Scientific Journals Zeszyty Naukowe

of the Maritime University of Szczecin Akademii Morskiej w Szczecinie

2018, 53 (125), 48–56
ISSN 1733-8670 (Printed) Received: 15.09.2017
ISSN 2392-0378 (Online) Accepted: 12.01.2018
DOI: 10.17402/265 Published: 16.03.2018

Hydrodynamic multidisciplinary optimization of a container

ship and its propeller using comprehensive HPSOP code

Hassan Zakerdoost, Hassan Ghassemi

Amirkabir University of Technology, Department of Maritime Engineering
Hafez Ave., 15875-4413, Tehran, Iran, e-mail: {h.zakerdoost; gasemi}@aut.ac.ir

corresponding author

Key words: hydrodynamic optimization, marine propeller, container ship, hull resistance, ship industry, open
water efficiency
Abstract
Hydrodynamic shape optimization plays an increasingly important role in the shipping industry. To optimize
ship hull and propeller shapes for minimum total (friction+wave) calm-water resistance and maximum open
water efficiency, respectively, the main particulars of a hull and propeller model are considered as design vari-
ables. The optimization problem is performed by using an integrated hull-propeller system optimization prob-
lem (HPSOP) code in a multi-level and multi-point methodology in early-stage ship design. Three numerical
methods with variable fidelity are employed to carry out the hydrodynamic performance analysis of a ship’s
hull and propeller. A ship and its propeller are selected as initial models to illustrate the effectiveness of the pro-
posed optimization procedure. The numerical results show that the developed technique is efficient and robust
for hydrodynamic design problems.

Nomenclature PV: Vapor pressure

K: Keller’s coefficient (0 < K < 0.2)
L: Length of ship
B: Breadth of ship Introduction
d: Draft of ship
D: Propeller diameter The conventional maritime industry approach
Z: Number of blades of the propeller is to optimize hull resistance and propeller perfor-
P/D: Pitch-diameter ratio of propeller mance separately. A ship’s hull resistance is mini-
EAR: Expanded area ratio of propeller mized by a naval architect and the propeller thrust
Fn: Froude number is maximized for a given power by the propeller’s
Rt: Ship total resistance designer. Once both the designs (hull and propel-
T: Propeller thrust ler) are combined, the actual performance of the
Q: Propeller torque system is found. The ultimate list of all the larg-
Kt: Thrust coefficient (Kt = T/ρn2D4) est container ships in the world is presented online
Kq: Torque coefficient (Kq = Q/ρn2D5) (Wikipedia, 2018). There, the names of their own-
Eta: Propeller efficiency (Eta = J·Kt/2πKq) ers – the world’s largest container shipping line
J: Advance coefficient (J = VA/nD) companies – can be seen. Details of one particular
N: Propeller rotating speed in RPM container ship (14,000 TEU) are as follows and are
n: Propeller rotating speed in RPS (= RPM/60) close to those of our case study container ship: Main
VA: Advance velocity engine power output: 72,240 KW, Speed: 22 kn,
PO: Atmospheric pressure Draft: 16 m, Breadth: 51 m, Overall Length: 352 m,

48 Scientific Journals of the Maritime University of Szczecin 53 (125)

 Hydrodynamic multidisciplinary optimization of a container ship and its propeller using comprehensive HPSOP code

