface; I
would include the SES and ground effect aerodynamics.
Regarding SES, we had a talk by Faltinsen on Cushion Acoustics, concerning the serious
problems of wave induced loads due to oscillations in the pressure under the vehicle. This
phenomenon must also involve structural response of the flexible seals, etc. We didn't hear
much about structural response here, as this is a meeting about fluid dynamics. However, we
should not forget that in the marine field, the loadings are very large and, in view of the
increase in speed, hydroelastic problems are getting increasing attention. It is obvious that
this will be the case here with vehicles operating in the interface. The design of compliant
structures might be in order, as Fischer suggested; i.e. compliant to water, but not to air. SES
cushion problems are hard to model, because there is a Mach number involved, the speed of
the craft over the speed of sound arising from the compressibility of the cushion. As a result,
these problems are notoriously hard to deal with experimentally, because we normally can't
test at full-scale speeds.
A third general area at this Symposium was stability and control. When we are operating in
the interface, load control and vertical control are very difficult, and extraordinarily
important. Much more important than when flying in the atmosphere in an airplane, as is
obvious. The problems of vertical control had earlier been crucial for submarines which
often have very little vertical room to maneuver under the sea surface, considering their
length. Here we heard here a number of talks dealing with the stability and control of
Ekranoplanes which are now designed to be inherently stable, which was very fascinating to
understand, and a wonderful achievement, i.e. to achieve that stability through purely passive
aerodynamic means. We heard a talk by Nebylov from Saint Petersburg on automatic
control, especially alleviation of wave motion, and I did see some figures put up by Kiril
Rozhdestvensky on accelerations. I remember from my own experience with hydrofoils
many decades ago that the tolerance of humans to vertical accelerations is very small. As I
recall, it is necessary to get the mean accelerations under 1/1 Oth of a G. Otherwise, you are
not going to carry anybody very long, and quarter of a G is probably too much. I saw
numbers like that quoted here, so that this is a very serious problem for hydrofoils and for the
SES, where we heard about the cobblestone effect, and for all interface vehicles in fact.
We heard some papers by Mr. Kummer from France on ship control with a new
implementation of the moving weight idea, automated, and including sophisticated fins. It is
always welcome to hear of advances like that. Finally, we heard of the problem of autopilot
landing of helicopters on decks, in the case of the unmanned helicopter by M. Reboulet.
Finally, we had lectures on foils and impact and bow splash hydrodynamics. We heard a
little bit about hydrofoil ventilation which is a very dangerous and destabilizing phenomenon
at the interface. At these high speeds the foils are going to cavitate and generally our
solution is to go into the supercavitating mode. This field has had an earlier very active life
and reached a peak; a great deal of research was done in that time. Some applications were
made, too. However, I think there are still new application opportunities and more research
and testing to do. In impact, more and more work is being done, as slamming of ships is an
important problem area, and more so at increased speeds. The 2D&T methods I showed you
GD-5
this morning can also be used here. When we consider ship moving in waves, the forefoot
emerges and slams down and these methods can be used to calculate the loads on the bottom.
There is quite a bit of interest in that subject now, combined with hydroelastic effects,
because of fast ship developments in the last 10 years. In these areas, as in many others
mentioned here, we have to keep doing research on these things, to put fundamental
information and applied information into the designer's reservoir of assets. If we don't
maintain this reservoir, design and innovation will suffer.
In my closing remarks, I would like to draw special attention to the Russian work,
particularly the Russian innovations. It is very remarkable to me that very bold engineering
innovations were made by Russians even in the Soviet regime, when I imagine the cost for
failure was not so pleasant. Nevertheless, remarkably bold things were done. I myself
started to work on supercavitating foils in 1952 and then someone put in my hands
information about supercavitating propeller developments in Russia, where a man named
Pozdunine had started developing supercavitating propellers by himself during WWII. He, in
fact, used the word superkavitatsia which I translated as supercavitation, that is how the word
came about. We have the phenomena of Alexeev, who, as somebody here said, made two
engineering revolutions. He was the man whose design bureau was responsible for the
wonderful Russian hydrofoil boats. I don't know if any of you have traveled on a Russian
hydrofoil boat; I had the pleasure in 1963 to travel on a big boat that was and still is running
on the Neva in Saint Petersburg. They are beautiful boats, beautiful interiors. They have
very shallow draft, they have a passive automatic system for altitude control. He also built
fast boats and sea keeping boats. Then he conceived the Ekranoplane. I understand that
there were people earlier in different countries talking about wings in ground effect and I
know there was a Finn who did pioneering work all by himself on ram effect wings when
hovercrafts first came out. However, to build the Caspian Sea Monster and all these other
Ekranoplanes is a remarkable thing. Alexeev and colleagues grappled with the fundamentals
and made stable boats. We also have Russian innovation with high speed supercavitating
missiles, Logvinovich and Professor Savchenko who you heard from this morning, in Kiev,
and these are really remarkable innovative developments. We owe a lot also for fundamental
fluid dynamics to Russian workers. To mention only a few: in turbulence, Kolmogoroff; and
Yaglov and Monin who wrote famous texts in the 60's; in waves, Köchin and Stretensky; and
in cavitation, Gurevich and Epstein. Professor Maslov told me that his Professor was
Epstein, who was a wonderful worker on problems of ventilation and cavitation and who
wrote a most interesting book which I don't think has been translated. I hope we can all
continue to work together and to profit from the knowledge of our Eastern colleagues and
especially from their innovative spirit.
