Académique Documents
Professionnel Documents
Culture Documents
Henrique M. Gaspar,* Donna H. Rhodes, Adam M. Ross, and Stein Ove Erikstad*
*Department of Marine Technology, Norwegian University of Science and Technology, Trondheim, Norway
This article approaches the complexity aspects of conceptual ship design from a
systems engineering point of view. We introduce the issue by defining the term com-
plexity in systems engineering, placing the conceptual ship design task as a complex
system problem and creating analogies between generic complex systems and a
ship. Five main aspects of complexity are presented, linking challenges of the con-
ceptual phase to each of the aspects. The structural aspect is related to the arrange-
ment and interrelationship of the physical objects in the ship. The behavioral aspect
derives from the form-function mapping. The external circumstances to which the
ship is subjected are approached through the contextual aspect. Uncertainties in the
scenarios and changes over time are related to the temporal aspect. The perceptual
aspect relates to how stakeholders perceive the value that they receive from a chosen
design. A theoretical study to address these five aspects is presented, applying the
responsive systems comparison method in the conceptual design of an anchor han-
dling a tug supply vessel. The last section discusses why decomposing the com-
plexities of the ship design task in five aspects is a benefit.
1. Complexity in systems engineering and the abstract boundaries in the meaning of the term and, by exclu-
conceptual ship design sion, all the infinite characteristics that the term does not contain.
A simple example is a ship. Is it a structure that floats? Does it
1.1. Defining complexity in systems engineering have its own propulsion? Are barges, canoes, and ships vessels?
Are vessel and ship synonymous?
THE FIELD known as complexity theory has been rapidly evolving
Second, to clarify the previous definition, it is necessary to
in the last few years, spreading complexity thinking in social, bio-
embrace a few arguments from the algorithm information theory
logic, and technical sciences from corporation management (Tasaka
and Kolmogorovs definition of complexity (Kolmogorov 1983):
1999) to the evolution of biologic organisms (Braha et al. 2006).
If any object is simply constructed, then for its description a
In the case of engineering, three seminal works can be used
small quantity of information is sufficient; but if it is compli-
as the basis for a complexity definition. First, Herbert Simon
cated, then its description must contain much information.
(Simon 1962) proposes that how complex or simple a structure
According to certain arguments, it is convenient to call the
is depends critically on the way in which we describe it. Simon
quantity thus introduced the complexity.
proposes a hierarchical approach to complexity, decomposing the
In simple terms, Kolmogorov introduces that the more infor-
system until it can be understood. It is, however, always difficult
mation an object has, the more complex it is. Our object is thus
to describe a structure; by selecting words, we are also defining
the system. The arguments that Kolmogorov refers to can be
understood as the other objects that interact with the system,
Manuscript received by JSPD Committee April 20, 2012; accepted because the specification task of an object is easier when another
July 9, 2012. object to which this object has a relation is already specified.
52
Figure 1 represents the link between complexity and information
with two essential principles. First, when a system requires more
information to be defined, it can be considered more complex.
Second, this information can be observed under many aspects, and
a good system definition will take into account only the relevant
ones. In the remainder of this section, we propose that the structural
and behavioral aspects have dominated when defining and evaluat-
ing a complex system. These two aspects are essential in engineer-
ing, but, when used alone, they exclude important information.
Table 1 Technical Systems Classified by Degree of Complexity (Hubka and Eder 1988)
I (simplest) Part, component Elementary system produced without assembly operations Bolt, bearing sleeve, spring, washer
II Group, mechanism, Simple system that can fulfill some higher functions Gear box, hydraulic drive, spindle head,
subassembly brake unit, shaft coupling
III Machine, apparatus, device System that consists of subassemblies and parts that Lathe, motor vehicle, electric motor
perform a closed function
IV Plant, equipment, complex Complicated system that fulfills a number of functions Hardening plant, machining transfer
machine unit and that consists of machines, groups and parts that line, factory equipment
constitute a functional and spatial unity
Traditional Structural: related to the form of system components and State of the practice: systems architecting and design, and emerging
their interrelationships model-based systems engineering approaches
Behavioral: related to performance, operations, and reactions
to stimuli
New aspects Contextual: related to circumstances in which the systems exists New constructs and methods: seek to advance state of the art,
Temporal: related to dimensions and properties of systems over time for example: set-based design, ship design and deployment problem,
Perceptual: related to stakeholder preferences, perceptions, epoch modeling, epoch-era analysis, multistakeholder negotiations,
and cognitive biases visualization of large data sets
information reflects in a characterization of the complexity in engi- acts with other structural components from the maritime transpor-
neering. As follows, we apply this characterization to the concep- tation system, a system of systems, in which it is included.