DWT: 165,887 tons. Figure 1 shows a model of the more efficient than the conventional approach of uti-
container ship. lizing just one. The scope of this study is to develop
a bi-level and bi-objective optimization code for the
hydrodynamic design of container ships propelled
by their propeller(s) as a combined system at two
operating points.
Hydrodynamic optimizations of ship hulls and
propellers as two independent systems have been
carried out by many researchers. Genetic algorithm
(GA) and Michell’s theory were used by Day and
Doctors (Day & Doctors, 1997) to find the minimum
total resistance of a mono-hull and catamaran. In
another work, Michell’s theory and evolution strat-
egy were respectively applied as a wave resistance
estimator and an optimization algorithm in the opti-
Figure 1. Model of container ship mization problem of a Series 60 hull form in calm
water (Zakerdoost, Ghassemi & Ghiasi, 2013). A GA
Two main parts of a ship system which have a sig- based optimization technique and the well-known
nificant effect on the ship’s total efficiency are its Dawson panel method were employed in research to
hull and propeller, thus it is necessary to optimize the optimize a ship’s hull form from a resistance point of
effective parameters of the propeller and hull using view (Dejhalla, Mrša & Vuković, 2002). Zhang et al.
a comprehensive numerical approach. Numerical (Zhang, Ma & Ji, 2009) applied nonlinear program-
techniques play a key role in the analysis of the flow ming to minimize the wave resistance of the bow-
field around ship’s hull and propeller, especially in body shape of an S60 hull evaluated by the Rankine
hydrodynamic shape optimization, because they are source method. A double-chine, planing hull form
potentially an appropriate replacement for towing optimization was implemented in another study
tank experiments. It is worth noting that the hull-pro- using evolutionary strategies with respect to hull
peller system optimization problem (HPSOP) code performance in calm and rough waters (Grigoro-
usually requires a large number of evaluations; thus, poulos & Chalkias, 2010). Other researchers (Jeong
selecting a less time-consuming and accurate numer- & Kim, 2013; Park, Choi & Chun, 2015; Kim, Choi
ical method is very important. In the early stages of & Chun, 2016) have also worked on the hydrody-
a multi-objective optimization problem (MOP) the namic optimization of ship’s hull form using poten-
design space is large; as the optimization algorithm tial flow solvers and optimization algorithms. The
progresses and approaches an optimal design area efficiency of a self-twisting composite propeller was
the design space becomes smaller. Therefore, using maximized by Pluciński et al. (Pluciński, Young
a fast numerical method with relatively good accura- & Liu, 2007) using a GA and BEM/finite element
cy in the initial steps, and a more accurate one in the method (FEM) solver. Benini (Benini, 2003) pro-
final stages of the optimization problem, is a rational posed a multi-objective evolutionary algorithm to
way to find optimal designs. optimize a B-series propeller, while the open water
Determination of the hydrodynamic character- performance was calculated using regression formu-
istics of a ship’s hull and propeller using compu- las. A program was developed by Burger (Burger,
tational fluid dynamics (CFD) techniques is one of 2007) to analyze propeller performance by using
the most important topics in naval architecture for the vortex lattice model based on GA. Gaafary et
computing ship performance in different operating al. (Gaafary, El-Kilani & Moustafa, 2011) present-
conditions. Among different CFD methods, the thin- ed a design optimization technique for B-series
ship theory of Michell is a simple and fast approach marine propellers to optimize their hydrodynamic
and the boundary element method (BEM) is a more performance (objective functions) at a single speed.
accurate but more time-consuming method than the In another work, a multi-objective optimization
former. These two potential-based hydrodynamic program was proposed to maximize efficiency and
solvers can respectively be applied in the concep- thrust coefficient by applying polynomial expres-
tual and preliminary design phases of ship system. sions and NSGA-II (Xie, 2011). A multi-objective
Simultaneously, optimization of a ship’s hull and Particle Swarm Optimization was developed to
propeller by using the two different fidelity solvers is maximize the efficiency and minimize the cavitation

