Docstoc

marnet-cfd final report

Document Sample
marnet-cfd final report Powered By Docstoc
					MARNET-CFD Final Report and State
Of the Art Review




             MARNET CFD FINAL REPORT
       AND REVIEW OF THE STATE-OF-THE-ART
           IN THE APPLICATION OF CFD IN
      THE MARITIME AND OFFSHORE INDUSTRIES


                                        Prepared by:

                WS Atkins Consultants (Thematic Network Co-ordinators)

                      And members of the Network Steering Committee

                 Sirehna, HSVA, FLOWTECH, VTT, Imperial College,
                       Astilleros Espanoles, Germanischer Lloyd




WS Atkins Consultants – Co-ordinators                                    Page   1
MARNET-CFD Final Report and State
Of the Art Review

                                           Summary

This document provides the final report on the activities of the MARNET-CFD Thematic
Network.

The MARNET-CFD Thematic Network has run over the period 1st October 1998 until 31st
March 2003. Its originally intended 3 year duration was extended to 4.5 years as the result of
under-spending of the budget by the participants, and in order to support the coordination of
FP5 and FP6 projects and their preparation for as long as possible.

The key objectives of the Network have undoubtedly been met in so far as:
•   It has provided a network within the European Union for shipbuilders, naval architects,
    offshore engineers and consultants, research institutes and towing tanks, marine
    manufacturers, Classification Societies and Universities, to co-ordinate their efforts in the
    development and exploitation of computational fluid dynamics for all marine
    applications.
•   It has established a shared database of experimental and computational research results
    for a variety of marine and offshore vehicle forms, generic geometries and operating
    scenarios.
•   It has coordinated efforts within the European marine and offshore industries in the
    validation of predictive techniques in hydrodynamic design.
•   It has provided best practice guidelines for the application of CFD to common marine
    problems.
•   It has provided a regular review of the state-of-the-art in Europe, including a survey of the
    CFD tools available (including interfaces to CAD) to the industry.
•   It has established an overview of the level of confidence in CFD within the industry that
    can be used to direct the development of future programmes.
•   It has stimulated Industry to pose the “grand challenge” problems, for all technical areas
    open to investigation using CFD, that generate progress in the research community, as
    evidenced in some of the papers presented at Network Workshops, and FP5 projects
    developed.
•   It has held 4 annual workshops with the purpose of comparing progress in the application
    and validation of numerical models, and subsequently produced proceedings that reflect
    well the state-of-the-art for the benefit of practising designers and hydrodynamicists.
•   It has developed a Web site on the Internet, containing all of the key deliverables and
    stimulating interest world-wide in its activities.
•   It has supported the development of new consortia for research within the Fifth
    Framework Programme to further European competitiveness in marine design and
    construction, many of which have led to projects with a direct impact on CFD validation
    and development.

The main achievements of the Network however can be summarised as:




WS Atkins Consultants – Co-ordinators                                                    Page   2
MARNET-CFD Final Report and State
Of the Art Review

•   The annual workshops, that have regularly been able to generate close to 20 good quality
    technical papers, and have supported the development of FP5 project consortia.
•   The best practice guidelines for marine CFD, developed in collaboration with the
    ERCOFTAC Industrial Advisory Committee, and available to al participants on the web.
•   To stimulate a significant inc rease in the amount of EU funded marine CFD research
    consortia.
•   The MARNET-CFD database of experimental and computational results.
•   Establishing a clear picture of the state of opinions and confidence in the use of CFD in
    marine applications.

The following report describes all activities carried out by the Network over its 4.5 year
lifetime. It provides and update to the technical state of the art review conducted in the first
year, and the results of a survey of views within the industry on the use of CFD in the
maritime sector.
The report also provides, in appendices, the final set of best practice guidelines for the
application of marine CFD, and the proceedings of the final workshop. All of this
information is also available on the MARNET-CFD web-site.

Finally, the following recommendations are made in light of the experience gained in
MARNET-CFD.
•   That the Thematic Network can be a strong mechanism for R&D support in this area but
    that a more flexible funding mechanism is required.
•   The state of the art and surveys of opinion within the industry suggest the following key
    priorities for future support in marine CFD:
           o The integration of tools and techniques across all aspects of the analysis
             process (potential flow and RANSE, rigid body and structural dynamics)
           o The improvement of solver speed and accuracy
           o Full scale validation of CFD through improved measurement techniques and
             equipment.
           o The introduction of tools for optimisation of form and performance.
•   It is unlikely that a single CFD tool suitable for all applications will be developed, but
    there are many synergies that could be better exploited.
•   The user remains the main area of uncertainty in application of marine CFD, and
    inconsistent practices the chief source of error and inconsistency. It is likely therefore
    that CFD will remain a specialist discipline in design analysis rather than become a
    mainstream process for naval architects.
Finally, we would recommend that consideration be given to the further support of marine
CFD in Europe through a second Thematic Network in marine CFD at some point in the
future. However, its modus operandi would need to be significantly different in order that
the full potential could be achieved.




WS Atkins Consultants – Co-ordinators                                                   Page   3
MARNET-CFD Final Report and State
Of the Art Review

                              MARNET-CFD FINAL REPORT

                                         CONTENTS

1.0    INTRODUCTION

2.0    OVERVIEW OF COORDINATION ACTIVITIES
       2.1 Meetings Held
       2.2 Framework Programme 5
       2.3 Framework Programme 6
       2.4 Technical Activities in CFD

3.0    DELIVERABLES
       3.1 Best Practice Guidelines
       3.2 Computational and Experimental Databases
       3.3 The MARNET-CFD Web-Site
       3.4 Proceedings of Technical Meetings

4.0    STATE OF THE ART REVIEW
       4.1 TA1 – Steady Ship Flow
       4.2 TA2 – Seakeeping and Manoeuvring
       4.3 TA3 – Propulsors
       4.4 TA4 – Offshore Floating Systems
       4.5 MARNET-CFD Questionnaire Results and Review.

5.0    REMAINING GAPS, RESEARCH AND INDUSTRY NEEDS
       5.1 Introduc tion
       5.2 Needs in each TA
       5.3 Industry Requirements For Greater Exploitation

6.0    BUDGET SUMMARY AND RECONCILIATION

7.0    CONCLUSIONS


Appendix I       Best Practice Guidelines in Marine CFD

Appendix II      Final State-of-the-Art Questionnaire Results

Appendix III     Proceedings of the Final Technical Meeting

Appendix IV      Membership and Management Structure of MARNET-CFD




WS Atkins Consultants – Co-ordinators                                Page   4
MARNET-CFD Final Report and State
Of the Art Review

1.0      INTRODUCTION
This document provides a final review of the activities of the MARNET-CFD Thematic
Network, (BRRT-CT98-5058), funded under Framework Programme 4 (FP4) Growth
Programme for sustainable surface transport. It provides an overview of the activities of the
Network and provides a final set of deliverables and exploitation plan. Complete details of
the reconciliation of the budget with the final spend on the project are also given by
participant and in summary. Finally, conclusions regarding the needs of the industry, the
success of the Network in addressing these needs, along with proposals for ways ahead, are
given.

MARNET-CFD was set up to support the needs of the European Shipbuilding and Offshore
Industries in computational fluid dynamics (CFD). The Network was started in October 1998
and ran for 4.5 years until March 2003. The main objectives of MARNET-CFD were:
•     To provide a network within the European Union for shipbuilders, naval architects,
      offshore engineers and consultants, research institutes and towing tanks, marine
      manufacturers, Classification Societies and Universities, to co-ordinate their efforts in the
      development and exploitation of computational fluid dynamics for all marine
      applications.
•     To establish, maintain and develop a shared database of experimental and computational
      research results for a variety of marine and offshore vehicle forms, generic geometries
      and operating scenarios.
•     To establish a common and concerted approach within the European marine and offshore
      industries to the validation of predictive techniques in hydrodynamic design, and provide
      guidance/codes of practice for application.
•     To provide a regular review of the state-of-the-art in Europe, including a survey of the
      CFD tools available (including interfaces to CAD) to the industry.
•     To stimulate Industry to pose the “grand challenge” problems, for all technical areas open
      to investigation using CFD, that generate progress in the research community.
•     To hold annual workshops with the purpose of comparing progress in the application and
      validation of numerical models, and subsequently produce summaries of the state-of-the-
      art for the benefit of practising designers and hydrodynamicists.
•     To publish a regular newsletter, supported by a Web page on the Internet, containing
      notification of research activities, latest publications, database updates and Network
      activities for all participants.
•     To stimulate the development of new consortia for research within the Fifth Framework
      Programme to further European competitiveness in marine design and construction.

All of these objectives were addressed successfully during the course of the Network. The
following report provides details of the achievements of the Network.




WS Atkins Consultants – Co-ordinators                                                      Page   5
MARNET-CFD Final Report and State
Of the Art Review

2.0    OVERVIEW OF COORDINATION ACTIVITIES

2.1    Meetings Held
The following meetings were held over the course of the Thematic Network.

Meeting Title       Date(s)             Location           Purpose
Kick-off meeting    13 /11/1998         WS Atkins - UK     Initiation of the Network and
                                                           definition of first year activities.
NSC Meeting         19/01/1999          Sirehna - France   Network steering committee meeting
                                                           to monitor progress and develop
                                                           plans for TA workshops
TA Workshops        13-                 Brussels           Thematic Area Co-ordinators
                    14/04/1999                             meetings to review initial half year
                                                           progress and FP5 calls for proposals.
TN Convention       15/04/1999          Brussels           Participation of the Network in the
                                                           Thematic Networks Convention
1st Technical       18-                 Barcelona          Full technical workshop and
Workshop            19/11/1999                             MARNET-CFD AGM.
NSC Meeting         27/01/2000          HSVA -             Network steering committee meeting
                                        Hamburg            to monitor progress and develop
                                                           plans for TA workshops
Spring Meeting      08/05/2000          Sirehna/ECN,       Network Thematic Area Group
                                        Nantes             Meetings and Plenary meetings on
                                                           Network deliverables
2nd Technical       17-                 TU Denmark,        Full technical workshop and
Workshop            18/10/2000          Copenhagen         MARNET-CFD AGM.
TN Convention       31/10/2000          Palais des         Participation of the Network in the
                                        Congress,          Thematic Networks Convention
                                        Brussels
NSC Meeting         11/06/2001          Chalmers           Network steering committee meeting
                                        University         to monitor progress, discuss budget
                                        Gothenburg         and the AGM in Crete.
3rd Technical       17-                 Crete Maritime     Combined TN Congress and
Workshop            19/10/2001          Technology         MARNET-CFD AGM and Technical
                                        Multi-             Workshop
                                        conference.
NSC Meeting         24/05/2002          London             Network steering committee meeting
                                                           to monitor progress, discuss budget
                                                           and related management matters.
Surface             3-6/06/2002         Valencia           Participation in the Conference on
Transport                                                  Surface Transport Technologies for
Technologies for                                           Sustainable Development, Chair –
Sustainable                                                Virtual Ship-Product Simulation.
Development
4th Technical       20-                 Haslar             Final technical workshop, AGM and
Workshop            21/03/2003          Hydrodynamic       FP6 meeting.
                                        Facilities, UK


