An Optimization Study for the Bow Form of High Speed

Displacement Catamarans

D. Bülent Danışman1, Ömer Gören1 (M), Mustafa Insel1 (AM), Mehmet Atlar2 (M)

Abstract

Interest in high-speed marine transportation has acquired a wide variety of hydrodynamic research
activities on catamarans, including shape optimization for minimum total resistance. This study attempts to utilize
the mathematical programming in optimizing the bow form of twin hulls for minimum total resistance as well as in
analyzing the optimization by physical model tests. The total resistance is assumed to be composed of wave-making
and frictional components, which are formulated using Michell’s thin-ship theory and ITTC-1957 friction line,
respectively. The optimized hull form is analyzed by means of a computational flow solver before going through the
experimental analysis. The study demonstrates the capabilities of the optimization procedure, presented for
catamarans, in reducing the total resistance, as well as its limitations to be used as a design tool in a relatively high-
speed zone.

1. INTRODUCTION resistance in formulating the objective function and then

Recent upsurge in their commercial applications
indicates that high-speed craft are gradually becoming
ships of interest for ship owners, engineers and
researchers as the importance of speed increases in the
seaborn trade. It appears that catamarans, and probably
trimarans in the near future, can satisfy the requirements
of marine transportation especially those of passenger
ferry market as the speed and safety are the main
concerns. Therefore, a better understanding of relatively
complex resistance characteristics of twin-hulls is of
crucial importance on the one hand, developing an
optimization procedure of these hull forms is an
indispensable goal on the other.
An investigation into the components of resistance
of catamarans has been addressed by Insel and Molland
[1]. Meanwhile a number of computational
investigations of the wave resistance of multihulls have
been reported recently in the literature, such as [2], [3],
[4]. A contribution to the problem of optimizing the
forebody geometries of catamarans to minimize the
resistance was first made by Hsiung and Xu[5],
followed recently by Doctors and Renilson [6], who
studied the influence of demihull separation and river
banks and by Papanikolaou et al. [7], who handled the
shape hull optimization problem as a component in a
global computer aided optimization procedure. Hsiung
and Xu[5] uses Lunde’s [8] formulation for the wave
resistance and ITTC-1957 formula for the frictional
solve the convex quadratic programming problem with
linear inequality constraints to obtain optimal forebody
shapes for catamarans. In their study the Froude
numbers chosen for optimization are not higher than
0.312 and the experimental data are given only to
compare the effects of demihull separation. In the work
of Papanikolaou et al. [7], the total optimization is
treated by the method of Lagrange’s multipliers where
the design constraints must be linear equalities. An
extension of the Michell integral is also made in [7] by a
normal dipole distribution on the centerplane to include
asymmetric demihull forms into the wave resistance
analysis of catamarans. Meanwhile the efforts to
minimize the wave resistance of catamarans lead, on the
other hand, to some unconventional solutions of
cambered catamarans by Chen [9] and of staggered
catamarans by Söding [10].
In the current work bow form optimization of high-
speed displacement-catamarans is tackled by extending
the previous work of Gören et al. [11]. The basic
purpose of this study is to test and to ascertain the
capabilities of the optimization procedure based on
convex quadratic programming problem as a result of
the use of thin ship theory, supported by a
computational wave resistance analyzer, through a
series of experimental analysis. The total resistance is
decomposed into skin friction resistance, calculated by
ITTC-1957 formula, and wave resistance based on
Lunde’s [8] formulation. The expression of total

