Professional Documents
Culture Documents
net/publication/282366386
CITATIONS READS
8 3,966
3 authors:
Yun Li
University of Glasgow
261 PUBLICATIONS 8,752 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Joo Hock Ang on 07 October 2015.
SUMMARY
‘Eco-friendly shipping’ and fuel efficiency are important issues and challenges faced by marine and offshore industry
today. This is primary due to increasing stringent environmental regulations and the pressure to reduce fuel cost to stay
competitive. Furthermore, the technical requirements of offshore vessels have also increased significantly. Recent
developments in advance optimisation techniques and High Performance Computing (HPC) are key enablers for
Simulation Based Design (SBD) in ship and offshore applications, where overall efficiency of design process and
hydrodynamic performance can be greatly improved. This paper integrates a geometry modification technique with
advance optimisation and evaluation algorithms, with the aim to improve the overall efficiency and hydrodynamic
performance of ‘ship-shaped’ offshore vessels. By incorporating concepts from Free-Form Deformation (FFD) and
Genetic Algorithm (GA), the proposed approach is applied to optimise the hull form design of a heavy-lift and pipe-lay
vessel to reduce the overall resistance while improving its sea-keeping performance.
unique operating requirements, both in transit and and G.F Clauss [8-11] applied hydrodynamic shape
stationary mode. Unfortunately, reducing resistance and optimisation procedure and parametric modeling to
improving sea-keeping are two seemingly conflicting improve sea-keeping qualities based on rational sea-
objectives, where improving one will worsen the keeping criteria.
performance of the other. Secondly, it was highlighted
earlier most design improvement are currently conducted With introduction of parallel computing, evolutionary
manually and hence makes it very difficult for the strategies such as Genetic algorithm (GA) proved to be
designer to investigate and obtain optimum design with very useful and became standard techniques used in
mainly conflicting objectives. Therefore, what is required many hull optimisation studies. GA was first developed
is the development of a fully automated optimisation by Holland et al. [12], which is a nature-inspired search
process that can effectively investigate and obtain the heuristic method based on Darwinian Theory of natural
optimal design in an efficient manner. Thirdly, there are selection and the ‘survival of the fittest’ principle [13].
common skepticisms within ship design firms and Y. Tahara et al [14] developed a hull form design system
shipyards on the use of optimisation and CFD procedure using GA, successive quadratic programming (SQP) and
in the design process due to lack of expert knowledge on Rankine-source panel method for minimum wave-
the concept behind the methods. At same time, ship making resistance of a container ship. Several other
design firms and shipyard uses different types of works also applied GA for ship design optimisation [15-
software for hull modeling and hydrodynamic or 17]. Particle swarm optimisation (PSO) is another
structural analysis. Familiarity and inter-portability of evolutionary strategy that is applied to solve global
computer aided design (CAD) model using different optimisation problem. It simulates the social behavior of
software for various analyses become an important swarm of birds or bees which are capable to share
consideration before new optimisation software solution information while looking for food. PSO was used by
can be adopted and applied to hull form optimisation in E.F. Campana et al. [18] in combination with filled
the design process. function based approach and Bezier patch to improve
sea-keeping performance of container ship. This
By automating an integrated geometry modification technique became increasingly popular for global
technique with an evolutionary algorithm and evaluation optimisation as applied in [19-21].
process, we demonstrate the usefulness of the proposed
framework to the hull form design optimisation of ‘ship- For any successful hull optimisation, geometry
shaped’ offshore vessel for reduced resistance and modification plays an important role to manipulate and
improved sea-keeping performance. This paper first transform the hull lines in order to investigate and
provides the background and challenges to hull form increase the chances of obtaining an ‘optimum’ hull. One
optimisation, followed by related literatures in Section 2. of the main techniques used is Free-from deformation
Section 3 will covers proposed HFDO framework and (FFD), which was first introduced by Sederberg and
the detail methodologies for optimisation, geometry Parry [22] based on scheme of trivariate Bernstein
modification and performance evaluation. Finally, a case polynomials. Y. Tahara et al [23] applied FFD and non-
study on optimising the hull form design of heavy-lift & uniform rational B-splines (NURBS) to modify the hull
pipe-lay vessel for reduced resistance and improved sea- geometry, in combination with GA and uniform covering
keeping will be carried out in section 4, followed by approach to reduce total resistance and improve sea-
conclusion. keeping merit function of a surface combatant. FFD was
also applied for geometry modification in [19, 20, 24].
