Computer-Aided Design 42 (2010) 1028–1044

Contents lists available at ScienceDirect

Computer-Aided Design
journal homepage: www.elsevier.com/locate/cad

Holistic ship design optimization

Apostolos Papanikolaou ∗
Ship Design Laboratory, National Technical University of Athens, Greece

article info abstract

Article history: Ship design is a complex endeavor requiring the successful coordination of many disciplines, of both
Received 6 February 2009 technical and non-technical nature, and of individual experts to arrive at valuable design solutions.
Accepted 14 July 2009 Inherently coupled with the design process is design optimization, namely the selection of the best
solution out of many feasible ones on the basis of a criterion, or rather a set of criteria. A systemic approach
Keywords: to ship design may consider the ship as a complex system integrating a variety of subsystems and their
Holistic ship design
components, for example, subsystems for cargo storage and handling, energy/power generation and ship
Multi-objective optimization
Genetic algorithms
propulsion, accommodation of crew/passengers and ship navigation. Independently, considering that
Minimization of resistance and wash ship design should actually address the whole ship’s life-cycle, it may be split into various stages that
Enhanced survivability are traditionally composed of the concept/preliminary design, the contractual and detailed design, the
ship construction/fabrication process, ship operation for an economic life and scrapping/recycling. It is
evident that an optimal ship is the outcome of a holistic optimization of the entire, above-defined ship
system over her whole life-cycle. But even the simplest component of the above-defined optimization
problem, namely the first phase (conceptual/preliminary design), is complex enough to require to be
simplified (reduced) in practice. Inherent to ship design optimization are also the conflicting requirements
resulting from the design constraints and optimization criteria (merit or objective functions), reflecting
the interests of the various ship design stake holders.
The present paper provides a brief introduction to the holistic approach to ship design optimization,
defines the generic ship design optimization problem and demonstrates its solution by use of advanced
optimization techniques for the computer-aided generation, exploration and selection of optimal designs.
It discusses proposed methods on the basis of some typical ship design optimization problems with
multiple objectives, leading to improved and partly innovative designs with increased cargo carrying
capacity, increased safety and survivability, reduced required powering and improved environmental
protection. The application of the proposed methods to the integrated ship system for life-cycle
optimization problem remains a challenging but straightforward task for the years to come.
© 2009 Elsevier Ltd. All rights reserved.

1. Introduction to holistic ship design optimization
Ship design was in the past more an art than a science, highly
dependent on experienced naval architects, with good background
in various fundamental and specialized scientific and engineering
subjects. The design space was practically explored using heuris-
tic methods, namely methods derived from knowledge gained
through a process of trial and error, often over the course of
decades.
Inherently coupled with the design process is design optimiza-
tion, namely the selection of the best solution out of many feasi-
ble ones on the basis of a criterion, or rather a set of criteria. A
systemic approach to ship design may consider the ship as a
complex system integrating a variety of subsystems and their
∗ Corresponding address: Heroon Polytechniou 9, 15 773 Athens-Zografou,
Greece. Tel.: +30 210 772 1416/1409; fax: +30 210 772 1408.
E-mail address: papa@deslab.ntua.gr.
URL: http://www.naval.ntua.gr/sdl.
components, for example, subsystems for cargo storage and han-
dling, energy/power generation and ship propulsion, accommoda-
tion of crew/passengers and ship navigation. They are all serving
well-defined ship functions. Ship functions may be divided into
two main categories, namely payload functions and inherent ship
functions (Fig. 1). For cargo ships, the payload functions are related
to the provision of cargo spaces, cargo handling and cargo treat-
ment equipment. Inherent ship functions are those related to the
carriage of payload safely from port to port with certain speed.
Independently, considering that ship design should actually
address the whole ship’s life-cycle, it may be split into various
stages that are traditionally composed of the concept/preliminary
design, the contractual and detailed design, the ship construc-
tion/fabrication process, ship operation for an economic life and
scrapping/recycling. It is evident that the optimal ship with respect
to her whole life-cycle is the outcome of a holistic 1 optimization of
1 Principle of holism according to Aristotle (Metaphysics): The whole is more than
the sum of the parts.