1
Istanbul Technical University, Dept. of Naval Architecture, Istanbul, Turkey.
2
University of Newcastle, Dept. of Marine Technology, Newcastle upon Tyne, U.K.
resistance is then taken as the objective function, which I ×J I ×J
is quadratic in ship’s half breadths. The linear inequality Cw = d mn y m y n = y T D y ( 2)
constraints used in the quadratic programming problem m =1 n =1
allow to implement the necessary design constraints into
the procedure. The standard quadratic programming where I×J denotes the total number of half-breadths
problem is then solved by Wolfe’s [12] algorithm. since I is the number of waterlines while J is the number
Although it is a well-known fact that in the high speed of stations. The coefficient matrix D is derived adopting
zone, the percentage of the wave pattern resistance in the same approach in Hsiung [14]. The quantity
total resistance does not generally exceed 15%, the calculated by equation (2) may not be considered as a
present study points out that by means of the good approximation of the wave resistance for high
optimization procedure it is possible to attain up to 5% speed vessels with transom stern. However this value,
gains in total resistance which may be regarded as which is of the same order as wave resistance, can
considerable for catamarans operating at high Froude relatively be minimized within the linear theory
numbers greater than 0.50. formulation.
The other component of the resistance that appears
in the objective function is the equivalent flat-plate
2. MATHEMATICAL PROGRAMMING frictional resistance, which depends on the wetted
PROBLEM FOR TWIN HULLS surface area approximated as
é 1 2 1 2 ù
The previous works of Gören and Calisal [13] and S =2 ê1 + 2 Fx ( x, z ) + 2 Fz ( x, z ) dxdz ( 3)
Gören et al. [11] is extended to cover twin hull ë
S0
geometries. Lunde’s [8] well-known formulation is used
to express twin hull wave resistance based on thin-ship Where, S0 is the projection area of the wetted surface of
theory; the ship on the center-plane. Fz is the derivative with
∞ respect to z. The use of tent functions is again made in
ρg (u 2 + 1) 2 (3) to give the frictional resistance coefficient, which is
RW = 4 2 1
(P 2 + Q 2 )
πc 2
(u + 2 ) 2
(1.a) based on the ITTC-1957 formula, in a quadratic form in
0
terms of the ship half breadths as in the following form:
ì é g 1 ùü
2 2
í 2 + cos ê2b 2 (u + 1)u (u + 2) 2 ú ýdu C F = c0 + c0 y + y T A y (4)
î ë c û F
In accordance with the assumption made for the total
Pü
T L
cos é g resistance, the objective function used in the
ù
ý= Fx ( x , z ) x(u 2 + 1) optimization procedure is taken as the sum of wave
Q sin êë c 2 resistance (2) and the frictional resistance (4):
0 0

ég ù CT = C F + CW = c0 + cb y + y T C y (5)
exp ê 2 ( z − T )(u 2 + 1) 2 dxdz (1.b)
ëc Note that C is a positive semi-definite, that is a
where c is the ship’s speed, ρ is the density of water and symmetric matrix.
g is the gravitational acceleration. L is the length of the The quadratic character of the objective
ship, T is the draft and Fx denotes the derivative of the function implies the employment of the quadratic
hull surface with respect to x. The coordinate axis used programming in which the design constraints can be
in the formulation is depicted in Figure 1. given by a set of inequalities. Thus the general form of
the quadratic programming is expressed as
to minimize p y + yT C y
( 6)
to satisfy A y ≤ B with y i ≥ 0

The system of equations (6) can be linearized and

solved by means of Wolfe’s [12] algorithm (see [11]
and [13] for details).

3. NUMERICAL STUDY FOR OPTIMIZATION

Before going through the optimization process for

Figure 1. The Coordinate axis and the arrangement of
the demihulls
One can discretize the underwater geometry by tent
functions (see Hsiung [14] for details). Thus (1.a) and
(1.b) is then nondimensionalized and given in terms of
the half breadths, yi, by the use of tent functions;
twin hulls, equation (1.a) needs a special numerical
treatment, since the integrand includes a highly
oscillatory term. This is overcome by a novel numerical
technique proposed by Sidi [15]. Any desired accuracy
can be attained by increasing the number of recursive
steps in Sidi’s algorithm. The intermediate integrations
between the roots of the cosine term are done by
Gaussian quadratures.
Since the primary goal of the present study is to
optimize the bow form of the demihulls, no effort has
been spent to determine the optimum hull spacing, s=2b.
However, the hydrodynamic interaction between the
two demihull is taken into account in the wave
resistance formulation. As stated in the introduction
section, this kind of an investigation can be found in [5]
for moderate Froude numbers. Therefore 3 different
“hull spacing to length” ratios, which provide a practical
range of importance for catamarans in service, s/L=0.2,
0.3 and 0.4, are chosen. Optimization efforts have been
focused on the spacing ratio, s/L=0.3. Figure 2a . Cross sections of the original demihull.
An NPL based form, Bailey [16], is selected as the form
to be optimized. The main characteristics of this NPL
form are given in Table I.

