Ocean Engineering 239 (2021) 109908

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

A decision-making process for the selection of better ship main dimensions

by a Pareto frontier solution
Onur Yurdakul a, Gözde Nur Küçüksu b, Ahmet Ziya Saydam c, *, Mehmet Sander Çalışal a
a
Piri Reis University, Turkey
b
Istanbul Technical University, Turkey and Hidro-Teknik Nautical Design, Turkey
c
Piri Reis University, Turkey and Hidro-Teknik Nautical Design, Turkey

A R T I C L E I N F O A B S T R A C T

Keywords: Initial ship design necessitates the evaluation of main ship parameters to obtain a feasible design solution
Ship Design satisfying the objectives. Traditionally, relationship between main dimensions and design parameters of proven
Main Dimensions designs provides a safe solution. This approach restrains the designer from improving the design with respect to
Pareto
conflicting design criteria. Pareto frontier technique has widely been utilized in the design of ships, mainly at
Seakeeping
Power
advanced design stages for optimization problems. As the principal dimensions of a ship are also vital on the
performance of the vessel, major improvements in performance may be achieved by selecting better principal
dimensions. This paper proposes to integrate the Pareto technique to the design cycle at an early stage as a
decision-making process for selection of better main dimensions with respect to multiple conflicting criteria A
Pareto Front has been observed for two case studies (a destroyer and a planing craft) indicating a technological
barrier for performance with respect to main dimensions. As the design process begins with main dimensions
along the technological barrier, performance is improved with respect to conflicting design criteria at the
beginning of the design process which serves as a better basis for further optimization studies to be conducted at
advanced stages.

etc … The priority of the commercial ship designer is to increase the

1. Introduction deadweight capacity of the ship and lower the fuel consumption. A fast
craft designer should ensure optimum weight distribution and simulta­
Traditionally, ship design was based on the duplication of the fea­ neously ensure safe operation at high speeds with reduced motion and
tures of successful ships that have already proven their better perfor­ acceleration levels without compromising for fuel efficiency.
mance. This was mainly true, during the times of sailing ships. Gradually With the tools and techniques readily available to the practicing ship
other concepts of design, such as ‘design cycle’ advanced, as described in designer, a compromised solution or in other words a balanced solution
modern textbooks. The ‘design cycle’ in fact, describes that the ship may be achieved satisfying the design objectives for all the ship types
design is a nonlinear process and in general, it is expected to converge to mentioned before. Self-sustaining ships that float and have sufficient
a reasonable solution, so that the ship floats and will have sufficient stability are added to the fleet as “proven designs”. These proven designs
stability, power, speed and structural integrity. However, the first phase are then added to the databases of designers for prospective regression
of the design cycle is still initiated with obtaining the main dimensions analysis for new designs. It may even be claimed that a spiral is created
and certain hull form parameters such as the block coefficient (CB) based lacking innovation and improvement.
on a regression analysis of successful “proven” designs. With a rather The main aim of the proposed Pareto-based decision-making
traditional approach as explained above, it is a challenge for the approach is to enable the designer to achieve a much broader perspec­
designer to improve the design with respect to multiple conflicting tive on the influence of key initial design parameters such as ship main
criteria at early stages of design. A common case that designers may dimensions and underwater volume distribution on numerous conflict­
encounter is to try to improve seakeeping characteristics of the design ing design criteria -as mentioned above-at a very early stage in ship
whilst also aiming to lower the fuel consumed by the vessel. This case is design, with particular emphasis on improving the design with respect to
the norm for designers of naval ships, fishing vessels, special work boats both conflicting criteria. This may be achieved with the assessment of

* Corresponding author.
E-mail address: [email protected] (A.Z. Saydam).

https://doi.org/10.1016/j.oceaneng.2021.109908
Received 4 March 2021; Received in revised form 21 September 2021; Accepted 22 September 2021
Available online 30 September 2021
0029-8018/© 2021 Elsevier Ltd. All rights reserved.
O. Yurdakul et al. Ocean Engineering 239 (2021) 109908

Nomenclature LC Chine Wetted Length (m)

LK Keel Wetted Length (m)
B Beam (m) LCB Longitudinal Center of Buoyancy (%LPP)
BM Beam at Midship (m) LCG Longitudinal Center of Gravity (m)
CA Correlation Allowance Lkeel Keel Length (m)
CB Block Coefficient LPP Length Between Perpendiculars (m)
CL Lift Coefficient nCG
̃ Average Center of Gravity Acceleration (g)
CL0 Lift Coefficient for Zero Deadrise Angle PE Effective Power (kW)
Cp Prismatic Coefficient RAO Response Amplitude Operator
CV Speed Coefficient RP Pressure Resistance (N)
CΔ Beam Loading Coefficient, Δ/wb3 S Wetted surface area (m2)
EHP Effective Horse-Power TEU Twenty-foot Equivalent Unit
FL0 Lift Force at Zero Deadrise Angle (N) V, VK Speed (m/s, kn)
FLβ Lift Force (N) Vt Instantaneous draft relative to calm water (m)
Fn Froude Number zmax z-coordinate of maximum pressure
FnB Beam Froude Number VCG Vertical Center of Gravity (m)
g Gravitational Acceleration (m/s2) β Deadrise Angle (◦ )
H1/3 Significant Wave Height (ft) ρ Density of Sea Water (t/m3)
ITTC International Towing Tank Conference λw Average wetted length
L Length (m) τ Trim Angle (◦ )
LOA Length Overall (m) τcrit Critical Trim (m)