WS Atkins Consultants – Co-ordinators                                                    Page   6
MARNET-CFD Final Report and State
Of the Art Review

The principal meetings of the Network were the annual technical meetings and AGM, held
over 2 days with accompanying Network Steering Committee meetings. The main aims of
these events were to assess developments in the MARNET-CFD deliverables, allow the
Thematic Area Groups to meet, provide opportunities for FP5 project groups to develop plans
and to have technical presentations of developments and applications of CFD.
The four meetings that were held were well attended, with over 50 participants on all
occasions. The peak attendance was achieved in the third annual meeting in Crete, which
was an event shared with other maritime TNs, SAFER-EuRoRo, CEPS and PRODIS. The
MARNET-CFD technical sessions were open to organisations outside of the main group of
participants, and this led to wider dissemination of results and participation in FP5 project
teams. The technical proceedings of these meetings can be found on the MARNET-CFD
web site.
The MARNET-CFD Network was also active in promoting and attending the annual
maritime Thematic Network conventions and latterly, the conference on Surface Transport
Technologies for Sustainable development. A session on technical developments was chaired
by the Coordinator and papers presented by various of the participants..
Overall, it can be concluded that the regular meetings of the Network in a conference format
were successful in achieving their aims, and gave MARNET-CFD the desired profile as,
fundamentally, a technical Network aimed at coordinating and advanc ing the state of the art.

2.2    Framework Programme 5
During the progress of MARNET-CFD, FP5 was commenced and projects in CFD for
maritime technology established. Certain of these projects were established with CFD as the
core technology to be developed. For others, CFD was seen as a means to an end, but
advances made in terms of developing expertise and confidence. The key projects of interest
in this respect are:

Project          Coordinator Outline
FANTASTIC        Fincantieri     Functional design and optimisation of ship hull forms,
                                 integrating CAD, CFD and computer based optimisation
                                 techniques.
FLOWMART Strathclyde             Development of practical tools for quantifying wake and
         University              wash effects using CFD.
OPTIPOD  Chantiers de            The development of design guidelines for podded propulsion
         l’Atlantique            systems, where CFD is one of a number of tools being
                                 applied.
NEREUS           University of   Development of tools and design methodologies to improve
                 Strathclyde     the survivability of Ro-Ro vessels, where CFD techniques for
                                 flooding are being used.
EFFORT           MARIN           Full scale hull flow measurement to derive flow velocity field
                                 data to support CFD validation.
EXPRO-CFD        WS Atkins       Development and validation of techniques to integrate CFD
                                 with offshore engineering hydrodynamic design tools.
SAFEFLOW         MARIN           Development of techniques, including CFD, to improve
                                 design for green water loading on FPSOs.
Leading Edge     VTT             The development of tools and techniques for the design and
                                 prediction of propeller cavitation.


WS Atkins Consultants – Co-ordinators                                                  Page   7
MARNET-CFD Final Report and State
Of the Art Review



At the time of writing, these projects are still underway and have not as yet reported their
results or deliverables.
During the early stages of MARNET-CFD it was proposed by the Commission that an
additional task, to form a “cluster” of these projects around the theme of CFD, be taken on by
the Network. At the same time, the Networks CEPS and PRODIS were active in the same
way, developing ideas for clustering and coordination across all FP5 maritime technology
projects. It was agreed therefore that MARNET-CFD would support the technical
dissemination of the results from these projects, particularly through the web-site and
MARNET-CFD database.
Although this was a generally agreed objective, few of the above projects have altered their
exploitation or Technological Implementation Plans (TIPs) to reflect this. Also, as the
projects are not yet fully reported, it has not been possible to assess the suitability of
experimental data or results in terms of their suitability for the site. This therefore represents
a gap in the potential benefit of the MARNET-CFD Network deliverables, to be addressed.
Throughout the period of the MARNET-CFD TN, there was good coordination with the
CEPS and PRODIS Thematic Networks. Meetings of these Networks were regularly
attended by the coordinator and certain members of the MARNET-CFD steering committee.
These Networks were particularly focussed on supporting the development of ideas for the
Maritime Industries Forum and COREDES, and were instrumental in the development of the
Maritime Industries Master Plan for Research and Development. There was direct input from
MARNET-CFD into this process, something which will be continued in part at least through
the coordinators involvement in the FP5 networks ERAMAR and ERASTAR.

2.3    Framework Programme 6
The development and initiation of Framework Programme 6 has coincided with the final 18
month period of MARNET-CFD. The new instruments, Integrated Projects and Networks of
Excellence have been examined by the MARNET-CFD steering committee with respect to its
role in supporting the development of new projects.
An expression of interest for an NoE was developed around the idea of continuing the
activities of MARNET-CFD, but with the inclusion of experimental work (MARNET-
CEFD). The coordinator received many requests for more information one the EoI had been
entered in the commission database. However, this idea has not been taken forward into the
work programme for FP6.
More recently, and in collaboration with the EVIMAR Network through Principia Marine,
and the hydrodynamic research organisations MARIN, SSPA, and HSVA, and Integrated
Project proposal for a virtual testing basin is being developed. The first full meeting of the IP
steering committee was held at the final MARNET-CFD meeting in March 2003, which also
featured a workshop for all participants in the Network, to support the development of
technical ideas.
The MARNET-CFD Co-ordinator will continue to participate in the work of the IP steering
committee, and provide links to members of the Network until suitable mechanisms exist
through the IP.




WS Atkins Consultants – Co-ordinators                                                     Page   8
MARNET-CFD Final Report and State
Of the Art Review

Finally, as an active participant in the ERAMAR and ERASTAR TNs, that will be running
throughout the FP6 programme, the Coordinator has the chief role of preparing the annual
review of the state of the art in marine CFD – and in particular RANS methods. In this way it
is hoped that the exploitation of the results and deliverables from MARNET-CFD can be
continued until a suitable opportunity exists to seek further funding for a follow on Thematic
Network.

2.3      Technical Activities in CFD
Throughout MARNET-CFD, the key coordinative activities revolved around technical issues
in the development, validation and application of CFD in the maritime sector. These
technical activities allowed interactions between participants which otherwise may not have
developed (particularly due to competitive pressures). Nevertheless the activities were in
themselves of direct benefit. The main technical activities in CFD carried out under
MARNET-CFD were:

•     Reviewing the state of the art in marine applications of CFD.
•     Surveying opinions among the marine CFD community regarding the state of the art and
      capabilities of the available tools.
•     Establishing the main gaps in capability limiting the wider use of CFD in the marine
      industry.
•     The preparation of best practice guidelines in the application of CFD to problems in
      marine hydrodynamics.
•     The development of a database of experimental and computational results.
•     The definition of so-called “Grand Challenge Problems” to help stimulate RTD projects.
•     The development of the MARNET-CFD web-site.

The achievements of the Network in each of these areas are described in more detail in the
following sections of this final report.




WS Atkins Consultants – Co-ordinators                                                 Page   9
MARNET-CFD Final Report and State
Of the Art Review

3.0    DELIVERABLES

3.1    Introduction
Progress in the development of the key deliverables during the fourth year of MARNET-CFD
has been satisfactory, although delayed by the funding issues described later. The main items
of concern to members of the network are the best practice guidelines, the database of
computational and experimental results, and the web-site from which these deliverables and
MARNET-CFD events can be accessed. Work to update the European state of the art review
first reported at the commencement of MARNET-CFD has also been started.

3.2    Best Practice Guidelines
Within MARNET-CFD, we are trying to develop procedures to guide the application of CFD
to maritime problems through providing two complementary sources that are:

1. General best practice guidelines (BPG) for marine CFD
2. Application specific guidance aimed at dealing with physical phenomena associated with
   particular problems, for example in cavitation.

This format matches closely that of the QNET-CFD Thematic Network as discussed in the
last annual report. The best practice guidelines used in QNET-CFD are those of the
ERCOFTAC Industrial Advisory Committee. In collaboration with QNET-CFD and
ERCOFTAC, the MARNET-CFD best practice guidelines have been developed as the first
example of industry specific best practice guidelines in CFD. There are, however, a number
of important differences between the MARNET-CFD BPGs and the current ERCOFTAC
SIG guidelines, as was described in the previous annual report.

During the fourth year of the Network, the guidelines have been modified to address a few
issues suggested by participants, particularly with respect to compatibility with International
Towing Tank guidance on matters relating to accuracy. It as expected that further
modifications would be needed following release of the full proceedings of the ITTC
conference held in Venice in September. However, on closer examination, these proceedings
have not added to the best practice suggested, and other than a few editorial changes, the
Steering Committee consider that the document is currently up to date. A copy may be found
in Appendix I of this report.

3.3    Computational and Experimental Databases
Two aspects relating to computational and experimental databases are reviewed, the
development of the database system to be used (via the MARNET-CFD web-site), and the
collection of available data entries for the system. Each of these is now discussed in turn.

The Database System on the Web
The database system has been running during the fourth year with two possible forms of
entry that are:

1. A Star Office platform for the primary database system.
2. A further web-tool that enables summary data available to be viewed independently of
   Star-Office.


WS Atkins Consultants – Co-ordinators                                                  Page 10
MARNET-CFD Final Report and State
Of the Art Review



The benefit of making the databases available also under Star-Office is that this system works
under all possible computer operating systems. Its performance under Windows 95/98 or ME
is not particularly fast, and we consider that the Access version is a better choice here.
However, both options work reliably, and the fact that Star Office is available at no cost
provides a considerable benefit to organisations who otherwise would have had to purchase a
licence for Microsoft Access.

Following discussions at the AGM in Crete, additional work has been done using the Atkins
Pronet system to allow easier viewing of the data available and selective access to files and
groups of data in the database. Given the size and download times of some of the items of
data, this is regarded by many of the participants as a welcome additional facility.

The Data on the Web
During the third year, a considerable amount of work was carried out by Sirehna, who added
to the examples available with contributions from TA co-ordinators. The database now
contains 12 entries which include experimental data for Series 60, and HSVA tanker forms,
links to the Gothenburg2000 databases, Osaka manoeuvring data, water entry (slamming)
data, three HSVA propeller data sets (varying skew), and tank sloshing (PIV) data. These
have been added to during the fourth year with links to the Surrey University led QNET-CFD
and ERCOFTAC databases, and direct entry of data from FP5 projects, including EXPRO-
CFD (vertical cylinder PIV data), Safer- EuRoRo (flooding PIV).