Table I. Main characteristics of the NPL form

Lwl 25.25 m
Bmax 2.6 m
T 1.4 m
CB 0.410

The forebody volume of the original NPL form,

which undergoes the optimization procedure, was
chosen as the volume between bow profile and the 18th
station. The half-breadths on the stations 20 (F.P.), 191/2,
19 and 18, each being described by 6 waterlines were
the unknowns in the optimization study. The resistance
coefficient matrices in (5) were computed only for the
design waterline at T=1.4 m and at the optimization
speed of 20 knots which corresponds Fn=0.65.
Among the set of design constraints, protruding
length of the bulb and the forebody volume were
employed as the variable design constraints, while the
other constraints such as a limiting value for waterline
slopes, waterplane area coefficient were kept constant.
Two protruding lengths, namely, lp=1.0 m and lp=1.5 m,
were taken into consideration while forebody block
coefficient CBf was increased gradually from an
acceptable minimum for every lp. As a consequence of
the solution of the system (6) for every combination of
the set of design constraints, a matrix of optimum hull
forms were obtained. Eventually a corresponding matrix
of numerical evaluation of the total resistance by means
of equation (5) was calculated. Ultimately the optimized
form having the best resistance performance was
selected from the matrix. In Figure 2a and 2b the cross-
sectional curves of the original and optimal hull forms
(lines faired) can be seen. This form proposed for tank
testing, has a protruding bulbous bow with a protruding
length of lp=1.5 m and the total volume was increased
by 5% as compared to the original NPL form.
Figure 2b. Cross sections of the optimized demihull. Aft
body remains unchanged.
Before the model experiments were conducted, more
accurate and reliable numerical tests and comparisons
were done by a computational wave resistance flow
solver reported in Goren and Atlar [17] based on
Dawson’s [18] algorithm. A sample of geometric
modeling of the optimized demihull form as it was used
in the flow solver is given in Figure 3. It is found from
the computed results that 12% of gain is expected at
least in the wave resistance at Fn=0.65 due to the
optimal form, with s/L=0.3. The same optimized form
shows 27% reduction in wave resistance at Fn=0.4.
The optimal demihull and the original NPL hull
were also compared with respect to their wave
deformations along the outer side of the demihull in
Figure 4. In Figures 5.a and 5.b computed contour plots
of the wave pattern of the original and optimal
demihulls are shown, respectively. The improvement in
the wave resistance is obvious due to relative reduction
in the bow wave elevation of the optimal form, both
quantitatively and qualitatively, as shown in Figures 4
and 5.
 20
18
Level 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Deformation (m): -0.60275 -0.5278 -0.45285 -0.3779 -0.30295 -0.228 -0.15305 -0.0781 0.0718 0.14675 0.2217 0.29665 0.3716 0.44655
16
14

10
6

7
6
7 5

11
9
8
12
6 5

10

12
7
10

8
7 5
9 9

11
8
9 6
8 10 5 6 10

12
8 7

7
9

11
8

9
6 6 5
10 11 9 5
6

7
11 10
11

13
4 7

12
8

13
5

6
4
12

14
1 2 13

13 4
9
76

1
2

1
10 8

10
9
11 11 10 6
12

11
4

8
-20 -10 0 10 20 30
X
Figure 3. Geometric modeling of optimized demihull and its free-surface vicinity.
Figure 5a. Computed wave pattern around the original demihull (Fn=0.65).
0.5
0.4
0.3
0.2
Wave Deformations (m)

0.1
20
NPL
0 18
-30 -20 -10 0 10 20 30 40 Optimal Level 1 2 3 4 5 6 7 8 9 10 11 12 13 14
F.P. A.P. 16 Deformation (m): -0.563 -0.492 -0.421 -0.350 -0.279 -0.208 -0.137 -0.066 0.076 0.147 0.218 0.289 0.360 0.431
-0.1 7 6