Zeszyty Naukowe Akademii Morskiej w Szczecinie 53 (125) 49

 Hassan Zakerdoost, Hassan Ghassemi

of marine propellers by using analytical and poly- major parts: first, a numerical technique used as an
nomial expressions (Mirjalili, Lewis & Mirjalili, analysis tool to indicate the values of the hydrody-
2015). NSGA-II was applied to a propeller opti- namic performances; second, a geometry modeling
mization problem by utilizing the BEM/FEM tech- approach to provide a connection between design
nique (Jiang et al., 2018). A hull-propeller system variables and body shape(s); and third, an algorithm
optimization was performed to minimize lifetime to solve the nonlinear optimization problem com-
fuel consumption by using blade element theory, posed of the objective and constraint functions.
Michell’s theory and the NSGA-II algorithm (Ghas-
semi & Zakerdoost, 2017). A mathematical model Numerical solver
of the external forces operating on a vessel and an
algorithm to solve the problem for calculating the The total calm-water resistance of a container
instantaneous speed of the vessel in selected weath- ship and the open water efficiency of the propeller
er conditions were proposed by Szelangiewicz et working behind it are our main hydrodynamic per-
al. (Szelangiewicz, Wiśniewski & Żelazny, 2014). formance parameters in this study. Keller’s inequal-
Moreover, a parametric model of a ship’s propulsion ity equation is one of the best known cavitation
system (screw propeller–propulsion engine) as well criteria for conventional marine propellers. This cri-
as a method, based on both the resistance and pro- terion may be used to obtain the expanded blade area
pulsion system models, of calculating the mean sta- required to avoid cavitation. The numerical solvers
tistical value of a ship’s service speed under seasonal employed to evaluate the wave resistance component
weather conditions occurring on shipping lines were of ship’s hull and hydrodynamic performance coef-
presented in research by Szelangiewicz and Żelazny ficients of a ship’s propeller are integrated in a code
(Szelangiewicz & Żelazny, 2015). of variable fidelity methods. The low-fidelity solv-
The main goal of this research is to present an ers are lifting line theory and Michell’s theory and
efficient tool (HPSOP code) for optimization of ship the medium-fidelity one is potential-based boundary
hull-propeller systems in the conceptual/prelimi- element method (BEM). The mathematical relation-
nary stage of ship design. A well-known optimiza- ships and validation of the numerical methods can be
tion algorithm integrated with two different fidelity found in the authors’ publications (Ghassemi, 2009;
methods in a multi-level procedure is used to mini- Ghassemi & Kohansal, 2010; Zakerdoost, Ghassemi
mize total ship resistance and maximize open water & Ghiasi, 2013).
propeller efficiency. The remainder of this study is
organized as follows: the forthcoming section brief- Geometry representation
ly describes the different parts of the HPSOP code,
including a numerical solver, geometry representa- Bezier and B-spline curves have a considerable
tion and optimization algorithm. Next section pres- shortcoming. They are polynomial-based and can-
ents a multi-level optimization strategy. Later, we not accurately represent implicit conic shapes, such
present a case study on a container ship driven by as circles, ellipses and hyperbolas; therefore, an
a well-known propeller and the results are discussed. extension of B-splines called Non-Uniform Rational
Finally, some conclusions are drawn. B-Splines (NURBS) was introduced to overcome
these shortcomings by using fractions of the same
Analysis and optimization techniques interpolation functions. Presently in computer aided
design (CAD), NURBS is one of the most common
Figure 2 shows a general scheme of the hydrody- geometric representation techniques. Because the
namic optimization methodology. It comprises three NURBS approach inherits the benefits of B-splines,
they exhibit excellent performance in curve manip-
Geometry modelling
ulation; most CAD systems have utilized them as
technique a powerful tool for generating curves and surfaces of
complex geometries. The family of curves that can
be represented with NURBS is much wider than that
with B-Splines or Bézier curves and also includes
Optimization
Numerical solver
algorithm conics. The algorithms associated with NURBS are
easier to implement, and these algorithms, as evalu-
Figure 2. General flowchart of the basic parts of the hydro- ations of positions or derivatives, are stable and fast.
dynamic optimization algorithm In this paper NURBS are used to build up the curves