0010-4485/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cad.2009.07.002
 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044
Structure Hull, poop, forecastle
Superstructures
Crew Facilities Crew spaces
Service spaces
Stairs and corridors
Ship Function
Machinery Engine and pump rooms
Engine casing, funnel
Steering and thrusters
Fuel & lub oil
Tanks
Water and sewage
Ballast and voids
Comfort Systems Air conditioning
Water and sewage
Outdoor Decks Mooring, lifeboats, etc.
Fig. 1. Ship functions, according to Levander [1].
the entire, above-defined ship system for its life-cycle. It is noted
that, mathematically, every constituent of the above-defined ship’s
life-cycle system itself evidently forms a complex nonlinear opti-
mization problem for the design variables, with a variety of con-
straints and criteria/objective functions to be jointly optimized.
Even the simplest component of the ship design process, namely
the first phase (conceptual/preliminary design), is complex enough
to be simplified (reduced2 ) in practice. Also, inherent to ship de-
sign optimization are the conflicting requirements resulting from
the design constraints and optimization criteria (merit or objec-
tive functions), reflecting the interests of the various ship design
stake holders: ship owners/operators, ship builders, classification
society/coast guard, regulators, insurers, cargo owners/forwar-
ders, port operators, etc. Assuming a specific set of requirements
(usually the shipowner’s requirements for merchant ships ormission
statement for naval ships), a ship needs to be optimized for cost ef-
fectiveness, for highest operational efficiency or lowest Required
Freight Rate (RFR), for highest safety and comfort of passen-
gers/crew, for satisfactory protection of cargo and the ship herself
as hardware and, last but not least, for minimum environmental
impact, particularly for oil carriers with respect to marine pollu-
tion in case of accidents and for high-speed vessels with respect to
generated wave wash. Recently, even aspects of ship engine emis-
sions and air pollution need to be considered in the optimization
of ship design and operation (see IMO 2008, [2]). Many of these
requirements are clearly conflicting and a decision regarding the
optimal ship design needs to be rationally made.
To make things more complex but coming closer to reality, even
the specification of a set of design requirements with respect to
ship type, cargo capacity, speed, range, etc. is complex enough
to require another optimization procedure that satisfactorily
considers the interests of all stakeholders of the ship as an
industrial product and service vehicle of international markets
or others. Actually, the initial set of ship design requirements is
2 The principle of reductionism may be seen as the opposite of holism, implying
that a complex system can be approached by reduction to its fundamental
parts. However, holism and reductionism should be regarded as complementary
approaches, as they are both needed to satisfactorily address complex systems in
practice.
1029
Containers
Cargo Units Trailers
Cassettes
Pallets
Bulk / Break Bulk
Payload Function
Holds
Cargo Spaces
Deck cargo spaces
Cell guides
Tanks
Cargo Handling Hatches & ramps
Cranes
Cargo pumps
Lashing
Cargo Treatment Ventilation
Heating and cooling
Pressurizing
the outcome of a compromise of intensive discussions between
highly experienced decision makers, mainly on the ship design and
shipbuilding side, and end-users who attempt to articulate their
desires and tradeoffs they are willing to allow. A way to undertake
and consolidate this kind of discussion in a rational way has been
advanced by the EU-funded project LOGBASED [3].
Since the middle 1960s, with the advance of computer hard-
ware and software more and more parts of the design process
have been taken over by computers, particularly the heavy calcula-
tory and drafting elements of ship design. Simultaneously, the first
computer-aided preliminary design software systems were intro-
duced, dealing with the mathematical parametric exploration of
the design space on the basis of empirical/simplified ship mod-
els for specific ship types or the optimization of design variables
for specific economic criteria by gradient-based search techniques
(Murphy et al. [4], Nowacki et al. [5]). Also, computer-aided stud-
ies on the optimization of a ship’s hull form for least resistance and
best seakeeping behavior (hydrodynamic design optimization), or
of a ship’s midship section/structural design for least steel weight
(structural design optimization) started being introduced to the
naval architectural scientific community until they led to matured
results in more recent years (see, e.g., Papanikolaou et al. [6], Valde-
nazzi et al. [7], Zalek et al. [8]).
With the further and faster advance of computer hardware and
software tools, along with their integration into powerful hard-
ware and software design systems, the time has come to look at
the way ahead in ship design optimization in a holistic way, namely
by addressing and optimizing several and gradually all aspects of
ship’s life (or all elements of the entire ship’s life-cycle system),
at least the stages of design, construction and operation; within
a holistic ship design optimization we should herein also under-
stand exhaustive multi-objective and multi-constrained ship de-
sign optimization procedures even for individual stages of ship’s
life (e.g. conceptual design) with least reduction of the entire real
problem. Recently introduced scientific disciplines in the general
framework of ‘‘design for X’’, namely ‘‘design for safety’’ and ‘‘risk-
based design’’ (SAFEDOR [9], Vassalos [10], Papanikolaou (ed) [11]),
‘‘design for efficiency’’, ‘‘design for production’’, ‘‘design for opera-
tion’’, etc. indicate the need for approaches and the availability of
matured methods and computational tools to address holistically
the ship design optimization problem (Papanikolaou et al. [12]).
1030 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044
The use of Genetic Algorithms (GAs), combined with gradient-
based search techniques in micro-scale exploration and with a util-
ity functions technique for the design evaluation, is advanced in
the present paper as a generic type optimization technique for
generating and identifying optimized designs through effective
exploration of the large-scale, nonlinear design space and a mul-
titude of evaluation criteria. Several applications of this generic,
multi-objective ship design optimization approach by use the de-
sign software platform of the Ship Design Laboratory of NTUA
(NTUA–SDL), integrating well-established naval architectural and
optimization software packages with various application methods
and software tools, as necessary for the evaluation of stability, re-
sistance, seakeeping, structural integrity, etc. may be found in the
listed references (Abt et al. [13]). The following examples, deduced
from recently completed or running EU-funded projects involving
NTUA–SDL, may be highlighted.
• Hull form optimization of a wave piercing high-speed monohull
for least resistance and best seakeeping (VRSHIP-ROPAX2000,
[14,15]).
• Hull form optimization of high-speed mono and twin hulls for
least wave resistance and wave wash (FLOWMART, [16,17]).
• Optimization of the compartmentation of a RoPax vessel for
increased damage stability and survivability and least structural
weight (ROROPROB, [18,19]).
• Optimization of a naval ship for increased survivability in case
of damage in waves and least structural weight [20].
• Optimization of an LNG floating terminal for reduced motions
and wave attenuation on the terminal’s lee side (GIFT, [21,22]).
• Logistics-based optimization of ship design (LOGBASED, [3,23,
24]).
• Risk-based design optimization of an AFRAMAX tanker for in-
creased cargo capacity and least environmental impact (SAFE-
DOR, [7,25]).
For the general concept and details of multi-objective optimiza-
tion by use of GAs and alternative procedures, reference is made
to Lucas [26] and Sen [27]. Comprehensive state of the art re-
ports on modern ship design methods and computer-aided design
procedures were recently presented by Andrews et al. [28] and
Nowacki [29].
In summary, the present paper provides a brief introduction
to the holistic approach to ship design optimization, defines
the generic ship design optimization problem and demonstrates
its solution by use of GAs and related techniques for design
generation, exploration and selection. It discusses proposed ship
design optimization methodology on the basis of two typical multi-
objective ship design optimization problems, namely the hull
form optimization of high-speed vessels for reduced powering and
environmental impact due to the generated wash of waves and the
optimization of roll-on roll-off (Ro–Ro) ferries for least structural
weight/increased transport capacity and enhanced survivability in
case of collision damage.
2. The generic ship design optimization problem
Within a holistic ship design optimization we should herein
mathematically understand exhaustive multi-objective and multi-
constrained optimization procedures with least reduction of the
entire real design problem. The generic ship design optimization
problem and its basic elements may be defined as follows (see
Fig. 2).
• Optimization criteria (merit functions, goals): This refers to
a list of mathematically defined performance/efficiency indica-
tors that may be eventually reduced to an economic criterion,
namely the profit of the initial investment. Independently, there
may be optimization criteria (merit functions or goals) that may
be formulated without direct reference to economic indicators;
see, e.g., optimization studies for a specific X ship function, like
ship performance in calm water and in seaways, ship safety,
ship’s strength including fatigue, etc. The ship design optimiza-
tion criteria are in general complex nonlinear functions of the
design parameters (vector of design variables) and are in gen-
eral defined by algorithmic routines in a computer-aided design
procedure.
• Constraints: This mainly refers to a list of mathematically
defined criteria (in the form of mathematical inequalities or
equalities) resulting from regulatory frameworks pertaining to
safety (for ships mainly the international SOLAS and MARPOL
regulations). This list may be extended by a second set of criteria
characterized by uncertainty with respect to their actual values
and being determined by the market conditions (demand and
supply data for merchant ships), by the cost of major materials
(for ships: cost of steel, fuel, workmanship), by the anticipated
financial conditions (cost of money, interest rates) and other
case-specific constraints. It should be noted that the latter set of
criteria is often regarded as a set of input data with uncertainty
to the optimization problem and may be assessed on the basis
of probabilistic assessment models.
• Design parameters: This refers to a list of parameters (vector of
design variables) characterizing the design under optimization;
for ship design this includes the ship’s main dimensions, unless
specified by the shipowner’s requirements (length, beam, side
depth, draught) and may be extended to include the ship’s
hull form, the arrangement of spaces and of (main) outfitting,
of (main) structural elements and of (main) networking
elements (piping, electrical, etc.), depending on the availability
of topological-geometry models relating the ship’s design