multitudinous design alternatives, from a few hundred to a couple that decreases greenhouse gas (GHG) emissions via ship scantling opti­
thousand depending on the nature of the design. mization. The design objective is lowering weight of steel while main­
There exist numerous applications of Pareto technique to engineer­ taining the cost of production at a tolerable level. The Pareto front has
ing problems in literature. Several examples may be found in the area of been mapped by using the repeated weighted sum solutions method and
ship design. 3 Pareto-optimum designs are obtained as a result of weight optimiza­
Kasprzak and Lewis (2001) have studied vehicle dynamics utilizing a tion, cost optimization and min-max solution. The weight-optimized
Pareto approach. A Pareto optimal solution has been achieved by using design has 12.72% saving on steel weight and 0.88% saving on pro­
Collinearity theorem to estimate the weighting of relative objectives by duction cost. On the other hand, the cost-optimized design has 4.52%
fitting polynomials. The whole mapping of the Pareto set results into the saving on production cost, but the steel weight increases by 5.1%.
relative objective weighting guarantees a huge quantity of data as well Min-max solution has 11.3% saving on steel weight and 1.58% saving on
as an easily applicable approach to the problem. production cost.
Gierson (2006) has published general guidelines for utilizing Pareto A joint research effort within the NATO RTO Task Group has been
approach for engineering problems with multiple design criteria. The conducted by Diez et al. (2015) joining CNR-INSEAN, University of
proposed Pareto approach has been based on basic economics; providing IOWA, Roma Tre University, ITU, NTUA and ECN. The overall idea of
trade-offs between design parameters and this paper explains the engi­ the project is to compare various methodologies in terms of optimization
neering application of Pareto approach clearly. A simple example has and improving the resistance and the seakeeping characteristics of
been shown on a structural plate design. It is shown that multiple con­ DTMB 5415. These methods provide hull variants and optimization of
flicting design criteria may be jointly improved in engineering problems each variants’ hydrodynamic performance, combining low and
with the utilization of Pareto approach. high-fidelity solvers, design modification tools and multi-objective
A holistic ship design process has been implemented by Papaniko­ optimization algorithms. INSEAN/IU, ITU and NTUA have used
laou (2010) and in general, utilization of Pareto frontiers has been low-fidelity solvers. ECN has verified the hull which is the optimum hull
considered as the possible output of multi-objective optimization of INSEAN/IU study by using a high-fidelity solver. The results have
studies. Wave wash and total resistance have been selected to form the provided significant optimization achievements with an average
Pareto frontiers for a monohull vessel. Several Pareto optimum designs improvement for calm water resistance of 10% and seakeeping perfor­
have been obtained and selection of the desired one has been left to the mance of 9%. The high-fidelity solver results indicate an overall 0.2%
designer. Also, during systematic case studies on a sample Ro-Ro pas­ reduction for calm water resistance.
senger ship, Pareto diagrams have been widely used between the pa­ Kim and Paik (2017)studied on a multi-objective design methodol­
rameters; structural weight, lane length and attained subdivision index. ogy for merchant cargo ships with a trade-off between structural weight
Diez and Peri (2010) have proposed a robust optimization approach and structural safety by using Pareto method instead of conventional
for a bulk carrier at concept design phase considering operational pro­ comparative approach which is mostly in use nowadays. Mentioned
files involving uncertainties and external effects arising due to the procedure has been applied to a very large crude oil carrier (VLCC). The
environment. A particle swarm optimization (PSO) algorithm has been reference ship and new design has been compared and according to this
utilized to provide solutions for the single-objective optimization study, new procedure lightened the structural weight by 3% while
problem which is minimizing the cost of transportation per unit. improving the safety factors for crucial structural parts from 3% to 5%
Multi-objective deterministic PSO method (MODPSO) has been used to and also reduced required man-hours by 20% during the design stage.
also minimize the standard deviation of transportation cost. Pareto The proposed Pareto frontier approach has been successfully applied
principle has been used for cross checking purpose. The numerical re­ to a fast containership design case by Priftis et al. (2018). The aim of
sults show that the robust formulation for an optimized design enables 6500 TEU container ship concept design optimization study has been
reaching to the finalized solutions comprising of least rate for the cost stated as demonstrating that the modern ship design optimization
and the lowest standard deviation from the mean. techniques may enable the minimization of the capacity ratio, the
Rigo and Caprace (2011) have published findings on an approach required freight rate, EEDI, ship resistance and the maximization of

2
O. Yurdakul et al. Ocean Engineering 239 (2021) 109908

stowage ratio within a holistic design approach. The techniques utilized multi-objective design in which he contributed the literature with a
are Sobol sequence for hull generation and a genetic algorithm with book chapter. He gives examples of Pareto illustrations and scatters from
Pareto front for optimization. The improvement of the attained EEDI is the studies of Papanikolaou (2010), (2011) and mentions the benefits of
3.29% and the RFR is 13.87%. Pareto curves or surfaces throughout multi-objective Models section.
An example of Pareto application for a structural design of ships has A packing approach developed by Van Oers (2011) improves the
been demonstrated by Na and Karr (2016). The optimization techniques early-stage ship design by exploring and generating a 3D design sets
are based on the size of step, the limit of convergence and a search with conflicting requirements. The packing approach is briefly based on
conducted in a random fashion. In this study, Pareto Strategy has been the description of a novel parametric ship to handle the 3D design
verified by comparing the results obtained to the results of other opti­ changes, search algorithm to generate a set of ship design, developing
mization techniques as Evolutionary Strategy and Random Search, by most promising feasible designs considering conflicting criteria, and the
means of superiority and effectiveness. The study has been performed on selection process.
the design of the double bottom structure of a double hull tanker, and Having reviewed the literature on the utilization of Pareto method on
the results show that Pareto Strategy is the best method in the viewpoint ship design, with particular emphasis on the design of the hull shape and
of accuracy and the success rate. its influence on performance, it is evident that majority of the research
Minimization of the required power and the optimum wake field of a effort has been concentrated on obtaining the optimized final hull shape
chemical tanker hull form has been studied by Veldhuis et al. (2016). with respect to certain single or multiple optimization criteria. Pawling
The calculation has been completed with a viscous flow solver. A Pareto et al. (2017) have investigated the relationship between the ship size
front has been used in the optimization process. To speed up the process, and the cost using the Design Building Block approach and spotted
Design of Experiment approach has been implemented. Hull forms have non-linear relationships between these parameters. Another example of
been generated by Lackenby shifts and Froude mode changes. The re­ a novel approach in determining main dimensions is proposed by
sults show that the generic basis hull shapes give a sufficient initial Jianping (2017) where fuzzy logic has been used to obtain main di­
design exploration. mensions for enhanced economic performance. Previous effort of Calisal
A multi-objective optimization of the DTMB-5415 have been studied (2017) is seen as another novel example of using Pareto approach as it
by Wu and others, by using an optimization tool named OPTShip-SJTU has been implemented at the very early stages of ship design, aiming at
(Wu et al., 2016). As a parametric hull modification, free-form defor­ achieving better main dimensions and form parameters when compared
mation (FFD) has been used and new hull surfaces have been generated to standard regression approach.
under geometric constraints. Both the resistance and seakeeping have The proposal of Calisal (2017) for the selection of better dimensions
been optimized and the Neumann-Michell (NM) theory and an extension should be regarded as a decision-making process for selecting ship main
of Bales seakeeping ranking method have been integrated for dimensions that uses Pareto frontier solutions rather than a
wave-making drag and Bales seakeeping rank factor R respectively. multi-criteria optimization problem with objective functions and con­
Pareto fronts have been produced with a multi-objective genetic straints. The present research campaign aims to extend the application
algorithm. of the Pareto based decision making process for selection of better main
An optimization of a hull form for container ships have been studied dimensions to a broader range of ships with different operating profiles.
by Liu et al. (2017). The optimization tool consists of three modules An initial design procedure for this decision-making process, enabling
which are hull deformation, hydrodynamic performance estimation and the evaluation of a substantially enlarged design space is proposed and
optimization. The modification of hull forms globally and locally has applied to the improvement of performance of a destroyer design and a
been performed by using the free-form deformation (FFD) and radial planing craft. The objective of the study is to aim for identifying a Pareto
basis function (RBF) methods. The Neumann-Michell (NM) theory and Frontier which indicates a technological barrier for performance with
an approximation model have been adapted to reduce the optimization respect to the conflicting performance criteria. Identifying the techno­
cost. The effect of each variable on other has been represented by the logical barrier and narrowing the design space toward a smaller zone
analysis of variance (ANOVA) method. The resistance coefficient has around the technological barrier at early stages of design would serve as
been minimized at three different speeds. A Pareto optimum solution has a better initial condition to local and global optimization studies which
been verified by using the NM theory and RANS-based CFD solver. would be conducted at advanced stages of design: In both cases, a
Esmailian (2017) have studied the ship propeller-hull optimization “better” design with improved main dimensions and hull parameters is
with the design parameters Lifetime Fuel Consumption (LFC) and a sought that would have enhanced the seakeeping performance whilst
function based on the cost calculation including thrust, torque, and open also lowering the fuel consumption.
water efficiency. NGSA-II, an evolutionary algorithm, has been
employed to optimize parameters that can find the Pareto front and the 2. Workflow of proposed design approach
results have concluded that the LFC and the cost have been minimized
efficiently. The process is initiated by the definition of the design space ac­
Calisal (2017) has applied the Pareto approach to fishing vessel cording to the design constraints. When selecting the range of the main
design. The approach was to account for resistance and motions in an dimensions for exploration and subsequently the combinations of these
algorithmic fashion and assess the effect of varying the ship main di­ dimensions to generate design alternatives, a quasi-random approach
mensions on the performance parameters within a simulation environ­ may come in as handy since the designer has a certain level of control
ment. The study has proven that while aiming for a fuel-efficient design, over the design space and at the same time, the generation process of
performance measures may also be incorporated into the design process design alternatives exhibits a fair degree of randomness.
without compromising from fuel efficiency. The design space is usually constrained either for the usage of the
Another multi-objective optimization problem with the objective machine or technical reasons. In ship application a constraint may be the
functions of minimization of oil outflow parameter, maximizing the depth of the water she must sail on or the dimensions of the harbour
cargo capacity and minimizing longitudinal bending moment have been berths. A good example could the beam of the ship to use the Panama
studied by Jafaryeganeh et al. (2019). The internal layout of a tanker Canal. These dimensions will reduce the design space in smaller zones
ship has been optimized with considering positions of the watertight where the acceptable design parameters must reside.
members as a design variable. Pareto frontier have been obtained with In this research campaign, the proposed approach is applied to
multi-objective genetic algorithm and the optimal solution have been existing designs, therefore constraints have been defined for the main
determined by using the utility function. dimensions and the form parameters as percentages of the initial di­
Nowacki (2019) has emphasized the use of Pareto approach in a mensions of the ships in consideration. This would yield alternative