50 Scientific Journals of the Maritime University of Szczecin 53 (125)

 Hydrodynamic multidisciplinary optimization of a container ship and its propeller using comprehensive HPSOP code

and surfaces of the geometric models describing population with the best members (selected using
a ship’s hull and propeller for allowing variation of non-dominated sorting and the crowded comparison
their forms during the optimization process. operator) of the combined population, 3) Output the
first non-dominated front of the population. A brief
Optimization algorithm description of the optimization algorithm is given in
Figure 3.
Among the most well-known stochastic multi-ob-
jective optimization techniques is the Non-dominat- Multi-level optimization procedure
ed Sorting Genetic Algorithm II (NSGA-II). NSGA
II (Deb et al., 2002) was proposed to remove several In an effort to reduce the overall computational
deficiencies of the first version (NSGA) that includ- cost of performing the multi-point optimizations and
ed the high computational cost of non-dominated thereby make it more efficient, we use the multi-lev-
sorting, lack of elitism and lack of sharing parame- el optimization approach to solve a container ship
ters. Currently NSGA-II is famous for its low com- system design problem. The optimization scheme is
putational complexity, simplicity and its ability to divided into two levels according to the variable-fi-
maintain a good spread of solutions. The non-dom- delity numerical methods. A flowchart of the hydro-
inated sorting method is an important characteristic dynamic design procedure of HPSOP is detailed
of NSGA-II. A general NSGA-II procedure to be in Figure 4. The first step in carrying out the shape
implemented in a routing problem is presented in the optimization is design space definition. The main
following steps: 1) Initialize the population, 2) While variables significantly influencing the hydrodynam-
the termination criterion is not met repeat the fol- ic performance of a HPS, i.e. ship length to beam
lowing: a) Evaluate each solution in the population ratio, beam to draft ratio, draft, propeller diameter,
by computing objective function values, b) Rank number of blades and pitch ratio have been selected
the solutions in the population using non-dominat- as the design variables vector, X = [L/B, B/d, d, D, Z,
ed sorting, c) Perform selection using the crowded P/D]. A Latin Hypercube Sampling (LHS) technique
binary tournament selection operator, d) Perform is used to perform the Design of Experiments (DOE)
crossover and mutation (as in conventional GA) to to generate the individuals. This technique can dis-
generate the offspring population, e) Combine the tribute the individuals throughout the given design
parent and child populations, f) Replace the parent space evenly. In this step, evaluation of the ship’s
hull resistance and propeller performance in calm
Initial population water are carried out by using coupled Michell’s
theory and ITTC-57 correction line formula and lift-
ing line theory respectively. The operating propeller
Non-dominated sorting revolution rate at the design condition is calculated
from the intersection point of required and avail-
able propeller thrust coefficient curves (Ghassemi,
Max generation Yes Pareto & Zakerdoost, 2017). The other hydrodynamic per-
number frontier
formance parameters of the propeller are obtained
No at this point. For estimating wake fraction and the
thrust deduction factor there exist some empirical
Toumament section
formulas, from which two well-known formulas for
single screw ships were selected. We took the for-
Genetic operators mulas from the trial results of more than 150 ships
and 65 tests performed respectively by Taylor and
Schoenherr (Ghose, & Gokarn, 2004). The wake
Non-dominating
sorting fraction is a function of the block coefficient and the
thrust deduction factor is related to the wake fraction.
The objective functions, which are a linear combina-
Elitist sorting
tion of the total hull resistance (f1) and the propeller
efficiency (f2) at two operating conditions, are penal-
Replace chromosome ized by a penalty function if the design constraints
of displacement (g1) and diameter to draft ratio (g2)
Figure 3. General schematic of NSGA-II algorithm are not satisfied. The HPSOP code will proceed to

Zeszyty Naukowe Akademii Morskiej w Szczecinie 53 (125) 51

 Hassan Zakerdoost, Hassan Ghassemi

Latin First level

Hypercube optimized results
Sampling

Design variables
vector (X)
Ship Propeller
Ship Hull Analysis
Analysis
(Michell’s theory)
(lifting line theory) Ship Hull and Propeller
Analysis (BEM)

Constraints Constraints
Updating (g1 and g2) Updating (g1, g2 and g3)
design design
variables variables
Objective Functions Objective Functions
(f1 and f2) (f1 and f2)

Optimization Algorithm Optimization Algorithm

(NSGA-II) (NSGA-II)