2. RELATED WORKS Another geometry modification technique that is applied
in hull shape transformation is blending or morphing
As computer becomes more readily available and method. In [25], Y. Tahara et al combines hull-form
powerful in the 1970s, there was rapid development and blending and prismatic-curve control using NAPA to
application of computer aided design (CAD) and modify the hull form of tanker to improve propulsion and
computational fluid dynamics (CFD) in ship design. maneuverability performance. This geometry
During the same period, optimisation was also modification technique was also applied in [26].
introduced and first applied in ship design by
Schneekluth, Nowacki and Soding in mid-1970s [4]. In hydrodynamic evaluation, potential flow theory are
Some early examples of techniques developed for hull used widely to estimate and analyse the flow around ship
form optimisation includes systematic hull variation by hull as well as determining the wave-induced resistance
directly correlated new hull form with parametric [27]. It is based on Laplace and Bernouli equation. H.
modification of prismatic curve (Cp curve) by H. Nowacki [28] combines potential flow solver with free
Lackenby [5]. In 1980s, C.C. Hsuing [6] applied Kochin surface and Navier stokes solver in an optimisation
function to evaluate wave resistance and optimisation problem to reduce wave and total resistance of a Wigley
using Lagrange multiplier. T. Ray et al [7] carried out hull. For sea-keeping analysis, various numerical
global optimisation using multi-start method and methods are developed and they includes strip theory,
simulated annealing for ship design optimisation as early unified theory, green function method, Rankine
as 1990s. In optimisation of offshore structures, L. Birk singularity method, as highlighted in [29]. Of which,
strip theory is still the most popular technique used today To begin the optimisation, the user first defines the
for calculation of wave induced motion of ship [30]. The problem in PF module where inputs such as design
approach is based on slender body theory and originated specifications and key design parameters are specified,
from Korvin-Kroukovsky and Jacobs in late 1950 [29]. including objective function and design variables.
Further variations of strip theories were developed, with Objective function is defined here as a function that can
the most widely used as proposed by Salvesen et al. in be maximize or minimise and the number of objective
1970 [31]. Example of sea-keeping evaluation include function can varies depending if it is single or multi-
work by Hassan Bagheri and Hassan Ghassemi [13] objectives problem. Design variables are used for
where they applied strip theory and GA to minimise controlling the hull geometry during the optimisation
vertical bow motion of a Wigley hull. procedure and constraints can be applied to restrict the
design variables in order to ensure it keeps within the
3. METHODOLOGY- HULL FORM DESIGN design limits. Depending if it is a new hull design or
OPTIMISATION FRAMEWORK modification from existing vessel model, the designer
will prepare the CAD model that would be used as an
Moving away from conventional approach towards hull initial design for the optimisation.
form design optimisation (HFDO), a new framework
which integrates the merits of evolutionary computation, II) DESIGN SPACE EXPLORATION (DSE)
geometry modification and advanced performance
evaluation techniques is proposed. The main aim here is At the core of the ‘Design Space Exploration’ (DSE)
to automate the hull form design optimisation process module is a Multi-Objective Genetic Algorithm (MOGA)
using minimal user intervention. The proposed based optimisation engine. The purpose of this module is
framework must be able to effectively explore the design twofold: it serves as an optimiser to explore the design
space to produce optimal and feasible solutions that search space and (ii) automates the entire design
meets industry standards. The framework described in optimisation process.
this section thereby represents an important step towards
this endeavor, as illustrated in Figure 2. Most real-world engineering problems consist of
multiple design objectives which often have conflicting
characteristics. To search for globally non-dominated
solutions that can satisfy each design criteria is not a
trivial task. This is a common pitfall for many classical
search algorithms [32]. MOGA, on the other hand, is
well suited for this class of optimisation problems [33].
