Sunteți pe pagina 1din 12

TOWARDS OPTIMAL DESIGN OF SHIP HULL SHAPES

J.J.MAISONNEUVE1, S.HARRIES2 , J.MARZI3, H.C.RAVEN4, U.VIVIANI5,H.PIIPPO6


1
SIREHNA, France, jean-jacques.maisonneuve@sirehna.ec-nantes.fr
2
TU Berlin, Germany, Stefan.Harries@ism.tu-berlin.de
3
HSVA, Germany, marzi@hsva.de
4
MARIN, The Netherland, H.C.Raven@marin.nl
5
FINCANTIERI, Italy, umberto.viviani@fincantieri.it
6
NAPA OY, Finland, heikki.piippo@napa.fi

ΑΒSTRACT

The recent increase in information technologies dedicated to optimal design, associated with the
progress of the numerical tools for predicting ship hydrodynamic performances, allows significant
improvement in ship design. A consortium of fourteen European partners – bringing together ship
yards, model basins, consultants, research centres and universities – has therefore conducted a three
years European R&D project (FANTASTIC) with the goal to improve the functional design of ship
hull shapes. The following key issues were thus considered: parametric shape modelling was
worked on through several complementary approaches, CFD tools and associated interfaces were
enhanced to meet efficiency and robustness requirements, appropriate design space exploration and
optimisation techniques were investigated. The resulting procedures where then implemented, for
practical assessment purposes, in some end-users design environments, and a number of
applications were undertaken.. Significant gains can be expected from this approach in design, in
term of time used for performance analysis and explored range of design variations.

1 OBJECTIVES

The principal objective of FANTASTIC is to improve ship design by applying parametric shape
modelling and state-of-the-art CFD analysis tools to predict ship hull performance. These functional
aspects are integrated in an optimisation environment. The approach is expected to i) allow for a
substantially larger coverage of design alternatives and ii) by using most recent CFD analysis tools
it will improve the quality of the design finally found optimal.

This was focused on the hydrodynamic prediction of steady performance, through two main types
of flow calculation tools: potential flow panel methods for a short term target in terms of practical
implementation, and RANSE methods, in view of a longer term application. Seakeeping was also
partially addressed as a major issue for some ship types. It was also implicit that the other aspects of
ship design assessment, like hydrostatics, manoeuvring, structural resistance, even if not explicitly
developed in the project, also had to be kept in mind during the optimisation process, for the sake of
realism.

The key issues with respect to these objectives are described in the following sections and in figure
1. The first one deals with the best way to model the ship with a restricted set of parameters so that
a wide range of variations can be investigated, while taking into account as many constraints as
possible regarding the shape and its feasibility. The second challenge is to make CFD methods
sufficiently accurate, robust, automatic and fast to be used in an efficient design search. The last
issue consists in defining the best tools, from the IT as well as from the algorithmic point of view so
that the integration of the parametric modelling and assessment tools be as easy as possible, and the
search for optimal design be as efficient as possible.

1
Design space
exploration

g
lin
el
----------

od
----------

n
---------- ----------
----------

tio
m
---------- ----------

er
---------- ----------

n
ra
---------- ric ----------

lv

tio
---------- ----------

ne
----------

So
et
----------

ac
---------- ----------

ge
----------
m

---------- ----------

tr
----------
ra

---------- ----------

ex
rid
----------
Pa

a
at
Ship parametric

D
Design criteria
representation

Assessment process: calculation chains


Figure 1: Description of the approach

2 PARAMETRIC MODELLING

Within the project the primary aim of parametric modelling has been to facilitate the modification
of hull shapes to be improved and, eventually, to optimise a ship’s hydrodynamic performance. A
parametric description captures the essence of the intended shapes and their possible variations and,
thus, reduces the number of unknowns within the modelling process. The shape is determined from
fewer data – e.g. points that define a specific location, control points that influence a certain hull
area – and/or data that represent information on higher levels – e.g. angle of entrance of the design
waterline, centre of floatation. This is achieved by formulating dependencies and constraints a
priori. Inherently, in a parametric approach the diversity of possible hull forms is confined by the
topology and the design rules established in the parametric set-up. Nevertheless, once a suitable set-
up is implemented, variations can be accomplished in less time and with higher quality.