11
10
14

9
8 7 6

11
6

8
12
-0.2
8 12

10
7

Y
6

9
10 5

11
6

8
7
9
-0.3 8

6
8 9

12
7
5

10
9
5 6 8

11
9 7
6 8 5 9 13
10

12
10 9 7 5

10
-0.4 9 24

12 12 13
4 6

10
11

7
13

14
11

9
8 7
2 111
0 8 8 21 1 3 11 2 3

8
6
9 2 1 1

13
5 1

10
4
-0.5 9 10 4

14
9
3

6
7

5
Distance From Mid-ship (m) -20 -10 0 10 20 30
X
Figure 4. Computed wave elevations along the outer side of the demihulls Figure 5b. Computed wave pattern around the optimal demihull (Fn=0.65).
(Fn=0.65).
 Monohull (Demihull in isolation) Twin Hulls, s/L=0.2
0.010
0.013
CT, optimized form (Exp.)
0.009 0.012 CT, optimized form (Exp.)
CT, original form (Exp.)
0.011 CT, original form (Exp.)
0.008 0.010
CT

0.009

CT
0.007
0.008
0.006 0.007
0.006
0.005 0.005
0.30 0.40 0.50 0.60 0.70 0.80 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Fn Fn
Twin Hulls, s/L=0.3 Twin Hulls, s/L=0.4
0.013 0.012
0.012 CT, optimized form (Exp.)
0.011
CT, optimized form (Exp.)
0.011 CT, original form (Exp.) CT, original form (Exp.)
0.010
0.010
0.009
0.009
CT

CT
0.008 0.008
0.007 0.007
0.006 0.006
0.005 0.005
0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Fn Fn
Figure 7. Comparison of experimental total resistance coefficients for original and optimized models.
 Mono hull (Demihull in isolation) Twin Hulls, s/L=0.2
0.09 0.12
Original form (Exp.)
0.08 Original form (Exp.)
Optimized form (Exp.)
0.07 0.10 Optimized form (Exp.)
CWP Distribution

0.06 CWP Distribution

0.08
0.05
0.04 0.06
0.03
0.04
0.02
0.01 0.02
0
0 20 40 60 80 0.00
Θ=(Wave angle) 0 20 40 60 80 100
Θ=(Wave angle)
Twin Hulls, s/L=0.3
0.16 Twin Hulls, s/L=0.4
0.14 Original form (Exp.) 0.12
0.12 Optimized form (Exp.)
CWP Distribution

Original form (Exp.)

0.1
0.10 Optimized form (Exp.)