In comparison to conventional optimisation method,
MOGA has many desirable traits and offer significant
advantages to efficiently navigate multi-dimensional,
large and challenging design search space to produce
globally optimal non-dominated solutions. Key to the
workings of the optimisation algorithm is the principle of
‘genes’ and ‘chromosomes’. Similar to biological
evolution, a chromosome is made up of a discrete
number of genes. It contains (encodes) all the
information to describe a potential solution to the design
problem. For any design problem, a number of
chromosomes are first randomly created to form a
population of candidate solutions. Through the use of
Figure 2: Proposed hull form design optimisation genetic operators - selection, crossover and mutation -
(HFDO) framework information exchange takes place between these
chromosomes over a number of iterations, typically with
Figure 2 provides an illustration of the proposed HFDO the fittest solutions replacing weaker ones, eventually
framework. It is made up of five modules: (1) Problem leading to a set of optimal solutions. In this paper,
Formulation (PF), (2) Design Space Exploration (DSE), solutions are ranked using Pareto dominance [34] as it
(3) Geometry Modification (GM), (4) Performance provides a non-bias way of ranking solutions. Figure 3
Evaluation (PE) and (5) Optimal Solutions (OS). [33] provides an illustration of the MOGA approach used
Modules 1 and 5 are input and output sections in the proposed framework.
respectively, while modules 2, 3 and 4 form the core
design components. Together, these modules are central
to automate the entire hull form design optimisation
process and are described in detail below.
∑ ( )( ) [∑ ( )(
) [∑ ( )( ) ]] (1)
stability, powering, maneuvering and so forth. Due to hydrodynamic damping coefficient and the spring
space constraint, only the two design objectives which coefficient.
are crucial to the case study presented in this paper will
be discussed. V) OPTIMAL SOLUTIONS (OS)
IV (a) Ship resistance Once the design criteria are met, the optimiser will
terminate the calculation and identify the optimal
Using Computation fluid dynamic (CFD), ship resistance solutions as part of optimisation output. This includes a
can be evaluated at early design stage. CFD are range of ‘non-dominated’ hull forms known as Pareto
interpreted to be a numerical simulation of fluid flow, optimal solutions, which had been evaluated in PE
which are modelled by solving a set of field equation module and identified by the optimiser to provide
describing the dynamic of fluid flow [36]. The most potentially superior performances relative to other design
common CFD technique used for evaluating resistance solutions found. The designer can then select the hull
are linear potential code and Reynolds Averaged Navier- form which best meets the design requirement. It is
Strokes Equation (RANSE). In this study, we will only recognised that the hull form may not be fixed during
focus on potential code due to its efficiency and fairly early design stage where other design parameters such as
good estimation for early ship design. powering or payload also play an important role. These
inter-dependent parameters often require constant
As the viscosity effect are often limited to small updating and modification during the entire project
boundary layer, potential flow model are particularly duration. For this purpose, proposed HFDO framework
useful for free surface flow, such as flow around ship can also be extended by adding more design parameters
hull [37]. The fluid is assumed incompressible, inviscid or objective functions so as to achieve an overall
and irrotational flow and the total velocity potential of integrated ship design optimisation solution.
the flow can be expressed below as per [38]:
4. CASE STUDY: MULTI-OBJECTIVE
( ) ( ) ( ) (2) OPTIMISATION OF PIPE-LAY AND HEAVY-
Subject to Laplace’s equation: (3) LIFT VESSEL
Where ( ) is the perturbation potential, ( ) is In order to demonstrate the efficiency of the proposed
the time-dependent velocity of translation and gradient framework, it is applied to optimise the design of a
operator ( ) heavy-lift and pipe-lay vessel. The proposed framework
is implemented in NAPA [39], a well-known commercial
IV (b) Sea-keeping performance ship design software that is used by many design
companies and shipyards around the world. This
Sea-keeping is an important design performance criteria demonstrates the feasibility to seamlessly integrate the
for ‘ship-shaped’ offshore vessels, due to its effect on i) proposed methodology into existing commercial software
comfort and safety of crew onboard, especially for long packages; offering the ease of use and familiarity to the
period of time during offshore deployment, ii) ship safety designer, which could potentially increase its adoption
such as risk of vessel capsizing, large roll motion and rate in the ship design community. More importantly, it
slamming and iii) operational limit for ships such as meets industry standards - the design solutions are
loading/offloading and landing of helicopter [29]. validated by commercial CAD modeling and simulation
tools.
