AN OPTIMIZATION PROCEDURE FOR THE PRELIMINARY DESIGN OF HIGH-SPEED

RORO-PASSENGER SHIPS
Sotirios Skoupas, National Technical University of Athens, Greece
George Zaraphonitis, National Technical University of Athens, Greece

SUMMARY
Αn integrated methodology for the preliminary design and optimization of high-speed monohull ROPAX vessels is pre-
sented. The core of the developed procedure encompasses the parametric design of ROPAX vessels, based on a selected
set of design parameters. Suitable macros have been developed for the design of the vessel’s hull form, internal layout
and hull structure. Additional modules are developed to perform relevant calculations for the evaluation of the design,
namely the preliminary evaluation of weights, stability in intact and damaged condition, powering and finally for the
economic assessment of the resulting designs. The parametric ship design application has been linked with an optimiza-
tion software, facilitating the design space exploration in a rational and efficient way. Results from the application of the
above procedure to the preliminary design and optimization of two high-speed ROPAX vessels are presented and dis-
cussed.

1. INTRODUCTION method is subsequently used to search the design pa-

rameter space.
During the preliminary Ship Design phase, important de-
An integrated methodology for the preliminary design,
cisions with significant impact on the vessel’s perform-
evaluation and optimization of conventional ROPAX
ance have to be made by the designer, usually based on
vessels, has been presented in (10). The extension of this
very limited information. In such cases, the designer has
work for the preliminary design and optimization of
to rely on his experience and engineering judgement, oc-
high-speed monohull ROPAX vessels is presented in the
casionally supported by the exploration of relevant data
following. The developed methodology has been applied
from past designs. An integrated design and optimization
for the design and optimization of a series of vessels of
methodology, facilitating the fast exploration of a series
various sizes on selected routes, connecting Greek is-
of design alternatives would be of great assistance to the
lands with the mainland. Typical results from these stud-
designer, in search of the ‘optimum’ solution subject to
ies are presented and discussed.
specific owner’s requirements.
The application of optimization methods in ship design is 2. NOMENCLATURE
not new. The work of Leopold (1) and Mandel et al. (2)
AC Vessel’s acquisition cost
and (3) goes back to mid sixties and early seventies. In
BOA Beam over all
recent years, optimization methods are increasingly used
BWL Beam at water line
in various ship design tasks, such as in the hull form op-
CB Block coefficient
timization, in association with advanced CFD software
CR Residual resistance coefficient
tools. Harries et al. in (4) apply parametric hull model-
D Depth
ling techniques along with CFD tools and formal optimi-
DWT Dead Weight
zation methods for the design of minimal wave resistance
FB Free Board
hullforms. Most design problems involve multiple objec-
GM Metacentric Height
tives, which quite often are mutually conflicting. This
GMcr GM critical
type of problems is in principle not tractable with the
GMm GM margin, above regulatory requirements
‘conventional’, single-objective optimization methods,
g Gravitational acceleration
unless a reformulation is introduced, some times based
LCB Longitudinal Centre of Buoyancy
on crude assumptions and simplifications. Alternatively,
LOA Length over all
this type of design problems could be formulated and
LBP Length between perpendiculars
treated as a typical multiple criteria optimization prob-
LWL Length at waterline
lems, as discussed by Sen in (5) and (6). A multicriteria
NPV Net Present Value index
optimization model for the design of Containerships is
RFR Required Freight Rate
discussed by Ray et al. in (7). Hutchinson et al. (8) apply
RT Total resistance
the Genetic Algorithm method for the optimization of
TD Design draught
RO-RO Passenger Ferries, based on the probabilistic sta-
TR Roll eigenperiod
bility standards. Brown et al. (9) apply the Multiattribute
VS Service speed in kn
Value Theory (MAVT) and the Analytical Hierarchy
Δ Displacement
Process (AHP) to synthesize an ‘effectiveness function’
η Water jet propulsion efficiency
for the evaluation of naval ships. The Genetic Algorithm
τ Running trim

227
3. PARAMETRIC DESIGN OF ROPAX VESSELS Since a linear transformation has no effect on the form
coefficients, all hard-chine vessels created by the present
An integrated methodology for the parametric design of application have a block coefficient equal to 0.395, with
high-speed monohull ROPAX vessels has been devel- their centre of buoyancy located at 0.382LWL from the
oped within the well known commercial ship design transom. The corresponding values for the round bilge
software NAPA (11), taking advantage of the program- hulls are CB=0.397 and LCB=0.436LWL.
ming capabilities of the NAPA macro language. The ves-
sel’s hull form and internal layout are generated auto- 3.2 Resistance and propulsion estimations
matically by a series of NAPA macros, followed by a
preliminary structural design. Subsequently, suitable For the Deep-V, double chine vessels, estimations are
macros or external software codes are called to perform based on the decomposition of total resistance in a fric-
the assessment of the technical, operational and eco- tional and a residual component. The corresponding fric-
nomic performance of each design. The basic tasks of the tional resistance coefficient is calculated applying the
developed parametric design methodology are: ITTC empirical formula. The residual resistance coeffi-
cient is calculated by a polynomial expression presented
1. Hull form development
in (14), derived from the regression analysis of experi-
2. Resistance and propulsion estimations mental measurements:
3. Development of internal layout 31
103 C R = ∑ α i xi (1)
4. Preliminary structural design i =1
5. Weights estimation
The variables xi, appearing in equation (1) are functions
6. Intact and damaged stability calculations
of the vessel’s main particulars. Expressions for their
7. Assessment of economic performance calculation, along with the values of the αi, coefficients
A brief description of the above tasks is presented in the are given in (14). Similar expressions are also given for
following. the calculation of the vessel’s running trim:
31
τ = ∑ bi xi (2)
3.1 Hull form development i =1