3.4    The MARNET-CFD Web-Site

During the fourth year the site has undergone an extensive update and shift of server. The
latter element is the result of some major changes in WS Atkins computing hardware
provision, but should allow much faster access and greater files storage capacity as a result.

The best practice guidelines have been released on the web-site as a simple Pdf file as noted
above. The final intention is to make this and the application procedures available in HTML,
to as to enable a more sophisticated linking of technical guidance, source references, and
illustrative examples.

Thematic Area Groups are making steady progress in preparing their own web pages. They
should be complete in time for the release of the Application Procedures part of the best
practice guidelines.

Links are now being established with each of the FP5 project web sites of relevance to
MARNET-CFD, as this is now seen as the most appropriate way to incorporate information
of this kind into the MARNET-CFD system.

Finally, an alternative URL which should allow access to certain parts of the MARNET-CFD
site via the world-wide web will be established at the next release. Being part of the iPronet
system, access is currently shielded from searches on www sites, which has been both a
benefit (to security) and a problem (in terms of dissemination) to the network.




WS Atkins Consultants – Co-ordinators                                                 Page 11
MARNET-CFD Final Report and State
Of the Art Review




4.0    STATE OF THE ART REVIEW

This section covers the period from 1999, the completion of the first state of the art review, to
the present day. The aim is to highlight advances made since then, rather than to give an
exhaustive review, and to focus on developments or changes that demonstrate whether CFD
techniques are being used with greater success in design. Nevertheless, we provide in each
area a brief overview of the techniques used reflecting the first MARNET-CFD SOA review
for completeness and to avoid the n     eed for the reader to refer back to the original review.
Each sub-section deals with the defined thematic areas, and the final sub-section reviews the
results of the state of the art questionnaire carried out during the final year of the network.

4.1    TA1 – Ship Performance

The area of ship performance primarily covers steady hull flow and the prediction of
resistance components in deep and restricted waters. There is some interaction with TA3
with respect to predicting the effect of propellers and rudders on the performance of the
vessel. As was the case in the first state of the art review, inviscid flow panel methods
remain in use for many applications in wave resistance and hull form optimisation, and so are
treated here separately from viscous flow modelling (RANSE) methods.

Inviscid Flows - non-linear free surface and wave resistance
Inviscid models of steady free-surface flow around ships are used to predict wave resistance,
wave patterns (including wash effects), and for the optimisation for minimum wave
resistance of certain hull form features such as bulbous bows. They may also be used to
calculate lift and induced drag on foil systems, rudders and stabilisers. Figures 1 and 2 below
illustrate these common applications with firstly, results achieved at MARIN using the
program RAPID for wave pattern and resistance, and secondly, and example of how such
models may be used to work on the optimisation of particular hull form features such as in
this case, a bulbous bow.




Figure 1. Typical computed wave pattern using       Figure 2. Example of local free surface flow
the RAPID program at MARIN                          calculations      using         SHIPFLOW




WS Atkins Consultants – Co-ordinators                                                    Page 12
MARNET-CFD Final Report and State
Of the Art Review

There are three primary forms of inviscid flow model:
   1. Pure panel methods,
   2. Combined panel/vortex lattice methods,
   3. Euler solvers.

The first two of these provide the additional constraint of irrotationality on the flow, being
primarily forms of boundary element method for the Laplace equation. The third method can
be regarded as a step toward the Navier Stokes Equations, being based on coupled solution of
the mass and momentum conservation equations in the absence of viscous effects.
The chief interest in the calculation of wave resistance is however the form of free surface
boundary condition applied regardless of which of the above representations of the flow are
used. It is now well understood that linear or simple linearisations of the full free surface
boundary conditions (which are equations describing the pressure and kinematics of the free
surface), are inadequate in most practical cases. Research in this area has therefore focussed
on the development of fully non- linear representations of the free surface, wherein both the
mathematical description of the boundary condition, and its application on the actual water
surface, taking into account variations in above water hull form, dry transom sterns, sinkage
and trim, are used.
For the first review, a number of examples of computer models to solve this problem in non-
linear wave resistance were mentioned, indicating that the technique had reached maturity
(see: Jensen 1989, Raven 1993, 1996 and 1998, Kim et al. 1994, Janson 1997, and Hughes
1997). All of the techniques used require iterative solution for the location of the free
surface, and typical modern desk-top PCs are adequate for these applications. The chief
limitations identified in the first MARNET-CFD SOA were the consistent representation of
wet transom sterns and the representation of bow breaking waves.
Since the first SOA review, there have been few new technical developments to these
techniques, which is not surprising given the previously noted level of maturity. Some efforts
have been directed at improving solutions in the far-field by using mixed Havelock/Rankine
based Greens’ functions, which implicitly give wave pattern solutions in the far field and
hence allow reductions in the total area of free surface modelled, and improvements in
computing time. Examples of this research were reported by Scragg (1999 – linear solutions,
2001 – non- linear extensions), and Yang et. al (1999).
In MARNET-CFD, a number of papers covering the use of these potential flow methods
within the design optimisation process have been presented, most notably by Vissoneau
(Crete - 2001) and Marzi (Haslar - 2003). These may be found on the MARNET-CFD web-
site. They demonstrate a growing confidence in the use of this form of CFD, particularly
with respect to the definition of hull form, in a domain previously exclusive to physical
testing.
A key element of the future use of panel methods lies as a tool for rapid hull form
optimisation, particularly for applications where wave resistance dominates (e.g. high speed
displacement craft) or factors such as wash minimisation are important within the design
process. The computation of wave resistance alone will increasingly be seen as a minor
benefit as RANSE solvers using VOF, or similar methods to represent the free surface, gain
acceptance.




WS Atkins Consultants – Co-ordinators                                                 Page 13
MARNET-CFD Final Report and State
Of the Art Review

Turbulent flows – RANSE and advanced methods
The first MARNET-CFD state of the art review noted the close relationship between the use
of CFD in wider industry and advances in marine applications. As will be identified in the
section dealing with the state of the art questionnaire, the increase in the use of CFD in
marine applications involves growth in the application of general purpose CFD codes, and
hence the benefits of wider advances have been exploited in the maritime sector. As with the
original SOA review, we consider here the main technical concerns of the ship
hydrodynamicist in grid generation, free surface flow, turbulence modelling, solvers,
validation and optimisation techniques.

Grid generation
During the first review, we identified the following types of grid structure in use,
   1.   Single block structured grids,
   2.   Fully unstructured grids,
   3.   Multi-block structured grids,
   4.   Hybrids of 2 and 3,
   5.   Overlapping or Chimera structured grids,
and described each in some detail. The outcome of the Gothenburg 2000 workshop allows us
to now take a view on the current popularity of each of these methods and other more recent
developments.
Firstly, it was observed that there was continued regular use of single block structured grids.
Many un-appended ship hull forms are, by appropriate use of numerical mappings, well
suited to this approach. The key benefits are those of simplicity, computational speed, and in
many cases high levels of boundary layer grid quality over most of the hull surface.
Multi-block structured grids are also in greater use, particularly where commercial CFD
codes (e.g. CFX4) offer this as a default methodology. In the USA, this approach is still
followed (Gothenberg 2000) along with finite analytic solver methods. However, although
structured multi-block offers many advantages (retaining good quality grids and natural
approaches to domain decomposition for parallel computing), they are not as flexible as
unstructured or hybrid approaches for highly complex geometries.
Fully unstructured grids were also used in G2K, albeit in the sense of approaches 3 and 4,
wherein particular regions of the flow domain are populated with regular, structured, finite
volume cells in order to take advantages of efficiency and greater numerical accuracy. Fully
unstructured grids are often subject to high overheads in data storage and solver speed. In
addition, certain finite volume cell types are not best suited to boundary layer flows (e.g.
tetrahedra).
It was noted in the previous review that a combination of structured multi-block and
unstructured methodologies would be ideal but that no suitable solvers were in existence. To
a large extent, this deficiency has now been overcome. Unstructured methods have now been
advanced to the stage where solvers are in use for cell volumes of almost arbitrary geometry
(i.e. general polyhedra). This allows very complex geometries to be modelled, but also
enables mixtures of structured and unstructured grids to be used together since issues relating
to the interfaces between regions can be solved using these arbitrary cell volumes. The finite
volume solvers used are formulated for their versatility rather than efficiency, but advances in
computing speed and parallel architectures have more than compensated for this. The main



WS Atkins Consultants – Co-ordinators                                                   Page 14
MARNET-CFD Final Report and State
Of the Art Review

concern with such methods relates to the formal order of accuracy achieved, not simply by
the originators, but by those using these methods in the commercial environment.
Finally, the previous state of the art reported on growing interest in the development of
overlapping multi-block grid structures, or Chimera grids. Specific tools available for this
purpose were referenced (Petersson, 1998, Larsson 1997, Chen & Huang 1998, Lin et. Al,
1998, Korpus et al., 1998). The commercial CFD code SHIPFLOW is now using the so-
called Chapman solver (Larsson, 1999), allowing these types of calculations although at the
present time, this appears still to be the only example of this technique specific to maritime
applications. There is a similar capability in general purpose commercial CFD codes wherein
so-called un- matched and overlapping grids may be used. These offer particular advantages
where relative grid movement is required (e.g. moving propellers). The main difference
appears to be in the degree of overlap of the grids, which is lower for the commercial tools
that for the full Chimera grid methodology.
In summary, practitioners of marine CFD continue to apply the latest developments in mesh
generation, and where appropriate, make use of the available commercial tools. Strategies for
grid generation appear now to be focussed on three approaches:
   1. Single block structured grids using numerical conformal mappings for single un-
      appended hulls, quick design studies or form optimisation.
   2. Fully unstructured grids, though used in such a way as to contain mixed cell types so
      as to balance considerations of accuracy with practicality.
   3. Though still in development, Chimera grids still offer many practical advantages and
      are being actively pursued.

Free surface flow
The presence of the free surface in hydrodynamics provides a major departure from
conventional CFD applications. The need to represent this fluid interface accurately presents
a considerable challenge, not least because its behaviour can vary considerably within the
computational domain, and as a function of hull form and speed.
The two key free surface flow phenomena which the CFD model would ideally be able to
represent are (a) surface wave generation and propagation and (b) breaking waves. At the
time of the previous state of the art review, the most commonly employed techniques in
steady ship flows were:
   • The Volume-of-Fluid or VOF Method (for RANSE solvers),
   • Free-surface fitting or grid adaptive techniques (most commonly applied to panel
       methods).