Strip theory is used to evaluate the sea-keeping
performance of the vessel. It computes the ship’s rigid- Through integration of macros and manager functions,
body motions, also known as motion for 6 degree of this is the first time automated design optimisation of an
freedom. The equation of motion for 6 degrees of offshore vessel for both resistance and sea-keeping
freedom is based on Newton’s second law of dynamics. performances have been conducted using NAPA.
Based on [30], taking origin of ship’s centre of gravity, Specifically, four separate modules in NAPA namely
these equations are given as: ‘General optimisation’, ‘FFD’, ‘Sea-keeping’ and ‘CFD’
functions are integrated, with each performing a core
∑ {( ) ̈ ̇ } (4) function within the proposed HFDO framework as
detailed in Section 2. Two design objectives which are
crucial for heavy-lift and pipe-lay vessel are selected in
Where represents the coupled surge, heave and this case study, namely ship resistance and sea-keeping
pitch motion, represents the coupled sway, performance.
roll and yaw motions, ̈ ̇ and are the acceleration,
velocity and displacement of harmonic oscillation in 4.1 OPTIMISATION PROBLEM SET UP
direction, harmonic exciting wave force or moment in
4.1 (a) Initial model
direction, solid mass or inertia coefficient,
hydrodynamic mass or inertia coefficient,
The initial model of the vessel is first created in NAPA represent the initial hull, solid line represents modified
using patch surface, based on the following principle hull, bolded line represent bounding box and dotted line
dimensions: represent the shifting or blending direction.
( ) (5)
Fig 6: FFD transformation at 180 (outboard), comparing
initial and optimised hull.
Where f is the vector of design objectives, x is vector of
design variables and X is the feasible design variable
4.2 RESULTS AND DISCUSSIONS
space.
We demonstrated in this case study the multi-objective
The GA settings are given as follows:
optimisation of a heavy-lift and pipe-lay vessel. The
proposed HFDO was implemented in NAPA. The entire
Number of individuals in a population: 10
design optimisation process took approximately 3 hours
Number of generations: 20 on a standard Intel i7 workstation. Figure 7 shows the
Number of genes per variables: 10 distribution of solutions found in the design search space
Mutation probability: 0.1 for the total pressure resistance coefficient and the roll
Crossover probability: 0.3 motion. Dominated solutions are denoted by empty
circles while non-dominated solutions are shown in solid
4.1 (c) Geometry modification- FFD diamond shape. The initial design solution is denoted by
a solid triangle. At the end of the optimisation, there are
Using binary coding scheme, FFD blending parameters altogether 5 non-dominated design solutions found.
which controls the amount of transformation applied to Although there were far fewer Pareto optimal solutions
the shape of the hull are optimised in the DSE module. found in comparison to the entire solution set generated
The new ‘control points’ are then parsed to the GM by the DSE module (~1%), it can be seen that the non-
module to manipulate the shapes. Two FFDs are applied dominated solutions offered a good trade-off between the
to increase the ‘variety’ of shape combination in order to two design objectives.
increase the search space for optimal solution. This is a
two-step process. Firstly, the bounding box is created
around the forward section of the ship model and control
point set along the Y-direction. Following this, FFD is
used to ‘drag’ the control points at 225 inboard as
illustrated in Figure 5 and then at 180 outboard as
shown in Figure 6. In both figures, the dashed line
Figure 8: Geometry comparison of forward hull between Figure 10: Wave profiles of hull form before and after
initial (INI) and optimised model (ID: 103) optimisation.
vessel using NAPA. Although the improvement in the space are, to some extent, limited and constrained as only
roll response is marginal, the improvement obtained in the FFD blending parameters are optimised. Our future
the reduction of the total resistance is particularly work will look into a more robust and effective shape
encouraging. The case study clearly demonstrated that by manipulation method that can be linked seamlessly to the
modifying the geometry of the hull form, the proposed optimiser and further improve on the application of
framework is able to effectively explore the design space HFDO process. With these new developments, we
to come up with design solutions that are able to reduce envisaged that significant advancements can be made to
the resistance while at the same time maintain good sea- revolutionaries the way ships are design in future;
keeping performances. moving from traditional Computer Aided Design (CAD)
to Computer Automated Design (CAutoD).