A set of Napa macros has been developed to facilitate the For the round bilge hulls, the total resistance is calculated
automatic development of the vessel’s external surface, directly, by a polynomial expression of the form:
based on two parent hulls forms: The first one is derived RT 27

from the parent hull of the NTUA high-speed, Deep-V, = α 0 + ∑ α i xi (3)

gΔ i =1
double chine systematic series, presented in (12). The
second alternative is a round-bilge hullform, derived The expressions for the calculation of the variables xi and
from the well known NPL series (13). Each hullform is for the coefficients ai appearing in equation (3) are given
derived from the corresponding prototype, according to by Radojcic et al. in (15). The vessel’s running trim is
the user’s selection (either double-chine or round bilge), calculated using expressions in the form of equation (2),
applying a linear transformation based on the specified also given in (15).
principal dimensions. The cross-sections of a typical
Deep-V, double chine hull with LWL/BWL=5.917 and For water-jet propelled vessels, the required propulsion
BWL/TD=5.13 derived by the developed Napa macros is power estimation is based on empirical expressions for
presented in Figure 1. The cross-sections of a round bilge the calculation of the overall propulsion efficiency (e.g.
hull with LWL/BWL=6.142 and BWL/TD=2.954 derived from equation (4) from (16)):
the NPL series is presented in Figure 2. η = aV 3 + bV 2 + cV + d (4)

Figure 1. Body plan of a Deep-V, double chine vessel Figure 2. Body plan of a round bilge NPL vessel

228
where V is the vessel’s speed in [kn] and formed applying the Det Norske Veritas rules for the
Classification of High-Speed Craft (17). The ship is lon-
a = 2.963e-19
gitudinally subdivided in a number of sections, between
b = -0.0003
successive transverse bulkheads. The sections are verti-
c = 0.0295
cally subdivided in sub-sections between successive
d = -0.0250
decks up to the strength deck, above which the super-
The resulting propulsion power is increased by a suitable structures sub-sections are defined. The construction ma-
power margin to derive the required propulsion engines terial is selected by the user. The available options in-
maximum continuous rating. clude construction of the entire vessel from high tensile
steel or aluminum alloys, or the partial use of HTS
3.3 Development of the internal layout grades for the hull, up to a deck defined by the user (ei-
ther the subdivision deck, or the top of the upper vehicles
The development of the vessel’s internal layout starts deck, herein considered as the strength deck), combined
with the definition the watertight subdivision bellow the with aluminum superstructure.
main vehicles deck. The first step consists of the defini-
tion of the bulkhead deck, the strength deck, the double
bottom and the remaining vehicles decks, according to
the particular design characteristics. The double bottom
height is accordingly defined to ensure effective protec-
tion in the event of racking damage, as defined in Chap-
ter 2, Regulation 2.6 of the HSC Code. In the second
step, the size and position of the main engines rooms and
the pump room is determined from the installed propul-
sion power, using suitable empirical expressions, derived
from existing vessels. Subsequently, the transverse wa-
tertight bulkheads, forward of the engine room are auto-
matically positioned.
The user has limited control on the resulting compart-
mentation, mainly by overriding the default values of the
relevant design parameters. The number of car decks is
either defined by the user, or is internally calculated ac-
cording to the required private cars transport capacity. In
the former case, the private cars transport capacity is cal-
culated by the design software, while in the later case it is
treated as a design parameter. Alternative layouts with
central or side casings may be specified by the user. A
number of upper decks are then created to provide the re-
quired accommodation and public space areas. Once
again, the number of accommodation decks may be ei-
ther defined by the user, in which case the design soft-
ware calculates the resulting passengers transport capac-
ity, or may be calculated according to the required Figure 3. General Arrangement of a ROPAX vessel
number of passengers. A typical General Arrangement of Following the definition of the geometry of the various
a relatively large vessel (LOA=135.6m, LWL=119.5m, parts of the ship structure, the local loading calculations
BOA=20.7m, TD=2.8m), developed by the parametric de- are performed and the maximum allowable stresses for
sign software is presented in Figure 3. This vessel, with a the local strength analysis are determined according to
transport capacity of 1600 passengers, 385 private cars, the Class rules. The required thickness for the plates
or 4 trucks and 290 private cars, has a main car deck of along with the section modulus and other cross-sectional
ample height and strength for the carriage of 30t trucks, characteristics of the attached stiffeners are subsequently
an upper car deck of reduced height for the carriage of calculated. Based on the above requirements, the selec-
private cars and two superstructure decks for the passen- tion of structural members is finalized, with an appropri-
ger spaces and the crew accommodation. ate corrosion thickness allowance. Secondary stiffeners,
minimising the combined plate-stiffener weight, are se-
3.4 Preliminary structural design lected from a data base. For the primary stiffeners, built-
up cross sections, minimising steel weight, while satisfy-
The preliminary structural design is performed by a set of
ing all the structural requirements and geometric con-
Napa macros, developed for the calculation of the re-
straints are evaluated.
quired plate thickness, along with the section modulus
and other cross-sectional characteristics of the attached Based on the results of the preliminary structural analysis
primary and secondary stiffeners. Calculations are per- module, a detailed structural arrangement is created