Several approaches were pursued regarding parametric modelling. They range from the
development of a ship parametric modelling tool (FRIENDSHIP-Modeler) via the integration of
parametric capabilities in a well established ship design system (NAPA) to more restricted but
simpler methods like the parametric definition of shape deformation functions (GMS/Facet). In
addition, the exploitation of general purpose CAD systems was investigated. This resulted in a set
of tools available for practical use, from which a user may choose depending on the design
environment and the optimisation task at hand.

2.1 Ship parametric modelling – FRIENDSHIP-Modeler

The FRIENDSHIP-Modeler [7] , developed by the Technical University Berlin since 1995, was
substantially extended and adapted to the needs of practical design environments. The approach
favoured within the modeller differs from the methodology pursued in most general purpose CAD
tools, for details see Abt et al. (2001). Traditionally, the hull geometry is built up from points via
lines to surfaces. A change in geometry is evoked by manipulating several points in space
concertedly which often is an interactive and time-consuming undertaking.

2
In the FRIENDSHIP-Modeler the classic naval architect’s technique is adopted, see [2]: A set of
longitudinal lines – so-called basic curves – is layed out from which all information can be retrieved
to subsequently establish the geometry. This is done in a three stage process:
1. Parametric design of a suitable set of basic curves such as deck, design waterline, flat-of-
side, flat-of-bottom, centerplane etc. The basic curves are built in agreement with a few
prominent transversal curves like the main frame section, the transom, and, optionally,
additional sections in the forward or aft body.
2. Parametric modelling of design sections derived from the basic curves.
3. Generation of a set of surfaces that interpolate or closely approximate the design sections.

All curves and surfaces are modelled as B-splines. All B-splines are calculated by geometrically
optimising for fairness while meeting the desired form parameters. In this way large flexibility and
high shape quality can be accomplished. The design of complex hull shapes featuring bulbs,
gondolas and skegs is provided for by attaching appendages to a bare hull, the appendages having
their individual and problem-oriented parameterisations. Efficient and effective form variations can
thus be achieved and the approach is well-suited for automated optimisation, see e.g. [3] and [6].

During the project the Technical University Berlin has improved the features available within the
FRIENDSHIP-Modeler with respect to the export of geometry via IGES files and several tool-
specific file formats, the import via offset files, a graphical user interface (GUI) – which was
implemented under GiD by CIMNE – and, most importantly, new modelling features to better
reflect local and global shape characteristics. The range of hull shapes that can be produced has
been extended from a topological point of view. Furthermore, the way the generation process is
handled has been improved by introducing an intelligent structure of parameters and their
dependencies. A human readable file format has been established which contains the entire ship
geometry in terms of form parameters. A form parameter file may incorporate just a few parameters
when designing from scratch and building up the hull geometry step by step. It may also contain
quite many parameters to fine tune the shape in a manual or, alternatively, an automated process.

Figure 2 depicts two hull forms – i.e., bare hull plus bulb with the skeg omitted – produced from the
FRIENDSHIP-Modeler. By varying just one parameter a substantial but well-behaved change in
geometry is obtained. As can be seen this enables the designer to manipulate specific features while
deliberately keeping others. In the example of figure 2 the area coefficient of the design waterline of
the forebody was reduced while the displacement was kept. The hull forms depicted are generated
fully automatically without touching any B-spline vertices by hand. Changes are exaggerated to
show the approach’s potential.

Figure 2: FRIENDSHIP-Modeler – Example parametric variation of forebody

3
2.2 Template Approach – NAPA

The second CAD system that played a central role in the project was NAPA, the Naval
Architectural Package from NAPA Oy [9]. NAPA is a CAE system for initial and basic ship design,
comprising, among other things, hull surface definition, production-level fairing, definition of the
ship’s compartmentation and naval architectural calculations.

In the NAPA system the hull surface is defined by a grid of curves, see figure 3. In the parametric
representation sets of geometric parameters have been utilised in the definition of these grid curves.
As the grid structure varies significantly for different ships, templates for several ship types have
been created within the project. The templates available comprise fast ferries, cruise liners,
passenger/Ro-Ro ferries, container ships and frigates as well as single screw tankers and bulkers.