While potential flow is widely used in ship resistance
evaluation, it is not very accurate in capturing the flow 6. ACKNOWLEDGEMENTS
on the stern of hull due to large turbulence [40] and
hence can only serve as a coarse estimate for resistance This research is funded under Economic Development
calculation. Nonetheless, for more accurate resistance Board (EDB) of Singapore and Sembcorp Marine Ltd.
analysis, RANSE can be applied under this framework, (SCM) under Industrial Postgraduate Programme (IPP)
which will be included in author’s future work. In grant no: COY-15-IPP/140002. The authors would also
addition, more design variables can be included in the like to thank NAPA for the technical support provided
GM module for a larger range of transformation which for the case study applied in this paper.
could lead to better designs. To further improve the
accuracy of the result, the sea-keeping performance can 7. REFERENCES
be further evaluated using response amplitude operator
(RAO) analysis for other ship motion such as heave and 1. ANG, J.H., C. GOH, AND Y. LI, ‘Key
pitch. Challenges and Opportunities in Hull Form
Design Optimisation for Marine and Offshore
Finally, this case study has demonstrated the feasibility Applications’, to appear in 21st International
of a fully automated optimisation process and Conference on Automation & Computing, 2015.
performance improvement as compared to conventional
manual hull improvement process. More importantly, the 2. ‘Ship design process’, available in MarineWiki
time taken to carry out this optimisation is significantly at URL:
lower from weeks to hours when compared to traditional http://www.marinewiki.org/index.php/SHIP_DE
methods where hull forms are individually modified SIGN_PROCESS, accessed 20 July 2015.
using conventional CAD tools.
3. PERI, D., A. PINTO, AND E.F. CAMPANA,
5. CONCLUSIONS ‘Multi objective optimisation of expensive
objective functions with variable fidelity
In this paper, a formal hull form optimisation framework models’, Nonconvex Optimization and Its
which integrates geometry modification techniques, Applications, 2006.
advanced performance evaluation process with an
evolutionary algorithm is introduced. By integrating this 4. HOCHKIRCH, K. AND V. BERTRAM, ‘Hull
processes into NAPA, we demonstrated the feasibility to Optimization for Fuel Efficiency – Past, Present
automate the hull form design optimisation of an and Future’, COMPIT, 2012.
offshore heavy-lift and pipe-lay vessel. Specifically, the
proposed framework was applied to two crucial design 5. LACKENBY, H., ‘On the systematic
objectives for such class of vessels, i.e. resistance and geometrical variation of ship forms’,
sea-keeping performances. Transactions of RINA, 1950.
Results obtained ascertained the usefulness of the 6. CHI-CHAO, H., ‘Optimal Ship Forms for
proposed methodology to effectively explore the design Minimum Wave Resistance’, Journal of Ship
space where design solutions found are in line with Research, 1981.
industry standards and are more superior to modified
designs generated based on designer’s experience. In 7. T. RAY, R.P.G., O.P. SHA, ‘A global
comparison to conventional design approaches, the optimization model for ship design’, Computers
proposed methodology also provides huge cost savings in Industry, 1995.
both in terms of time and computational resources.
8. BIRK, L., ‘Application of Constrained Multi-
Despite the demonstrated efficiency of the proposed Objective Optimization to the Design of
framework, it is important to note that while the Offshore Structure Hulls’, Journal of Offshore
integration of FFD and MOGA has been demonstrated to Mechanics and Arctic Engineering, 2009.
be a useful tool, it was observed that the design search
9. BIRK, L. AND G.F. CLAUSS, ‘Automated hull 22. T, S. AND P. S., ‘Free-form deformation of
optimization of offshore structures based on solid geometric models’, ACM SIGGRAPH,
rational seakeeping criteria’, 11th International 1986.