Regarding the future of Ekranoplanes, we had a good forecast by Aframeev and further heard
some hope about the future in the last talk by Dr. Fischer. I don't know the reality myself,
but several speakers here obviously believe in the possibility of transporting in the speed
range of Ekranoplanes, say 200 kts, and it is true, there is a transportation gap there. Many
of us in the same field have though about that; we studied, in my own company, 100 kt SES's
when they were under naval development, it did appear difficult though to beat a 747 for
overseas transport at high speed with the SES. You have to make realistic economic studies,
including sea state operability. So the future isn't clear yet. Perhaps disaster response, as
proposed by Aframeev is worth pursuing.
GD-6
The other thing I want finally to comment on the fluid dynamics side is the breadth of fluid
dynamics at the interface. In the United States, one gets the impression sometimes that all
that we don't know in fluid dynamics is turbulence. This is of course a joke, actually. If you
look at fluid dynamics in this meeting, you will see we started off with some problems
involving separated flows and turbulence, even though the Euler equations can apparently
make wakes and everything else without any Reynolds stresses. Overall, we see the
enormous range of problems which are not turbulent but have to do with working at the
interface, things we don't yet understand despite our advances, like cavitation, re-entrant jets
and so forth. I myself work on waves and wind waves and I have a big facility where I blow
wind over waves just to understand the ocean. That is complicated and interesting enough,
but here we are putting vehicles into that complex situation. In addition, to the problems
which were hinted at here, propulsion at the interface is also very difficult. We heard very
clearly in Fischer's talk about the problem of takeoff propulsion because of the big hump
drag there. This is also the case with hydrofoil boats where the water propeller has to work
very hard, and is going to cavitate even at low speeds because of the high thrust loadings, and
become very inefficient. I am sure that the audience could obviously add more items to this
list for the future. This completes my technical remarks.
I do think it is a nice idea to do this again some time. After some years Professor Masure's
questions may come closer to being answered, and we will see future developments, and it
might be productive to review progress and add some new subjects. In the interface, I don't
think anybody else is dealing specifically with these vehicles with this breadth.
We all owe a great debt to the organizers and the funders for this meeting. Thank you.
I will now very rapidly show you a few hydroplane photos. Here is an old hydroplane
constructed by Dornier. In the past, there was strong interest and activity in these
hydroplanes. Mr. Stilton, an English specialist in seaplanes, could eventually discuss this.
Here is the celebrated Russian seaplane, the Albatros which weighed 90 tons, the largest jet
powered hydroplane in the world. It is quite recent, and the question is whether there is a
future for this type of seaplane. Now, I will show you the BE 200 from Beriev design
bureau. You see it, I think, in the New York sky, but this photo is an artist view. This is a
civilian hydroplane which made its first flight 10 days ago, with great success. It was built
by a consortium including Russia, Switzerland, Singapore, and the US - a real international
project.
In the U.S., as Mr. Tulin told us, the Lockheed Company, represented here by Michael
Swick, is working on a floatplane version of the C130 Hercules. You see here this navalized
GD-7
version of this plane. That is to say, concerning the military aspects, it could be of interest to
develop hydroplanes. In Canada, everyone knows the seaplanes developed by Canadair-
Bombardier which are used to control fires. The latest of these seaplanes from Canadair, is
the CL 415. The Russian design bureau Beriev proposes, with a lot of imagination, the
development of giant hydroplanes of about 1000 to 5000 tons.
I wrote to all of these organizations to request information and papers on all the seaplanes,
which, of course, have problems dealing with the sea/air interface, and I had no response!