229
within the NAPA environment. Following a bottom-up ings, both in the departure and arrival condition. The se-
procedure, the plate elements with the attached stiffeners lected vehicles loadings correspond to:
are combined to form planar or curved panels. The ele- a. private cars loading at 100%, 30% and 0% of
mentary panels are grouped together to form sub- the vessel’s total capacity without trucks and
arrangements corresponding to larger parts of the outer b. for the ships with truck carrying capacity, the
shell, entire decks and bulkheads. The sub-arrangements specified number of trucks, combined with
are then combined to create the structural arrangement of 100%, 30% and 0% of the remaining private
the entire steel structure (Figure 4). Based on the defini- cars capacity.
tion of the structural arrangement, the attained section
modulus amidships is calculated, using standard tools 3.7 Assessment of economic performance
available within the Napa Steel module. Compliance with
the Class requirements with respect to vertical bending in Building and operating costs are decomposed in major
calm waters and in waves is verified. If the attained sec- items and sub-items and suitable expressions have been
tion modulus is less than the required, the bottom and derived for their calculation. Crew synthesis and the cor-
strength deck scantlings are accordingly increased and responding crew costs are determined according to the
the procedure is repeated. Greek statutory regulations. The annual income is calcu-
lated for the particular service conditions specified by the
user. Based on these calculations, the vessel’s economic
3.5 Weights estimation
performance is assessed using appropriate economic in-
The vessel’s light weight is divided in the following ba- dices, such as the Required Freight Rate, or the Net Pre-
sic weight categories: structural, propulsion, auxiliary, sent Value. These are complex criteria, encompassing in
deck machinery and outfitting, electrical, piping, heating a rational way the building and operating costs as well as
and air-conditioning, accommodation and miscellaneous. the annual revenues. Transport capacity and propulsion
With the vessel’s detailed structural arrangement readily power are therefore accounted, via the annual income,
available, the structural weight is obtained by direct cal- the fuel costs and the acquisition cost of the propulsion
culation. Machinery weight is further divided in sub- plant.
items (main engines, gear-boxes, shafting, water-jets).
The remaining basic weight categories are also further 4. OPTIMIZATION ENVIRONMENT
divided in sub-items and relevant expressions have been
developed for the estimation of the corresponding The developed parametric design software has been
weights and weight centres. The results of the above pro- linked to a commercial multi-objective optimisation
cedure have been compared with available data from a code, namely modeFRONTIER (19), to form an inte-
number of existing vessels and appropriate correction co- grated design and optimization environment. The method
efficients have been derived. Payload is determined, sub- of Genetic Algorithms was selected as the most suitable
tracting the light weight and the various DWT items choice for the specific problem, for its inherent capability
(consumables, provisions, stores etc.) from the vessel’s to deal with multi-objective optimisation problems with
displacement. mixed continuous-discrete variables and discontinuous
and non-convex design spaces.
3.6 Intact and damaged stability calculations
4.1 Objective functions
Stability calculations are performed for the vessel in in-
tact and damaged condition (both for side damage and The objective functions, available to the user are of eco-
bottom racking) to verify compliance with the require- nomic or technical type. Building and operational costs
ments of the 2000 High-Speed Craft Code (18). A series and annual revenue may be used. Other economic crite-
of macros has been prepared to control the process flow, ria, such as the Required Freight Rate (RFR) or the Net
while the actual stability analysis is performed using the Present Value (NPV) however are considered more suit-
calculation capabilities provided by the NAPA software. able, revealing the vessel’s economic performance on a
Calculations are performed for a predetermined range of selected route in a more rational way. Propulsion power
initial draughts at zero trim and also for specific loading can be used as an objective function, although its effect is
cases, with 100% passengers and variable vehicles load- indirectly accounted in the RFR and the NPV.