In order to be able to work with the parametric hull description in the NAPA system, a user friendly
and efficient graphical interface has been developed. The graphical user interface offers the end
users an easy way of adapting the templates to their own needs and, moreover, supports the creation
of totally new parametric hull surface definitions. Partial surfaces or local areas of the ship hull, for
instance the bulb area, can also be parameterised while leaving the rest of the hull surface without a
parametric definition.

The parametric approach has been made part of NAPA’s hull editor which is intended for the
modification of hull surfaces or any surface defined by a curve grid. The curves and surfaces are
defined by alphanumeric descriptions. Instead of purely manipulating these alphanumeric
description, however, the hull editor works directly with the geometric components as shown on the
graphics display, and the effect of each change can be seen immediately. Further to the basic actions
applicable to individual points and angles, the hull editor has other editing functions for executing
more complex actions.

Figure 3 presents the main window of the hull editor with the parametrically described hull of a
container ship. The stem curve is highlighted to show its definition. Via the parametric description
concerted changes in geometry can be brought about with fewer entities than with NAPA’s
conventional shape representation.

Figure 3: NAPA – Main window of hull editor

4
2.3 Shape Transformation Functions – GMS/Facet

A third option developed and applied in the project does not rely on a complete parametric
description of the hull shape itself, but on a shape definition consisting of the original shape plus
parametric deformations applied to it. This development has been done by MARIN, and
implemented in GMS, Marin's CAD system for hydrodynamic hull form design.

The incentive for this alternative development was as follows. Suppose that one already has an
initial hull form given as a surface description in the CAD system. Optimisation using a parametric
description then would first require a redefinition of the same hull form, which takes time and may
mean a loss of detail and precision.

Instead, a procedure has now been created in which a designer defines parametric deformations to
the original hull. Rather than manipulating control points individually, a group of control points is
selected that will be affected by the deformation. Secondly, the deformation is specified as a set of
single-parameter translation functions that will be applied to the selected control points.

This being done, a one-parameter family of hull forms is defined. By subsequent further
deformation specifications, more parameters can be involved. Once a family of hull shape
variations has been defined that is considered sufficient for the next optimisation step, the
interactive design session is closed, and any hull form variation, including a suitable hull panelling,
can be generated in batch by a single command giving the desired values of all parameters.

Figure 4 shows some examples of variations of an initial bulbous bow shape, all defined by
different single-parameter deformations. It has appeared that the system provides a large flexibility
and freedom in shape generation, retains the smoothness of the original hull, and can be applied
immediately to practical cases. On the other hand, the deformation modes need to be specified for
every case anew, and efficiency of this process is crucial; a requirement satisfied by the
implementation in the Graphical User Interface of GMS, with instant visualisation of the
corresponding hull shapes. Current experience indicates the approach is very well suited to
incorporation in a design cycle in which a limited number of parameters is addressed at a time, and
the designer still has an important role in the decision on what variations to consider.

Figure 4 : GMS/Facet - a variety of bulbous bow shapes, defined by different single-parameter


deformations to an original shape.

3 CFD ENHANCEMENT

For hydrodynamic assessment of ship hulls in the optimisation process, CFD codes play an
important role. Their use in the optimisation chain requires they are accurate, fast, reliable,
automatic, and that they communicate with the other components in the chain. In the first stage of
the project, (potential flow) panel codes have been used, which already largely satisfied these
criteria and provided immediate results. Due to the substantially higher computational effort
associated with RANSE computations, these required a greater effort to achieve solutions that lend
themselves to practical application in an automated optimisation environment. Today, practical

5
optimisation applications can already be envisaged using panel codes and mean term exploitation
can be expected for RANSE methods. This work concerned a set of the most widespread CFD tools
from several European institutes. A summary of the main tools involved in the project is given in
table 1.