Since then, the level set method has been developed for RANSE solvers, and is seeing
application across a number of areas, and a hybrid formulation known as Coupled Level Set
Volume of Fluids (CLSVOF).
Descriptions of the first two approaches can be found in the first state of the art review, and
in a number of texts on general CFD.
Some advances have taken place in the VOF method since the original MARNET-CFD
review, most notably with the introduction of High Resolution Interface Capturing (HRIC)
due to Peric et al (2000). Examples of the use of HRIC can be found in Azcueta et al. (1998)


WS Atkins Consultants – Co-ordinators                                                  Page 15
MARNET-CFD Final Report and State
Of the Art Review

and Woodburn and Gallagher (2001), reporting on results from the FP5 project Nereus. VOF
methods still represent the most common approach to free surface modelling in RANSE
codes.
The level set method involves solving (numerically), an advection equation for a function (Φ)
representing the normal distance from the interface, and represented by a series of points on
the free surface for which Φ = 0. Discretization of the advection equation is a source of
numerical error and may lead to loss or gain of fluid mass. Schemes to combat this problem
through re- initialisation and other methods have been developed however, and examples of
application can be found by Vogt & Larsson (1999), Sussman & Dommermuth (2000), and
Rhee & Hino (2000). Level set is not yet available in commercial CFD codes, which remain
focussed on VOF methods.
Since the last review, methods to improve accuracy though grid adaptation have made
advances with respect to free surface flow. In the context of EU projects Nereus and
EXPRO-CFD, methods that use a combination of non- linear potential flow (to guide grid
refinement local to the free surface) and RANSE VOF solvers (using HRIC) have
demonstrated both the flexibility needed to model breaking waves and the accuracy required
to propagate waves without excessive numerical diffusion errors.
Finally, the Gothenburg 2000 workshop has demonstrated that VOF methods employing such
techniques are now of sufficient accuracy, when used on appropriate grids, to provide
accurate ship wave predictions. Key to this is to maintain gr id resolution in the far field
above that needed for the viscous or boundary layer/wake part of the RANSE solution, and
appropriate instead to having sufficient grid cells to resolve the wavelengths of interest.

Turbulence modelling
The first MARNET-CFD state of the art review described in some detail the various types of
turbulence models used for steady flow prediction for ship hulls, and this information may
also be found in the MARNET-CFD best practice guidelines.
The modelling of turbulence is possibly the most actively researched subject in computational
fluid dynamics. The current consensus of opinion remains that the tuning of turbulence
models to deal with specific types of flow is perhaps the most practical way forward. Since
the first state of the art review therefore, the majority of the literature and related publications
has been aimed at evaluating and validation of turbulence models for ship flows.
The alternative two equation formulation of Wilcox, known as the k-ω model (Wilcox 1988),
and a more recent variation of it, known as SST, due to Menter (1993), represents the current
state-of-the-art in this approach. Deng and Vissoneau (1996), Watson & Bull (1998), Hino
(1998), and Vissoneau et al. (1999), have all experimented with these approaches and
obtained much improved results over those possible with the classic high Reynolds number
k-ε model. With these approaches, it is possible to solve all equations without the use of wall
functions, and this is itself a significant improvement which could feed through to practical
applications within the next few years.
Perhaps the most definitive inter-comparison of turbulence models for the steady ship flow
problem can be found within the G2K workshop (2001) and summarised in the 2002
proceedings of the 23rd International Towing Tank Conference. These proceedings present
the somewhat enigmatic result that, in certain cases, quite simple turbulence models (e.g.
Baldwin-Lomax or Spalart & Allmaras), when well applied, are able to produce results for
wake and boundary layer which are of comparable accuracy to the more sophisticated


WS Atkins Consultants – Co-ordinators                                                       Page 16
MARNET-CFD Final Report and State
Of the Art Review

Reynolds Stress models. At the same time, results achieved with two equation models using
the eddy viscosity approach appeared to be of more variable accuracy, particularly in the
wake.
In all of these observations, it should be taken into account that it is extremely difficult to
make comparisons of turbulence models unless the calculation are done using the same grids
and solvers, and this was not the case for the G2K workshop. Neither could it be said that the
contributors followed similar or consistent best practice in application of their CFD models,
rather they used their own experience and were subject to the particular limitations of their
own computing resources. Of course, this is a true reflection of real life, and the value of the
G2K workshop and its predecessors is that it gives this realistic view of the state of the art. It
would be interesting to compare the results achieved using the different CFD codes when all
participants are able to follow agreed best practice guidelines.
Solvers
The development of equation solving techniques for RANSE applications in ship
hydrodynamics reflects the trends found in wider applications in CFD. The two key elements
are the approach to the coupling of pressure and velocity (for the purposes of ensuring
continuity or mass conservation), and the form of equation solver used. The most common
approaches to the former remain the so-called SIMPLE formulation and variations thereof,
and the more recent PISO algorithm. However for ship hydrodynamics (and specifically
where aerospace practice is followed), artificial compressibility methods (Cowles &
Martinelli 1998, and Hino 1997) have also been used.
The key area of interest however is the development of methods for solving the resulting sets
of algebraic equations ever more rapidly, and the scaling of computational speed with
problem size. The main areas of research identified in the previous state of the art relate to
the use of multi- grid techniques and the more direct approach of preconditioning the systems
of numerical equations priors to their solution. The former multi- grid methods are now well
established and can be found in use in commercial CFD codes. A variant, known as the
Algebraic Multi-Grid method (AMG), appears to have gained in popularity where large
unstructured grids are used. These methods also have benefits where the grid structure
naturally produces cells of high aspect ratio, such as in boundary layers.
Since the previous state of the art review, there appears to have been little progress with pre-
conditioning methods.
One further new development with solvers is the availability of so-called fully coupled
solvers. This method eliminates the need for an iterative coupling of velocity and pressure
and instead, solves for the three mean velocity components and pressure simultaneously. The
main advantage appears to be a significant reduction in the number of global sweeps required
to satisfy the conservation equations, and improved accuracy. However, the amount of
overall computational effort remains high, and these techniques appear still to take longer to
run in practice.
Optimisation techniques
There is considerable interest in the use of optimisation techniques coupled with CFD in
order to link directly the process of hydrodynamic analysis and design.
At the time of the previous state of the art review, techniques had been developed primarily
for implementation with respect to the wave resistance problem, with some simple empirical
calcula tion of viscous resistance. Typical examples were cited as Janson and Larsson (1996)


WS Atkins Consultants – Co-ordinators                                                     Page 17
MARNET-CFD Final Report and State
Of the Art Review

                                                                                  on-
and Chou et. al. (1998) (using an inverse method). The review also mentioned n linear
programming methods that had been investigated by Hamasaki et. al. (1996), and by Tahara
et. al. (1998), in combination with a non- linear panel code for the wave field and RANS
calculations for the stern flows.
It was concluded at the time that the use of such optimisation techniques in hull form design
remained some considerable way off.
Since the first state of the art review, significant amounts of new work have appeared in the
literature, and some classification of the methodologies has now appeared. Two general
approaches have developed, being based either on derivative free methods (e.g. genetic
algorithms (GA)), or derivative based methods (using finite differences (FDM) or sensitivity
equations (SEM)).
Within MARNET-CFD, the results of deign optimisations using GA methods were reported
by Vissoneau at the 3rd Workshop in Crete (2001), and at the final workshop. Recent
examples of the FDM approach may be found in Valorini et al (2000) for the optimisation of
a tanker hull at 16 Kts, and Peri et al (2001a), also for a tanker (but with added experimental
data for comparison).
In the above examples, optimisation is a carried out with respect to a single parameter,
usually a design speed, and frequently to minimise resistance. Perhaps of more interest to the
overall design is so-called multi-disciplinary optimisation, which as its name suggests, seeks
to balance a number of different aspects of the design problem. An interesting example of this
can be found in Peri et al (2001b), in which resistance and sea-keeping parameters are
combined in a multi-disciplinary optimisation.
As the result of this revised review of the state-of-the-art, it now appears that tools for design
optimisation are maturing, and that it is feasible to use some of these techniques in practice,
although the computational effort required puts much of this work beyond the capability a
typical design office. Further research efforts should be directed at finding practical ways to
bring this technology to the market place.


4.2    TA2 – Unsteady Ship Flows

Within MARNET-CFD, the subject of unsteady flows covered two areas of interest to the
hydrodynamicist. The first involves the response of the vessel in waves, and includes both its
rigid body response and structural behaviour. The second concerns ship manoeuvring and the
ability to simulate or predict the flow around and fluid forces and moments acting on ship or
its control surfaces as the result of course changes.

In certain areas, there is some overlap where manoeuvring in waves is of interest, such in the
modelling of phenomena such as broaching. For the purposes of this update to the state of
the art however, we shall review each of these areas separately as before.

Vessel responses in waves
It was reported in the first state of the art that the application of CFD to predicting ship
motions and the hydrodynamic loads imposed on the hull structure was relatively new, and it
remains the case that full unsteady RANSE solutions of ships in waves is rare and un-
validated.



WS Atkins Consultants – Co-ordinators                                                     Page 18
MARNET-CFD Final Report and State
Of the Art Review

As described in the previous state of the art, traditional approaches to calculating ship
motions and loads, developed in the late 1950s and early 1960s, (Salvesen, Tuck & Faltinsen,
1963) were based upon the concepts of classical hydrodynamics. These methods led to so-
called “strip theory” computer programs, working primarily in the frequency domain, to give
solutions of the linearised 2 degree of freedom (coupled pitch and heave), equations of
motion. Later, the development of 3 dimensional approaches for wave diffraction problem
the frequency domain provided further generalisation, such that all 6 degrees of freedom of
ship motions could be addressed. These techniques produce results which are efficient and
sufficiently accurate for engineering purposes, over a wide range of vessel types, speeds and
environmental conditions. Moreover, probabilistic approaches to load prediction in spectral
seas, aimed at calculating long term loading statistics, are a key requirement in naval
architecture and CFD simulations are not well suited to such applications. There has n     ot
therefore been a significant driver behind the development of RANSE based CFD for basic
sea-keeping calculations.
As previously reported, there are two general areas where CFD based simulations methods
have received attention in sea-keeping however, and these are:
   §   Large vessel motions and extreme loads in steep waves, and slamming, for which the
       linear theories break down,
   §   Roll motions, damping and the actions of active fin stabilisers and bilge keels.
For the first of these, the literature shows that CFD has been applied to both quantifying local
loads and flow phenomena, and complete non- linear ship motions. The former particularly
includes slamming, water impact, and the shipping of green water, to which unsteady
RANSE methods have been applied. The prediction of large scale ship motions is essential to
the calculation of extreme loads and appears to have been approached mainly through the use
of non- linear inviscid and potential flow methods.
The first sate of the art review described the applicatio n of CFD to slamming problems
following both inviscid and RANS solution approaches.
The former is primarily restricted to 2D calculations of the flow and loads resulting from the
impact of wedges on still water (Zhao & Faltinsen, 1993), and more recently axi-symmetric
bodies and those with chines (Zhao, 1996). Typically, boundary element methods are
applied, solving for the flow field local to the wedge, and the near-by free surface. Similar
boundary element techniques are used in the prediction of bow flare slamming loads (Wang
et al. 1996) and Fontaine and Cordier (1997), with some good comparisons with experiments.
Coupled structural response and hydrodynamic impact simulations have been reported by
Kim et al. (1996), Xu, Troesch and Peterson (1998), and Wu & Greaves (1999). The
majority of these simulations have thus far only been compared qualitatively with results
from rigid body tests however.
Various applications of VOF based RANS solvers to the water impact problem have been
reported also, by Muzaferija & Peric (1998) for example, and for bow flare by Schumann
(1998). The chief advantage of these methods is that the VOF technique can deal effectively
with the complex free surface shapes which result in these simulations. In three dimensions,
Sames et al. (1998) have demonstrated the application of VOF techniques full ship forms in
regular waves, producing credible overall load distributions. Since the first state of the art
review, no additional reports of the application of CFD specifically to slamming are worth
remark, and it is not clear whether there is sufficient confidence in CFD as yet for these
applications.