tual ship design problem. An overview of the structural aspect is presented in Fig. 2 with
The following aspects are explained through the product point the ship as the main system and the surrounding interconnections.
of view, in which the process is just a means to achieve the desired
physical object. Although one could also address complexities 2.2.2. Behavioral. The behavioral aspect derives the function
in the design task (process), as briefly commented by McDonald from the form, and it is handled by the technical analysis per-
(2010), the focus of this article is on the product (ship). formed during the design phase such as resistance, propulsion,
seakeeping, maneuverability, and stability. It can also be observed
2.2. Five aspects in conceptual ship design as the interaction between the system and a stimulus, either inter-
In the following subsections the five aspects of classification nal stimuli from a subsystem (such as the propulsion engine) or
are applied to the conceptual ship design problem. An introduc- external stimuli from the environment (such as the waves).
tion to the complex system application is discussed by Gaspar Figure 3 illustrates the behavioral aspects.
et al. (2012), but the authors focus mainly on the relation between During conceptual design, the type of analyses required to esti-
uncertainties and the temporal aspect. In this section we extend mate the ship behavior relies to a large degree on empirical formu-
this application to the five aspects as a whole. lation and advanced engineering tools such as regression analysis
models, finite element methods, and computational fluid dynamics.
2.2.1. Structural. The structural aspect is related to the form The traditional performance is thus estimated by the correct
in the basic form-function mapping of the design. It includes mapping between form and function to ensure in the conceptual
the physical objects of the ship, as observed by Levander in phase that a ship X will perform a task Y. Over the past decades,
his system-based ship design methodology (Levander 2006; model-based approaches have been developed to handle the
Erikstad & Levander 2012). structural/behavioral aspect and are now at a relatively mature
Primarily, this aspect relates to the ship as a large, self- practice level (Rhodes & Ross 2010a).
contained system with several highly integrated subsystems such In addition to traditional technical/economical tradeoffs, con-
as propulsion, hull, outfit, and so on. Each of the subsystems ceptual ship design nowadays requires estimation of the behavior
contains many components, which also consists of a physical across a broad number of areas, including risk, safety, and envi-
structure, and interacts with other components, similar to the ronmental performance. It results in a more complex and multi-
classification presented in Table 1. All these subsystems must objective evaluation function.
be provided by the vessel itself within a limited volume, in which
changes on one part may influence other parts through highly 2.2.3. Contextual. The contextual aspect consists of external enti-
interactive relationships. This large structure, the ship, also inter- ties, interfaces, and factors outside of the control of ship designers,
Fig. 2 Structural aspect for the ship as a system, with subsystems and system of systems.
which may affect the behavior of the system and should be taken
into account while designing it. Traditional contextual aspects are These different possible contexts can be interpreted as what if
usually considered fixed and predetermined during the require- situations. If the set of parameters leads to context A, the desired
ments of the elucidation phase. It usually includes a specific scenario system behavior/structure is obtained in design A; if uncertain
with a predefined set of market variables, regulations, and rules. context shift creates a different context B, then another ship may
Andrews and Pawling (2009) defend an extension of these perform better, leading to design B. This idea of the temporal
fixed requirements, extending the context to include those ele- aspect is captured in Fig. 5.