parameters to a generic ship model to be optimized.
• Input data: This includes first the traditional owner’s specifica-
tions/requirements, which for a merchant ship are the required
cargo capacity (deadweight and payload), service speed, range,
etc., and may be complemented by a variety of further data
affecting ship design and its economic life, like financial data
(profit expectations, interest rates), market conditions (demand
and supply data), costs for major materials (steel and fuel), etc.
The input data set may include besides numerals of quantities
also more general types of knowledge data, like drawings (of
the ship’s general arrangements) and qualitative information
that needs to be properly translated for inclusion in a computer-
aided optimization procedure.
• Output: This includes the entire set of design parameters
(vector of design variables) for which the specified optimization
criteria/merit functions obtain mathematically extreme values
(minima or maxima); for multi-criteria optimization problems
optimal design solutions are on the so-called Pareto front and
may be selected on the basis of tradeoffs by the decision
maker/designer. For the exploration and final selection of
Pareto design solutions a variety of strategies and techniques
may be employed.
• In mathematical terms, the multi-objective optimization prob-
lem may be formulated as
min[µ1 (x), µ2 (x), . . . , µn (x)]T , subject to g (x) ≤ 0 and
h(x) = 0 and xl ≤ x ≤ xu
where µi is the i-th objective function, g and h are a set
of inequality and equality constraints, respectively, and x is
the vector of optimization or vector of design variables. The
solution to the above problem is a set of Pareto solutions,
namely solutions for which improvement in one objective
cannot be achieved without worsening of at least one other
objective. Thus, instead of a unique solution, a multi-objective
optimization problem has (theoretically) infinite solutions,
namely the Pareto set of solutions.
 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044
Fig. 2. Generic ship design optimization problem.
The use of Multiple Objectives Genetic Algorithms (MOGAs),
combined with gradient-based search techniques in micro-scale
exploration and with a utility functions technique for the design
evaluation, is advanced in the present paper as a generic type
optimization technique for generating and identifying optimized
designs through effective exploration of the large-scale, nonlinear
design space and a multitude of evaluation criteria occurring in
ship design. Several applications of this generic, multi-objective
ship design optimization approach by use of NTUA–SDL3 ’s design
software system, integrating the naval architectural software
package NAPA
r
, 4 the optimization software modeFRONTIER
r5
and various application software tools, as necessary for the
evaluation of stability, resistance, seakeeping, etc. may be found
in the listed references (see Fig. 3, a sketch of the general approach
to the generic ship design optimization problem).
Some typical examples of application of the introduced generic
ship design optimization procedure of NTUA–SDL are presented
and briefly commented on in the following.
3. Typical ship design optimization problems
3.1. Hull form optimization of high-speed vessels with respect to
powering and wash
3.1.1. Overview of the problem
A ship’s hydrodynamic performance in terms of speed, power-
ing, seakeeping characteristics, maneuverability is of paramount
importance, especially for High-Speed Craft (HSC). Wash wave
generation has worried neither the designers nor the ship oper-
ators until very recently. It is the introduction of numerous large
high-speed vessels that is currently driving maritime authorities
to consider applying to an extent possible rational wash criteria to
the operation of HSC, because of the impact on the marine envi-
ronment and the safety of activities in coastal areas. Therefore, at
least for HSC designs, wash reduction has become a major require-
ment of the vessel’s hydrodynamic performance, along with other
traditional hydrodynamic objectives.
3 National Technical University of Athens–Ship Design Laboratory, NTUA–SDL,
http://www.naval.ntua.gr/sdl.
4 NAPA Oy (2005), NAPA software, http://www.NAPA.fi/.
5 E.STE.CO (2003), ‘‘modeFrontier software v.2.5.x’’, http://www.esteco.it/.
1031
From the conceptual point of view, long and slender hull forms
are recognized for their favorable resistance and wash chara-
cteristics. Increased separation distance of twin-hull vessels will
generally result in wave resistance and wash wave reduction.
Unfortunately, the selection of a vessel’s main particulars is a com-
promise of numerous considerations and constraints, and thus can-
not be dictated only by low wash requirements. Therefore, the
integration of a wash minimization methodology in the design pro-
cess, preferably in the very first stages, when the vessel’s main par-
ticulars are defined and the hull form is developed, is becoming a
prerequisite to reduce the impact of regulatory speed limitations
that will drastically impair the vessel’s ultimate economic poten-
tial. If such a methodology is to be efficient, a reliable wash predic-
tion numerical method has to be available. Although wash wave
prediction is not at all a simple problem, particularly for vessels in
the semi-planning and planning condition, recent progress in CFD
resulted in the development of software tools, either based on the
Kelvin or Rankine source distributions that can be used with a good
degree of confidence. Incorporation of such numerical tools within
an integrated design environment is the main goal of the work
presented herein. Formulation of the ship design procedure in the
framework of a multi-objective optimization problem, where wash
reduction is one of the objective functions, allows the application
of formal optimization methods to derive the optimum hull form
subject to the owner’s requirements and technical and regulatory
constraints. Other objective functions might be the vessel’s total
resistance, seaworthiness, dynamic stability and so on, provided
that adequate numerical tools are available for their reliable and ef-
ficient calculation. In addition, optimization criteria reflecting the
vessel’s economic potential, like the building and operational costs,
transport capacity, net present value or required freight rate, may
also be used.
The present study at NTUA–SDL is focusing primarily on the
minimization of powering and the environmental impacts caused
by excessive wash waves. Thus the selected objective functions are
limited to total resistance and wash wave minimization. To further
simplify the calculations, the effect of the vessel’s propulsion
system, either water-jets or propellers, on the generated wash
waves has been omitted. Omission of objective functions reflecting
the economic performance of the vessels is partly justified by the
imposed condition of constant transport capacity. In practice this
is ascertained by the requirements for a specified minimum Ro–Ro
cargo deck area and constant displacement.
1032 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044
Fig. 3. Generic procedure for the ship design optimization problem.
The selected objectives have been approached as follows:
(a) The total resistance is approximated by the sum of frictional
plus wave resistance, where the frictional resistance is calculated
by use of the ITTC frictional drag coefficient formula. Shipflow
r
, Length at waterline
6 Overall beam
a well-known commercial CFD code of Flowtech, is employed for
the wave resistance and wash wave calculation. Nonlinear iterative
calculations are performed, since it is considered necessary to take
into account the effect of sinkage and running trim on the wave
resistance and wash waves.
(b) For the second objective function, an appropriate wash
wave measuring criterion should be selected for each particular
application, depending on the kind of wash effects to be assessed.
In the present study, basically aimed at demonstrating the
potential of the optimization concept, a simple wash measure has
been adopted, in the form of an ‘average’ wave height W along a
longitudinal wave cut at a certain distance from the vessel’s center
line:
s Z x2
1
W = ζ (x, y)2 dx
x2 − x1 x1
where ζ (x, y) is the wave elevation, while x1 and x2 are the
starting and end points of the integration interval along a wave
cut. Alternative wash criteria can be easily introduced in the
optimization procedure, such as the maximum occurring local
wave height. Wave period or wave length may be also introduced,
combined with wave height to obtain a wash criterion expressing
the local wave energy density. For the solution of this optimization
problem, the generic procedure outlined in Fig. 2 has been applied.
3.1.2. Reference vessels
Two reference high-speed vessels have been selected for the
demonstration of the outlined optimization procedure, namely a
high-speed monohull and a twin-hull vessel. Relevant work has
been conducted within the EU-funded project FLOWMART [16,17].
The selected monohull vessel is the Corsaire 11000, by Leroux-
Navale. The vessel’s main technical characteristics are listed in
Table 1.
6 http://www.flowtech.se.
Table 1
Main characteristics of the selected monohull vessel Corsaire 11000.
Overall length 102 m Transport capacity 566 passengers and 148 cars
87.5 m Propulsion power 4 × 6500 kW
15.4 m Main engines 4 MTU 20V 1163TB73L
Draught 2.5 m Diesel engines
Service speed 37 kn Propulsors 4 KaMeWa waterjets
Model tests for the above vessel were performed by SIRHENA
within the FLOWMART project at a model scale of 1:30, in a
towing tank with a beam of 5 m and depth of 3 m corresponding
to a depth Froude number Fnh = 0.641. Due to the narrow
beam of the towing tank, significant reflections were expected
to affect the measured wash waves. Therefore, calculations have
been performed for the vessel in unrestricted water width and
90 m depth (full scale) and also in a channel of width and depth
corresponding to the dimensions of the towing tank. Comparison
of the predicted vs. measured results at 0.25L and 0.5L transverse
(1) distance off centerline (CL) are presented in Figs. 4 and 5.
In the first part of the wave cuts and for approximately three
ship lengths from the bow, the effect from the limited chan-
nel width on the numerical predictions is comparatively weak.
Further aft, this effect increases significantly with the predictions
for the vessel in the channel comparing much better with the ex-
perimental measurements. A very steep wave crest, approximately
two ship lengths from the bow, can be observed in the experimen-
tal results for the wave cut at 0.25L. This wave crest is approx-
imately 50% higher compared to the numerical predictions. The
same phenomenon is visible in the wave cut at 0.5L, where a steep
wave observed in the experimental measurements between 300 m
and 400 m from the bow is significantly underpredicted by the nu-
merical results.
The second selected vessel is the high-speed catamaran Red Jet
III, designed by FBM. The vessel’s main technical characteristics are
listed in Table 2.
With a length Froude number equal to 0.97 this vessel is
operating in the planning region. Model tests for the above vessel
were performed by MARINTEK, also within FLOWMART [16], at
a model scale of 1:12.5 for a speed range from 10 kn to 33 kn
and water depths equal to 3.75 m, 7.5 m, 15 m and 37.5 m
 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044 1033