WS Atkins Consultants – Co-ordinators                                                   Page 19
MARNET-CFD Final Report and State
Of the Art Review

Within the FP5 programme, it is worth noting that the SAFEFLOW project has demonstrated
the use of the COMFLOW CFD programme the simulation of green water impact on FPSOs.
COMFLOW uses VOF techniques and solution algorithms taken from the early research
carried out at the Los Alamos Laboratories that led to the commercial FLOW-3D code (Hirt
et al). COMFLOW does not include the simulation of the air above the free surface, neither
does it include viscous or turbulent flow effects. The results of these studies are to be
reported in early 2004 when the SAFEFLOW project ends.
The other area where CFD is applied to sea-keeping and loads prediction is that associated
with the simulation of large motions and steep waves. These methods seek to simulate the
behaviour of the vessel by calculating the complete fluid loading and its distribution over the
instantaneous wetted surface at each instant. Even with current computer technology, this
represents a considerable challenge.
The current state of the art therefore is to carry out such calculations using inviscid flow
models. Perhaps the most well reported example of this is due to Sclavounos and Nakos
(1993) with their SWAN code. This model is based on using a 3D Rankine source based
boundary element method, with the hull and free surface discretised into quadrilateral panels.
A fully non- linear formulation of the free surface kinematic and dynamic boundary
conditions are applied on the free surface. The technique is able to predict well the
hydrodynamic coefficients and wave loads on a ship at forward speed. Similar techniques
have been developed by Beck et. al. (1994), Scorpio (1997), and Gallagher (1995). Most
recently, the WASIM code, a development of the original SWAN methodology by Det
Norske Veritas, has been made commercially available. It is also the subject of work by
DNV within the EXPRO-CFD FP5 project, in which it is being integrated with unsteady
RANSE CFD methodologies. This too is due for reporting in early 2004.
As noted above, RANS calculations of the complete ship sea-keeping problem remain a
considerable challenge. Numerous 2D simulations of the roll motions of ship sections have
been carried out with the aim of predicting the viscous contribution to roll damping (Graham
et. al. 1992; Korpus and Falzarano, 1997; Sarkar 1999). Fully 3D RANS simulations of ship
sea-keeping have been most recently reported by Wilson et. al. for the well known Wigley
hull form in head seas. It seems likely that in the near future research will continue in this
area, particularly with respect to solving the diffraction problem in the RANSE solver
environment.
In summary therefore, the current state of the art in CFD for sea-keeping is much the same as
it was for the first MARNET-CFD report. Potential flow BEM approaches for the non- linear
simulation of motions and loads are in active use for conventional hull forms, and
commercial tools have become available to this end. Linear diffraction methods and strip
theory are still in extensive use however in industry. RANSE methods remain for research
only. High speed craft or other novel hull forms appear not to have been addressed directly.
Both BEM potential flow and RANSE calculation methods have been used for studies of
detailed elements, such as slam loading or roll damping in 2D.
For the future, it would appear unlikely that unsteady RANSE methodologies will be deemed
suitable for general use in design and analysis. It is more likely that some hybrid approach
based upon an integration of non- linear potential flow and RANSE methods will be
developed (as in the EXPRO-CFD project for offshore applications) for ships with forward
speed. As this revised state of the art review is in preparation, an FP6 Integrated Project
directed at the development of a so-called Virtual Testing Basin is being prepared also and
this may well adopt such an approach to sea-keeping and ship motions.


WS Atkins Consultants – Co-ordinators                                                  Page 20
MARNET-CFD Final Report and State
Of the Art Review



Manoeuvring
As described in the previous state of the art calculation of manoeuvring forces and moments
using CFD techniques has made considerable advances in recent years with the ever
increasing computing power available making unsteady flow simulations more practical.
Again, both potential flow solvers and RANSE techniques have been developed. In addition,
researchers have chosen to study both complete ship hull flows and model the behaviour of
individual appendages. Each of these is dealt with in turn below, following a short word on
traditional approaches to manoeuvring, and the need for improved predictive capabilities.
The important forces and moments in manoeuvring arise from inertial, viscous and lifting
effects. The manoeuvring derivatives are even now mostly derived by carrying out specific
forms of experiment in which a model is made to carry specific motions in a towing tank, and
the forces and moments acting on it are measured. Such experiments are quite involved
however, since there are a large number of manoeuvring derivatives required for the
equations of motion, and frequently recourse is made to empirical formula and past databases
for certain of the terms.
The most recent ITTC report (Proceedings 23rd ITTC) has provided a number of examples of
research into the use of CFD methods for manoeuvring, falling into the area of inertial and
lifting effects (potential flow models), RANSE bases methods, and work on propulsors,
appendages and their interaction.

Potential flow CFD methods
Potential flow methods are only appropriate for the modelling of inertial and lifting effects in
manoeuvring. The former is straightforward, since classical hydrodynamics deals easily with
the computation of added mass or rotational inertia effects.
The prediction of lift presents more of a problem. There are numerous publications for
idealised geometries, such as surface piercing wing sections, flat plates and the Wigley hull,
produced by Landrini & Campana (1996), Zou (1996), Zou & Soeding (1995), Wellicome
et.al. (1995), Kose et. al. (1996 & 1997), using various combinations of lifting surface and
slender-body theory. These are useful publications in so far as they confirm the relevant
theories for the hull forms and geometries that they are intended for, and give good overall
predictions. More recent examples and extensions can be found in the work of Karasuno et
al (2000), Kijima & Kishimoto (1999) and Kijima and Kaneko (2000).
These models are based on slender body theory and are not applicable to full hull forms such
as VLCCs that present the greatest challenges in manoeuvring. None of the above deal with
the interaction of trailing vortex sheets shed from appendages with the hull, or viscous
effects.
Calculations with panel methods of the promise of being able to deal with full hull forms, but
are often restricted to quasi-steady simulations of ships in the steady, oblique towing,
situation. Examples of such calculations can be found in the work of Ando et. al. (1997) and
Nakatake (1998), wherein in both cases, surface panel methods have been developed and
tested on VLCC hulls. Some reasonable results were achieved for side force prediction, and
for small yaw angles, yaw moments were also good. However, for large yaw angles (above
10 degrees), poor agreement was demonstrated for yaw moment, with the most likely reason
being a poor representation of flow separation at these angles of attack.



WS Atkins Consultants – Co-ordinators                                                   Page 21
MARNET-CFD Final Report and State
Of the Art Review

So in summary regarding these potential flow methods, they can more easily be verified for
slender or classical lifting geometries, where the results are less sensitive to the absolute
accuracy of the resulting vortex field. This is not the case for full hull forms however, where
significant amounts of tuning are required in order to achieve reasonable agreement with
measured manoeuvring forces and moments.

RANSE based methods for manoeuvring
One clear way to remove the above difficulties in the prediction of manoeuvring forces is
through the application of RANSE methods which in principle should capture the complete
details of the flow field and vortex structures, and their influence on the forces and moments
on the hull. The main difficulties in achieving such improvements lie with the ability of the
RANSE techniques to predict flow separation and capture properly the intensity of the
longitudinal vortices present in these flows. As noted above in the case of steady flows, the
standard k-ε method, when used with wall functions, achieves only qualitative accuracy for
flows of this kind.
Nevertheless, good agreement with experimental results for certain well established test cases
(a Series 60 hull, the Esso OSAKA, and the SR221 tanker forms) has been achieved using
RANSE methods.         Publications of particular interest are those of Alessandrini &
Delhommeau (1998), Cura Hochbaum (1998), Tahara et. al. (1998), Berth et. al (1998),
Ohmori et. al. (1996) and Makino & Kodama (1997). Going one step further, Sato et. al.
have coupled rigid body equations of motion for a ship with a RANSE model to give a
complete simulation.
Most recently, Zhang and Wu (2001) have applied the finite analytic method to calculation of
hull forces in oblique flow, and Yamasaki et al. have described the application of CFD
methods to an appended hull at the early stages of design.
This would therefore appear to be a promising area for the development of RANSE solvers in
ship hydrodynamics, although as ever, there is some need for further validation, mesh and
turbulence modelling sensitivity studies, before confidence and trust can be established.

Methods for dealing with hull/propeller/rudder interaction
Both RANSE and potential flow based methods have been used in research and practical
design of rudders, thrusters and podded propulsion systems.
The previous state of the art report cited the work of Molland & Turnock (1996, 1998) and
Wang et. al. (1994) in using potential flow and lifting surface methods for
hull/rudder/propeller interaction, and the viscous flow models developed by Hinatsu (1995)
and Suzuki (1996) and Chau (1998).
More recently, Takada & El Moctar (2000), and El Moctar (1999), have demonstrated the
application of CFD multiblock RANSE methods to the calculation of rudder/propeller
interaction. The performance of the rudder and staff effects are reported as well predicted.
El Moctar also presented a complete investigation of hull/propeller/rudder interaction using
RANSE methods (2001a and 2001b). These results were also presented at the MARNET-
CFD Annual Meeting in Crete (2001) and can be found on the MARNET-CFD web site.
It would appear that the use of unsteady RANSE methods for hull/propeller/rudder
interaction is well established, but requires considerable care by users. The computational
methods and best practice (in the sense of grid resolution, choice of turbulence models etc.)



WS Atkins Consultants – Co-ordinators                                                  Page 22
MARNET-CFD Final Report and State
Of the Art Review

are similar to those developed for aerodynamics applications and similar rotating flows, and
so there is considerable opportunity for validation.

We conclude therefore that the technology is in a position where it is ready to be exploited in
the design and optimisation process, albeit that considerable CFD expertise is required if
reliable results are to be achieved.