3
O. Yurdakul et al. Ocean Engineering 239 (2021) 109908

combinations of parameters in a pre-determined interval based on the The discretization of the design space gives the efficient Pareto
initial design and random algorithms (pre-defined algorithm in MATLAB frontier. A polynomial description of the frontier can be obtained from
and Sobol Sequence) have been used for the definition of the design this Pareto curve. The determination of the optimum depends on the
space. For application to new designs, the process may be initiated from selection of the utopia point. Utopia point is the point that satisfies both
principal dimensions obtained from traditional regression analysis. criteria ideally and simultaneously. There are some other methods to
Sobol sequence is a type of “quasi-random low-discrepancy choose the optimum point from Pareto set as well. As this procedure is
sequence”. The details of its algorithm are well defined by Bretley and considered to be a decision-based design theory, to choose the best-
Fox (Bretley and Fox, 1988). When the Sobol algorithm is applied on matched point with the criteria is seen as the most efficient method.
creating a design space, the created design space is uniform and The smallest distance from the utopia point to the candidate design
randomly generated. The sequence generates the numbers as binary could also be sought in the search of the “better” design (Çalışal, 2017).
fraction from a set. In comparison to the random distribution, this Utopia point could be an unreachable limit in the design but could be
method is avoiding the production of clusters or voids. This smart space considered as an ideal performance or dimension for the design. In the
samples and uniform distribution makes the sequence more efficient for current studies no fuel consumption and no acceleration were usually
the optimization process (Turan et al., 2016). The Sobol sequence also taken as the utopia point. However, any point away from the Pareto
avoids the generation of grid lines, this provides a stable basis for frontier may be selected in principle as a perfect target design point.
following steps such as an optimization (Fassardi and Hochkirch, 2006). The search for the best candidate is not a well-defined process even if
Following the design space definition, the performance criteria that there is a Pareto front. In general, equal weight could be given for the
will be driving the decision-making process needs to be defined. A ship performances used for the selection criteria. However, one may possibly
like any other machine must have performance criteria. For ships such argue that one performance criteria could have a relatively higher
criteria are determined by the owner and the regulatory agencies. These weight, the procedure used allows for such a weight assignment. In the
parameters may be the payload, speed for the owner, fuel consumption current studies, equal weights are assumed for the considered criteria.
and pollution or wake quality. For a naval ship survivability, speed,
power demand, reduced motions and signature may be demanded with 3. Case studies with the proposed design approach
equal importance. A planing craft would need to consume little fuel
while motions would need to be restricted. It is evident that the per­ 3.1. USCG 5628 based planing vessel
formance criteria set by the owner/operator/regulatory body or end
user are conflicting with one other; improvement of the performance In this part of the work, a design procedure as described in Chapter 2
with respect to one parameter would normally yield degradation ac­ is implemented on the design study of a planing craft case. The con­
cording to the other such as high payload vs low fuel consumption or low ceptual design of a planing craft, owing to the nature of its design is
power vs lowered motions. In this study, multiple design performance more suitable to algorithmic analyses when compared to other types of
criteria that are naturally conflicting with each other are aimed for marine craft such as naval ships and commercial ships. The ease of
improvement for both considered cases simultaneously. defining a planing craft’s shape and linking its performance to its main
As soon as the design space and performance parameters are defined, dimensions, deadrise angle and LCG makes it easier to utilize the Pareto
the candidate designs will need to be analysed with respect to the per­ based design procedure that could enable the investigation of consid­
formance criteria by suitable methods and tools. Experimental, analyt­ erably higher number of candidate designs in a design office environ­
ical or numerical tools provide now to the designers excellent ways to ment. Initially, a review on procedures and algorithms for planing craft
make such analyses. The choice for utilization of specific tools for the performance prediction is presented, followed by the validation study.
studies depends on the ship type and the phase of the design as well as For this procedure, the amount of energy required to maintain a
time and budget constraints. At initial phases of design, algorithmic planing or semi planing speed, or the effective power demand should be
methods and tools with lower fidelity that provide quicker turnaround quantified. For this purpose, the procedure described in Faltinsen (2005)
time when compared to high-fidelity alternatives may be utilized has been used. Faltinsen assumes four steps for the calculation of the
enabling exploration of a substantially enlarged design space. There is a horizontal component of the total resistance therefore of the EHP (Fal­
trade-off between accuracy and exploration of enlarged design spaces tinsen, 2005). First, the average wetted length λw is calculated solving
during this selection and if necessary, hybrid approaches may be the nonlinear equation below:
adopted where the performance of the initial and final (and maybe a few
LCG 1
intermediate) candidates may also be assessed by the higher-fidelity − 0.75 + [ ]=0 (3)
λw B /
methods for further verification. In this study, two different ap­ 5.21 Fn2 2 + 2.39
λw
proaches such as algorithmic methods and low-fidelity solvers have been
utilized for two case studies. In the second step, the trim angle is calculated. Trim angle τ is ob­
Pareto procedure requires some statistical data or possible simula­ tained using the two equations below:
tions and algorithms. As pointed above there may be multiple design ( )
alternatives competing in the design. The designer must have a decision 0.067 − CL0 − 0.0065βCL0 0.6 = 0 (4)
procedure to recommend the ‘better hull’. The Pareto frontier seems to ( )
define a technical barrier which identifies a performance boundary for 0.0055λw 2.5 FL0
CL0 = τ 0.012λw 2.5 + = (5)
that design. Fnβ 2 0.5ρV 2 B2
The Pareto design procedure used in this study is a combination of In the third step, chine wetted length LC and keel wetted length LK are
the Pareto design procedures described by Gierson (2006) and Kasprzak calculated:
and Lewis (2001). /
The challenge for Pareto approach is the minimization of the vari­ λw =
0.5(LK + LC )
= 0.5(Xs + 2LC ) B (6)
ables in a conflicting criteria space called as X. The conflicting criteria B
can be denoted by fi and j. A vector solution x* may be stated to
dominate another solution x if (Das and Dennis, 1998): where:
​ Xs = LK − LC (7)
fi (x* ) ≤ fi (x) for ​ all ​ i ​ є ​ (1, ​ 2, ​ ... ​ , ​ n) (1)
( ( ) Effective Horse-Power is then calculated by using speed, viscous drag
fj (x* ) < fj (x) for ​ some ​ j ​ є ​ 1, ​ 2, ..., ​ n (2) and pressure drag.