1.50
Calculated, NTUA-SDL (infinite water width)
Calculated, NTUA-SDL (in a channel)
Measured, SIREHNA (in a channel)
1.00

Wave Elevation (m)

0.50

0.00
0 100 200 300 400 500 600 700

-0.50

-1.00
Distance from Bow (m)

Fig. 4. Comparison of measured and predicted wave cuts at 0.25L off CL for the monohull vessel.

1.50
Calculated, NTUA-SDL (infinite water width)
Calculated, NTUA-SDL (in a channel)
Measured, SIREHNA (in a channel)
1.00
Wave Elevation (m)

0.50

0.00
0 100 200 300 400 500 600 700

-0.50

-1.00
Distance from Bow (m)

Fig. 5. Comparison of measured and predicted wave cuts at 0.5L off CL for the monohull vessel.

0.40
Experimental Results (MARINTEK)
0.30 Numerical Predictions (NTUA-SDL)
Wave elevation (m)

0.20

0.10

0.00
0 25 50 75 100 125 150 175 200
-0.10

-0.20

-0.30
Distance from bow (m)

Fig. 6. High-speed catamaran, comparison of measured vs. predicted wave cuts at 0.845L off CL and 7.5 m water depth.

Table 2 Hull form development

Main characteristics of the high-speed catamaran vessel Red Jet III. The developed optimization procedure is based on the paramet-
Overall length 32.9 m Service speed 33 kn ric generation of alternative hull forms by use of NAPA r
. Careful
Length at waterline 29.58 m Transport capacity 120 passengers identification of the most suitable design parameters, along with
Overall beam 8.32 m Propulsion power 2 × 1360 kW their appropriate range of variation, is needed to ascertain the gen-
Demihull beam 2.27 m Main engines 2 MTU 12V 396 TE 84
Draught 1.133 m Propulsors 2 MJP 650 water-jets
eration of feasible and efficient hull forms. For the monohull case
the hull form generation is controlled by a set of points and inclina-
tion angles. Through these points a grid is created that defines the
(full scale). Comparison of the predicted vs. measured results at a hull. In Fig. 8 a perspective view of a typical hull form is presented,
speed of 30 kn and at 0.845L transverse distance off centerline are where the grid and the definition points are shown.
presented in Figs. 6 and 7. For the twin-hull case, a two-step hull form development
Satisfactory agreement between the numerical results and the procedure was adopted. First, an auxiliary hull form is derived,
experimental measurements is obtained up to 5.5 ship lengths again using appropriate definition points (see Fig. 9). This hull
from the bow. Further aft, considerable discrepancies between the is characterized by a long knuckle line, throughout the vessel’s
two curves can be observed, possibly due to the proximity to the length, and by transverse sections with straight-line segments at
aft limit of the free-surface panelization area. the bottom and the side. Two transition curves are then projected
1034 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044

0.40
Experimental Results (MARINTEK)
0.30 Numerical Predictions (NTUA-SDL)

Wave elevation (m) 0.20

0.10

0.00
0 25 50 75 100 125 150 175 200
-0.10

-0.20

-0.30
Distance from bow (m)

Fig. 7. High-speed catamaran, comparison of measured vs. predicted wave cuts at 0.845L off CL and 15 m water depth.

Fig. 8. Grid definition and resulting hull form for monohull vessel.

Fig. 9. Grid definition and resulting hull form for catamaran vessel.

on the side and the bottom (dashed lines in Fig. 9). A new grid Scatter Chart. Rt vs. Wash
0.25
defining the final hull is created, rounding the transverse sections
between the two transition curves. Appropriate macros have been 0.24
developed to facilitate the parametric modeling of the hull forms,
0.23
making full use of the NAPA macro language. After creating the
Wash

hull forms, other macros are executed to check compliance with 0.22
the geometric constraints and to prepare appropriate output files,
describing the geometry in a form suitable to be processed by 0.21
Shipflow.
0.2

3.1.3. Results of optimization 0.19

450 460 470 480 490 500 510 520
Typical results of the hull form optimization of the above-men- Rt
tioned semi-displacement monohull and the high-speed catama-
ran vessel are discussed in the following sections. Fig. 10. Monohull vessel, total resistance RT (kN) vs. wash W (m).

3.1.3.1. Optimization of the monohull vessel. In Fig. 10, a scatter di-
agram of the wash wave measure W versus total resistance RT
(approximated by RT = RF + RW ) for the generated designs is
shown. The corresponding values of the original vessel (according
to Shipflow calculations) are presented with the thick solid circle at
the upper right corner of the diagram.
A number of designs with favorable hydrodynamic character-
istics are identified. The obtained reductions in resistance, wash
wave measure and maximum wave height, compared to the orig-
inal vessel, are presented in Table 3. Similar comparisons were
obtained from the results calculated for the wave cuts located at
0.25L and 0.75L transverse distance from the vessel’s centerline.
The boundary line (‘Pareto Frontier’) shown in Fig. 10 corresponds
to the best obtainable results. All the designs located on that line
are considered ‘optimal’, since it is impossible to improve the ves-
sel’s performance with respect to one criterion without impairing
 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044 1035

Table 3
Comparison of obtained results.
RT (kN) Diff. % W (m) Diff. % H max (m) Diff. %

Original vessel 500.5 0 0.205 0 1.0515 0

Hull no 47 449.3 −10.2 0.173 −15.6 0.8840 −15.9
Hull no 118 464.3 −7.2 0.160 −22.0 0.789 −24.9
Hull no 282 494.4 −1.2 0.155 −24.4 0.7473 −28.9

0.60 0.15
Original Hull
Hull 47 Original
Wave Elevation at 0.5L (m)

0.40 Hull 118

Hull 282

0.20 0.14

Wash (m)
0.00
-100 0 100 200 300 400 500 600
-0.20 0.13

-0.40
Hull 98B15W
-0.60 0.12
62 62.5 63 63.5 64 64.5 65 65.5 66
Distance from Bow (m)
Rt (t)
Fig. 11. Monohull vessels, comparison of free-surface elevation at LPP /2 off CL.
Fig. 12. High-speed catamaran vessel, total resistance RT (kN) vs. wash W (m).

its performance with respect to the others. It is the designer’s re- 0.40
sponsibility to select the most preferable solution among the de- Original Hull
signs located on this line, based on his experience and possibly 0.30 Hull 98B15W
further evaluation criteria. Decision support tools, like the Utility Hull 98B125W
Wave Elevation (m)