WS Atkins Consultants – Co-ordinators                                                  Page 23
MARNET-CFD Final Report and State
Of the Art Review

4.3    TA3 – Propulsors

Introduction
This section reviews the state-of-the-art of CFD in the domain of ship propulsors. The
emphasis in this section is on Reynolds-Averaged-Navier-Stokes (RANS) equation codes, the
turbulence models used, and the correlation between calculated and measured results. The
correlation comparisons are given in table form, if only possible. The appendix part of this
section summarises the propulsor CFD work in the EU funded OPTIPOD, FAST POD, and
LEADING EDGE projects. Regarding CFD methods for extreme off-design conditions and
optimisation techniques the reader is referred to the First Annual Report of MARNET-CFD.

The reports of the technical committees of the 21st, 22nd, and 23rd International Towing
Tank Conferences form detailed high quality state-of-the-art reports in the domain of ship
hydrodynamics (ITTC, 1996, 1999, 2002). In addition to the committee reports printed in the
proceedings, the 22nd ITTC Propulsion Committee and the 20th ITTC Propulsor Committee
organised workshops of comparative calculations. In the 1998 Workshop both steady state
Reynolds-Averaged-Navier-Stokes (RANS) equation codes and unsteady panel method codes
were compared (Gindroz et al., eds., 1998). In the 1992 Workshop panel method codes were
compared (Koyama, 1993). The proceedings of the two workshops contain complete
presentations of the participants and model scale or ship scale experimental results of the test
propellers. Additionally, Streckwall (1999, 2003) and Szantyr (2000) have published
propulsor CFD overview papers recently.

Panel Methods

Panel methods for conventional propellers. Potential based panel methods are most widely
used type of panel code. In the 1990’s the development emphasis of vortex wake models has
been in a more accurate modelling of the tip region flow and in the inclusion of simple
viscous considerations in the model. Panel methods have been extended to problems of the
unsteady performance and cavitation prediction of propellers. The recent improvements in
panel method codes have resulted in good agreement between the measured propeller and
predicted performance and pressure distributions on the blade for most types of propellers
(e.g. Lee & Kinnas, 2001). Panel method predictions of the details of slipstream velocity field
are not accurate (ITTC, 1999, p. 39 & 44).

The key for the accurate prediction of unsteady hull pressures is accurate prediction of time
variation of cavity volume (e.g. Young & Kinnas, 2001). The present day computational
methods are not yet fully validated for practical use. In some cases, the predictions agree very
well with the measurements. But significant discrepancies are seen in other cases.

It is often observed in the experiments that the trailing edge of the blade cavity of a highly
skewed propeller is lifted away from the blade surface near the tip. The present day
computational methods do not take this fact into account in their blade modelling. Also, the
effect of tip vortex cavity would not be negligible. More accurate modelling of the tip vortex
cavity should be developed.

Panel method applications for unconventional propellers. Both steady and unsteady vortex
lattice methods are capable of predicting the overall performance of a contra-rotating
propeller set at the design point with at least satisfactory but, in some cases, with good


WS Atkins Consultants – Co-ordinators                                                   Page 24
MARNET-CFD Final Report and State
Of the Art Review

accuracy. Observed cavity patterns on the aft propeller often differ distinctly from predictions
obtained by vortex lattice methods.

None of the published design and analysis methods for end plate propellers includes the
calculation of the interference drag of the tip plate to the blade. All published end plate
propeller design procedures need empirical corrections in order to obtain the specified design
performance.

Panel methods are helpful in the design of propeller boss cap fins, but still require further
developments to obtain accurate performance predictions. The existing design methods
provide practical design methods for propellers with pre-swirl vanes. These devices are
subject to significant scale effects, the Reynolds number of the stators being significantly less
than that of the propellers. In the 1990's and early 2000's panel m     ethod codes have been
extended to the analysis of complicated interaction configurations such as cavitating flow
around a horn-type rudder in the race of a propeller (Han et al., 2001) and podded propulsor
flows (e.g. Szantyr, 2003). Advances have also been made in panel method applications to
super-cavitating propellers and surface piercing propellers (e.g. Young & Kinnas, 2000).

Reynolds-Averaged-Navier-Stokes (RANS) Equation Methods

Prediction accuracy. Reynols-Averaged-Navier-Stokes (RANS) equation codes are capable
of predicting accurately the propeller open water performance and pressure distributions on
the blade at the design advance number. Table 1 shows a sample correlation comparison for a
ducted propeller. The correlation is seen to be good. The accuracy of performance prediction
for off-design advance numbers that are about half of the design advance number is good or
acceptable. The main features and the main variables of the flow field, streamlines and
boundary layer are generally well predicted.

                            Experiment       1157632 cells      144704 cells
                 KTP        0.197            0.197              0.220
                 KTD        0.048            0.046              0.042
                 KT         0.245            0.243              0.263
                 KQ         0.0345           0.0361             0.0418

Table 1. Comparison of ducted propeller performance for the advance number J = 0.5 (Sánchez-Caja
           et al, 2000). FINFLO code with low Reynolds number k-ε turbulence model.

RANS equation codes form an excellent tool for developing scaling methods. Tables 2a and
2b depict calculated viscous resistance coefficients of a tractor type podded propulsor in two
scales (Lobachev et al., 2001). Lobachev et al modelled the propeller by body forces. This
averaging procedure allowed the solutions to be obtained more quickly than by using time-
accurate modelling.

Sánchez-Caja et al (2003) investigated the scale effects of a tractor type podded propulsor
flow by using sliding mesh technique with circumferential averaging. In the calculations 7.5
million cells were used for both model and ship scale. The radial stretching in the two scales
were different. The scaling results of Sánchez-Caja et al differ from those of Lobachev et al
due to different housing geometry.


WS Atkins Consultants – Co-ordinators                                                    Page 25
MARNET-CFD Final Report and State
Of the Art Review



           Model scale              CXVP*103           CXVF*103           CXV*103
           Without propeller        3.705              4.438              8.143
           J = 0.879                9.453              6.811              16.264
           J = 0.750                14.512             8.637              23.149

   Table 2a. Calculated viscous resistance coefficients of the pod and the strut (fin) in model scale
 (Lobachev et al., 2001). KSRI RANS code with wall function k-ε turbulence model; 632896 cells.

           Ship scale               CXVP*103           CXVF*103           CXV*103
           Without propeller        3.299              1.835              5.133
           J = 0.879                8.349              2.847              11.196

   Table 2b. Calculated viscous resistance coefficients of the pod and the strut (fin) in ship scale
 (Lobachev et al., 2001). KSRI RANS code with wall function k-ε turbulence model; 632896 cells.

Hsiao and Pauley (1999) studied the marine propeller flow using the INS3D-UP code with
the Baldwin- Barth turbulence model. A comparison of calculated and measured KT and KQ
results in Table 3a shows good agreement for the higher advance number. The discrepancy
increases when the advance number is decreased. The computations over-predicted the level
of eddy viscosity in the vortex core, which led to an overly diffusive and dissipative tip
vortex. The conclusion was that since the turbulence around the tip vortex is thought to be
highly anisotropic and the Coriolis force in the rotating propeller can also induce anisotropic
turbulence, anisotropic turbulence models such as Reynold s Stress Models may be required
to resolve accurately the tip vortex of a rotating propeller.

       J           Computed KT          Measured KT       Computed KQ        Measured KQ
       0.98        0.404                0.371             0.0901             0.0888
       1.27        0.238                0.229             0.0606             0.0618

 Table 3a. Comparison of propeller performance for two advance numbers (Hsiao & Pauley, 1999).
  INS3D-UP code with Baldwin-Barth turbulence model. 2400000 cells in all, in the vortex core
                              domain 20x20 cells in cross section.

    J           Reynolds number           -Cpmin             σi                 x/R
    0.98        3.4*106                   4.67               4.88               0.054
    1.27        3.9*106                   1.36               1.11               0.059

Table 3b. Comparison of calculated pressure coefficient and tip vortex cavitation inception index for
  two advance numbers (Hsiao & Pauley, 1999). INS3D-UP code with Baldwin-Barth turbulence
        model. 2400000 cells in all, in the vortex core domain 20x20 cells in cross section.

Effect of turbulence model. Kaarlonen (2002) investigated the effect of turbulence models on
the leading edge flow around rounded delta wing. Table 4 gives the calculated lift and drag
coefficients. The report also includes diagrams that compare measured and calculated
pressure coefficients. The explicit algebraic Reynolds stress model (EARSM) is seen to yield
the best correlation with measurements.



WS Atkins Consultants – Co-ordinators                                                         Page 26
MARNET-CFD Final Report and State
Of the Art Review




       Grid cell number             Turbulence model        CL                CD
                                    (Measured)              0.255             0.0359
       2949120                      EARSM                   0.235             0.0337
       2949120                       k-ω SST                0.235             0.0317
       2949120                      Laminar flow            0.239             0.0353
       5505024                      EARSM                   0.238             0.0335

  Table 4. Lift and drag coefficient for rounded delta wing at the angle of attack of 12 degrees and
   Reynolds number 7.9*106 (Kaarlonen, 2002). FINFLO code with different turbulence models.

Hull-propeller-rudder interaction flow. Ohashi et al (2002) compared the measured
performance of a twin screw container ship model with two CRPs wit h RANS predictions
obtained by the NEPTUNE code. The propellers were modelled as body forces. Table 5
compares the results. The prediction accuracy for thrust deduction is seen to be about 6 %.
The relative trends of the three cases investigated, i.e., without propeller, CRP, and CP, seem
reasonable.

                                    Without propeller      CRP               CP
        CV*103                      3.200                  3.440             3.496
        1 - t (calculated)          -                      0.868             0.844
        1 - t (measured)            -                      0.836             0.800

   Table 5. Calculated and measured viscous resistance coefficients and thrust deduction for a twin
screw container ship (Ohashi et al., 2002). NEPTUNE code with Spalart-Allmaras turbulence model;
                                            1016064 cells.

Abdel-Maksoud et al (2000) compared the measured performance of a single screw container
ship model with RANS predictions obtained by the CFX code. In the calculations the
propellers had finite number of blades. In the calculations the option of an interface for non-
matching points was utilised. Tables 6a and 6b give the results. The prediction accuracy for
propeller performance is seen to be about 6 %. The relative trends of in C V Table 5 and CF in
Table 6 make physical sense.

                             Without propeller      With propeller           Change
        CT*103               3.65                   4.02                     10.4 %
        CP*103               0.93                   1.27                     37.4 %
        CF*103               2.72                   2.75                     1.18 %

Table 6a. Calculated container ship resistance coefficients with and without propeller in model scale
 (Abdel-Maksoud et al., 2000). CFX code with wall function k-ε turbulence model; 400303 cells
                            without propeller & 921158 with propeller.