4
O. Yurdakul et al. Ocean Engineering 239 (2021) 109908

Wetted area is seen in two parts:

⎛ ⎞
tan 2 β ⎜ B2
S1 = ⎝ ( ) ⎟⎠ (8)
sin β
4 1 + zmax/V τ
t

B
S2 = LC (9)
cos β
Horizontal component of the viscous resistance is calculated using
the sum of above values for wetted surface and the ITTC 57 formula.
Pressure resistance Rp is calculated from:
RP = FLβ τ (10)

A MATLAB based code, developed for this purpose was validated

Fig. 2. Trim angle in degrees from the Faltinsen procedure and the experi­
using the data provided in the above reference and other published al­
mental results of NSS.
gorithms by Taunton (2010) and the University of Naples Planing hull
series (NSS) by Luca De F. (2017) for the range of accuracy and trend in
the calculations. Reasonable estimates of the effective power, dynamic Table 1
trim, and dynamic waterline length are obtained for a preliminary The main particulars of US Coast Guard Motor Life Boat
design stage, even though the design algorithm by Faltinsen does not (MLB) 5628.
include advanced features of these series such as steps or variable Mass (kg) 18144
deadrise angle. Figs. 1 and 2 below show example comparisons of the
LCG 0.38 Lpp
predicted effective power and the dynamic trim angle τ by the algo­
Lpp (m) 14.33
rithms provided for the hull C3 of the Naples Systematic Series and the Beam (m) 4.42
Faltinsen’s procedure. Deadrise angle (Degrees) 16.61
The seakeeping analysis of planing hulls is rather complicated as the
dynamic lift characteristics are changing due to ship motions. Rather
was linear and the interval for beam and dead rise angle were ran­
advanced tests are reported by Bagovic (2014) on the head sea motion of
domized. A total of 1000 hulls were generated randomly within the
planing crafts, such as that of the Napoli series (NSS). The algorithmic
design space by MATLAB.
development for this series is understood to be continuing.
Within the developed approach, the design mass and the location of
Taunton et al. (2011) provide an algorithm for the acceleration levels
the initial the center of gravity of the hull and the length of the hull in
of their own series. They show a nonlinear dependence of acceleration
still water are required as inputs. The algorithmic resistance and ac­
on the significant waves. For simplicity, a simple acceleration algorithm
celeration level of the hull at the centroid of mass are then calculated.
provided by Savitsky (2003) is used for the early development of this
The hull dimensions and speed range foreseen by the algorithms are then
design procedure.
checked, against the speed to length ratio for semi planing and planing
For the testing of the overall procedure outlined above, a case study
hulls. In this study, speed to beam ratio V/√B is set for the interval of
is defined based on the US Coast Guard Motor Life Boat (MLB) 5628
1.6–4. When this length ratio is not satisfied, that specific design is not
series as given in Table 1.
included in the design population and in the final selection or suitability
For the design alternatives, the beam interval was set to be between 4
as a “better hull”.
to 8 m, dead rise angle between 10 and 20◦ and the speed range was
Another issue for suitability of the planing hull is a dynamic property
restricted between 26 and 30 knots for input to the design program. In
of the hull identified as porpoising. The prediction of porpoising was
this case, the speed of the vessel has been considered as a design variable
studied by Day J.P, Haag R.J (1952), Celano (1998) and a critical trim
as ship resistance and acceleration in waves are functions of the
angle algorithm is provided in this study against the speed coefficient
encounter frequency therefore, are functions of the speed. It should be
V/√gB. The relationship includes the dead rise angle, and the lift co­
noted that, as the design procedure is flexible, one can narrow the speed
efficient of the hull. The critical trim angle at the design speed coeffi­
to a much smaller interval or keep it as a constant. The interval for speed
cient is then compared with the trim of the designed hull at the same
speed coefficient. The porpoising and nonlinear sea keeping of planing
hulls make the additional calculations seemingly not reliable at this
stage of the design and more work is possibly justifiable in this area.
The experience gained from this study suggests that major hull pa­
rameters for a better planing hull, could be identified with the help of
existing algorithms and a design logic based on the Pareto front concept.
The likelihood of porpoising could also be included in the selection of
the main parameters with the identification of a critical trim angle. The
work done confirms that additional research needs to be done for the
better understanding of planing hull performance in regular and irreg­
ular waves.
As depicted in chapter 2, the process is initiated with the definition of
a relatively large number of principal dimensions of boats in a random
fashion. The design criteria have been set as:

• Minimum effective power requirement,

• Minimum acceleration level at the center of gravity.
Fig. 1. The prediction of Faltinsen method as compared to experimental values
of Luca De F. and Pensa (2017) for hull C of Naples Systematic Series (NSS).