3.1 Panel methods

Panel codes are probably the most widely used CFD tools in ship design. Given their robustness,
versatility and price / performance ratio they have been adopted by a large number of model basins,
shipyards and consultants. Ease of use and especially the speed of computations make them ideal
candidates for use in the automated optimisation environment in FANTASTIC. The latter aspect is
of great importance, especially in view of computer capacity usually available among the
companies and institutes in the marine branch.

Three widely used non-linear panel codes for wave resistance predictions have mainly been used in
the project: RAPID from MARIN, X-PAN / SHIPFLOW from FLOWTECH , ν-SHALLO from
HSVA

All of these have already a long history of development and a long record of successful
applications. Although the theoretical background of all three codes is quite comparable – all of
them are using singularity distributions to model the ship hull and further de-singularised sources
for the modelling of the free surface – the individual prerequisites for setting up a computational
model and running the code for a specific case are quite different.

The panel codes need as input a description of the hull form (variations) as generated by a CAD
system; usually in the form of a panel mesh. The generation of suitable hull panel meshes is a major
issue. Some of the codes imbed own mesh generators, other rely on external tools, while also some
of the CAD systems have mesh generation capabilities.

As the project involves two major CAD systems for marine applications, NAPA and FRIENDSHIP,
as well as additional tools in use with other partners, the first issue was to interface the CAD tools
with the panel codes. Table 1 shows the range of CAD-, mesh generation and CFD (panel codes) -
tools finally used in the context of FANTASTIC. Two kinds of connections could be established:
either, via an IGES file written by the CAD system and read in by the panel generator; or by direct
generation of a panelling suitable for the different panel codes inside the CAD system.

In a second step, the automation of the panel mesh generation has been addressed by the code
providers. This aspect is of utmost importance when using the CFD tools in an optimisation
environment which would be almost useless when the process would be dependent on a manual or
user driven mesh generation. Templates and macros have been developed that allow for mesh
generation of parametric variants of a given design. These have been applied to several test cases in
the course of the project.

Due to the work in this field, a significant number of combinations of the tools mentioned in Table
1 was made possible, especially between the parametric modellers and panel codes, with direct or
IGES links. Consequently, a variety of optimisation chains could be composed. These were then
used in the project in massive, automated, calculation of design variations, and proved to be robust
enough for actual exploitation.

6
3.2 RANSE methods

Hullform optimisation for a ship is a complex problem involving a range of dependencies. Certainly
the viscous dominated flow at the stern of the ship and the propeller inflow are decisive factors for
the final performance of the ship. Therefore, parallel to the work on panel codes the enhancement of
RANSE codes to be used in an optimisation environment has been addressed.

Again three main codes have been selected by FANTASTIC’s CFD partners. These encompass
MARIN’s PARNASSOS, SHIPFLOW/CHAPMAN from FLOWTECH and COMET which has
been used by HSVA. RANSE codes require 3-d volume grids as inputs. Today these grids do not
lend themselves easily to totally automatic grid generation. The work in the project has addressed a
partial automation of the process, and modification of volume grids based on templates. In addition,
due to developments on the solver, a greater flexibility in grid generation has been created to reduce
the time spent on grid generation.

Computational performance is a major issue when speaking RANSE codes. Other than panel codes
which can perform non-linear free surface computations in the order of a few minutes only, here
several hours are usually required for a single case computation, which is hardly bearable in an
automatic optimisation when a large number of design variants must be computed. Thus
performance and efficiency matters have been addressed. Improvement in the Parnassos solution
algorithms have substantially speeded up multiblock computations. Also parallelisation is an
appropriate means to increase speed. For the Parnassos solver, the turnaround time has been
reduced to 20 minutes, for a full-scale RANS computation (without free surface) on a grid of 1
million cells, using 8 processors on a supercomputer. Such computation times bring optimisation
within reach, and steps in that direction are being taken now. Significant advances have been made
running RANSE codes on parallel PC-clusters, e.g. at HSVA.