WS Atkins Consultants – Co-ordinators                                                          Page 27
MARNET-CFD Final Report and State
Of the Art Review



                                          Measured            Calculated
                      CT*103              3.56                3.65
                      KT                  0.158               0.170
                      KQ                  0.0303              0.0288

Table 6b. Calculated and measured total resistance, thrust, and torque coefficients for a single screw
 container ship (Abdel-Maksoud et al., 2000). CFX code with wall function k-ε turbulence model;
                            Calculations without free surface effects.

McDonald & Whitfield (1996) developed a method for the trajectory prediction of fully
appended self-propelled underwater vehicle with rotating propeller. The numerical solution
of unsteady incompressible turbulent Navier-Stokes equations and a six-degree-of-freedom
(6-DOF) computational method were coupled. The RANS equations were solved using a
multiblock multigrid scheme with relative motion structured subblocks for handling the
rotating propellers. No correlation data were given in the paper.

El Moctar (1999) and Laurens & Grosjean (2002) simulated the unsteady viscous propeller-
rudder interaction flow. In both cases the computed forces and moments were reported to
agree well with measurements.

Waterjet. The papers of the third RINA Waterjet Conference in 2001 illustrate the status of
the applicability of RANS codes for the prediction of waterjet flows. RANS codes form a
good tool to investigate hull-propulsor interaction. Allison et al (2001) used the UNCLE code
to obtain information of the flow around the ship without and with waterjet. Unlike the bare
hull, where the streamlines passed downstream rather benignly, a large portion of the
upstream flow is drawn into the inlet with the waterjet operating.

Seil (2001) used the FLUENT code with the k-ε turbulence model to investigate the effect of
shaft rotation and Reynolds number on the waterjet inlet flow. For validation Seil found
RANS predictions and experimental data for the velocity distribution at the duct exit to be in
good qualitative and quantitative agreement. After this validation the effect of shaft, shaft
rotation and scale effect (Reynolds number) on the flow in a waterjet inlet was investigated.
Seil found that shaft rotation has a significant effect on distorting the wake at the duct exit.

Huntsman and Rothersall (2001) developed a software for the design of waterjet impeller and
stator. The quasi Q3D design method is based on a streamline curvature method, a geometry
creation program, and a panel method. A 3D RANS solver is used for the analysis. The
RANS code was made in house. The Baldwin-Lomax turbulence model is used. The
diagrams depicting the correlation between measured and calculated results showed some
discrepancy. This may partly be due to the turbulence model used.

Hu and Zangeneh (2001) used different commercial CFD codes such as FLUENT, UNS,
RAMPANT, and TASCflow to calculate waterjet impeller torque. Torque is obtained from
( (rVθ ) LE - (rVθ )TE ), where Vθ is the tangent ial velocity. The predicted torque values are
compared with measurements in Table 7. Generally the prediction accuracy is seen to be very
good.



WS Atkins Consultants – Co-ordinators                                                          Page 28
MARNET-CFD Final Report and State
Of the Art Review



       Code               (rVθ ) LE          (rVθ )TE            (rVθ ) LE - (rVθ )TE
       FLUENT             0.00924            0.1021             0.0928
       UNS                0.01042            0.1064             0.0959
       RAMPANT            0.00712            0.1060             0.0989
       TASCflow           0.00600            0.1006             0.0946
       Experiments                                              0.0944

          Table 7. Calculated and measured (rVθ ) LE and (rVθ )TE (Hu & Zangeneh, 2001).

Modelling of cavitation inception. The simplest criterium for cavitation inception is to set the
negative pressure coefficient equal to the cavitation index. In Table 3b (Hsiao & Pauley,
1999) of this Section the calculated pressure coefficient differs from the observed cavitation
index by -5 % and +18 % for the two cases shown.

Farrell (2001) developed an Eulerian/Lagrangian procedure for the prediction of cavitation
inception by event rate. The event rate is governed by the number distribution of nuclei, the
instantaneous pressure field in the flow, the trajectory of the nuclei, and bubble dynamics.
The development of the procedure utilised an experimental database for an axisymmetric
headform known as the "Schiebe" body. The carrier-phase flow field was computed using a
RANS solver. The trajectories were computed using Newton's second law with models for
the various forces acting on the bubble. The growth was modelled using the Rayleigh-Plesset
equation. The simulation results in Table 8 indicate agreement with experimentally observed
trends.

     Free stream        Experimental             Predicted cavitation    Predicted cavitation
     velocity [ft/s]    cavitation index for     index at visual         index at 2000
                        visual inception         event rate              events per second
     30                 0.55                     0.49 - 0.53             0.56
     50                 0.56                     0.59 - 0.61             0.50
     60                 0.53                     0.63 - 0.64             0.51

    Table 8. Comparison of experimental and simulated cavitation indices vs. free stream velocity
               (Farrell, 2001). UNCLE code with k-ω and k-ε turbulence models.

Hsiao and Chahine (2001, 2002) developed a spherical and a non-spherical bubble dynamics
model to study cavitation inception, scaling, and dynamics in a vortex flow. The spherical
model is a modified Rayleigh-Plesset model modified to account for bubble slip velocity and
for non-uniform pressures around the bubble. The non-spherical model is embedded in an
unsteady RANS code and a moving Chimera grid around the bubble. Hsiao and Chahine
showed that non-spherical deformations and bubble/flow interaction are important for
accurate prediction of cavitation inception.

Modelling of developed cavities. Chen (1996) solved unsteady RANS equations using the
CFDSHIP code and the Baldwin- Lomax turbulence model for the four-quadrant ma rine
screw propeller flow. For the crashahead and crashback conditions, a steady flow assumption
and a cavitating correction were used. Tables 9a and 9b show performance comparison in




WS Atkins Consultants – Co-ordinators                                                        Page 29
MARNET-CFD Final Report and State
Of the Art Review

crashahead conditions. Large differences are seen between experiments and predictions
especially for the higher loading. After the cavitation correction the correlation improved.

     J           Measured KT       Calculated K T    %          Calculated K T %
                                                                after correction
     0.80        0.662             0.866             31         0.753            14
     0.55        0.411             0.845             106        0.415            1.0

   Table 9a. Comparison of propeller performance (K T) for two advance numbers in crashahead
   conditions (Chen, 1996). CFDSHIP code with Baldwin-Barth turbulence model; 270000 cells.


     J           Measured KQ       Calculated KQ     %          Calculated KQ %
                                                                after correction
     0.80        0.1104            0.1476            34         0.133            20
     0.55        0.0736            0.1440            96         0.0772           4.9

   Table 9b. Comparison of propeller performance (K Q) for two advance numbers in crashahead
   conditions (Chen, 1996). CFDSHIP code with Baldwin-Barth turbulence model; 270000 cells.

The cavitation model used by Coutier-Delgosha et al (2001) was based on one fluid,
characterised by a density that varies in the computational domain according to a state law.
When the density in a cell equals the liquid one, the whole cell is occupied by liquid, and if it
equals the vapour one, the cell is full of vapour. Between these two extreme values, the cell is
occupied by a liquid/vapour mixture that is also considered as one homogeneous single fluid.
The void fraction corresponds to the local ratio of vapour contained in the mixture. In this
model the void fraction is related to the state law. The model assumes that in each cell the
velocities are the same for the liquid and the vapour. The model utilises an empirical
                                                                         o
barotropic law. The purely liquid or purely vapour fluid states are j ined smoothly. The
maximum slope of the derivative of pressure with respect to density can be interpreted as the
second power of the minimum speed of sound in the mixture.

Coutier-Delgosha et al used the FINE/TURBO(TM) RANS CODE with the Baldwin- Lomax
turbulence model to study of the effect of the leading edge shape on cavitation around
inducer blade sections. For two 2D leading edge shapes at two angles of attack the correlation
of cavitation number vs. dimensionless cavity length was good. For the larger angle of attack
used in the calculations the observed cavities adopted an unsteady behaviour with large
vapour cloud shedding. The numerical model failed to simulate this behaviour. This
limitation of the model was linked to the use of a standard turbulence model.

Song et al (1998, 2001) developed a numerical code based on the Navier-Stokes equations of
compressible fluid and a virtual single-phase equation of state to capture the highly dynamic
nature of cavitating flows. To capture the radiation of a pressure wave and its effect on the
global flow field, it is necessary to include the compressibility effect of liquid phase and the
gas phase. Compressibility effect may dominate the flow even if the Mach number is very
small provided the flow changes rapidly enough. For this reason Song et al used a Large
Eddy Simulation (LES) method. Strictly the LES method assumes that the computational
grids should be so small that the sub grid scale turbulence is isotropic. However, Song et al



WS Atkins Consultants – Co-ordinators                                                    Page 30
MARNET-CFD Final Report and State
Of the Art Review

considered that the method can accurately simulate the primary and secondary eddies as long
as the grids are sufficiently smaller than the size of the eddy to be simulated. The model was
applied to study 2D flow patterns over a NACA0015 hydrofoil. Only limited amount of
correlation data was given in the two papers. The photographs showed similarity to the
instantaneous vorticity.

Kubota, Kato, and Yamaguchi (1992) developed the Bubble Two-Phase Flow (BTF) cavity
model to explain the interaction between viscous fluid and bubble dynamics. In the
macroscopic view BTF treats the cavity flow field as a compressible viscous fluid whose
density varies greatly. The fluid inside the cavity and the fluid outside the cavity are treated
as a single continuum. Therefore detached cavitation clouds can be modelled. In the
microscopic view BTF treats the cavity as bubble clusters. This is because bubbles play an
important role in the inception of cavitation and also in the cavity collapse. The interaction
between viscous fluid and bubble dynamics is obtained by coupling the macroscopic and
microscopic views. There are two cavity types in the microscopic view, the vapour film and
the bubble cluster with vortices. This structural microscopic model cannot basically be
applied to the vapour film type of cavitation. In the macroscopic view contour lines of void
fraction give the shape of the cavity. The local void fraction is obtained from the bubble
density and the typical bubble radius. The interaction between the bubbles occurs through the
local pressure in the liquid as bubbles grow. At the center of a cell the influence of other
bubbles in the cell is obtained as the sum of the velocity potentials of the other bubbles.

The computational procedure of Kubota, Kato, and Yamaguchi was closely similar to that of
the Marker-and-Cell method. The method was applied to the NACA0015 hydrofoil. In the
test calculations boundary layer velocity profiles were predicted well, but not the pressure.
The predicted lift coefficient was only 58 % of that obtained in the measurements. Cavity was
defined as a region where void fraction is larger than 0.1. In another test calculation the
unsteady features predicted agreed with the observed sheet cavitation patterns.

Yakushiji, Yamaguchi et al (2001) used the STAR-CD code, in which the BTP cavity model
has been implemented, to simulate in 2D the observed sheet-cloud cavity on a NACA0015
section. The computational method was VOF. The simulation reproduced unsteady shedding
of cloud cavity, although some differences appeared in flow details. Lindenau et al (2002)
applied STAR-CD to a NACA 16-206 a=0.8 (mod) 2D hydrofoil section and a NACA 16-206
a=0.8 (mod) 3D hydrofoil. Lindenau et al considered the correlation of 3D calculations with
experiments reasonably good. The simulations also captured the beginning of tip vortex
cavitation.