Functions Technique, are available within modeFrontier to assist 0.20

the designer in this selection procedure.
The results presented in Fig. 11 were calculated using 2×686
panels on the wetted surface and 2 × 3345 panels on the
free surface. A maximum number of four iterations was set,
resulting in approximately 630 s CPU time per vessel using a
DEC XP1000 workstation. The results presented in Table 3 have
been recalculated with a larger free-surface panelization area,
corresponding to almost 100% of the computer capability (2 ×
7742 free-surface panels: 7602 s CPU time). Convergence has been
obtained after nine iterations for all vessels. Note that the wash
measure W in Table 3 was also recalculated for a larger integration
interval (with its aft end at the zero down-crossing point of the
wave cut, situated closer to the point at 500 m behind the vessel’s
bow). The corresponding free-surface elevation at a wave cut
located LPP /2 from the vessel’s track is presented in Fig. 11.
3.1.3.2. Optimization of the catamaran vessel. A series of optimiza-
tion studies was performed for the selected high-speed catamaran
vessel. A diagram showing the obtained wash wave measure W
versus the total resistance RT for each vessel tested during the final
stage optimization is presented in Fig. 12. The corresponding val-
ues for the original vessel (according to Shipflow calculations) are
indicated by a thick solid circle. It should be noted that the orig-
inal Red Jet III hull form was already carefully designed for low
resistance and wash waves and it was anticipated from the be-
ginning that it would be rather difficult to obtain further improve-
ments by the applied optimization. This has been confirmed by the
obtained results, at least regarding the resistance, where the maxi-
mum obtained reduction is in the order of 0.7%. However, a number
of hull forms can be identified in Fig. 12 with quite favorable wash
characteristics. Among them, Hull no 98B15W, with a reduction of
13.8% in wash measure W , and a total resistance practically equal
to the original, may be regarded as a strong alternative to the origi-
nal hull form. The obtained reductions in resistance, wash measure
and maximum wave height, compared to the original vessel, are
presented in Table 4.
The calculated running trim of Hull no 98B15W was equal to
0.163◦ by the bow. To investigate the effect of running trim on
0.10
0.00
0 50 100 150 200
-0.10
-0.20
-0.30
Distance from Bow (m)
Fig. 13. Catamaran vessels, comparison of free-surface elevation at LPP /2 off CL.
the generated wash waves, a series of variants of this hull have
been tested. Results from one variant (Hull no 98B125W) with zero
running trim are included in Table 4.
The results presented in Fig. 12 were calculated using 2×348
panels on the wetted surface and 2 × 1763 panels on the free
surface. A maximum number of five iterations was set, resulting in
approximately 310 s CPU time per vessel. The results presented in
Table 4 have been calculated with a larger free-surface panelization
area (2 × 8456 free-surface panels: 34960 s CPU time) and for ten
iterations. The wash measure W in Table 4 was again calculated
for a larger integration interval, resulting in a significantly lower
value in comparison with the results of Fig. 12. The corresponding
free-surface elevation at a wave cut located LPP /2 from the vessel’s
center line is presented in Fig. 13.
3.1.4. Conclusions
It has been demonstrated that the developed procedure is a
valuable design tool for the hull form development of a variety of
high-speed vessels. Applying local shape variations and under the
assumptions of constant speed, displacement, length at waterline
and separation distance (for the twin-hull case) the obtained
reduction of the maximum wave height according to the numerical
wash waves predictions is up to 30% for the monohull vessel
and up to 15% for the catamaran, noting that both original
vessels are carefully optimized designs, with good hydrodynamic
1036 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044
Table 4
Comparison of obtained results.
RT (kN) Diff. %
Original vessel 64.09 0 0.116
Hull no 98B15W 63.65 −0.7
Hull no 98B125W 63.76 −0.5
and market performance. The method allows the introduction of
further objective functions and constraints in accordance to the
designer’s needs in daily practice. The application to other types of
ships designed for lower speed appears straightforward and even
less problematic, because the simpler calm water hydrodynamics
can be better assessed by suitable hydrodynamic software tools.
The validity of the results of the presented optimization procedure
is strongly dependent on the accuracy of the numerical approach
employed for the hydrodynamic evaluation of the alternative
hull forms. Although results from the comparison of numerical
predictions with available experimental measurements indicated
encouraging agreement, it is however recognized that further
work is necessary to improve the numerical approach employed.
While employing potential theory for the assessment of the
vessel’s resistance and wash characteristics represents a major
simplification, it is however anticipated that even if the method
fails to provide very accurate results on the absolute values of the
objective functions (particularly regarding the vessel’s resistance)
it still remains a quite useful tool, particularly in the initial design
stage, enabling a fast design space exploration and assisting the
designer to distinguish good from bad designs and to classify
alternative design solutions in the right performance order.
3.2. Optimization of compartmentation of Ro–Ro passenger ships for
enhanced safety and efficiency
3.2.1. Overview of the problem
The introduction of the probabilistic damaged stability regula-
tory concept (A.265 [30]), about 35 years ago, as an alternative to
the deterministic SOLAS 74 requirements, has been considered as
a major step towards the rationalization of the procedure of as-
sessing a ship’s stability following damage. The new SOLAS damage
stability regulations that entered into force on January 1, 2009 and
which is applicable to all new buildings of any passenger ship or
dry cargo type is entirely based on the probabilistic concept; thus
designers are now forced to learn to work with the probabilistic
concept, which is very complex and less transparent, compared to
the traditional deterministic concept. The related computational
effort is quite significant and can be carried out only by use of spe-
cialized software programmes, which need to be interfaced with
other ship design software tools and optimization procedures.
This lack of design experience and systematic research moti-
vated the set-up of an EU-funded project on the ‘‘Probabilistic
Rules-Based Optimal Design of Ro–Ro Passenger Ships, RORO-
PROB’’ [18]. The project was completed in 2003, and it aimed to
develop and implement an integrated design methodology for the
optimal subdivision of Ro–Ro passenger ships, based on the prob-
abilistic damage stability regulations.
The following outline is related to the work of NTUA–SDL
within the ROROPROB project and refers to the development
of a formalized multi-objective optimization procedure for the
internal compartmentation of Ro–Ro passenger ships, based on the
probabilistic approach for the damage stability assessment [19].
The objectives of the optimization are the maximization of the
ship’s resistance against capsize, expressed by the Attained
Subdivision Index and of her transport capacity, in terms of both
increased deadweight and garage deck space. Alternatively, the
W (m) Diff. % H max (m) Diff. %
0 0.546 0
0.100 −13.8 0.462 −15.4
0.105 −9.5 0.496 −9.2
Attained Subdivision Index7 may be treated as a constraint (in
the form Attained Subdivision Index ≥ Required Subdivision Index)
and the optimization may be performed with respect to the
maximization of the transport capacity and minimization of the
building cost, an approach closer to a ship-owner’s perspective.
Building cost reduction is herein considered mainly as the
result of steel weight minimization. The reduction of the number
of watertight compartments below the subdivision deck is also
considered to have a significant impact besides structural weight,
also on equipment costs.
3.2.2. Outline of the procedure
The adopted procedure is based on the integration of a well-
known commercial ship design software package (NAPA) and a
general-purpose optimization software package (modeFRONTIER),
in the frame of the generic solution procedure outlined in Fig. 2.
The vessel’s watertight subdivision is automatically generated,
assuming the hull form and the main layout concept given, based
on a number of design variables and design parameters. For each
design variant the Attained Subdivision Index, along with the total
vehicle lane length in the lower hold and main garage deck, and the
steel weight up to the top of the main garage deck are calculated.
The main features of the adopted procedure are outlined in the
following sections.
3.2.2.1. Parametric development of internal arrangement. Appro-
priate NAPA macros have been created for the generation of the
ship’s internal watertight arrangement based on a set of design
variables, forming the so-called ‘design space’, and in addition on
a set of design parameters supplied by the user. The design vari-
ables are systematically updated during the optimization, using
appropriate utilities within modeFrontier to perform the design
space exploration. The user-supplied design parameters are used
to define the vessel’s intact loading conditions in partial and full
draught, and to provide necessary data for a variety of calculations
(specific weights for the structural weight calculation, vehicle di-
mensions for the lane length calculation, etc.). The design parame-
ters are kept constant during the optimization. Selected quantities
may be treated either as design variables or parameters, depend-
ing on the user’s intentions or the specific requirements of each de-
sign case. For example, in the special case in which the watertight
subdivision optimization is restricted to the area of the vessel for-
ward of the Main Engine Room, the user may treat the correspond-
ing design variables defining the aft ship compartmentation as
parameters.
Following the generation of the internal layout, the procedure
continues with the assessment of each design variant, making
full use of the calculation capabilities available within NAPA.
Appropriate NAPA macros have been developed to control the
damage stability analysis, to calculate the structural weight and
transport capacity (both in terms of DWT and lanes length) and to
verify the consistency of each design.
Characteristic designs with both central and side casing ar-
rangements on the main deck, generated by the above procedure,
are shown in Figs. 14 and 15. Apart of the main casing, small side
7 See the list of notions in the Appendix.
 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044 1037