Shape generation and variation


Napa ;Friendship ;GMS ;E4
;Catia
Grid generation
Specific; GiD; Facet; Newpangeo;
ICEM; GridGen
Calculation
ν−shallo (HSVA): Rapid (Marin);
Shipflow (Flowtech); Parnassos;
Chapman; Comet; Warp; Soap;
Fluent
Data extraction-Post-processing
Specifically developed; Tecplot;
GiD; FieldView
Optimisation
Xopt(MMA); Chwarismi ;
modeFRONTIER
Main involved tools and links

Table 1: Overview of main tools and links worked on in the project (left),
list of tools involved in applications (right)

7
An alternative approach to reduce the computational effort is to reduce the number of computations
needed in an optimisation. With this objective, Chalmers University investigated the use of adjoint
equation approaches for the assessment of sensitivities of the flow to elementary variations, with
Ranse methods. This is to be used to outline trends around given design points, and also in gradient
based optimisation methods.

4 OPTIMAL DESIGN TECHNIQUES

The objective within the project was to call for optimisation techniques in view of the enhancement
of ship design procedures, especially applied to hydrodynamics. It was thus necessary to find
approaches that ensure the global validity of optimal design approaches, and that allow more
dedicated and refined hydrodynamic design optimisation.

As regards optimal design approaches, and after some years of investigations [5], it is now obvious
that each optimisation problem has its own specificity that cannot be dealt with by any universal
algorithm, and that a real optimisation problem is basically multi-objective. So it appears that
relevant systems must offer an environment and a set of tools that allow the best choice of tools, or
combination of tools, to achieve possibly conflicting goals and allow decision making. This must
include:
- intuitive tools for the integration of calculation chains, allowing complex logics, and thus able to
address real life cases where conditional process must be ensured (eg stopping calculation as
soon as some criteria are not satisfied, to avoid useless computation…)
- exploitation of most recent technologies in order to address remote calculations in
heterogeneous environments, concurrent calculations, maintenance and readability of full design
database, portability of software and data, etc…
- tools and algorithms for an efficient exploration and analysis of the design space, including:
- design of experiments
- response surfaces, including classical approximation functions but also more elaborate ones,
able to deal with non regular functions involved in real problems (e.g. neural networks)
- optimisation algorithms, adapted to local search involving smooth functions (e.g. gradient
based methods…) but also able to search in complicated design spaces involving non
regular functions (e.g. genetic algorithms), and to really deal with multiple objectives
problems. This last point is of prime importance for practical design problems, which are
very rarely naturally mono-objective ones.
- multi-criteria decision making tools, that help the designer to investigate pareto fronts and
outline trade-offs.
- efficient analysis tools for a quick and good understanding of the design space.
All these features have to be easily combined to best fit to requested search strategies.

Numerous tools and algorithms were investigated in the project, and used in optimisation searches
(e.g. Method of Moving Asymptotes, by Chalmers university). Some were specifically developed in
existing environment (e.g. simplex and moga, by Napa Oy). More integrated optimisation
environments, that meet most of the above requirements, were also investigated. One of these, the
modeFRONTIER package, initially developed on the basis of an EC funded project [8], was chosen
as the central optimisation package in most applications in the project.

5 APPLICATIONS

The development done in parametric modelling, CFD enhancement, optimal design approaches, and
the communication between all these components, made it possible to set up several ship

8
optimisation chains in practical design environments. A number of end users thus have validated
and evaluated the approach. Examples are given below.

5.1 Optimisation of a fast monohull

Applications tested at FINCANTIERI concerned fast ro-ro vessels, frigates and fast ferries. While
for some ship types, the hull form is often optimised for resistance and propulsion only, in the case
of a fast ferry seaworthiness is an important goal and needs to be tackled early in the design
process. The hull shape needs to have both low resistance and good seakeeping qualities, with
particular attention to passengers’ comfort.

We here consider a Medium Deep Vee (MDV) monohull developed by FINCANTIERI, that in the
last years has designed and built a number of MDV's for the Mediterranean. The one considered
here has Lpp between 128 m and 138 m. (varied during the optimisation), a required transport
capacity of about 1800 passengers and 460 cars, and a design speed of 40 knots. The ship was
modelled as shown in table 2. The optimisation was carried out with modeFRONTIER on the
FRIENDSHIP-NEWPANGEO-WARP/SOAP calculation chain, and consisted of two phases.