CWP Distribution
0.08
0.08
0.06 0.06
0.04 0.04
0.02 0.02
0.00 0
0 10 20 30 40 50 60 70 80 0 20 40 60 80 100
Θ=(Wave angle) Θ=(Wave angle)
Figure 8. Measured wave spectrums for original and optimized forms, (Fn=0.65).
4. EXPERIMENTAL WORK
In parallel to the numerical investigation, the
physical model tests were carried out to investigate the
resistance characteristics of the original NPL form and
the optimized hull form in Ata Nutku Ship Model
Testing Laboratory (160m x 6.0m x 3.40m) of I.T.U.,
which is an ITTC Class Facility. The experiments were
performed for two sets of measurements, namely, total
resistances and wave pattern resistances.
Associated models were built with a length of 2.5
meters and tripwires were used for the turbulence
stimulation, as shown in Figure 6. The test cases
covered were the mono hull (i.e. one demihull in
isolation) and twin hulls having s/L ratios of 0.2, 0.3
and 0.4, [19].
Figure 6. Bow form view of the two models: Original
NPL form and its optimized counter part (at the back).
The measurements in total resistance tests were
acquired by using electronic dynamometer and by
means of computer aided data acquisition system, for a
range of Froude numbers of 0.1≤Fn≤0.7. Models were
tested free to sink and to trim. The results of the total
resistance tests are presented in Figure 7. As shown in
this figure, up to 24% of reductions can be observed in
total resistance, CT, for the mono-hull configuration
beside the fluctuations around the low speed range. The
reduction in total resistance attains its maximum value
around Fn=0.5. For the twin-hull configurations,
reductions in wave resistance are observed for s/L=0.2
as 10%, s/L=0.3 as 11% and for s/L=0.4 as 12% at the
speed which corresponds to Fn=0.65. Note that
optimization was performed for Froude number 0.65
and for the gap ratio s/L=0.3. In order to make a direct
comparison of wave resistance with that of the
numerical analysis, wave patterns generated by the twin
hulls were also measured by 4 wave probes located at
one side of the tank so as to give the longitudinal cut
data. The wave cut analysis depends on the wave form
expression:
N 0.007
ζ = [ ] æ 2πyn ö
ξ n cos(ω n x ) + ηn sin (ω n x ) cosç
è W
(7)
n=0
where W is the width of the tank, ωn = K n cos(θn ) ,
2πn g 0.002
K n sin (θn ) = and K n − 2 sec2 (θn ) = 0.
W V
Here n, Kn and θn denote the nth harmonic, wave number
and wave angle, respectively. Thus, wave pattern
resistance is given by;
ρgW é 2
( ) ( )æçç1 − cos 2(θ ) ö÷ù (8)
N 2
2 2 2 n
RWP = ê ξ 0 +η 0 + ξ n +η n ÷
2 ëê n =1 è ø
In this case, resistance components are expressed
as: RT = RV + RWP = (1 + k ) RF + RWP , (1+k) is being the
form factor. The results of the wave pattern resistance
are shown in Figure 8 in terms of wave spectra only for
the case of s/L=0.3. The optimal form reduced the wave
pattern for the wave angles up to 60°. It is also effective
for all tested demihull separations. In Figure 9, it is
possible to compare the numerical and experimental
results for the gap ratio s/L=0.3. It should be remarked
here that there is 5% excess of wetted surface area in
optimized form as compared to that of the original form.
The experimental and numerical analyses give the same
order of reductions in wave pattern resistance, where as
a shift is observed at crest and troughs of the resistance
curves. This may be attributed to the inherent phase
shifts in Dawson’s algorithm, which appear
considerably when there is interference. It is interesting
to note here that the present optimization is more
successful in reducing divergent waves rather than
reducing the interference effects in high-speed zone.
This can be seen in Figure 10. Ultimately, the effective
power requirements of the original form and the
optimized form for s/L=0.3 are given in Figure 11 for
comparison. It is clear that there is at least 4% reduction
in required effective power at Fn=0.65
5. CONCLUSION
An attempt is made to optimize the bow forms of
high-speed displacement catamarans. The optimization
procedure presented here is able to produce reasonable
protruding bulbs suitable for catamarans. The numerical
and experimental studies show that the optimized bow
form is effective in reducing the wave resistance
successfully. The results show that at least 4% of
reduction is attainable in total resistance which may be
regarded as considerable for catamarans operating at
high Froude numbers greater than 0.50.
On the other hand experimental work points out
that the resistance reductions are attained mostly due to
the reductions in divergent wave patterns. Although it is
known that asymmetric forms may be helpful in
reducing interference resistance, a further study is
required to search for obtaining reductions in the
interference resistance as well.
Twin Hulls, s/L=0.3
0.009
0.008 CW, optimal (Exp.)
CW, original (Exp.)
0.006
0.005
CW,optimal (Comp.)
CW, original (comp.)
CW
0.004
0.003
0.001
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Fn
Figure 9. Comparative wave pattern resistance
coefficient
 [6] DOCTORS, L. J. and RENILSON, M. R.,
1.80 “The Influence of Demihull Separation and
Optimized form (Exp.), s/L=0.3

1.60
Original form (Exp.) s/L=0.3 River Banks on The Resistance of a
Catamaran”, Proc. FAST’93, Yokohama, 1993.
RW cat/(2RWmono )