ments that traditionally would be considered in another phase of The context parameters are normally related to some general
the project and/or during the design of another system/subsystem categories of key exogenous uncertainties such as: demand (opening
related to the ship. For instance, estimation of operational profiles of a new market/route); technology development (a new type of
is used in the design of machinery configuration (Gaspar et al. fuel or engine, changes in the hull or ballast system); policy and
2010) and should be taken into account in the estimation of regulations (a new emission control area [ECA] or an SOX /NOX
possible scenarios for the design of complex mission vessels tax); and market shifts (alterations in fuel and freight price, high
such as offshore support vessels. Fleet estimation data such or low global economic growth).
as optimal velocity, total capacity, and a possible contract
scenario also can be fed into the conceptual phase for envi- 2.2.5. Perceptual. The perceptual aspect is related to how the
ronmental assessment (Winnes & Ulfvarson 2006) and risk system is interpreted through the perspective of system stake-
(Berle & Asbjrnslett, 2011). Hagen and Grimstad (2010) also holders. In general terms this aspect intends to answer the question
comment about the addition of contextual information in terms of How is decision X perceived by stakeholder Y? For instance,
of new technologies and environmental rules/milestones in the how the decision of installing a technology to reduce air emission
early phases. (e.g., selective catalytic reduction [SCR]) is considered by:
Figure 4 illustrates contextual elements that should be taken
into account when designing the system. Society ! reduce emission in atmosphere
Shipyard ! more components during construction
Shipowner ! higher construction cost
2.2.4. Temporal. The dynamic properties of the context lead to Ship User ! higher maintenance cost
changes of the external entities that may affect the system over Design Office ! more elaborated mathematical model to
time. The temporal aspect characterizes those shifts of the context estimate air emission during design
during the system lifespan. Uncertainties in how contexts may
unfold are also incorporated as part of this aspect, for instance The perceptual aspect also deals with the perceived good-
the uncertainty related to the operational profile of the ship or ness of design concepts when a stakeholders preferences shift
as a result of the estimation of future contract scenarios. over time. The what if situation is extended, adding other
points of view into the design, to achieve value robustness of
the system.
A key point to realize per this aspect is that the same system
could be perceived to change in value solely as a function of
changes in the stakeholders preferences. That means, the system,
its context, behavior, and everything else could remain the same,
but the system could still fail if people change their perception
and expectations (i.e., preferences) on what they want from the
system. This is distinct from requirements, which are just aggre-
gated, derived snapshots of needs early in the life cycle.
To answer whether design A is more efficient than design B
will require a formal construction of the system key perfor-
mance indicators (KPIs), customized to each stakeholders pref-
erences, without changing the core system. As an example,
Fig. 4 Contextual aspects of the system consider the recent concern with energy efficiency. A solution
Fig. 9 Anchor handling tug supply vessel overall problem and needs statement
Attribute Name Units Utility (0 to 1) Epoch Variables Units Range (minimummaximum) Levels
2
Assumptions:
Supply: cargo tanks volume proportional (/) 0.4 * the cubic number
(CN) [m3]; cargo deck area (/) 0.4*total deck area [m2]
Towing and anchor handling: / bollard pull [ton]
ECA: comply with ECA regulations [y/n]
Availability: / perform a task under a certain time window [% time
window] energy efficiency: (/) fuel consumption given a operation profile
[ton/year]
Energy efficiency: (/) fuel consumption given a operation profile [ton/year]
L/B and B/D ratio: / regression analysis data [adimensional]
Bollard pull: / installed power (182[CN]0.4368).
Extra installed power: / percentage of 182pCN[0.4368 [%]
CAPEX: / size of the ship (10e3 USD/GT) plus extra capabilities
(installed power, ECA, energy efficiency, availability) [M USD]
OPEX: / fuel consumption cost within a given contract [M USD]
Contract revenue / to the contract requirements. A more demanding
contract will pay more. Fig. 10 Tradespace evaluation process
sis of the good design solutions among the epochs. Our design
set was based on static utility; however, it could be obtained from
the performance of any single or set of epochs.
Figure 12 presents the epoch space with all 3168 epochs. An
epoch set is selected to compare epoch variables (i.e., contract
requirements). The life cycle cost was selected rather than the
CAPEX, because now we are taking into OPEX (fuel consump-
tion) and epoch revenue.