Figure 4. Structural Arrangement of a typical high speed ROPAX vessel

230
The maximization of stability reserve, in excess of the pacity, with an additional requirement of four trucks ca-
regulatory requirements may be also used as objective pacity with an average weight of 30t (an average number
function. An alternative and more conventional approach of two trucks per one-way trip was used for the calcula-
would be to treat stability as a constraint, instead of an tion of the vessel’s revenue). A minimum metacentric
objective function, by requiring the fulfilment of the height margin GMm≥0.30m, above the requirements of
regulatory requirements, preferably with a small safety the intact and damaged stability regulations has been
margin, to account for the possibility of an underestima- specified. The calculation of the Net Present Value index
tion of the light ship’s centre of gravity height. has been performed for an average fare of 50€ per pas-
Seakeeping performance is also an important issue, espe- senger, and a freight rate of 75€ for the private cars and
cially for high-speed passenger vessels; therefore appro- 165€ for the trucks. The Required Freight Rate calcula-
priate links have been established between the design en- tions were performed for a standard ratio of the private
vironment and a seakeeping software, enabling its use for cars freight rate to the passenger’s fare equal to 1.5. The
the assessment of the vessel’s response in a seaway. trucks to private cars freight rate ratio was equal to 2.2.
Once again, the user has the choice of treating seakeep-
Table 1. Vessel’s main particulars and operating re-
ing performance as a constraint instead of an objective
quirements for the Piraeus-Naxos route
function, specifying acceptable limits regarding the ves-
sel’s performance (motions, accelerations or sea-sickness Main Particulars From To
index) in specific sea-states. Other objective functions LBP [m] 100.0 120.0
can also be used if required, since the design software ar- BOA [m] 18.0 20.0
chitecture is quite flexible. TD [m] 2.5 3.5
FB [m] 2.5 3.5
4.2 Demonstration studies The optimization exercise has been performed for the fol-
lowing objective functions:
The developed methodology has been applied to the de-
sign and optimization of a series of high-speed ROPAX a. minimization of the Required Freight Rate and
vessels of various sizes and on selected routes, between a
b. maximization of the attained stability margin in
number of Aegean islands and the Greek mainland. In the
excess of the requirements of the intact and
following, two examples of the obtained results are pre-
damaged stability regulations.
sented and briefly discussed.
The second example corresponds to a smaller vessel, op-
The first example corresponds to an optimization exer-
erating between the port of Rafina, 27km east of Athens
cise for a relatively large vessel, operating between the
and the island of Andros (Figure 5). The length of the
port of Piraeus and the island of Naxos in the central Ae-
route is 45 sea miles. With a required service speed of
gean Archipelagos (Figure 5). The length of the route is
35kn the time at sea for a one-way trip is around 90min.
125sea miles. With a required service speed of 40kn the
The upper and lower bounds for the vessel’s main par-
time at sea for a one-way trip is around 190min. The up-
ticulars are presented in Table 2. A transport capacity of
per and lower bounds for the vessel’s main particulars
750 passengers has been defined. Instead of specifying
are presented in Table 1.
the private cars transport capacity, the number of car
decks has been given (i.e. two vehicles decks) and the ac-
tual transport capacity of each design is calculated by the
design software. In this case also, a minimum metacen-
tric height margin GMm≥0.30m has been specified. The
NPV calculations have been based on an average fare of
30€ per passenger and a freight rate of 48€ for the private
cars. The RFR calculations were carried out assuming a
standard ratio of the private cars freight rate to the pas-
senger’s fare equal to 1.6.
Table 2. Vessel’s main particulars and operating re-
quirements for the Rafina-Andros route
Main Particulars From To
50 km LBP [m] 75.0 95.0
BOA [m] 16.0 18.00
TD [m] 2.0 3.0
Figure 5. Studied routes FB [m] 2.5 3.5
Instead of requiring a specific transport capacity in terms In this case a different formulation of the optimization
of passengers and private cars, the number of decks has problem has been selected: In order to improve perform-
been given (i.e. two vehicles decks and two passenger ance in beam and quartering seas, vessels with relatively
decks). The actual transport capacity of each design al- large roll eigenperiod were searched for. An approxima-
ternative is calculated by the design software, based on tion of the roll eigenperiod was obtained according to the
the available deck area and the vessel’s load carrying ca- IMO Resolution A-749(18) by the following expression:

231
 2C ⋅ BOA are presented in Table 6. The attained RFR of the feasible
TR = (5) designs varies from 39.20€ to 66.75€, while their GM
GM
margin varies from 0.322m to 2.121m. The design mini-
where: mising RFR is found in the 47th generation. Alternative
BOA L designs of almost equally low RFR where identified
C = 0.373 + 0.023 − 0.043 WL (6) however quite earlier (e.g. the design 222 in the 12th gen-
T 100
eration, with an RFR of 40.33€, or the design 371 in the
Therefore the second optimization exercise has been per- 19th generation with an RFR of 39.46€). Figure 6 presents
formed for the following objective functions: the history diagram of the attained RFR for the first 40
generations. The circles correspond to the feasible de-
a. the minimization of the Required Freight Rate
signs, while the triangles to the unfeasible ones (i.e. those
and
failing to satisfy at least one of the specified constraints).
b. the maximization of the roll eigenperiod.
Figure 7 presents the history chart of the attained GM
In both studies the vessels have been considered operat- margin of the first 40 generations. In the first generations
ing for 12 years. The price of each vessel at the end of a number of unfeasible designs may be observed, some
the 12 years period was set at 33% of the corresponding of them with negative GM margin (i.e. designs not meet-
acquisition cost. A 50% loan with a 7% interest and a ten ing the stability requirements) and others with
years payback period has been considered for both stud- GMm<0.3m (the minimum acceptable GM margin was set
ies. The discount rate was set at 8%, while a 40% tax rate at 0.3m). After some generations however, the unfeasible
was assumed. The Fuel and Diesel oil prices were set at designs practically disappear and high values of the GM
600€/t and1250€/t respectively.Some additional assump- margin are obtained. Figure 8 presents the scatter dia-
tions regarding the seasonal conditions for the two routes gram of the GM margin versus RFR. The Pareto designs
are summarized in Table 3 and Table 4. are highlighted. The scatter diagram of the RFR vs. the
Installed Propulsion Power is presented in Figure 9. The
Table 3. Seasonal conditions (Piraeus – Naxos route)
corresponding diagram of the RFR vs. the vessel’s Ac-
Season High Medium Low quisition Cost is presented in Figure 10. Finally, the scat-
Months per year 3 3 5 ter diagrams of RFR vs. LBP and BOA are presented in
Round trips per day 2 1 1 Figure 11 and Figure 12 respectively and that of GMm vs.
Passengers loading 70% 40% 20% BOA in Figure 13.
Private cars loading 70% 20% 10%
Trucks number 2 2 2 Table 7summarises the main characteristics of the ob-
tained Pareto designs from the second design exercise
Table 4. Seasonal conditions (Rafina – Andros route) (the Rafina-Andros route). The main characteristics of
Season High Medium Low the 120 feasible designs with the smaller RFR value are
Months per year 3 3 5 presented in Table 8. The attained RFR of the feasible
Round trips per day 3 2 1 designs varies from 24.40€ to 44.78€, while their GM
Passengers loading 70% 30% 15% margin varies from 0.303m to 1.614m. The design mini-
Private cars loading 70% 20% 10% mising RFR is found in the 43th generation. Alternative
In both studies the vessel’s hullform was derived from designs of almost equally low RFR where identified in
the Deep-V, double chine, NTUA systematic series. The the 5th generation (design number 83 with RFR=24.87€),
5083 aluminum alloy was selected as the construction or in the 15th generation (design number 297 with
material both for the main hull and the superstructures. RFR=24.58€). Figure 14 presents the history diagram of
the attained RFR for the first 60 generations. The circles
The Multiple Objectives Genetic Algorithm optimisation correspond to the feasible designs, while the triangles to
scheduler of modeFRONTIER has been applied for the the unfeasible ones (i.e. those failing to satisfy at least
optimisation studies. Using the Steady-state GA algo- one of the specified constraints). Figure 15 presents the
rithm MOGA, 120 generations were derived with a popu- history chart of the attained roll eigenperiod. Figure 16
lation of 20 designs per generation. The genetic opera- presents the scatter diagram of the roll eigenperiod versus
tions were executed with a 5% probability of selection, RFR. The Pareto designs are highlighted. The scatter dia-
50% probability of directional crossover, 10% probabil- gram of the RFR vs. the Installed Propulsion Power is
ity of mutation, 10% DNA string mutation ratio and pe- presented in Figure 17. The corresponding diagram of the
nalize objectives for treating constraints. RFR vs. the vessel’s Acquisition Cost is presented in
Results from the first optimization example are presented Figure 18. Finally, the scatter diagrams of RFR vs. LBP
in the following. and BOA are presented in Figure 19 and Figure 20 respec-
tively and that of the attained roll eigenperiod vs. BOA in
Table 5 summarises the main characteristics of the de- Figure 21.
signs that meet the Pareto optimality: A design is Pareto
After the Pareto Front is obtained, a multiple criteria
optimal if it satisfies the constraints and is such that no
ranking is applied to capture the preferences of the deci-
criterion can be further improved without causing at least
sion maker. The decision maker is asked to compare
one of the other criteria to decline. The main characteris-
pairwise a collection of design attributes. Using these
tics of the 15 feasible designs with the smaller RFR value