Fig. 14. Design variant with forward and aft lower hold and central casing.

Fig. 15. Design variant without aft lower hold and side casings.

casings at the aft end of the main deck can be seen in both arrange-
ments. The position of the transverse bulkheads, the longitudinal
and transverse extent of both lower holds (forward and aft, if any),
the vertical position and the extent of double bottom and all the
other details of the internal arrangement are controlled by the set
of design variables. In the studies presented herein, 43 design vari-
ables are used to define the vessel’s internal layout along with 28
design parameters. According to the user’s selection, a subset of
the design variables is used to define the design space, while the
remaining variables are kept fixed during the optimization.
3.2.2.2. Damage stability calculations. The calculations for the At-
tained Subdivision Index have been performed according to the
probabilistic damage stability concept. Although we are dealing
with the design of Ro–Ro Passenger vessels, the results presented
are based on Regulation 25 of SOLAS Part B-1, originally applica-
ble to cargo ships. However, the developed optimization procedure
can be easily extended to account for the latest harmonized proba-
bilistic damage stability formulations (SOLAS 2009) considered by
the NAPA software package.
3.2.2.3. Optimization procedure implementation. The coordination
of the optimization procedure is performed using the modeFrontier
software package, providing the means for the definition and
control of the calculation chain and for the integration of the
necessary external software packages. A graphical user interface
is used for the implementation and review of the optimization
logical scheme (see Fig. 16). The input variables, along with
their variation interval and the necessary design parameters, are
defined in relevant input files. Links to the appropriate external
applications are established with the help of batch files, permitting
modeFrontier to control the procedure’s execution and to perform
the required data transfer between the various directories and/or
computers on the network. The selection of the appropriate
1038 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044
Fig. 16. Logical scheme of the applied procedure for the multi-objective optimization of the watertight compartmentation of Ro–Ro passenger ships for enhanced efficiency
and safety.
Fig. 17. Hull form of the selected Ro–Ro passenger ferry.
optimization scheduler depends on the particular problem to be
solved. In our case studies both SIMPLEX and a Multiple Objectives
Genetic Algorithm (MOGA) have been used. For the analysis of the
output of the optimization procedure, the various options provided
by modeFrontier (parallel graphs, scattered diagrams and Student
plots) were used. The latter are used to evaluate the importance of
each input variable with respect to the output values.
3.2.3. Case studies
Systematic case studies have been performed applying the
above procedure to a sample Ro–Ro passenger ship (Fig. 17). The
vessel’s particulars and the definition of the two initial loading
conditions (full and partial load) are presented in Table 5. The
calculation of the heeling angles was limited to 60◦ and no down-
flooding openings were defined in the case studies presented
herein. The permeability of the garage spaces is set equal to 0.90;
for the engine rooms it is 0.85, and for the rest of the spaces it is set
equal to 0.95.
3.2.3.1. Multi-objective optimization for maximum A and transport
capacity and minimum structural weight. In the presented optimi-
zation study the maximization of both the Attained Subdivision
Index A and the lane meters is addressed, while the structural
Table 5
Ship’s particular and loading conditions.
Overall length 193.6 m
Length b.p. 176.0 m
Breadth 25.0 m
Depth (reference) 9.100 m
Design draught 6.550 m
Full load draught 6.520 m
Full load displacement 17 520 t
Full load reference GM 2.440 m
Partial load draught 5.884 m
Partial load displacement 14 880 t
Partial load reference GM 1.830 m
weight is minimized. It is obvious that the first two objectives are
contradicting because the maximization of the A requires dense
compartmentation, while this will limit the lower hold length and
thus the total lane meters. The minimization of the structural
weight is also a contradicting objective against the maximization of
A. For demonstration of the applied methodology, the results for a
central casing configuration on the main vehicle deck are presented
in the following (see 3.2.3.2 for the assessment of the side casing
configuration).
The vessel’s internal arrangement optimization is restricted
to the area forward of the Main Engine Room, keeping the aft
part arrangement fixed. The logical scheme of the optimization
procedure is shown in Fig. 18. Seven design variables describing
the compartmentation forward of the Main Engine Room were
selected to define the design space (free variables). A constraint
for the minimum acceptable value of the Attained Subdivision
Index was imposed. The Multiple Objectives Genetic Algorithm
(MOGA) optimization scheduler has been selected for the actual
optimization, and an initial population of 42 designs was randomly
generated. The number of initial designs was estimated by a rule
of thumb suggesting ‘‘2 × number of variables × number of
objectives’’. The optimization process was subsequently initiated
for 30 generations with a probability of directional crossover of 0.5,
probability of selection 0.5, and a probability of mutation of 0.1. A
total of 950 designs were created and evaluated. The Pareto design
results of this study are presented in Figs. 19–21.
 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044 1039

Fig. 18. Logical scheme of the developed procedure for the case study.

Scatter Chart- outA vs. OutWeight

1.021

1.018

1.015

1.012

1.009

1.006

1.003

1.000

1.997

1.994
OutWeight

1.990

1.987

1.984

1.981

1.978

1.975

1.972

1.969

1.966

1.963

1.960

1.957

1.953
0.791 0.796 0.801 0.806 0.810 0.815 0.820 0.825 0.830 0.835 0.839 0.844 0.849 0.854 0.859 0.864 0.868 0.873
outA

Fig. 19. Pareto scatter diagram of maximum A index vs. minimum structural weight.