5
O. Yurdakul et al. Ocean Engineering 239 (2021) 109908

At this step, an additional porpoising check is necessarily added to are not the same, these design parameters, as in Fig. 3, are non-
the sequence. dimensionalized. That is, each value of the elements of a set is divided
To be able to observe the existence of the Pareto front, analyses have by the maximum value of the elements of that set. In this case, the
been conducted for the effective power and the acceleration levels as maximum value of a variable such as EHP is non-dimensionalized by the
explained previously. All algorithms have been coded in MATLAB. maximum value obtained for EHP in the search. This provides a non-
The resistance and trim simulation for a planing hull start with an dimensional decision space, of one unit in each direction. However,
assigned location of the center of gravity and the mass of the load and some weights may be given by the designer as to provide the order of
follows the step-by-step procedure outlined by Faltinsen (2005). The magnitude of importance, seen by the designer or included in the
total resistance is converted into a more practical, easily conceivable owner’s requirements if necessary.
effective horsepower value this study. The fuel consumption is assumed In this study it was observed that such a front indeed exists for
to be proportional to effective power demanded by the ship. The planing hulls. The displacement type ships moving at relatively slow
calculated effective power is then used as one of the performance pa­ speeds are eliminated from the population of ships.
rameters in the definition and search of Pareto front. Fig. 3 suggests that the set of minimized acceleration levels at a
The algorithm used in this study for the simulation of the accelera­ constant effective power can be seen as a Pareto Front. The Utopic point
tion at the center of gravity is from Savitsky (2003). The average ac­ in this study is taken as the origin of the design space; the zero accel­
celeration at the center of gravity is formulated as: eration and the zero effective power point, is defined as the ‘utopic
( / ) ( )( )2 ( )/ point’. In this study, the design with minimum distance from the utopic
/
nCG = 0.0104 H13 + 0.084 τ 4 5 3 − β 30 VK √̅̅̅
̃ /
/ / /
L/b point to the candidate design is defined as the better ship. The suitable
b L C Δ
number of designs on the Pareto front are then located by a numerical
(11) search method.
Fig. 3 shows the performance space and the “better” design is shown
From more general algorithms, such as a response spectrum, one can
with a quadrilateral marker. All the computer-based study and the
of course calculate additional statistical values, for an assigned sea state.
programming are done in MATLAB, and it takes about 20 s to run the
One can possibly use the assumptions in Jensen (2004) to calculate the
program including the graphical work. Table 2 below shows some of the
combined motion at the bow of the ship and the relative bow motion
design parameters calculated for the better point and 4 other points near
with respect to the wave elevation. The main assumption in Jensen
the better point. The Frontier also depicts a clear technological barrier
(2004) is that the interactions between different modes of motions can
for the given design constraints. It is seen that existing designs may be
be neglected and or ignored. In this study the sea state is taken as sea
improved up to the technological barrier with the proposed procedure.
state 3.
The applied procedure suggests that the importance of the two per­
One thousand combinations of beam values, hull speeds and dead
formance criteria studied by the method is of equal importance. How­
rise angles are used to obtain a design population for this planing hull
ever, as recommended by Kasprzak et al. (2001), if one criterion is more
study. After the completion of the calculations, the existence of a specific
important than the other one by a factor n, the coordinate of the
Pareto front is sought. This front is seen as technological barrier in the
candidate designs could be increased in that direction by the same fac­
design. In this case 1000 design points, are plotted as in Fig. 3. As the
tor, reflecting the increased importance of that criteria or that objective.
units of the design parameter such as acceleration and effective power
Table 2 suggests that increased deadrise angle and slightly decreased
beam provides a better compromise solution when both effective power
demand and accelerations are aimed for improvement. Best design
(Best1) has 6% less effective power requirement as well as 6% less ac­
celeration at LCG.
The critical degree to avoid porpoising suggested by Day et al. (1952)
as:
( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
)
CL − 0,2629
τcrit = 0, 1197 β 0.7651
exp 15.7132 β (12)
2

where
Δ
CL = (13)
0.5ρV 2 B2
Savitsky (1993) provides a figure for the critical angle, a curve fit to
those curves, expressed as:
( )
τcrit = 16.47 CA 2 + 14.97CA + 2.38 (0.010β + 1) (14)

Table 2
Design values obtained for US Coast Guard 5628.
Best1 Best 2 Best3 Best 4 Best 5 Original

Trim (degrees) 4.96 4.67 4.99 4.75 4.74 4.70

FnB 2.13 2.03 2.13 2.05 2.05 2.10
Beta (degrees) 18.69 18.69 19.10 18.69 18.69 16.61
Beam (m) 4.02 4.43 4.02 4.31 4.33 4.27
Lkeel (m) 11.66 11.89 11.70 11.82 11.83 11.53
Effective Power (kW) 290 312 303 308 300 310
Acceleration at LCG 0.81 0.87 0.81 0.85 0.86 0.86
(g)
Fig. 3. A Pareto front for planing hull design.