Analyzing across many epochs allows one to identify the pas-
sive value robust designs, that is, designs that retain high value
over many different epochs. As an example, Fig. 13 plots the total
utility for Epoch 1032, considered as if the whole life cycle
(25 years) consisted of no contextual changes. The plot also
presents the decomposition of the total utility within the four
selected utilities.
It is also possible to analyze the performance though one spe-
cific set of epochs with common characteristics, for instance, only
epochs with ECA requirements or epochs with high nontransport
capability requirements. From the analysis of many Pareto Frontiers
of each, set it is possible to obtain a multiepoch Pareto Set, exem-
plified by Fig. 14.
4.7. Anchor handling tug supply vessel life cycle path analysis
The last process of the RSC method deals with the perceptual
aspect, focusing on what kind of result is needed to achieve value-
robustness. It includes the comparison of solutions, tradeoffs,
and time-based strategies to select a set of designs that will be
carried forward to be refined after the conceptual phase.
In this example we plot the profit per each of the epochs (the
Fig. 14 Pareto set with efficient design across epochs
revenue of a contract minus the OPEX) and the life cycle profit for
two eras (sum of the revenue of all epochs, minus the CAPEX)
Table 6 Storytelling Process for Construct Era Space
presented in Fig. 15. Net present value is used to calculate the
profit with a symbolic discount rate of 3%, combining the revenue
Expectations Epoch Start End
of each contract, ship CAPEX, and OPEX.
Era 1 An extension of the model may contain the calculation for a
Minimal contracts for the first 10 years 0 2013 2023 given return of investment (ROI) period, which varies from 5 to
SECA requirements is added 8 2023 2028 12 years in shipping. This means the revenue of future contracts
5 years later NECA requirement is added 12 2028 2038 deployed after this period should have much less importance
Era 2 in the selection of the design than the ones deployed on the
Minimal contracts for the first 5 years 0 2013 2018 first years. One way to address ROI in the current model is to
Increase of Bollard Pull (190 tons) 64 2018 2028
construct eras with the duration of only the ROI period, that is,
5 years later, increase of availability (97%) 66 2028 2033
modifying the era length from 25 years to the desired period, for
Increase of Bollard Pull (240 tons) 146 2033 2038
instance 10 years.
The paper shows the application of RSC methodology to The big challenge and opportunity for the authors of this paper
conceptual design of the non-conventional ship. is to get their ideas instantiated in the world of practicing
systems engineers, particularly in government acquisition
The authors considered several performance attributes for ships programs where they are most applicable. The conceptual ship
during this analysis, which is divided in some main utilities. design example in this paper is a good start. A reasonable next
Among these non-transport utilities is included availability, step could be to map this papers ideas more directly to a
which is a key point for the design. specific government agencys formal policies, regulations and
processes for acquisition of complex systems to show how the
Authors Response
We are glad that Professor H. Hopman and Dr. M. A. Yukish ship design, we corroborate with Andrews requirement
each raised a thoughtful critique on the concept of complexity elucidation rather than requirement engineering proposition. The
that we use. When we decided to include it in the paper, we strategy is to tackle the wicked problem of determining what is
knew that any kind of definition could raise different wanted of a ship and what can be afforded [3] via a three stage
interpretations. As stated by Mitchell [2], this is not only a basis: concept exploration, concept studies and concept design.
systems engineering issue but is applicable to the whole
complex theory field. We support Professor Hopmans suggestion for a deeper study
in the relation of human factors and its relation with the
Professor H. Hopman and Dr. M. A. Yukish provide critiques taxonomy. It was not considered in the scope of our work, and
that are primarily related to the classic product/process we agree that attention should be given to how much of the
distinction. While in the structural aspect it is easier to human system is part of (and affects) each of the aspects.
distinguish between the product (ship) and the process
(construction), for other aspects this is not so easily delineated. The decision making problem is another point commented on by
Probably the most overlapping aspect is the perceptual, where Professor Hopman, which is not strictly related to our approach,
the distinction between the product (ship) and the process that it but probably to all proposed design methodologies. Indeed
is placed in (operation, commodities, part of a fleet) varies nowadays we have, given a large design space, a suite of
significantly for each of the stakeholders. In this sense, approaches and metrics to intelligently select a set of good
Professor Hopmans criticism is valid, and a risk-assessment of designs that should be further investigated, such as Pareto Trace
the contextual parameters, for instance, would bring more [4], Filtered Outdegree [5] and Fuzzy Logic [6]. These methods
complexity to the process, but not necessarily to the system. bring some rationale for making an informed decision.