The designer’s selection of the ‘best’ out of the generated Pareto
designs may be supported by the Utility Functions Technique of
Multi-Criteria Decision Making [E.STE.CO (2003), ‘‘modeFrontier
software v.2.5.x’’, http://www.esteco.it/]. Assuming equal weights
for the three objectives, the ranking of the studied central casing
designs is shown in Fig. 22, whereas in Fig. 23 the general
arrangement of the resulting optimum design (number 782) is
shown.
When giving a higher preference to the cargo capacity (lane
meters), which is considered closer to the classical design
expectations of a potential ship owner, as the lane length has an
immediate impact on the economic value of the ship, while the
Attained Subdivision Index should be just over the limit set by the
safety regulations, the results change, as is shown in Figs. 24 and 25.
3.2.3.2. Comparison between the central and side casing configura-
tions. A comparison of obtained results, when optimizing for both
the central and side casings configurations, is presented in Figs. 26–
28. From these figures it may be observed that the side casings
configuration results in considerably increased transport capacity,
combined with an appreciable increase of the Attained Subdivision
Index. The increase of transport capacity is mainly attributed to the
more efficient utilization of the main deck area. In addition, the ex-
istence of side casings has a positive impact on the vessel’s stabil-
ity characteristics following damage, enabling the increase of the
lower hold area while fulfilling the requirements for an increased
Attained Subdivision Index. No significant differences between the
two design concepts regarding structural weight may be observed
from the comparison. However, it should be noted that the struc-
tural weight calculations were herein based on predefined specific
weights per square meter for the various sections of the ship, re-
gardless of the selected design concept. In this respect the com-
parison may be to a certain degree biased towards the side casings
configuration, considering that the central casing configuration
inherently disposes an increased structural stiffness. In practice,
heavier transverse beams, deep longitudinal girders and a number
of pillars are necessary to support the deck weight of side casings
concepts.
1040 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044

Scatter Chart- outA vs. OutLaneMeters

1.071

1.065

1.059

1.053

1.047

1.041

1.035

1.029

1.023

1.017
OutLaneMeters

1.011

1.005

1.999

1.993

1.987

1.981

1.975

1.969

1.963

1.957

1.951

1.945

1.939
0.791 0.796 0.801 0.806 0.810 0.815 0.820 0.825 0.830 0.835 0.839 0.844 0.849 0.854 0.859 0.864 0.868 0.873
outA

Fig. 20. Pareto scatter diagram of maximum A index vs. maximum lane length.

Scatter Chart- OutLaneMeters vs. OutWeight

1.021

1.018

1.015

1.012

1.009

1.006

1.003

1.000

1.997

1.994
OutWeight

1.990

1.987

1.984

1.981

1.978

1.975

1.972

1.969

1.966

1.963

1.960

1.957

1.953
0.939 0.948 0.956 0.964 0.972 0.981 0.989 0.997 1.005 1.013 1.022 1.030 1.038 1.046 1.055 1.063 1.071 1.079
OutLaneMeters

Fig. 21. Pareto scatter diagram of maximum lane length vs. minimum structural weight.

3.2.4. Conclusions
A multi-objective optimization procedure has been presented,
aiming to assist the designer of Ro–Ro Passenger ships in the
preliminary design stage, when the layout of the internal water-
tight subdivision is investigated, considering the impact of pro-
babilistic damage stability regulations and aspects of efficiency
and building cost. The developed procedure is based on the
integration of modeFrontier, an environment for Multi Objective
and Collaborative Design Optimization with NAPA, a well-known
and versatile naval architectural design software package. Results
from the application of the above procedure revealed its potential
as a useful and practical design tool, enabling the designer to assess
systematically and in very short time hundreds of alternative
layouts, subject to a variety of constraints and objective functions
related to the ship’s efficiency and safety. The developed procedure
can be used either to generate a vessel’s internal subdivision
from scratch, or to improve significantly an existing design. It
allows the designer to gain a better overview of the design space
 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044 1041

Fig. 22. Ro–Ro ship central casing design ranking: Uniform weights of objectives.

Fig. 23. Ro–Ro ship central casing optimum design (number 782): Uniform weights of objectives.

and to obtain a better compromise of the contradicting design
objectives. The huge amount of calculations for a vessel’s damage
stability assessment required by the probabilistic approach leads
to a calculation time of about 3.5 min using a PC with a Pentium IV
microprocessor at 2.4 GHz, for the evaluation of each vessel. This
calculation time could be significantly decreased by the use of more
powerful PC computers that have become available in recent years,
noting that the results shown were originally generated in 2003.
The extension of the above procedure to other types of ship appears
straightforward, especially for ships with fewer complicated
arrangements, like cargo ships, as their compartmentation can be
generated by significantly fewer design parameters.
4. Conclusions and the way ahead
The present paper provided a brief introduction to the holistic
approach to ship design optimization, defines the generic ship
design optimization problem and demonstrates its solution by
use of Genetic Algorithms and a developed integrated ship design
optimization procedure. This was applied to two distinct examples,
1042 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044

Fig. 24. Ro–Ro ship central casing design ranking: Non-uniform weights of objectives.

Fig. 25. Ro–Ro ship central casing optimum design (number 651): Non-uniform weights of objectives (enhanced lane meters).

namely the optimization of hydrodynamic performance and
environmental impact of high-speed vessels and the optimization
of Ro–Ro ships for enhanced survivability and transport efficiency.
It was shown that multi-objective mathematical optimization
approaches are very valuable tools that greatly enhance the quality
of ship design, even if applied to vessels already optimized by
traditional methods. In the list of references further applications of
the presented approach to other design problems may be found. It
is pointed out that this list is to a great extent limited to references
of work of the Ship Design Laboratory of NTUA and is indeed not
exhaustive.
A final comment on the way ahead: though the generic solution
approach to the holistic ship design problem appears well estab-
lished, it remains for researchers to develop and integrate a long
 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044 1043

1.03 any such knowledge, information or data, or of the consequences

Central Casing
1.02 Side Casing
thereof. The author would like to thank his associates, Ass. Prof. G.
Zaraphonitis, Dr. E. Boulougouris, Dr. E. Eliopoulou and Dipl.-
Weight (nondimensional)

1.01
Eng. D. Mourkoyiannis, for their contribution to the presented
1.00 work. Finally, the author would like to express his gratitude to
0.99 his mentor Professor Horst Nowacki for introducing him, during
0.98 his studies and work at the Technical University of Berlin, into
the philosophy, methods and practice of ship design optimization
0.97
at a time when only very few scientists around the world were
0.96 dealing systematically with this in many respects very demanding
0.95 discipline.
0.94
0.90 0.95 1.00 1.05 1.10 1.15 Appendix
Lane meters (nondimensional)

Fig. 26. Scatter diagram of lanes vs. weight for both central and side casing Important notions used in the paper
configurations.
• Holism (from Greek óλoς , meaning entire, total)
1.03 • Reductionism-reduction: Sometimes interpreted as the opposite
Central Casing
1.02 Side Casing of holism.
• ‘‘A complex system can be approached by reduction to its
Weight (nondimensional)

1.01
fundamental parts.’’
1.00 • Holism and reductionism need, for proper account of complex
0.99 systems, to be regarded as complementary approaches to
system analysis.
0.98
• Risk (financial): ‘‘A quantifiable likelihood of loss or of less-than-
0.97 expected returns.’’
0.96 • Risk (general): ‘‘A quantifiable likelihood of loss of an acceptable
0.95 state or of a worse-than-expected state condition.’’
• Safety: May be defined as ‘‘An acceptable state of risk’’.
0.94
0.74 0.76 0.78 0.80 0.82 0.84 0.86 0.88 0.90 • Survivability: In engineering, survivability is the quantified abil-
Attained Index ity of a system, subsystem, equipment, process, or procedure to
continue to function during and after a natural or man-made
Fig. 27. Scatter diagram of attained index vs. weight for both central and side casing disturbance; a ship’s survivability may be defined as the abil-
configurations.
ity of the ship to continue to function after an environmental
1.15
disturbance (e.g. effect by seaway) or a damage to her hull or
Central Casing equipment caused by collision, grounding or weapon impact
Lane meters (nondimensional)