6
O. Yurdakul et al. Ocean Engineering 239 (2021) 109908

where: been established for reliability analysis and is competent to be used for
√̅̅̅̅̅̅ optimization problems and robustness evaluation. It allows the link of
CA =
CL
(15) different solvers and the construing of the input and the output pa­
2 rameters (Bucher et al., 2001). It has previously been used for the
Celano (1998) has extended Day et al.‘s studies with larger models optimization problems of the variable speed turbine designs (Tengs
and higher deadrise angles as seen in recent designs. Analysing the et al., 2018) and the derivation of the mechanical properties of bonded
extensive test data followed by a regression study resulted with the joints (Bernardin et al., 2013).
below formulation for critical trim angle: A Sobol sequence as defined by (Bretley and Fox, 1988) has been
( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) implemented on MATLAB for obtaining the design space. 30 parent
τcrit = 0.1197β0.7651 exp 15.7132
CL − 0.2629
β (16) ships with varying length and beams have been developed with the
2 above-mentioned sequence. The variation of length and beam has been
restricted to 10% compared to the initial hull. Even though 10% varia­
Celano’s algorithm was utilized in this study to identify if the design tion may be considered to be a large range when seeking improved
is in or near a critical area. Fig. 4 shows that the designs and specifically design alternatives, keeping the variation range reasonably large en­
the ‘better’ hull has a trim less than the critical trim given by the algo­ ables the designer to detect outside-the-box better designs -if any-that
rithms. In the Figure, trim angles are plotted against CV which is ob­ would normally not be considered unless randomness is involved in
tained with the below formulation: the process. The developed ships all have the same payload capacity; the
V lightship weight has been assumed to be varying with the product of the
CV = √̅̅̅̅̅ (17)
gb main dimensions (length, beam, and depth) and the payload of the
parent ship has been added to the lightship of each candidate ship to
obtain its displacement and draft. A total number of 270 ships define the
3.2. DTMB 5415 based destroyer design space that are variants of 30 parent ships. The length-beam
variation of the parent ships obtained with the Sobol algorithm are
This case study considers the DTMB 5415 hull which is a ship model given in Fig. 5. All ships are generated from the parent ships with var­
tested at David Taylor Model Basin based on an early DDG 51 Arleigh iations given below:
Burke class destroyer as the initial hull form for seeking better main
dimensions and form parameters. As this hull has been considered as an • LCB: 50%, 51%, 52%,
initial conceptual design that has been dimensionalized for some pre- • CP : 0.59, 0.607, 0.624.
determined set of performance criteria, it is evident that setting a new
pair of design criteria that may be said to be conflicting with each other The main dimensions and the LCB values created by the Sobol al­
would possibly yield a different trade-off solution than that of the initial gorithm are used to modify the initial DTMB 5415 hull in Maxsurf
hull. Modeller. Main dimensions are modified by scaling and parametric
In this study, a new set of improved main dimensions and form pa­ transformation has been used to vary LCB position. This process is car­
rameters based on the DTMB 5415 hull are sought with respect to the ried out in batch-mode.
following criteria: Following the generation of candidate hulls, they have been exported
in IGES format and fed into the meshing software. GID software has been
• Minimization of effective power demand, used for meshing. A structured meshing strategy has been adopted. A
• Minimization of the pitch acceleration at the bridge. sample mesh is depicted in Fig. 6.
The effective power demand of the ships has been estimated with a
In contrast to the planing craft case presented in section 3.1 above, potential flow solver nuShallo at 18 knots of speed. Nushallo is a non-
this case study required the use of dedicated solvers for resistance and linear free-surface potential flow solver that incorporates the method­
seakeeping calculations. The handling of all the creation of the geome­ ology presented by Jensen et al. (1986) in terms of capturing the
try, the meshing, the performance calculations, and the post-processing free-surface. The ship hull is discretized by triangular and rectangular
of the data for an enlarged design space required the use of dedicated cells, the mesh of the hull is blended to the obtained free surface at every
software for linking all the components that are utilized in the process. iteration loop, leading to an improved numerical accuracy. The trim and
In this study, Opti-Slang has been used as the control platform of the the sinkage of the hull are also accounted for during the calculations
procedure. Opti-Slang is an optimization platform that has principally (Marzi, 2003).
The validation study for nuShallo has been based on an extrapolation

Fig. 4. Critical Trim angle check for porpoising. Fig. 5. Length and beam variations of the parent ships.

7
O. Yurdakul et al. Ocean Engineering 239 (2021) 109908

Table 3
Seakeeping calculation parameters.
Sea Spectra Pierson-Moskowitz

Sea State SS5

Heading Angles 180◦ (head sea)
Wind Speed 24 knots (Sea State 5)
VCG 7.54 m
Roll Gyradius 40% BM
Fig. 6. Sample mesh utilized for calculation. Pitch Gyradius 25% LOA

from the resistance test results of the DTMB 5415 (Olivieri et al., 2001),
effective power values are non-dimensionalized by dividing with
which is the initial hull utilized in this study. Fig. 7 shows the effective
maximum value. Fig. 9 shows the obtained Pareto front.
power results obtained from nuShallo in comparison to the
The dimensions, the effective power and the acceleration values of
experiment-based data. A good agreement is observed for the desired
the initial and the better hull are given in Table 4. In this case, a larger
speed range. As the aim of this study is to explore an enlarged design
and a slightly narrower hull has shown better performance when the
space at an early design stage with reasonable effort and sufficient ac­
multiple criteria are considered for the decision of the main dimensions.
curacy, it is anticipated that utilization of a potential solver is a sensible
A reduction of 1% in effective power has been obtained while also
approach.
improving the pitch induced acceleration by 9%.
The acceleration values at the bridge location have been obtained
The chosen main dimensions of the destroyer influence both the
from Maxsurf Motions that utilizes a strip theory-based solver for the
seakeeping performance and the effective power demand. However, the
calculation of the motions and the accelerations. Seakeeping calculation
effect of choosing of the different main dimensions is more pronounced
conditions and crucial modelling parameters are given in Table 3.
on the seakeeping responses when compared to the power demand.
The utilized seakeeping software for the optimization process needs
Local refinement of the hull shape could lead to further reductions in
to have sufficient capability for capturing the differences in responses of
resistance as the derived hull forms may still be keeping certain features
the alternative hulls in the heave, the pitch motions and in the more
of the initial hull which have not been adjusted to the variants (Diez
comprehensive evaluations in the added resistance due to wave action.
et al., 2015). Therefore, there is further potential of improvement by
For the currently utilized seakeeping software, three validation cases are
single or multi-criteria optimization studies which may be conducted at
presented in the User Manual (Formation Design Systems, 2011) namely
the progressive stages of design, where the hull form could be varied
for NPL Round Bilge Series, Series 60 and a fast freight vessel of SA van
locally with a suitable algorithm. At this stage, it can be concluded that
der Stel. The limitations and assumptions of the utilized strip theory
the proposed approach for the selection of the main dimensions is not a
approach is presented as well.
replacement for single or multi-criteria optimization studies but a sup­
Additionally, a validation study has been conducted with the utilized
plementary approach that may be adopted at the very early stage of the
software for the vessel in consideration. For this purpose, experimental
design.
data for the subject vessel has been utilized from (Olivieriet al, 2001
When the improvements obtained by this study are compared with
cited in Cakici et al., 2017). The considered case is at a speed corre­
the optimization studies cited in Chapter 1 (Diez et al., 2015) and (Wu
sponding to Fr = 0.41 and the heave and the pitch responses at the
et al., 2016) it is seen that the improvement in the seakeeping perfor­
center of gravity are obtained for the head seas condition. Results ob­
mance achieved by selecting better dimensions is comparable to that of
tained from the computations in Maxsurf are presented along with the
the optimization studies. The seakeeping improvements varied between
experimental data in Fig. 8. It is seen that there is good agreement be­
3 and 13.5% in the studies. The reduction in the resistance is signifi­
tween the predictions and the experimental data for the speed and the
cantly higher for the optimization studies as the local hull form is more
heading combination. The method is capable of predicting the trend in
influential on the resistance characteristics. Current modification of the
the variation of the RAO’s across the wave frequency range.
main dimensions and the underwater volume distribution has a certain
Results of the potential flow and the seakeeping analysis indicate the
effect on the resistance as well; a significant improvement in seakeeping
formation of a Pareto front for this case as well. The acceleration and the
performance has been achieved while also reducing effective power
demand by 1% by altering the main dimensions, LCB and Cp.