No Termination Yes No Termination Yes Final optimized

condition condition designs

Figure 4. Comprehensive flowchart of the hydrodynamic design procedure of HPSO

the second level if one of the termination condi- Results and discussion
tions is satisfied. The output generation of the first
level is used as the input generation of the second The present multi-objective optimization prob-
one. In this level, the design space is small and we lem to find the minimum total resistance and the
need to apply a higher fidelity tool, BEM, to evalu- maximum open water efficiency was performed on
ate hull-propeller systems (HPSs). In addition to the a typical 14,000 TEU container ship, the Duisburg
two constraints used in the first level, the Keller’s Test Case (DTC), propelled by the well-known sin-
cavitation criterion (g3) is employed as a third design gle propeller DTMB P4119, as the initial ship sys-
constraint. This process is repeated until it reaches tem design. As already expressed, NSGA-II is used
the maximum generation and finally arrives at the as an optimization algorithm. The system param-
Pareto front of optimal solutions. The formulation of eters of NSGA-II are as follows: crossover rate
HPSOP is as follows: = 0.9, mutation rate = 0.05, population size = 50,
Minimize: maximum generation number = 150. The character-
istics and operating conditions of the optimization
F  X    f1  X , f 2  X  , X  R 6
T
(1) problem are presented in Table 1. The upper and
where lower bounds of the design variables are depicted in
Table 2. The Pareto-optimal front of the optimiza-
2 
Rt  2
  tion problem in each generation and also the final
f1  X      ; f 2  X     0  (2)
i 1  RT0 i i 1   i Pareto front obtained in 150 generations are repre-
sented in Figure 5. The figure shows that NSGA-
Subject to: II can promote the spreading of individuals along
the Pareto front. Thus, it can be concluded that the
  0
g1  X    0.05 diversity of the algorithm, which is one of the main
0 factors in multi-objective evolutionary algorithms, is
D appropriate. The horizontal and vertical axes show
g 2  X   0.55   0.75 (3) the objective functions f1 and f2 respectively.
d
1.3  0.3Z T  K  EAR Table 3 indicates the main characteristics of the
g3  X   initial design against five optimal designs which have
PO  PV D 2 been obtained by employing the NSGA-II algorithm.

52 Scientific Journals of the Maritime University of Szczecin 53 (125)

 Hydrodynamic multidisciplinary optimization of a container ship and its propeller using comprehensive HPSOP code

Table 1. Characteristics of HPSOP Among the main dimensions of the optimized

hulls, the breadth and draft values have decreased in
Parameters Value comparison to those of the initial hull. The hull length
Hull type DTC and block coefficient have increased. As can be seen,
Propeller type P4119 the variations of the optimized hull dimensions are
1st operating condition Fn1 = 0.22 such that the total hull resistance reduces while the
2nd operating condition Fn2 = 0.27 ship’s displacement remains unchanged. The table
1st weight coefficient w1 = 0.65 also illustrates that the number of propeller blades
2nd weight coefficient w2 = 0.35 and the propeller diameter of the optimal designs
are larger than those of the initial one but the pitch
ratio has not changed significantly. As was expected,
Table 2. Limits of design variables vector these changes in hull and propeller geometries have
improved the values of the objective functions. Fig-
Parameters Case: DTC-P4119
ures 6 and 7 show the wave-making resistance and
Design variable Lower limit Upper limit
Number of blades 3 7 ×104
2.5
Pitch ratio 0.95 1.05
Propeller Diameter[m] 8.5 9.5 Ini.
Op.1
Draft[m] 14 14.5 2 Op.2
Breadth to Draft ratio 3.0 4.5 Op.3
Length to Breadth ratio 6.5 8.5 Op.4
1.5 Op.5
Rw [kN]

0.955
1

0.945
OP.2 0.5

0.935
0
f2

0.1 0.15 0.2 0.25 0.3 0.35

Fn
0.925
Figure 6. Wave-making resistance of the optimized hulls for
DTC-P4119 HPSOP
OP.4
0.915
OP.1 ×104
OP.3 2.5
OP.5
0.905
0.65 0.7 0.75 0.8 0.85 0.9
f1 2

Figure 5. Evolution of the Pareto fronts during DTC-P4119

HPSOP
1.5
Rt [kN]