Side Casing
(naval ships).
1.10
• Optimization: ‘‘The identification of the best out of a series of
many feasible options.’’
1.05 • Holistic ship design optimization: ‘‘The multi-objective optimiza-
tion of ship design considering simultaneously all (holistically)
1.00 design aspects of the ship system and for the entire ship’s life-
cycle.’’
– Major design objectives: Performance, safety-risk-surviv-
0.95
ability, cost.
– Major design constraints: Safety regulations, state of market
0.90 (demand, supply, cost of steel, fuel, etc.), others largely case
0.74 0.76 0.78 0.80 0.82 0.84 0.86 0.88 0.90
specific.
Attained Index
• Holistic Risk in Optimization: Considering the risk investing in
Fig. 28. Scatter diagram of attained index vs. lane meters for both central and side new shipbuilding, the design of which should be holistically
casing configurations. optimized, we might interpret the Holistic Ship Design Opti-
mization also as a generic Risk-based Ship Design Optimization,
list of application algorithms and related software, addressing the where the risk of an investment with specific profit expectation
great variety of ship design for the life-cycle. This is a long-term is minimized, or the profit maximized for an acceptable risk.
task of decades, requiring profound skills and understanding of the • The Attained Subdivision Index, A, is a measure for the probability
physics and design of ships, a domain requiring properly trained of survival of a ship in case of a statistically probable damage. It
naval architects and scientists of related disciplines. should be less than the so-called Required Subdivision Index, R,
which is the minimum value for the Attained Subdivision Index
Acknowledgements A and represents a generally accepted (imposed in regulations)
survival level for the ship under consideration, corresponding
The presented work has been partly financially supported by to her size and the number of people onboard exposed to the
the European Union under FP5 and FP6, projects FLOWMART, collision hazard. Thus, through the direct comparison of A and
ROROPROB and SAFEDOR. The European Community and the R of a ship, her level of relative safety with respect to her
author shall not in any way be liable or responsible for the use of survivability in case of collision is established.
1044 A. Papanikolaou / Computer-Aided Design 42 (2010) 1028–1044
pi vi si , where
P
• The general formulation of the index A is A =
the sum has to be taken over all watertight compartments or
group of compartments. Herein, the factor pi represents the
probability that the compartment or group of compartments
under consideration (i) is flooded, without the consideration
of possibly fitted horizontal subdivisions (boundaries) of com-
partment i, and vi the probability that a space above an existing
horizontal boundary is not flooded. Both the above factors di-
rectly depend on the geometry of the ship’s construction and
are determined by a statistical analysis of systematically col-
lected damaged ship data. The factor si represents the proba-
bility of survival after flooding of the compartment or group of
compartments under consideration, including the possible ex-
istence of horizontal boundaries.
References
[1] Levander K. Innovative ship design – can innovative ships be designed in
a methodological way. In: Proc. 8th int. marine design conference-IMDC03;
2003.
[2] International Maritime Organisation (IMO). Prevention of air pollution from
ships. Report of the working group on greenhouse gas emissions from ships.
MEPC 58/WP. 8; October 2008.
[3] LOGBASED, Logistics based design. EU FP6 project, Contract number TST3-CT-
2003-001708; 2004–2007.
[4] Murphy RD, Sabat DJ, Taylor RJ. Least cost ship characteristics by computer
techniques. Journal Marine Technology 1965;2(2).
[5] Nowacki H, Brusis F, Swift PM. Tanker preliminary design – an optimization
problem with constraints. Trans SNAME 1970;78.
[6] Papanikolaou A, Kaklis P, Koskinas C, Spanos D. Hydrodynamic optimization
of fast displacement catamarans. In: Proc. 2lst int. symposium on naval
hydrodynamics, ONR’ 96. 1996.
[7] Valdenazzi F, Harries S, Janson C-E, Leer-Andersen M, Maisonneuve J-J,
Marzi J. et al. The fantastic RoRo: CFD optimisation of the forebody and its
experimental verification. In: International conference on ship and shipping
research — NAV 2003. 2003.
[8] Zalek SF, Parsons MG, Papalambros PY. Multicriterion design optimization of
monohull vessels for propulsion and seakeeping. In: Proc. 9th international
marine design conference. 2006.
[9] SAFEDOR (2005–2009). Design, operation and regulation for safety. EU project.
Contract number 516278. http://www.SAFEDOR.org.
[10] Vassalos D. Risk-based design: Passenger ships. In: Proc. SAFEDOR midterm
conference. 2007.
[11] Papanikolaou A, editor. Risk-based ship design — methods, tools and
applications. Springer Publishers; 2009.
[12] Papanikolaou A, Andersen P, Kristensen HO, Levander K, Riska K, Singer D. et
al. State of the art design for X. In: Proc. 10th int. marine design conference-
IMDC09. 2009.
[13] Abt C, Harries S. FRIENDSHIP-framework — integrating ship-design modelling,
simulation, and optimisation, The Naval Architect, RINA. 2007.
[14] VRSHIP-ROPAX2000. A virtual environment for life-cycle design of ship
systems. EU FP5 project. Contract Number G3RD-CT-2001-00506; 2001–2005.
[15] Boulougouris E, Papanikolaou A. Hull form optimization of a high-speed wave
piercing monohull. In: Proc. 9th int. marine design conference-IMDC06. 2006.
[16] FLOWMART. Fast low wash maritime transportation. EU FP5 Project. Contract
number G3RD-CT 1999-00013. 2000–2003.
[17] Zaraphonitis G, Papanikolaou A, Mourkoyiannis D. Hull form optimization of
high speed vessels with respect to wash and powering. In: Proc. 8th int. marine
design conference. 2003.
[18] ROROPROB. Probabilistic rules-based optimal design of Ro–Ro passenger
ships. EU FP5 project. Contract Number G3RD-CT-2000-00030. 2000–2003.
[19] Zaraphonitis G, Boulougouris E, Papanikolaou A. An integrated optimisation
procedure for the design of Ro–Ro passenger ships of enhanced safety and
efficiency. In: Proc. 8th int. marine design conference. 2003.
[20] Boulougouris E, Papanikolaou A. Optimisation of the survivability of naval
ships by genetic algorithms. In: Proc. 3rd int. Euroconference on computer
applications and information technologies in the maritime industries,
COMPIT’04. May 2004.
[21] GIFT, Gas import floating terminal. EU FP6 project. Contract number TST4-CT-
2004-12404. 2005–2007.
[22] Boulougouris E, Papanikolaou A. Multi-objective optimization of a floating LNG
terminal. Journal Ocean Engineering 2008; Elsevier Publishers.
[23] Boulougouris E, Gkohari C, Papanikolaou A. Ship design optimisation in the
multimodal logistics framework. In: Proc. EAMARNET int. conference on
ship design, production and operation. Republished in Journal of HARBIN
Engineering University; 2007.
[24] Brett PO, Boulougouris E, Horgen R, Konovessis D, Oestvik I, Mermiris G. et al.
A methodology for the logistics-based ship design. In: Proc. 9th Int. Marine
design conference. 2006.
[25] Papanikolaou A, Tuzcu C, Tsichlis P, Eliopoulou E. Risk-based optimization of
tanker design. In: Proc. 3rd design for safety conference. October 2007.
[26] Lucas C. Practical multiobjective optimisation.
http://www.calresco.org/lucas/pmo.htm.
[27] Sen P, Yang JB. Multiple criteria decision support in engineering design.
Springer-Verlag London Limited; 1998.
[28] Andrews D, Papanikolaou A, Erichsen S, Vasudevan S. IMDC2009 state of the
art report on design methodology. In: Proc. 10th international marine design
conference - IMDC090. May 2009.
[29] Nowacki H. Developments in marine design methodology: routes, results and
future trends, keynote. In: Proc. 10th international marine design conference,
IMDC090. 2009.
[30] RESOLUTION A.265 (VIII). Regulations on subdivision and stability of
passenger ships as an equivalent to Part B of Chapter II of the int. convention
for the safety of life at sea, 1960. IMO 1973.