Phase I was a preliminary investigation at the beginning of the design process, when the target hull
was defined only in terms of transport capacity. The objective was to identify the main particulars
to minimise the calm water resistance while maximising the comfort of the passengers. The comfort
is measured through the Motion Sickness Index (MSI) evaluated at 4 different locations on the ship
and considering all headings and the probability of occurrence of each Sea State in the geographical
area where the vessel will operate. The calm water performance, i.e. RT/cars, is evaluated at the
design speed of 40 knots, while the MSI is evaluated for 30 knots to take into account the voluntary
speed reduction in the higher Sea States.

Free variables
Length +/- 10%
Beam +/- 1 car
Draft +/- 10%
KNdraftTran +/-15%
KNzMaxBeam +/-200%
KNmaxBeam 0.85 Beam
Constraints Figures of merit
Displacement +/-8% RT/cars at 40 knots RT/ ∆ at 40 knots
GMT Monitored Averaged MSI at 30 knots
Table 2: problem definition

The investigation of the design space started with a Design of Experiments (DoE) of 100
individuals created by the parametric modeler and evaluated with the CFD and seakeeping codes.
Fig. 5 clearly shows the effect of the main dimensions. The best design here is nr. 50 (highlighted in
the figure). Compared with the worst feasible designs, design nr. 50 means a gain in RT/cars of
26%, while the improvement in terms of MSI amounts to 19%.

Phase II consists in a refinement of design 50. Here the main dimensions are fixed, the
displacement is allowed to vary in a +-1% range and the variable parameters are related to more
local features of the ship geometry. In this case the measures of merit are total resistance (RT) at 40
knots, vertical acceleration at stern and bow in sea state 4, at 30 knots; and number of bow

9
slamming events per hour, in sea state 4 at 30 knots; and are, therefore, mostly related to seakeeping
characteristics.

RT/cars vs Length RT / ∆ vs Draft

MSI vs KnzMaxBeam MSI vs Length


Figure 5: Effect of main parameter on performances

In this case the optimisation was made using a multi-objective genetic algorithm (MOGA) followed
by a gradient search. For the latter a global figure of merit has been used, combining all measures of
merit mentioned above, non-dimensionalised with respect to the values of design 50.

The result of this combined search is a design that has a seakeeping improved by 6% in terms of
vertical accelerations and 10% in terms of number of slamming events, but has a 3% increase in
total resistance compared with design 50. This is considered acceptable, being a reasonable trade-
off between seakeeping and resistance.

These studies have been carried out at the technical departments of FINCANTIERI and CETENA.
The optimisation approach proved to be robust, to lead to realistic designs and to give relevant and
promising improvements. While verification in a towing tank has not been done for this particular
optimisation, past experience has already proved the realism of CFD calculations.

5.2 Other applications

IZAR has carried out an optimisation of a frigate type ship, with goals addressing resistance,
calculated with the FRIENDSHIP+FACET+RAPID calculation chain, as well as manoeuvring and
seakeeping performance assessed by formulas specific for this type of ship, all being involved in the
modeFRONTIER environment. The search was done using MOGA algorithms, response surfaces,
and a number analysis tools like t-student analysis that gives information on the sensitivity to free
variables. Multi-criteria decision making tools were also used to help defining trade-offs among the
Pareto set. Efficient calculation chains could be set up in the existing design environment, and the
benefit of optimisation approaches could be assessed, and ways of improvement defined.

10
Many other practical applications, e.g. for ro-ro, naval, and container ships, were made to
investigate the capabilities and performance of the design approaches; many different combinations
of the tools listed in Table 1 were applied. As regards RANSE methods, most of the work addressed
improvement of performance and automation; but some first restricted, optimisations were
successfully performed. This in particular demonstrated the applicability of genetic algorithms,
combined with response surface techniques, to CPU-intensive applications.

All these applications brought, for the first time, a significant experience on the different aspects of
the developed approaches.