232
preferences a set of step-wise marginal utility functions is Methodology for Ships and Other Complex Systems’,
derived analytically for all the attributes of a design. The Philosophical Transactions, Royal Society of London,
marginal utility functions are then combined to form a A273, 1972.
composite fitness function used by multi-objective GA. 4. Harries, S. and Abt, C., ‘Formal Hydrodynamic Op-
Following the above multi objective decision making timization of a Fast Monohull on the Basis of Para-
procedure, the design number 1129 with RFR=25.18€ metric Hull Design’, Proceedings of the 5th Interna-
and TR=7.07sec has been identified as the preferable so- tional Conference on Fast Sea Transportation – FAST
lution. ’99, Seattle, USA, 31 Aug.-2 Sept. 1999.
5. Sen, P., ‘Marine Design: the Multiple Criteria Ap-
5. CONCLUSIONS proach’, Transactions of the Royal Institution of Na-
val Architects, vol. 134, Part B, pp 261-276, 1992.
An integrated methodology for the preliminary design, 6. Sen, P. and Yang, J.B., ‘Multiple Criteria decision
evaluation and optimization of high-speed ROPAX ves- Support in Engineering Design’, Springer-Verlag
sels has been presented. The core of the developed pro- London, ISBN: 3540199322, 1998.
cedure encompasses the parametric design of ROPAX 7. Ray, T. and Sha, O.P., ‘Multicriteria Optimization
vessels within the commercial ship design software Model for a Containership Design’, Marine Technol-
NAPA, based on a selected set of design parameters. ogy, vol. 31, no. 4, 1994.
Suitable macros have been developed for the design of 8. Hutchinson, K.W., Sen, P., Buxton, I.L. and Hills,
the vessel’s hull form, the development of the internal W., ‘Multiple Criteria Design Optimisation of RO-
layout and the preliminary structural design evaluation. RO Passenger Ferries with Consideration of recently
Additional modules perform relevant calculations for the proposed Probabilistic Stability Standards’, The 7th
evaluation of each design alternative, including stability International Symposium on Practical Design of
analysis in intact and damaged condition, powering, Ships and Mobile Units, PRADS ’98, The Hague,
weights and transport capacity estimation and finally for The Netherlands, 20-25 Sept. 1998.
the assessment of the economic potential of the vessel in 9. Brown, A.J. and Thomas, M., ‘Reengineering the
a selected route. All these modules are integrated in a de- Naval Ship Concept Design Process’, Research to
sign environment facilitating the investigation of various Reality in Ship Systems Engineering Symposium,
design alternatives in very short time. The parametric de- ASNE, Sep. 1998.
sign and evaluation of a relatively large vessel 10. Zaraphonitis, G., Papanikolaou, A., Loukakis, T.,
(LWL≈120m) takes approximately 100sec in a personal Tsouvalis, N., Skoupas, S., Mitsotakis, K. and Anto-
computer with an Intel® Core™ 2 CPU 6600 processor niou, A., ‘Development and demonstration of an inte-
at 2.4GHz. The most time consuming tasks are the pre- grated methodology for the design of RoRo-
liminary structural design, requiring approximately 75sec Passenger ships for the needs of short sea shipping in
and the damaged stability analysis, with approximately Greece’, Proceedings of the 12th International Con-
15sec. gress of the International Maritime Association of the
Mediterranean (IMAM 2007), Varna, Bulgaria, 2-7
To further increase the effectiveness of the developed Sep. 2007.
procedure, the parametric ship design application has 11. Napa Oy, the Naval Architectural Package (NAPA),
been linked to a commercial multi-objective optimization http://www.napa.fi
software (i.e. mode FRONTIER), facilitating the design 12. Grigoropoulos, G.J. and Loukakis T. A., ‘Resistance
space exploration in a rational and efficient way. To and Seakeeping Characteristics of a Systematic Series
demonstrate its potential, the developed design method- in the Pre-planing Condition (Part I)’, SNAME
ology has been applied for a series of design and optimi- Transactions, vol. 110, pp. 77-113, 2002
zation studies. Results from the optimization of two high- 13. Bailey, D., ‘The NPL high speed round bilge dis-
speed ROPAX vessels were presented and discussed. placement hull series’, Marine Technology Mono-
graph No. 4, Royal Institution of Naval Architects,
6. ACKNOWLEDGMENTS 1976.
14. Radojcic, D., Grigoropoulos, G., Rodic, T., Kuvelic,
Mr Skoupas contribution to the present work has been T., Damala, D., ‘The Resistance and Trim of Semi-
partially supported by the Greek State Scholarships Displacement, Double-Chine, Transom-Stern Hull
Foundation (IKY). Series’, Proceedings of the 6th International Confer-
ence on Fast Sea Transportation – FAST2001, South-
7. REFERENCES ampton, UK, Sept. 2001.
1. Leopold, R., ‘Mathematical Optimization Methods 15. Radojcic, D., Rodic, T. and Kostic, N., ‘Resistance
Applied to Ship Design’, MIT, Department of Naval and Trim Predictions for the NPL High Speed Round
Architecture and Marine Engineering, Report 65-8. Bilge Displacement Hull Series’, International Con-
2. Mandel, P. and Leopold, R., ‘Optimization Methods ference On Power, Performance and Operability of
Applied to Ship Design’, SNAME Transactions, vol. Small Craft, Southampton, UK, Sept. 1997.
74, 1966. 16. Knox, J., ‘Ferries’ in Ship Design and Construction,
3. Mandel, P. and Chryssostomidis, C., ‘A Design vol. 2, edited by T. Lamb, ISBN: 0-939773-41-4,

233
 published by the Society of Naval Architects and Ma- 18. IMO, ‘International Code of Safety of High-Speed
rine Engineers, SNAME, 2004. Craft’, International Maritime Organization, MSC
17. DNV, ‘Rules for the Classification of High-Speed, 73/21/Add.1, Annex 5, 2000.
Light Craft and Naval Surface Craft, Part 3 & 5’, Det 19. E.STE.CO, modeFRONTIER, http://www.esteco.it
Norske Veritas, 2002.

Table 5. Pareto designs, Large Vessel

Des Pass. Cars Inst.
Id LBP BOA TD FB Numb. Numb. DWT Δ Power AC GMcr GMm RFR
(m) (m) (m) (m) (t) (t) (kW) (m€) (m) (m) (€)
119 116.00 20.00 2.70 2.60 1606 287 591 2251 35621 55.09 8.108 1.651 45.96
191 120.00 20.00 2.65 2.50 1692 235 545 2286 39569 56.44 8.100 1.821 49.18
318 113.20 20.00 2.55 2.50 1570 158 434 2075 36812 54.46 8.418 1.961 51.94
391 120.00 20.00 2.60 2.50 1677 195 494 2244 39016 57.95 8.186 1.925 50.63
486 120.00 19.85 2.65 2.50 1668 298 605 2269 33105 52.86 8.308 1.516 41.20
588 100.00 20.00 2.70 2.50 1333 198 462 1940 38856 51.00 7.734 2.121 60.65
760 108.60 19.90 2.65 2.60 1494 253 532 2057 32776 49.34 8.304 1.533 45.63
771 100.80 20.00 2.70 2.50 1357 201 469 1956 38943 51.33 7.734 2.112 59.75
822 101.20 20.00 2.70 2.50 1358 205 473 1963 38796 51.36 7.749 2.110 59.37
849 105.00 20.00 2.70 2.50 1423 238 516 2037 37162 51.73 7.904 2.022 53.80
929 118.80 19.95 2.65 2.50 1646 327 635 2258 31422 51.72 8.604 1.447 39.20
932 112.80 19.95 2.70 2.50 1538 265 558 2183 36048 53.86 7.921 1.812 48.57
960 116.00 19.95 2.70 2.50 1601 271 572 2245 36036 55.06 7.916 1.798 46.88
1024 117.20 20.00 2.70 2.70 1625 285 591 2275 35703 55.73 8.108 1.549 45.75
1050 114.40 19.95 2.65 2.50 1575 313 610 2174 31534 50.33 8.596 1.464 40.89
1116 113.80 20.00 2.70 2.50 1588 284 585 2208 35632 54.20 8.127 1.708 46.26
1254 118.80 19.95 2.55 2.50 1661 174 465 2174 38102 55.10 8.416 1.853 50.15
1262 117.20 19.95 2.55 2.50 1624 181 469 2144 37719 54.22 8.491 1.908 50.42
1454 102.60 20.00 2.70 2.50 1384 221 493 1990 37924 51.33 7.830 2.058 56.60