Table 3. Characteristics of the initial and optimal designs for

DTC-P4119 HPSOP 1 Ini.
Op.1
L [m] B [m] d [m] CB Z D [m] P/D f1 f2 Op.2
Op.3
IS 355.00 51.04 14.50 0.661 3 9.00 1.00 1.00 1.00 0.5
Op.4
OP1 362.43 43.35 14.03 0.751 4 9.45 0.95 0.77 0.910 Op.5
OP2 362.46 44.15 14.15 0.733 5 9.37 1.02 0.67 0.933
0
OP3 370.45 43.58 14.15 0.728 5 9.45 0.97 0.68 0.922 0.1 0.15 0.2 0.25 0.3 0.35
Fn
OP4 364.70 44.10 14.18 0.745 4 9.41 0.98 0.74 0.915
Figure 7. Total resistance of the optimized hulls for DTC-
OP5 366.23 43.55 14.09 0.742 5 9.38 1.05 0.70 0.916
-P4119 HPSOP

Zeszyty Naukowe Akademii Morskiej w Szczecinie 53 (125) 53

 Hassan Zakerdoost, Hassan Ghassemi

total resistance of the initial and optimized solutions It is worth noting that all the optimized HPSs on
versus Froude number. As can be seen from these the Pareto front (as shown in Figure 3) are candidates
figures, the wave-making resistance and total resis- for the designers’ final choice. Selection of the final
tance of the optimized solutions are significantly HPS is based on the designers’ conditions. In this
lower than those of the initial one at all values of paper, from the final optimal results, we choose one
Froude number, including the operating speeds. The HPS, called compromise solution, by using a deci-
remarkable increase in the optimal length to breadth sion-making technique. In this technique, the objec-
ratio confirms the decrease in the hull wave-making tive functions are normalized and then the solution
and totals resistances shown in Figures 6 and 7. which has the minimum distance to the utopia point
Figure 8 compares the hydrodynamic perfor- is selected as the best optimal solution (Ghassemi, &
mance of the initial and optimized propellers for Zakerdoost, 2017). Based on this technique, the indi-
DTC-P4119 HPSOP. The open water efficiency of all vidual OP3 is the final optimal or compromise solu-
the optimized propellers is relatively larger than that tion. If diagrams of the required and available thrust
of the initial one at most advance coefficients which coefficients are drawn at the two design Froude num-
means the optimized propellers produce a higher bers we get two intersection points for each of the
thrust value than the initial one for a given torque. initial and compromise solutions. These two inter-
The changes in propeller diameter, number of blades section points, J1 and J2, and the hydrodynamic per-
and pitch ratio lead to these results. In other words, formance of the hulls and propellers at these points
the increase in the diameter has a relatively stronger are reported in Table 4. As can be seen from the table,
influence on the hydrodynamic performance of the the advance coefficients of the compromise solution
propeller than a decrease in pitch ratio and increase have been increased compared to those of the initial
in number of blades. one which indicates getting closer to the location of

×104
KTIni 3
0.8
10KQini
ηini
0.7 2.5
KTOP1
Hydrodynamic performance

10KQOP1
0.6 Fn = 0.22 Fn = 0.27
ηOP1
2
KTOP2
0.5
Rt & T [kN]

10KQOP2
ηOP2 1.5
0.4
KTOP3
10KQOP3 RtIS
0.3
ηOP3 1
KTOP4 TIS
0.2
10KQOP4 RtCS
ηOP4 0.5
0.1 TCS
KTOP5
10KQOP5
0 0
0.3 0.5 0.7 0.9 1.1 ηOP5 0.15 0.2 0.25 0.3 0.35
J Fn
Figure 8. Open water performance of initial and optimized Figure 9. Hull resistance and propeller thrust of initial and
propellers for DTC-P4119 HPSOP compromise solutions