No one replied. Knowing the special interest of this Symposium and this subject, I find it
regrettable that no one replied. The second question I would like to discuss concerns
Ekranoplanes. I will try to provoke some comments here. Personally, I doubt that in the
future Ekranoplanes will become the general means of transport replacing airplanes. I doubt
this and I open the discussion for your responses.
The prospects. Today, we are already able to build Ekranoplanes of up to 800 or perhaps
1000 tons, because the only critical factor in such projects is the power system.. The
necessary propulsion systems are already available, and the turbines currently developed, in
fact, enable the construction of such machines, today. We already have the materials needed
for the hulls, as well as the avionics, and there is therefore, no practical reason for not
building these machines. In my opinion, further development is technologically feasible up
to 2000-3000 tons, 4000 at the most. These Ekranoplanes will be able to provide
transoceanic transport throughout the Pacific and Atlantic oceans. There is no need to look
GD-8
any further. These machines will have the range and capacity to carry loads which are
impossible to transport by aircraft : steel structures of huge dimensions, heavy bulk goods,
which can be handled only by Ekranoplanes. Thank you.
WIG craft can fill the gap between the fast but expensive air, and slow but cheap sea
transport. I think this is the answer in the end, and since WIG craft with their best lift-drag
ratio and the biggest useful room, they have a good chance for selected missions.
When such craft have sidewalls, or floats which act as sidewalls, they satisfy Hoerner's
definition of a ram-wing - which is why I prefer that term to WIG. Low altitude flight at
high speed for prolonged periods is exhausting. For many years I was a professional test
pilot. While at the then Royal Aircraft Establishment at Farnborough in the early 1960s, I
was involved in experimental testing at very low altitude and high speed, in connection with
the TSR 2, designed to operate at Mach 1.0 at 200 ft altitude (60 m). While that is faster
than an Ekranoplane, they, nevertheless, must be capable of 300 or 400 knots at even lower
altitudes, to be economically useful. To cruise more slowly involves carrying around an
excess of 'wetted area' (airframe surface in contact with air in relative motion). The point
about flying at 400 knots and low altitude is that in rain special windscreen clearance
systems are needed. I found that during tests at 565 knots and 200 ft (60 m) in heavy rain off
Singapore, the water repeatedly cut the glass of a navigation light. All-weather operation is
essential if Ekranoplane-type aircraft are to be cost-beneficial.
Later, during twenty years certification test flying with the Civil Aviation Authority, I also
tested floatplanes and flying boats. They are useful aircraft, but they have their limitations,
forced by operation from water. Yet, during the 1939 war the small single-engine
Supermarine Walrus biplane flying boat, which had a crew of three, weighed about 3.5 tons
and was designed to be catapulted from warships, once squeezed 9 on board. It could operate
in Sea State 3, and rescued more than 1000 aircrew, both allied and enemy. The flying boat
was much used along the air routes of the former British Empire during the 1930s. Had there
been no World War it is possible that they would have continued in a more advanced state of
development today. Even so, two four-engine Short Sunderlands (developed out of the Short
Empire boats before that war) rescued the crew of the Kensington Court, torpedoed by a U-
boat off southwest Ireland. One, loaded to the edge of the door, leaving no further freeboard,
is said to have taken 15 miles to become airborne. What killed off the flying boat was the
same war and the heavy bombers which needed hard-surface runways. Imagine attempting
to load even a 2 ton bomb into a high-performance seaplane on water, to leave via a bomb-
door in the keel, let alone a 10 ton bomb, designed to penetrate the reinforced concrete of a
U-boat pen. The same bombers were modified to become the first airliners after the war,
while airfields grew to become major terminals, like Gatwick, Heathrow and Manchester, in
the UK.
GD-10
A second 'killer' of the flying boat was the jet engine. These could be used more
conveniently on long-range land planes, like the De Havilland Comet (the world's first jet
airliner) and the Boeing 707, than in seaplanes which threw up vast quantities of blister and
ribbon spray, depending upon weight, altitude and sea state. Then, no seaplane designer had
tried installing the engines above and behind the wings, near the tail, being developed now
by Beriev, for example, in Russia.