Table 6. Minimum RFR designs, Large Vessel

Des. Pass. Cars Inst.
Id LBP BOA TD FB Numb. Numb. DWT Δ Power AC GMcr GMm RFR
(m) (m) (m) (m) (t) (t) (kW) (m€) (m) (m) (€)
929 118.80 19.95 2.65 2.50 1646 327 635 2258 31422 51.72 8.604 1.447 39.20
644 119.00 19.95 2.65 2.85 1662 317 624 2262 31453 52.32 8.644 1.224 39.26
640 118.80 19.95 2.65 2.75 1660 311 617 2258 31422 52.24 8.577 1.255 39.39
924 119.00 19.95 2.65 2.75 1660 312 618 2261 31453 52.30 8.580 1.257 39.41
659 120.00 19.95 2.65 2.80 1675 321 631 2280 31941 52.76 8.608 1.257 39.42
371 119.00 19.95 2.65 2.80 1661 310 616 2262 31453 52.38 8.575 1.219 39.46
838 120.00 19.95 2.65 2.70 1673 320 630 2281 31941 52.70 8.581 1.310 39.47
811 118.60 19.95 2.65 2.80 1660 306 611 2254 31397 52.27 8.567 1.217 39.49
656 118.40 19.95 2.65 2.60 1635 321 626 2250 31371 51.72 8.595 1.378 39.54
931 118.60 19.95 2.65 2.85 1661 302 606 2254 31397 52.39 8.549 1.177 39.58
558 118.40 19.95 2.65 2.80 1649 308 612 2250 31371 52.14 8.569 1.210 39.61
744 117.60 19.95 2.65 2.65 1633 312 615 2235 31286 51.58 8.583 1.330 39.67
690 118.20 19.95 2.65 2.70 1636 314 618 2246 31351 51.84 8.578 1.294 39.68
698 118.40 19.95 2.65 2.70 1636 315 619 2250 31371 51.91 8.581 1.300 39.69
677 120.00 19.95 2.65 2.90 1676 308 616 2281 31941 53.08 8.551 1.157 39.76

234
Table 7. Pareto designs, Small Vessel
Des. Cars Inst.
Id LBP BOA TD FB Numb. DWT Δ Power AC GMcr GMm TR RFR
(m) (m) (m) (m) (t) (t) (kW) (m€) (m) (m) (sec) (€)
147 79.60 17.65 2.75 2.95 262 448 1352 22377 30.63 6.385 0.303 7.24 26.86
223 84.60 18.00 3.00 3.50 293 491 1483 25339 34.02 6.167 0.307 7.32 29.08
239 83.00 17.60 2.80 2.90 272 462 1402 23344 31.91 6.146 0.303 7.31 27.71
297 94.00 16.55 2.20 2.50 162 317 1232 16402 28.61 6.982 0.550 6.54 24.58
331 75.00 17.85 2.80 2.70 245 426 1276 21876 29.05 6.541 0.420 7.18 26.40
365 82.00 17.70 2.75 3.50 286 478 1431 22926 32.31 6.279 0.370 7.26 27.08
476 75.00 18.00 2.75 3.05 259 442 1324 21722 29.74 6.614 0.714 7.08 26.06
496 75.00 18.00 2.75 3.30 257 440 1333 21722 30.08 6.573 0.596 7.16 26.25
571 75.00 18.00 2.75 3.50 252 434 1334 21722 30.36 6.543 0.491 7.23 26.59
585 84.40 18.00 3.00 3.50 293 491 1478 25294 33.92 6.176 0.309 7.31 29.01
855 94.00 17.95 2.35 2.50 269 452 1426 19524 30.38 8.406 0.996 6.28 24.40
925 94.00 16.50 2.20 2.50 154 308 1228 16650 28.74 6.878 0.516 6.56 25.13
1129 75.60 17.30 2.65 2.50 234 410 1243 20644 27.38 6.353 0.341 7.07 25.18
1228 75.60 17.80 2.70 2.70 248 429 1288 21378 29.09 6.635 0.552 7.08 25.92