4. Conclusions

A Pareto based decision-making approach for the selection of the

ship main dimensions has been proposed. The aim of this approach is to
enable the designers to explore their design space in such a way that they
may see the influence of ship main parameters on multiple performance
criteria, enabling them to optimize or improve the design and most
importantly give the designers the opportunity of thinking outside-of-
the-box. By utilizing the proposed methodology, the designer would
have the chance to push the design towards the technological barrier
instead of relying on a particular design solution balancing the design
objectives. Traditional design approach usually dictates the minimiza­
tion of the power demand. This study was aimed to see if a better hull
could be designed by including more design objectives and still save
some fuel. It is seen that the initial designs in both case studies could be
improved up to the technological barrier. The efficacy of the proposed
process required to enhance the initial design up to the technological
barrier is notable and useful results may be obtained in a cost-effective
Fig. 7. Effective power predictions compared to experimental data (Olivieri fashion if the utilized tools are sufficiently quick and have satisfactory
et al., 2001).

8
O. Yurdakul et al. Ocean Engineering 239 (2021) 109908

Fig. 8. Comparison of heave and pitch RAO’s predicted by the utilized software and experiments for subject vessel.

the proof of the universality of the proposed Pareto approach. However,

it should be noted that further diversified evaluation of the proposed
approach for different ship types would be beneficial for generalizing its
universality.
The proposed method is shown to find rather efficiently a techno­
logical barrier for the performance criteria such as the fuel efficiency
and the acceleration levels of designs. In both presented cases, the
technological barrier could easily be achieved with the proposed
approach. Of course, the results are limited with "today’s knowledge” on
the type of a ship in terms of the independent variables the designer has
control of, such as the main dimensions. With this knowledge of the
technological barrier, which is limited by the information side of the
design process, a design decision is still needed. This selection decision
needs a design criterion which is the selection of the “Utopic point “. For
different usage of the hull, as for a passenger ship or for a naval ship,
different utopic points or maximum levels could be considered as an
owner requirement. The existence of the technological barrier also
suggests that new technologies are needed to further improve this bar­
rier by changes such as: the ventilated hulls, stepped hulls for planing
Fig. 9. Pareto front for the DTMB 5415.
craft and by bulbs or parabolized waterlines for displacement type ships.
The proposed approach may be used for selecting the main di­
mensions in a more refined fashion compared to traditional approaches.
Table 4 The introduced randomness enables the assessment of combinations that
Main dimensions and the performance parameters of the initial and the best hull. may generally be considered outside of the regular design alternatives
Parent Ship (Hull 000) Better Ship (Hull 264)
that would otherwise not be considered.
In the current study, the process was based on improving the main
L (m) 142.171 149.289
dimensions for two basic performance requirements, namely the effec­
B (m) 19.079 18.249
LCB, % L from Aft Perpendicular 0.51 0.5 tive power at one speed and the seakeeping performance which was
Cp 0.607 0.607 assumed to be represented by the acceleration in head seas. Evidently,
Effective Power (kW) 3364 3323 there are many more important performance aspects that may be driving
Pitch Acceleration (rad/s2) 0.02617 0.02393
the process of the selection of main dimensions and certain initial hull
parameters. The selection of the performance parameters that are said to
fidelity. be driving the process is dependent on the operation profile, the per­
It is evident from the results presented above that the ship main di­ formance requirements, the type of vessel and the relevant ranking of
mensions, the distribution of weight and the distribution of underwater the importance defined by either the designer, or the owner, or the
volume have notable effect on the performance of the vessel in terms of operator of the vessel. Therefore, the designer using this procedure
the fuel consumption and the ship motions in the seaway. The additional could easily and freely choose other hull parameters for progressing
effort that would be put in the early design process to evaluate an with and obtaining “better” main dimensions.
enlarged design space with different main dimensions and form pa­ Although the procedure shows a potential as a prospective approach
rameters is found to be worthwhile for the tested cases. This approach for use in a design office environment, further improvements are
should be regarded as a supplementary effort at the very initial phase of beneficial. Currently, the importance of each design parameter has been
the design, and it should be noted that it is not an alternative to local or assumed as equal. The design criteria or the importance of ‘weights’ of
global optimization studies that would be conducted at the progressive design performance measures could be changed by the owner or the
stages involving single or multiple criteria. Combining the two by designers. Other measures or combination of measures could be added
selecting better main dimension by the proposed method and perform­ into the design study. The procedure developed for two performance
ing the necessary optimization studies thereafter could enable the actual objectives could be extended into multiple performance objectives in a
potential of the design to be unveiled. systematic way.
Two case studies have been presented using the proposed approach
that are based on DTMB 5415 and a planing craft. A Pareto frontier has CRediT authorship contribution statement
been observed for all the cases considered. This finding is beneficial for
Onur Yurdakul: Methodology, Validation, Formal analysis,