The Wright Brothers flew their aircraft of 1903 in ground-effect, needing a catapult by
means of a falling weight to accelerate it to flying speed. If someone had told those two
bicycle-makers that they had started something big, they might not have believed it. Yet,
within less than 100 years we have massive aeroplanes in operation, especially in Russia and
in the USA. To make Ekranoplane and surface-effect machines equally successful, they must
be fast, with high-order (computerised) controls. One dare not trust hand-flying by a human
pilot at high speed and low altitudes for long periods. There is the ever-present danger of
pilot-induced oscillations to contend with, leading to contact with the sea in a split second. It
happens. For such low altitude and high-speed operations, more advanced materials are
needed. Composites currently available will be dented, and shredded. I lost the wooden
leading-edge of the fin of an experimental aeroplane in rain and hail at high speed. It wore
away. Special metal leading-edges or other forms of reinforcement are needed. A hole in
the gel coat of a composite material lets water get inside the structure of the material, with
unpleasant consequences. Personally, I do not think that the IMO is adequately equipped
and experienced enough yet to cope with certification of large, heavy, fast Ekranoplane/ram-
wing/surface-effect aircraft for public transport operation. A mixture of airworthiness
thinking is needed when carrying passengers - ten of which, in the West, weigh around 1 ton
with baggage. I believe that a joint organization must be created out of ICAO (International
Civil Aviation Organization and IMO (International Maritime Organization), both of which
are organs of the United Nations, if their regulation is to be practical and commercially
useful.
First of all, it has been explained by Dr. Sinitsyn that his estimates for his marine passenger
Ekranoplane show that the fuel consumption can be reached of the order of 30 grams per seat
per kilometer. That is the first thing. Secondly, if you say that this is a future projection, we
may refer even to the history. You know that Lippisch XI13 craft had a fuel consumption
comparable to an automobile, which is about 20 grams per seat per kilometer. This is quite a
remarkable fuel consumption. So, it is a concrete example that fuel consumption matters can
be resolved when you go for efficient aerodynamic configuration. Of course, although we all
have a lot of respect and give credit to naval Ekranoplane developments in Russia,
nonetheless, these naval Ekranoplanes were intended to fulfill another task. This was a task
of destroying rather than getting commercial viability. Economically viable aerodynamic
configurations come with commercialization of Ekranoplanes and can be developed within
aerodynamic concepts of the second generation.
Secondly, speaking about feasibility studies, I should say that if one estimates direct
operating costs and takes into account the necessity to return invested (loaned) money, one
finds that the building cost is the most important. Therefore, even if the fuel consumption is
halved, this would be economy within fraction of the fuel costs, which is only a small part of
total direct operating costs. So, what is really important concerns reduction of building
costs, and that has been explained by Mr. Fischer, and it is also prompted by our own
estimates. In particular, there was an article by Dr. Kirillovykh from the Central Hydrofoil
Bureau, where the building costs for Ekranoplane were estimated to be of the order of 70%
of a similar aeroplane. So 30% better as far as economic feasibility is concerned.
Among other economic factors related to feasibility, one can single out transport
productivity, defined as a product of payload by speed. This is where, evidently, the
Ekranoplane can even get comparable with aeroplanes. Although the speeds are lower, the
capacity to carry large weights and large volumes is very important. And this has been
indicated again by Mr. Fischer. Just in time and large cargo - I think this is an obvious
advantage of the future applications of Ekranoplanes.
We should also take into account that you don't need runways. When thinking about costs,
the price should be considered of not just a craft but that of the whole system and
infrastructure.
Lastly, but not less important, - and it brings us to a comment by Mr. Stinton - is the safety
issue. Note, that we are very exigent towards the Ekranoplane as far as safety is concerned,
whereas all of us are flying aeroplanes, and we know once something happens, then the
casualty percentage nears 100. We do not question that because we all are prepared to die
once there is something. Here, where the casualty percentage will be very low (maybe of the
order of several percent), there have not been casualties during the flights of the large
Ekranoplanes. Of course, there were some cases where people suffered, there was some
damage to health, but your runway, your airport is always beneath. To add, in the case of
emergency, you do not drop with severe impact, there is a cushion underneath which one
sustains, either by dynamic or static air cushion. So, I think issues of safety of the
GD-12
Ekranoplanes are largely exaggerated by people who are probably not quite acquainted with
the possibilities of this emerging novel means of very fast sea transportation.
In my opinion, the most important impact on economics does not come from the technical
parameters, even if this seems strange, but from the infrastructure which accompanies the
operation of Ekranoplanes. If we consider the dispatching services, air traffic control, airport
maintenance, we see that these constitute the lion's share of operating expenses incurred for
transporting these difficult loads. That is why I wish to particularly stress this aspect.
Thank you.