5.3 Common demonstration case

In order to demonstrate the applicability and potential benefits of the developed methodology, a
final common optimisation exercise was undertaken, on a “generic” hull form for a medium speed
ROPAX vessel (Fig. 6), supposed to be representative for a class of modern ships, at a Froude
Number FN = 0.311. The initial design was a transformed version of an existing ship, but main
parameters and design speed were modified. So the initial design was sub-optimal and provided
sufficient room for improvement. Three parametric hull form modification approaches and three
panel codes were used to optimise the hull form from the resistance point of view. The initial design
and the – agreed – optimised version will be compared in a model test, which was not accomplished
at the time of this paper. Some results will be presented at the conference.

Initial shape

Alternative 1

1
Alternative 2 2

Total resistance vs design id

Figure 6: common fanta_ro test case, some variations and first investigations

6 CONCLUSIONS

A large amount of work was dedicated to the enhancement of tools and links, that can be combined
in many different calculation chains which can be operated with reduced manpower. This is already
a very positive outcome of the project, even if it is only a necessary condition to go further in the
field of optimal design.

11
Major progress was made in parametric modelling, which is of prime importance regarding
capabilities to consistently investigate numerous variations during design, and especially when
optimisation techniques are involved. Several approaches, each one corresponding to specific needs,
are now available. The FRIENDSHIP-Modeler describes and generates the entire hull surface from
a limited parameter set related to higher-level geometrical properties. NAPA, meanwhile, assembles
control points by prescribing explicit relations between control point coordinates, and puts these in
a template which then defines a family of shapes for a limited number of parameters. GMS/Facet
groups control points in terms of a user-selected box and applies user-defined transformation
functions with a few parameters.

The next step consisted in providing tools that actually exploit these modelling and calculation
tools, towards optimal design. Advantage was there taken of the most recent technologies in this
area, that allow efficient integration of calculation chains, extensive exploration of the design space,
including multi-objective search that are of prime importance for real life problems. The project
made it possible to largely investigate all the possible techniques, identify best strategies for given
problems, and produce guidelines for further exploitation.

The first applications to relevant test case show the relevance of these approaches, and their
capacity to be actually used and to bring benefits, in the ship design areas where calculation tools
are already valid and current practice. Immediate use of this can be expected after the project.

As regards more complicated applications, eg involving Ranse solvers, some work is remaining to
make it current practice, but some promising applications let expect it in the near future.

7 ACKNOWLEDGEMENT
The work presented in this paper was performed within the EC funded Project FANTASTIC,
G3RD-CT 2000-00096, a R&D project on ‘Functional Design and Optimisation of Ship Hull
Forms’. The Project Partners are FINCANTIERI, SIREHNA, FSG, IZAR, HSVA, TUB, NAPA,
MARIN, FLOWTECH, SSPA, CHALMERS, CIMNE, SINTEF, CETENA.

8 REFERENCES
1. Abt, C.; Bade, S.D.; Birk, L.; Harries, S. (2001) “Parametric Hull Form Design – A Step
Towards One Week Ship Design,” 8th International Symposium on Practical Design of Ships
and Other Floating Structures ·PRADS 2001, Shanghai.
2. Harries, S. (1998) “Parametric Design and Hydrodynamic Optimization of Ship Hull Forms,”
Dissertation, Technische Universität Berlin; Mensch & Buch Verlag.
3. Harries, S.; Valdenazzi, F.; Abt, C.; Viviani, U. (2001) “Investigation on Optimization
Strategies for the Hydrodynamic Design of Fast Ferries,” FAST'01, Southampton.
4. Larsson, L. et. Al., (2000) “Gothenburg 2000 – A workshop on numerical ship hydrodynamics”,
Gothenburg, Sweden, Chalmers University, Report CHA/NAV/R-02/0073
5. Maisonneuve, J.J., (2002), "Sirehna's background in optimal design", internal report.
6. Valdenazzi, F.; Harries, S.; Viviani, U. Abt, C. (2002) “Seakeeping Optimisation of Fast
Vessels by Means of Parametric Modeling,” HSMV02, Naples.
7. FRIENDSHIP-Modeler Web Site: http://www.FRIENDSHIP-Systems.com
8. modeFRONTIER Web Site: http://www.esteco.it
9. NAPA Oy’s Web Site: http://www.napa.fi

12

S-ar putea să vă placă și