Offshore and Polar Engineering Conference,
2001. 23. TAHARA, Y., ET AL., ‘Computational fluid
dynamics based multiobjective optimization of
10. CLAUSS, G.F. AND L. BIRK, ‘Hydrodynamic surface combatant using global optimization
shape optimization of large offshore structures’, method’, J Mar Sci Technol, 2008.
Applied Ocean Research, 1996.
24. PERI, D., ET AL., ‘Potential Flow Based
11. BIRK, L, ‘Parametric Modeling and Shape Optimization Of A High Speed, Foil-Assisted,
Optimization of Offshore Structures’, Semi-Planning Catamaran For Low Wake’, 10th
International Journal of CAD/CAM, 2006. International Conference on Fast Sea
Transportation (FAST), 2009.
12. HOLLAND, J.H., ‘Adaptation in Natural and
Artificial Systems’, University of Michigan 25. TAHARA, Y., S. TOHYAMA, AND T.
Press, 1975. KATSUI, ‘CFD based multiobjective
optimization method for ship design’,
13. BAGHERI, H. AND H. GHASSEMI, International Journal For Numerical Methods
‘Optimization of Wigley Hull Form in order to In Fluids, 2006.
Ensure the Objective Functions of the
Seakeeping Performance’, J. Marine Sci. Appl., 26. FENG, B., ET AL., ‘Ship hull automatic
2014. optimization techniques research based on
CFD’, IEEE, 2009.
14. TAHARA, Y., ET AL., ‘Development of
CAD_CFD optimizer integrated hull form 27. BRONSART, R., G. KNIELING, AND M.
design system’, J. Kansai Soc, 2003. ZIMMERMANN, ‘Automatic generation of a
panel-based representation of ship hulls for
15. JACQUIN, E., ET AL, ‘Hull form optimisation wave resistance calculations’, Practical Design
using free surface RANSE solver’, 25th of Ships and Mobile Structures (PRADS), 2004.
symposium on naval hydrodynamics, 2004.
28. NOWACKI, H., ‘Hydrodynamic design of ship
16. BAIWEI, F., ET AL., ‘Ship resistance hull shapes by methods of CFD’, Progress in
performance optimization design based on Industrial Mathematics at ECMI, 1996.
CAD_CFD’, 3rd International Conference on
Advanced Computer Control, 2011. 29. BERTRAM, V., ‘Practical Ship
Hydrodynamics’, Butterworth-Heinemann,
17. KIM, H., ‘Multi-Objective Optimization for 2000.
Ship Hull Form Design’, George Mason
University, Computational and Data Sciences 30. JOURNÉE, J.M.J., ‘Quick strip theory
Department, 2012 calcuations in ship design’, Practical Design of
Ships and Mobile Structures (PRADS), 1992.
18. CAMPANA, E.F., ET AL., ‘New global
optimization methods for ship design problems’, 31. SALVESEN, N., E.O. TUCK, AND O.M.
Optimization and Engineering (OPTE), 2009. FALTINSEN, ‘Ship motions and sea loads
Trans’, SNAME, 1970.
19. TAHARA, Y., ET AL., ‘Single and
multiobjective design optimization of fast 32. BACK, T., FOGEL D.B., MICHALEWICZ, Z.
multihull ship’. J Mar Sci Technol, 2011. (Editors), ‘Handbook of Evolutionary
Computation’, IOP Publishing Ltd, 1997.
20. XI CHEN, ET AL., ‘High-fidelity global
optimization of shape design by dimensionality 33. KONAK, A., D.W. COIT, AND A.E. SMITH,
reduction, metamodels and deterministic particle ‘Multi objective optimisation using
swarm’, Engineering Optimization. 2014. GA_tutorial’, Reliability Engineering and
System Safety, 2006.
21. CHUN, H. H., ‘Hull form parameterization
technique with local and global optimization 34. ZITZER, E., THIELE, L., ‘Multiobjective
algorithm’, International Conference on Marine evolutionary algorithms: a comparative case
Technology (MARTEC), 2010 study and the strength Pareto approach’, IEEE
8. AUTHORS BIOGRAPHY