9
O. Yurdakul et al. Ocean Engineering 239 (2021) 109908

Software, Writing – original draft. Gözde Nur Küçüksu: Formal anal­ Jianping, C., Jie, X., You, G., Li, X., 2017. Ship hull principle dimensions optimization
employing fuzzy decision-making theory. Math. Probl Eng. 1–9. 2017.
ysis, Software, Writing – original draft. Ahmet Ziya Saydam: Method­
Kasprzak, E.M., Lewis, K., 2001. Pareto analysis in multiobjective optimization using the
ology, Project administration, Supervision, Validation, Visualization, colinearity theorem and scaling method. Struct. Multidiscip. Optim. 22 (3),
Writing – original draft. Mehmet Sander Çalışal: Conceptualization, 208–218.
Methodology, Software, Validation, Formal analysis, Writing – review & Kim, D.H., Paik, J.K., 2017. Ultimate limir state-based multi-objective optimum design
technology for hull structural scantlings of merchant cargo ships. Ocean Eng.
editing. 318–334.
Liu, X., Wan, D.-C., Zhao, M., 2017. Hull form multi-objective optimization for a
Declaration of competing interest container ship with neumann–michell theory and approximation model, 04 Int. J.
Offshore Polar Eng. 27, 423–432.
Luca, F.D., Pensa, C., 2017. The Naples warped hard chine hulls systematic series.
The authors declare that they have no known competing financial Journal of Ocean Engineering 205–236.
interests or personal relationships that could have appeared to influence Marzi, J., 2003. Use of CFD methods for hullform optimisation in a model basin. In:
MARNET CFD Workshop, pp. 149–164 (Haslar, UK: MARNET – CFD Workshop).
the work reported in this paper. Na, S.-S., Karr, D.G., 2016. Development of Pareto strategy multi-objective function
method for the optimum design of ship structures. International Journal of Naval
References Architecture and Ocean Engineering 8 (6), 602–614.
Nowacki, H., 2019. On the history of ship design for the Life cycle. In: Papanikolaou, A.
(Ed.), A Holistic Approach to Ship Design. Springer Publishers, pp. 43–73.
Bagovic, E., Berorello, C., Pennini, S., 2014. Planing hull seakeeping in irregular head
Olivieri, A., Pistani, F., Avanzini, A., Stern, F., Penna, R., 2001. Towing Tank
seas. Trans. FAMENA 38, 1–12.
Experiments of Resistance, Sinkage and Trim, Boundary Layer, Wake and Free
Bernardin, P., Vacik, J., Kroupa, T., Radek, K., 2013. Determination of the mechanical
Surface Flow Around a Naval Combatant Insean 2340 Model. IIHR Technical Report
parameters of a bonded joint between a metal and a composite by comparing
No. 421.
experiments with a finite-element model. Mater. Technol. 417–421.
Papanikolaou, A., 2010. Holistic ship design optimization. Comput. Aided Des. (42),
Bretley, P., Fox, B.L., 1988. Algorithm 659: implementing Sobol’s quasirandom sequence
1028–1044.
generator. ACM Transactions on Mathematical Software.
Papanikolaou, A., Harries, S., Wilken, M., Zaraphonitis, G., 2011. Integrated Design and
Bucher, C., Will, J., Riedel, J., 2001. Multidisciplinary Non-linear Optimization with
Multiobjective Optimization Approach to Ship Design. Proceedings of ICCAS,
Optimizing Structural Language OptiSLang. 19th CAD-FEM Users’ Meeting,
Trieste.
International Congress on FEM Technology. Potsdam: International Congress on FEM
Pawling, R., Andrews, D., Percival, V., 2017. A study into the validity of the ship design
Technology.
spiral in early stage ship design. Journal of Ship Production and Design 81–100.
Cakici, F., Sukas, O., Kinaci, O., Alkan, A., 2017. Prediction of the vertical motions of
Priftis, A., Boulougouris, E., Turan, O., Papanikolaou, A., 2018. Parametric design and
DTMB 5415 ship using different numerical approaches. Brodogradnja 29–44.
multi-objective optimization of containerships. Ocean Eng. 156, 347–357. https://
Çalışal, S., 2017. Fuel or Acceleration: A Pareto Design Procedure for Ships. IMAM.
doi.org/10.1016/j.oceaneng.2018.02.062.
International Maritime Association of the Mediterranean, Lisbon, Portugal.
Rigo, P., Caprace, J., 2011. Optimization of ship structure. 1st international Conference on
Celano, T., 1998. The Prediction of Porpoising Inception for Modern Planing Craft.
maritime Technology and engneering, pp. 925–944 (Lisbon: International Conference
United States Naval Academy, Annapolis, Maryland.
of Maritime Technology and Engineering).
Das, I., Dennis, J.E., 1998. Normal-boundary intersection: a new method for generating
Savitsky, D., 1993. The Evolution and Contribution of Model Tests to Understanding
the Pareto surface in NonLinear multicriteria optimization problems. SIAM J. Optim.
Planing Hull Hydrodynamics.
631–657.
Savitsky, D., 2003. On the Subject of High-Speed Monohulls. Greek Section Of SNAME.
Day, J., Haag, R., 1952. Planing Boat Porpoising. Webb Institute of Naval, New York.
Greek Section of the Society of Naval Architects and Marine Engineers (SNAME),
Formation Design Systems, 2011. Seakeeper User Manual.
Athens, Greece.
Diez, M., Peri, D., 2010. Robust Optimization for Ship Conceptual Design. ElsevierLtd,
Taunton, D., Hudson, D., Shenoi, R., 2010. Characteristics of a series of high speed hard
Rome, Italy.
chine planing hulls Part I performance in waves. International Journal of Small Craft
Diez, M., Serani, A., Campana, E.F., Gören, Ö., Danışman, D.B., Grigoropulus, G.,
Technology 55–75.
Stern, F., 2015. Multi-objective Hydrodynamic Optimization of the DTMB 5415 for
Taunton, D., Hudson, D., Shenoi, R., 2011. Characteristics of a series of high speed hard
Resistance and Seakeeping. International Conference on Fast Sea Transportation,
chine planing hulls Part II performance in calm water. Intarnational Journal of Small
Washington DC.
Craft Technology 153, 153.
Esmailian, E., Ghassemi, H., Zakerdoost, H., 2017. Systematic probabilistic design
Tengs, E., Storli, P.-T., Holst, M., 2018. Optimization Procedure for Variable Speed
methodology for simultaneously optimizing the ship hull-propeller system.
Turbine. Engineering Applications of Computational Fluid, pp. 652–661.
International Journal of Naval Architecture and Ocean Engineering 9, 246–255.
Turan, O., Mizzi, K., Demirel, Y.K., Banks, C., Atlar, M., Kaklis, P., 2016. Design
Faltinsen, O., 2005. Hydrodynamics of High Speed Marine Vehicles. Cambridge
optimization of propeller boss cap fins for enhanced propeller performance. Appl.
University Press, Cambridge.
Ocean Res. 210–222.
Fassardi, C.D., Hochkirch, K., 2006. Sailboat design by response surface optimization. In:
Van Oers, B., 2011. A packing approach for the early stage design of service vessels
2nd High Perfomance Yacht Design Conference. Auckland: Proceedings of the
(Doctoral Dissertation, TU Delft, Netherlands). Retrieved from. https://repository.tu
International Conference on High Performance Yacht Design.
delft.nl/islandora/object/uuid%3A6be7582c-63b1-477e-b836-87430bcfb43f.
Gierson, D., 2006. Pareto analysis of Pareto design. In: Joint International Conference on
Veldhuis, C., Scholcz, T.P., Gornicz, T., 2016. Ship Optimization Using Viscous Flow
Computing and Decision Making in Civil Building Engineering. Montreal, Canada.
Computations in Combination with Generic Shape Variations and Design of
Jafaryeganeh, H., Soares, C.G., 2019. Multi-objective optimization of internal
Experiments. Practical Design Of Ships And Other Floating Structures. Copenhagen:
compartment layout of oil tankers. Journal of Ship Production and Design 374–385.
Practical Design of Ships and Other Floating Structures.
Jensen, G., Mi, Z.X., Söding, H., 1986. Rankine methods for the solution of the steady
Wu, J., Liu, X., Wan, D., 2016. Multi-objective hydrodynamic optimization of ship hull
wave resistance. In: Sixteenth Symposium on Naval Hydrodynamics. California:
based on approximation model. In: International Ocean and Polar Engineering
Symposium on Naval Hydrodynamics, 16th.
Conference. Rhodes: International Society of Offshore and Polar Engineers.
Jensen, J., Mansour, A., Olsen, A., 2004. Estimation of ship motions using closed form
expressions. Ocean Eng. 31 (1), 61–85.

10