This is exemplified by Dr. M. A. Yukishs case, of a 1940s ship
analyzed under modern techniques: the ship would be the same, We should keep in mind as well that the best design is not
while the information about it would be greater (i.e. more necessarily the most profitable or with best KPIs, but the one
complex). We agree with this distinction, the design process is presenting higher value robustness, strongly connected to the
thus more complex, and not necessarily the product. However, perceptual aspect. It is not wrong to assume that this aspect is
we must add an obvious statement to show how fuzzy the the most uncertain of all, dealing with (the human perception of)
distinction between the product and process is in the conceptual the lack of knowledge from all other aspects. It may not be
phases: if we use the current information level, the actual wrong to assume that the degree of value robustness that a
(complex) design process would certainly lead to a very system must have is strongly connected to the amount of
different ship. In other words, it would be unlikely one would uncertainty that this system needs to (will) face during its
consider the simpler 1940s solution a good one. lifetime.
As for the complexity of the problem, it seems more a question In this sense, we thank Professor Hagen for his discussion. It
of boundaries of the problem rather than the five-aspects touches the core of the question, reminding us how much
taxonomy. Given the basic assumption that the design should residual information exists in a real design process. As stated,
describe a ship able to perform its mission, the challenge seems accepting the low residual (or lower uncertainty) means to spend
to be related to the types of constraints the designer is (or reallocate) the time and money to other points of the design
considering during the process. In this sense, for the conceptual process in order to obtain value robustness, and therefore select
We chose to introduce the method presenting the ship as a single [2] MITCHELL, M. Complexity: A guided tour. Oxford
unit, however it does not limit use of the five aspects in University Press (2009).
considering the ship as made up of subsystems, with each [3] ANDREWS, D. Marine Requirement Elucidation and the
subsystem having its own design variables, and the ship as a Nature of Preliminary Ship Design, Trans. RINA, Vol 153,
combination of modules. We believe that the methodology has Part A1, 2011
potential to be combined with other design methodologies, such [4] ROSS, A.M., Rhodes, D.H., and Hastings, D.E., Using
as the building block approach, both upstream and downstream Pareto Trace to Determine System Passive Value
in the process. Robustness, 3rd Annual IEEE Systems Conference,
Vancouver, Canada, March 2009
Concluding, we are grateful to all of the discussants for their [5] VISCITO, L., Chattopadhyay, D., and Ross, A.M.,
thoughts, comments, and insights. We are pleased that the five- Combining Pareto Trace with Filtered Outdegree for
aspects taxonomy was considered a useful approach to address Identifying Valuably Flexible Systems, 7th Conference on
complexity during the early phases of ship design. We hope that Systems Engineering Research, Loughborough University,
this work can serve as stimulus to more research towards the UK, April 2009
questions raised by the discussants that we were not able to fully [6] GRAY, A. W., Daniels, A. S and Singer, D. J. Impacts of
answer, specifically related to the human system incorporated in Fuzzy Logic Modeling for Constraints Optimization, Naval
the methodology, the overlap of product/process during the Engineers Journal, 2010
conceptual design phase and the impacts of non-rational
information during the early stages of design.
ADDITIONAL REFERENCES
[1] GASPAR, H. M., Balland, O., Aspen, D. M., Ross, A. M. and
Erikstad, S. O., Assessing air emission for uncertain
lifecycle scenarios via responsive systems comparison
method (unpublished) in Gaspar, H. M - Handling Aspects
of Complexity in Conceptual Ship Design, NTNU, PhD
Thesis, 2013.