Table 4. Characteristics of initial and compromise designs for DTC-P4119 HPSOP

Initial design Compromise design

Fn1 0.22 Fn2 0.27 Fn1 0.22 Fn2 0.27
Rt1[N] 5764439 Rt2[N] 13273775 Rt1[N] 3940328 Rt2[N] 9218147
J1 0.648 J2 0.558 J1 0.774 J2 0.684
N1 [RPM] 98.69682 N2 [RPM] 140.6647 N1 [RPM] 76.27231 N2 [RPM] 105.9236
Kt1 0.232305 Kt2 0.262911 Kt1 0.230096 Kt2 0.275562
10Kq1 0.39259 10Kq2 0.42511 10Kq1 0.411433 10Kq2 0.468943
Eta1 0.610257 Eta2 0.549242 Eta1 0.688924 Eta2 0.639698

54 Scientific Journals of the Maritime University of Szczecin 53 (125)

 Hydrodynamic multidisciplinary optimization of a container ship and its propeller using comprehensive HPSOP code

maximum efficiency. Figure 9 demonstrates the total of the High Performance Computing Research Cen-
resistance and thrust of the initial and compromise ter (HPCRC) at Amirkabir University of Technolo-
solutions and also confirms that the resistance has gy (AUT), their support is gratefully acknowledged.
been balanced by the propeller thrust at a wide range The authors received no direct funding for this
of Froude numbers, especially in design conditions, research.
Fn = 0.22 and 0.27.
References
Conclusions
1. Benini, E. (2003) Multiobjective design optimization of
B-screw series propellers using evolutionary algorithms.
This paper concerns multidisciplinary optimiza- Marine Technology 40(4), pp. 229–238.
tion of a typical 14,000 TEU container ship, DTC, 2. Burger, C. (2007) Propeller performance analysis and
driven by the single propeller DTMB P4119, by multidisciplinary optimization using a genetic algorithm.
using two different numerical methods with variable Dissertation, Auburn University, Alabama.
fidelity. This methodology is used to integrate the 3. Day, A.H. & Doctors, L.J. (1997) Resistance optimization
of displacement vessels on the basis of principal parameters.
conceptual and preliminary stages of ship design and Journal of ship research 41(4), pp. 249–259.
optimizes HPSs automatically in one stage. 4. Deb, K., Pratap, A., Agarwal, S. & Meyarivan, T. (2002)
The obtained results demonstrated the effec- A fast and elitist multiobjective genetic algorithm: NSGA-II.
tiveness and capability of the NSGA-II algorithm IEEE Transactions on Evolutionary Computation 6(2), pp.
182–197.
for finding the optimal solutions which were uni- 5. Dejhalla, R., Mrša, Z. & Vuković, S. (2002) A genetic
formly distributed over the Pareto front. Regard- algorithm approach to the problem of minimum ship wave
ing the main dimensions of the optimized hulls, resistance. Marine Technology 39(3), pp. 187–195.
the length was increased and the breadth and draft 6. Gaafary, M., El-Kilani, H. & Moustafa, M. (2011) Op-
were decreased which led to a significant reduc- timum Design of B-Series Marine Propellers. Alexandria
Engineering Journal 50(1), pp. 13–18.
tion in wave-making and, hence, resistance com- 7. Ghassemi, H. (2009) The effect of wake flow and skew angle
pared to those of the initial hull. The increase in on the ship propeller performance. Scientia Iranica 16(2),
propeller diameter has a relatively stronger influ- pp. 149–158.
ence on the hydrodynamic performance than an 8. Ghassemi, H. & Kohansal, A. (2010) Hydrodynamic anal-
ysis of non-planing and planing hulls by BEM. Scientia
increase in number of blades or decrease in pitch Iranica. Transaction B: Mechanical Engineering 17(1), pp.
ratio which usually have negative effect on propel- 41–52.
ler performance. Shifting the advance coefficients 9. Ghassemi, H. & Zakerdoost, H. (2017) Ship hull–propeller
to the location of maximum efficiency confirms system optimization based on the multi-objective evolution-
an increase in the efficiency of the optimized pro- ary algorithm. Proceedings of the Institution of Mechanical
Engineers, Part C: Journal of Mechanical Engineering Sci-
pellers. The comparison between the initial and ence 231 (1), pp. 175–192.
compromise solutions showed an improvement in 10. Ghose, J. & Gokarn, R. (2004) Basic ship propulsion. New
the total hull resistance and propeller efficiency. Delhi: Allied Publishers.
The propeller thrust of the initial and compromise 11. Grigoropoulos, G.J. & Chalkias, D.S. (2010) Hull-form
optimization in calm and rough water. Computer-Aided De-
solutions satisfied the total hull resistances across sign 42 (11), pp. 977–984.
a wide range of Froude numbers. 12. Jeong, S. & Kim, H. (2013) Development of an efficient hull
It is important to note that shortcomings are form design exploration framework. Mathematical Prob-
attributed to the inability of the solvers to capture lems in Engineering, Article ID 838354, 12 pages.
the physics of the problem, especially 3D viscous 13. Jiang, J., Cai, H., Ma, C., Qian, Z., Chen, K. & Wu, P.
(2018) A ship propeller design methodology of multi-ob-
flows and discretization of the bodies and free sur- jective optimization considering fluid–structure interaction.
face near the parts with severe curvature changes. Engineering Applications of Computational Fluid Mechan-
All obtained results lead to a conclusion that the ics 12 (1), pp. 1–13.
optimization strategy developed in the present 14. Kim, H.J., Choi, J.E. & Chun, H.H. (2016) Hull-form op-
timization using parametric modification functions and par-
study is efficient and worthy of further investi- ticle swarm optimization. Journal of Marine Science and
gation and can aid practical ship design at early Technology 21(1), pp. 129–144.
design stages. 15. Mirjalili, S., Lewis, A. & Mirjalili, S.A.M. (2015)
Multi-objective optimisation of marine propellers. Procedia
Acknowledgments Computer Science 51, pp. 2247–2256.
16. Park, J.H., Choi, J.E. & Chun, H.H. (2015) Hull-form
optimization of KSUEZMAX to enhance resistance per-
The numerical computations presented in this formance. International Journal of Naval Architecture and
paper have been performed on the parallel machines Ocean Engineering 7(1), pp. 100–114.