1.40 [7] PAPANIKOLAOU, A., KAKLIS, P.,

1.20
KOSNIKAS, C. and SPANOS, D.,
“Hydrodynamic Optimization of Fast
1.00 Displacement Catamarans”, Proc. 21st Symp.
on Naval Hydrodynamics., Trondheim, 1996.
0.80
[8] LUNDE, J. K., “On the Linearized Theory of
0.60 Wave Resistance for Displacement Ships in
0.3 0.4 0.5 0.6 0.7 Steady and Accelerated Motion”, Trans. of
Fn SNAME, Vol. 59, 1951.
Figure 10. Effect of interference [9] CHEN, X.-N., “Theoretical Foundations of
Wave Resistance Elimination at Supercritical
Speeds, Especially by Use of Cambered
14000
Catamarans”, 18th Duisburg Colloquium, G.
12000 Mercator Univ., Duisburg, 1997.
10000 [10] SÖDING, H., “Drastic Resistance Reductions
in Catamarans by Staggered Hulls”, Proc.
EHP(PS)

8000
6000
4000
EHP Optimized (Exp.), s/L=0.3
2000 EHP Original form (Exp.), s/L=0.3
0 Technology Research, Vol. 27, No. 3, 1997.
6.0 11.0 16.0 21.0
V(knot)
Figure 11. Effective power for full-scale twin hulls
ACKNOWLEDGEMENT
Funding for this project was partially provided by
Istanbul Technical University (ITU) R&D Funds. A
joint support of Scientific and Technological Research
Council of Turkey (TUBITAK) and the Royal Society,
given to the second author for his visit to the University
of Newcastle, is gratefully acknowledged.
REFERENCES
[1] INSEL, M. and MOLLAND, A. F., “An
Investigation into the Resistance Components
of High Speed Displacement Catamarans”,
Trans. RINA, April 1991.
[2] BRUZZONE, D., CASELLA, P., PENSA, C.,
SCAMARDELLA, A. and ZOTTI, I., “ On the
Hydrodynamic Characteristics of High Speed
Catamarans with Round-Bilge Hull”, Proc.
FAST’97, Sydney, 1997.
[3] SUZUKI, K., NAKATA, Y., IKEHATA, M.
and KAI, H., “Numerical Prediction on Wave
Making Resistance of High Speed Trimaran”,
Proc. FAST’97, Sydney, 1997.
[4] LARSSON, L., JANSON, C.-E. and BRUN,
F., “A Numerical Investigation of Trimaran
Configurations”, Proc. FAST’97, Sydney,
1997.
[5] HSIUNG, C.C. and XU, H.,” Determining
Optimal Forms of a Catamaran for Minimum
Resistance by the Mathematical Programming
Method”, Schiffstechnik, Vol. 35, No.1, 1988.
FAST’97, Sydney, 1997.
[11] GÖREN, Ö., HELVACIOĞLU, Ş. and INSEL,
M., “Bow Form Optimization of Displacement
Ships by Mathematical Programming”, Ship
[12] WOLFE, P., “The Simplex Method for
26.0
Quadratic Programming”, Econometrica, Vol.
27, No.3, 1959.
[13] GÖREN, Ö. and CALISAL, S. M., “Optimal
Hull Forms for Fishing Vessels”, 13th STAR
Symp., SNAME, Pittsburgh, 1988.
[14] HSIUNG, C.C., “Optimal Ship Form for
Minimum Wave Resistance”, J. Ship Research,
Vol.25, No.2, 1981.
[15] SIDI, A., “The Numerical Evaluation of Very
Oscillatory Infinite Integrals by Extrapolation”,
Mathematics of Computation, Vol.38, No.158,
1982.
[16] BAILEY, D., “The NPL High Speed Round
Bilge Displacement Hull Series”, Marine Tech.
Monograph, Nr.4, RINA, 1976.
[17] GÖREN, Ö. and ATLAR, M., “A
Computational Study for the Wave Resistance
Analysis of Multi-Hull Forms”, University of
Newcastle, Department of Marine Technology
Report: MT-1998-031, August 1998.
[18] DAWSON, C. W., “A Practical Computer
Method for Solving Ship-wave Problems”, 2nd
Int. Conf. Numerical Ship Hydrodynamics,
Berkeley, 1977.
[19] DANISMAN, D. B.; “Bow form Optimization
of Twin Hulls for minimum Total Resistance”,
(in Turkish), MSc Thesis, I.T.U. School of
Graduate Studies, Istanbul, 1997.