GD-13
Now that we come to the closing, I would like to thank a number of people. First, I want to
thank Prof. Masure, the Chairman of the Program Committee for this Symposium. He has
done a tremendous amount of work, and I thank him very much for this. I would also like to
thank the National Coordinator from the Netherlands, Leo Sombroek who has organized this
for us in these nice surroundings of the Academy of Sciences in the Netherlands. Also, thank
you very much for the social events that you have planned for us. Very many thanks to his
assistant Mayke Langelaar. I would like to thank all of you who have presented papers and
who have taken part in the discussion. Very cheerful thanks to our Technical Evaluator,
Prof. Tulin from Santa Barbara. A very special thanks to our interpreters who have done a
terrific job here, changing from one language to the other. These are Mrs. Goffinet,
Mrs.Celie, Mrs. De Susbielle, Mrs. Pasco, Mr. Bartchenkov and Mr. Kinley. Furthermore, I
would like to thank the technicians who brought all the videos here in time, and organized
the audiovisual system. These are Mr. Yundsen, Mr. Parker and Mr. Allen. Last, but not
least, I would like to thank the staff from the RTA, Research and Technology Agency in
Paris. Our secretary Danielle Pelat, and a very special thanks to Jack Molloy, who has
served the Fluid Dynamic Panel for 5 years and is now serving the Applied Vehicles
Technology Panel. He is leaving us by the end of the year. Jack, thanks very much for the
tremendous work you did for the Fluid Dynamics Panel and for the Applied Vehicles
Technology Panel. With this I would like to close this meeting. At 14:00 hours, we have our
Technical Advancement Exchange Forum, and everybody here is invited to attend this
meeting. In this meeting, we will discuss the future activities of the AVT Panel.
REPORT DOCUMENTATION PAGE
1. Recipient's Reference 2. Originator's References 3. Further Reference 4. Security Classification
of Document
RTO-MP-15 ISBN 92-837-0004-X UNCLASSIFIED/
AC/323(AVT)TP/9 UNLIMITED
6. Title
Fluid Dynamics Problems of Vehicles Operating Near or in the Air-Sea Interface
7. Presented at/sponsored by
The RTO Applied Vehicle Technology Panel (AVT) Symposium (organised by the
former AGARD Fluid Dynamics Panel), held in Amsterdam, The Netherlands,
5-8 October 1998.
8. Author(s)/Editor(s) 9. Date
Multiple February 1999
12. Distribution Statement There are no restrictions on the distribution of this document.
Information about the availability of this and other RTO
unclassified publications is given on the back cover.
13. Keywords/Descriptors
Fluid dynamics Shipboard landing
Air water interactions Computational fluid dynamics
Aerodynamics Wakes
Surface ships Surface effect ships
Three dimensional flow Hydrofoils
Boundary layer separation Hydrofoil craft
Ground effect Ground effect machines
Drag Hydrodynamics
Reduction
14. Abstract
The papers prepared for the RTO Applied Vehicle Technology (AVT) Symposium on "Fluid
Dynamics Problems of Vehicles Operating Near or in the Air-Sea Interface" which was held
5-8 October 1998 in Amsterdam, The Netherlands, are contained in this report. In addition, a
Technical Evaluator's Report aimed at assessing the success of the Symposium in meeting its
objectives, and an edited transcript of the General Discussion held at the end of the Symposium
are also included.
In addition to presentations from the NATO Countries, this Symposium included several
presentations by Russian and Ukrainian authors. In total, 30 papers were presented during
sessions on the following subjects:
• Aerodynamics and Flight Dynamics around Ships
• Stabilization and Control Techniques for Ships
• Non-Classical Aircraft Flying Near the Air-Sea Interface
NORTH ATLANTIC TREATY ORGANIZATION
SALES AGENCIES
NASA Center for AeroSpace The British Library Document Canada Institute for Scientific and
Information (CASI) Supply Centre Technical Information (CISTI)
Parkway Center Boston Spa, Wetherby National Research Council
7121 Standard Drive West Yorkshire LS23 7BQ Document Delivery,
Hanover, MD 21076-1320 United Kingdom Montreal Road, Building M-55
United States Ottawa K1A 0S2
Canada
Requests for RTO or AGARD documents should include the word 'RTO' or 'AGARD', as appropriate, followed by the serial
number (for example AGARD-AG-315). Collateral information such as title and publication date is desirable. Full bibliographical
references and abstracts of RTO and AGARD publications are given in the following journals:
Scientific and Technical Aerospace Reports (STAR) Government Reports Announcements & Index (GRA&I)
STAR is available on-line at the following uniform published by the National Technical Information Service
resource locator: Springfield
http://www.sti.nasa.gov/Pubs/star/Star.html Virginia 22161
STAR is published by CASI for the NASA Scientific United States
and Technical Information (STI) Program (also available online in the NTIS Bibliographic
STI Program Office, MS 157A Database or on CD-ROM)
NASA Langley Research Center
Hampton, Virginia 23681-0001
United States
ISBN 92-837-0004-X