Zeszyty Naukowe Akademii Morskiej w Szczecinie 53 (125) 55

 Hassan Zakerdoost, Hassan Ghassemi

17. Pluciński, M.M., Young, Y.L. & Liu, Z. (2007) Optimi- 20. Wikipedia (2018) List of largest container ships. [Online]
zation of a self-twisting composite marine propeller using Available from: https://en.wikipedia.org/wiki/List_of_larg-
genetic algorithms. 16th International Conference on Com- est_container_ships [Accessed: February 10, 2018]
posite Materials ICCM-16, Kyoto, Japan. 21. Xie, G. (2011) Optimal preliminary propeller design based
18. Szelangiewicz, T., Wiśniewski, B. & Żelazny, K. (2014) on multi-objective optimization approach. Procedia Engi-
Forecasting operating speed of the ship in the selected neering 16, pp. 278–283.
weather conditions. Scientific Journals of the Maritime 22. Zakerdoost, H., Ghassemi, H. & Ghiasi, M. (2013) Ship
University of Szczecin, Zeszyty Naukowe Akademii Morskiej hull form optimization by evolutionary algorithm in order to
w Szczecinie 38 (110), pp. 89–95. diminish the drag. Journal of Marine Science and Applica-
19. Szelangiewicz, T. & Żelazny, K. (2015) An approximate tion 12 (2), pp. 170–179.
method for calculating the mean statistical service speed of 23. Zhang, B.-j., Ma, K. & Ji, Z.-s. (2009) The optimization
container ships on a given shipping line and its application of the hull form with the minimum wave making resistance
in preliminary design. Scientific Journals of the Maritime based on Rankine source method. Journal of Hydrodynam-
University of Szczecin, Zeszyty Naukowe Akademii Morskiej ics, Ser. B 21 (2), pp. 277–284.
w Szczecinie 44 (116), pp. 34–42.

56 Scientific Journals of the Maritime University of Szczecin 53 (125)