Table 8. Minimum RFR designs, Large Vessel

Des. Cars Inst.
Id LBP BOA TD FB Numb. DWT Δ Power AC GMcr GMm TR RFR
(m) (m) (m) (m) (t) (t) (kW) (m€) (m) (m) (sec) (€)
855 94.00 17.95 2.35 2.50 269 452 1426 19524 30.38 8.406 1.368 6.28 24.40
297 94.00 16.55 2.20 2.50 162 317 1232 16402 28.61 6.982 0.550 6.54 24.58
560 93.80 16.55 2.20 2.50 158 313 1228 16417 28.62 6.969 0.547 6.39 24.73
762 94.00 16.55 2.20 2.60 159 314 1231 16402 28.72 6.978 0.476 6.43 24.73
958 93.60 16.55 2.20 2.50 157 311 1226 16435 28.60 6.963 0.549 6.39 24.78
500 93.40 16.55 2.20 2.50 156 310 1224 16451 28.58 6.958 0.546 6.39 24.81
83 93.40 16.55 2.20 2.55 155 309 1223 16451 28.62 6.961 0.508 6.40 24.87
1050 94.00 17.95 2.35 2.70 262 444 1426 19524 31.59 8.359 1.242 6.32 25.02
542 92.20 16.60 2.20 2.50 149 302 1211 16500 28.51 6.962 0.588 6.40 25.07
353 94.00 16.55 2.20 2.70 151 304 1231 16402 28.90 6.964 0.391 6.45 25.08
85 93.40 16.55 2.20 2.75 150 303 1223 16451 28.82 6.961 0.356 6.47 25.13
570 95.00 16.50 2.20 2.50 159 314 1241 16765 28.93 6.917 0.528 6.38 25.13
925 94.00 16.50 2.20 2.50 154 308 1228 16650 28.74 6.878 0.516 6.56 25.13
784 94.00 16.55 2.20 2.80 150 303 1231 16402 28.98 6.973 0.308 6.48 25.16
1129 75.60 17.30 2.65 2.50 234 410 1243 20644 27.38 6.353 0.341 7.07 25.18
1010 92.20 18.00 2.35 2.55 253 433 1403 19902 30.42 8.375 1.363 6.30 25.19
326 81.20 17.60 2.20 2.60 119 264 1132 15951 27.47 7.969 1.198 6.33 25.23
748 92.60 16.55 2.20 2.50 149 302 1213 16695 28.57 6.920 0.572 6.39 25.28
101 94.00 16.95 2.25 2.90 181 342 1289 17456 29.88 7.291 0.514 6.49 25.35
252 84.20 17.85 2.25 2.95 155 310 1217 17007 28.99 8.167 0.864 6.47 25.51

235
 66 2.0

62 1.5

58
1.0

GMmarg (m)
RFR (Euro)

54
0.5

50

0.0
46

-0.5
42

38 -1.0
-80 0 80 160 240 320 400 480 560 640 720 80 0 80 160 240 320 400 480 560 640 720 80
Design ID Design ID

Figure 6: RFR history diagram (first 40 generations) Figure 7. GMm history diagram (first 40 generations)

2.0 66

62
1.5

58
1.0
GMmarg (m)

RFR (Euro) 54

0.5

50

0.0
46

-0.5
42

-1.0 38
38 42 46 50 54 58 62 66 25000 30000 35000 40000 45000 5000
RFR (m) Installed Power (kW)

Figure 8. Scatter diagram, GMm vs. RFR Figure 9. Scatter diagram, RFR vs. Installed Power

66 66

62 62

58 58
RFR (Euro)
RFR (Euro)

54 54

50 50

46 46

42 42

38 38
43 45 47 49 51 53 55 57 59 61 98 100 102 104 106 108 110 112 114 116 118 120 12
Acquisition Cost (mEuro) LBP

Figure 10. Scatter diagram, RFR vs. Acquisition Cost Figure 11. Scatter diagram, RFR vs. LBP

66 2.0

62
1.5

58
1.0
GMmarg (m)
RFR (Euro)

54

0.5

50

0.0
46

-0.5
42

38 -1.0
17.8 18.2 18.6 19.0 19.4 19.8 20 17.8 18.0 18.2 18.4 18.6 18.8 19.0 19.2 19.4 19.6 19.8 20.0 20
Beam (m) Beam (m)

Figure 12. Scatter diagram, RFR vs. BOA Figure 13. Scatter diagram, GMm vs. BOA

236
 36 8.5

34
8.0

32

7.5
RFR (Euro)

To (s)
30

7.0

28

6.5
26

24 6.0
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
Design ID Design ID

Figure 14: RFR history diagram (first 60 generations) Figure 15. TR history diagram (first 60 generations)
8.5 36

34
8.0

32

7.5
RFR (Euro)
To (s)

30

7.0

28

6.5
26

6.0 24
24 26 28 30 32 34 36 14000 16000 18000 20000 22000 24000 26000 28000
RFR (Euro) Installed Power (kW)

Figure 16. Scatter diagram, TR ratio vs. RFR Figure 17. Scatter diagram, RFR vs. Installed Power
36 36

34 34

32 32
RFR (Euro)
RFR (Euro)

30 30

28 28

26 26

24 24
24 26 28 30 32 34 36 74 78 82 86 90 94
Acquisition Cost (mEuro) LBP (m)

Figure 18. Scatter diagram, RFR vs. Acquisition Cost Figure 19. Scatter diagram, RFR vs. LBP
36 8.5

34
8.0

32

7.5
RFR (Euro)

To (s)

30

7.0

28

6.5
26

24 6.0
15.8 16.2 16.6 17.0 17.4 17.8 18.2 15.8 16.2 16.6 17.0 17.4 17.8 18.2
Beam (m) Beam (m)

Figure 20. Scatter diagram, RFR vs. BOA Figure 21. Scatter diagram, TR vs. BOA

237