Abdel-Maksoud (2003) used the CFX-TASCflow code with k-ε turbulence model to simulate
cavitating propeller in homogeneous onset flow. The volume void fraction field may vary
continuously from 0 to 1 in the cavitation zone covering many grid elements. The added
complexity in comparison with the traditional VOF approach is that a source term is now
added to the volume fraction equation to model the creation and destruction of vapour. The
source term depends on the local pressure difference. The rate of vapour production of the
source term is obtained from a linearised Rayleigh-Plesset equation. In practice, the
vaporisation and condensation processes have different time scales.

In the test calculation Abdel-Maksoud used 1310000 cells after a grid refinement. The drop
of thrust and torque for low cavitation numbers was well predicted. Some oscillation in KT


WS Atkins Consultants – Co-ordinators                                                   Page 31
MARNET-CFD Final Report and State
Of the Art Review

and KQ as function of cavitation number was predicted for cavitation numbers higher than 2.
The measured results show no such wavyness. With regards to cavitation patterns good
agreement between calculated and observed was achieved for the cavitation numbers of 1.36
and 1.99. The numerical results showed root cavitation at the cavitation numbers of 3.89 and
4.52, but the volume of cavitating flow is very limited in comparison with experimental
results.

Lindau et al (2002) applied the UNCLE-M code to the analysis of attached cavitation. These
cavities are presumed to be amenable to a homogeneous approach, are generally unsteady,
and contain regions of separated flow. In the test cases interface curvatures were small, and
pressures and velocities approximately continuous across the interface. This implies that
nonequilibrium interface dynamics are of negligible magnitude, and the effect of surface
tension is not incorporated. The physical model treats each phase as a separate species. A
liquid, a vapour, and a non-condensable gas phase are modelled. Mass transfer between the
liquid and the vapour phases is achieved through a finite-rate model. A high Reynolds
number form of two-equation models with standard wall functions provides turbulence
closure. Separate models are used to describe the transformation of liquid to vapour and the
transformation of vapour back to liquid. One of the test cases was a hemispherical cavitator.
The computed averaged pressure coefficients generally agreed with the measured data. The
differences were at their largest near the end of the cavity.

Ahuja et al (2001) formulated a cavitation model that yields acoustically and
thermodynamically consistent behaviour of multi-phase systems. The importance of
thermodynamic consistency is that the cavitation model can be used for simulations from
marine screw propellers to cryogenic pumps. The cavitation formulation was implemented in
the CRUNCH code. The numerical formulation was implemented on a hybrid unstructured
framework which permits tetrahedral/prismatic cells. The two 2D sample simulations were
computed as steady-state calculations using the high Reynolds number k-ε turbulence model.
A cylindrical head- form formed one of the test cases. Since the calculation was performed
with a steady RANS turbulence model, the wake simulated corresponds to the time-averaged
solution, which reproduces the mean mixing rate. If the calculations were performed with an
unsteady LES model, an unsteady wake would result in vortices being shed off the cavity tip.
A NACA 66 hydrofoil at a 4-degree angle of attack with sheet cavity formed the second test
case. The numerical results underpredicted the length of cavity and overpredicted the aft
recovery pressure for the cavitation numbers 0.89 and 0.91.

Validation Data for RANS and Panel Methods
Numerous LDV measurements of propeller and hydrofoil tip vortex flows have been
performed in the late 1990's. Turbulence has been measured in a number of studies utilising
multi-component coincidence. PIV (particle image velocimetry) has the advantage of
capturing highly unsteady or transient flows. In the early 2000's some PIV measurements
have been made on propellers (ITTC, 1999, p. 56; Calcagno et al., 2003).

Additional experimental data of blade surface pressures and slipstream velocities are needed
for the validation of RANS equation and panel method codes. Particularly little flow field
data for off-design operation conditions have been published (ITTC, 1999, p. 39).




WS Atkins Consultants – Co-ordinators                                                Page 32
MARNET-CFD Final Report and State
Of the Art Review

The results from a questionnaire sent to ITTC members show a large range of criteria for the
determination of cavitation inception data (ITTC, 1999, p. 59). For that reason, comparing
cavitation inception results conducted in different facilities with their own water quality
characteristics is difficult. This fact should be accounted for when utilising the cavitation
inception data to validate cavitation inception predictions.


Conclusions
In the period 1998-2003 substantial advances have been made in the application of RANS
codes to ship propulsor flows. The number of correlation tables included in this section
reflect these advances. Not many such correlation data were available in 1998. In 2003 the
analysis of hull-propeller-rudder interaction flow is within the state-of-the-art. In 2003 the
number of cells required for such calculations limits more widespread use of RANS to
interaction studies in shipyard design work. In 2002-2003 two-phase and cavity models have
been added to the leading commercial RANS codes. Because of the large calculation times
RANS predictions of unsteady propeller induced hull pressure have not yet been reported.

RANS equation codes are capable of predicting accurately (up to within 1 % of measured
values) the propeller open water performance and pressure distributions on the blade at the
design advance number. The accuracy of performance prediction for smaller off-design
advance numbers is good or acceptable. The main features and the main variables of the flow
field, streamlines and boundary layer are generally well predicted. Global variables of
interaction flow are predicted with an accuracy of about 6 %.

In the period 1995-2003 panel method codes have been extended to treat quite complicated
interaction flows such as podded propulsor and cavitating propeller-rudder interaction flow.

Unsteady vortex lattice and panel methods are the main CFD tools used for predicting
unsteady cavitation on propulsors and propulsor induced hull exitation forces. The recent
improvements in panel method codes have resulted in good agreement between the measured
propeller and predicted performance and pressure distributions on the blade for most types of
propellers. However, significant discrepancies in the predicted hull pressures are seen in
some cases.

4.4    TA4 – Offshore Floating Systems

In MARNET-CFD, Thematic Area 4 was intended to cover the application of CFD to all
aspects of offshore engineering above and below water. However, among the participants in
the Network, it was found that there was little experience in platform aerodynamics,
dispersion and explosion modelling. Following some brief reviews as reported in the first
state of the art document, it was concluded that platform aerodynamics were not sufficiently
critical to the design of floating systems to continue their inclusion within the remit of
MARNET-CFD. In addition, the wind and safety engineering literature contained evidence
of a thriving R&D community and extensive use of these techniques in practice. There was
thus little that could be added by MARNET-CFD.
Hydrodynamic loading however proved to be a key area where MARNET-CFD could make
an impact, and hence this became the key subject for development in terms of the state-of-
the-art, best practice and FP5 research projects. The main emphasis therefore is the


WS Atkins Consultants – Co-ordinators                                                 Page 33
MARNET-CFD Final Report and State
Of the Art Review

calculation of wave loads on stationary or very slow moving offshore floating production
systems.
                                            e
As reported in the first state of the art r view, historically, there has been considerable
emphasis on predicting the loads on fixed pla tforms in the form of tall lattice structures
known as “jackets”, piled into the seabed, and with heights of around 200 metres at the limit.
The primarily structural members used are circular cylinders, and turbulent drag loads tend to
be of leading order in the design process. Other forms of early fixed “gravity” platforms
were built in concrete, e.g. the CONDEEP structure, and consequently took the form of a
large diameter circular section tower, held down on the seabed by its own weight. Loads on
these structures are affected by wave diffraction, and hence 3D boundary element/panel
methods were developed to calculate these loads.
Between these two extremes, floating exploration and production systems, in the form of
semi-submersibles, Tension leg platforms, SPAR buoys and FPSOs have also been
developed. For these vessels, problems related to their motions in waves and second-order
effects such as drift, tether springing, air-gap and steady tilt, require the application of 3D
diffraction analysis. Viscous effects have a role also, to varying degrees depending upon the
type of systems and its location.
In all of the above cases, the previous state of the art review described how computational
tools were developed which met the primary need to estimate wave loads and responses. For
the slender jacket structures, the so-called Morrison equation approach was developed.
Morrison’s equation provides a simple and reliable method, but numerous uncertainties
remain in its application, including appropriate values of drag coefficient, its application in
areas close to joints or nodes, and in the wave zone. However, it is not normally regarded as
a suitable application for CFD.
The previous state of the art review identified the numerous technical gaps in the application
of numerical hydrodynamics to offshore platforms and systems. Some of these gaps appear
as the result of the shortcomings of linear wave theory, and some as the result of uncertainties
in viscous and turbulent flow effects. Thus for our purposes, CFD involves 3D fluid
mechanics models for which non-linearities arising from either non- linear free surface
behaviour, lifting effects, viscosity or turbulence, require iteration (to steady state) or
simulation methods to be applied.
The principle areas of interest in offshore hydrodynamics and CFD identified originally were:
   1.   Non-linear wave diffraction,
   2.   Wave run-up and impacts, green water
   3.   Vortex shedding and vortex induced vibration (risers and tethers),
   4.   Viscous effects in the wave-zone,
   5.   Low frequency responses.

The current state of the art in potential flow diffraction modelling can be summarised as
follows:
   1. Linear, frequency domain diffraction analysis is regarded as a mature and well
      exploited computational tool. Best practice is well understood and the computer
      programs well validated.
   2. The next level up is second order diffraction analysis. Additional potentials due to
      sum and difference frequency compone nts are evaluated using extensions of first
      order techniques. Here too, the theory and modelling techniques are sufficiently


WS Atkins Consultants – Co-ordinators                                                   Page 34
MARNET-CFD Final Report and State
Of the Art Review

       mature for the improved calculation of low frequency motions (drift) and high (sum)
       frequency effects such as tether ringing. Improved estimates of air- gap as the result
       of closer modelling of near- field diffraction and interference effects are also possible.
   3. Fully non-linear time domain diffraction analysis is also possible, and some models
      exist for, primarily, research purposes (Ferrant et al…, Wu & Eatock Taylor).
      However, it would appear that there are inherent difficulties with the boundary
      element method for step waves that prevent these models predicting wave breaking.
      Indeed they are prone to numerical instability in such cases. It could be argued that
      such methods hold no great advantage over second order calculations, whilst at the
      same time requiring considerably greater computational resources. However, certain
      higher order effects in very steep waves indicate the need for robust theories beyond
      second order.




               Figure 5. Example of results from a fully non-linear free surface model
                          of wave diffraction (Wu & Eatock-Taylor, 1999)


Wave run-up and green water loading calculations provide the ultimate challenge in free
surface dynamics, not least as the result of the severe wave breaking and spray forming
effects that occur. Potential flow models are unsuitable in such cases and, though less well
mathematically formulated, VOF simulation techniques appear to offer the only feasible
approach at the present time. Combinations of classical diffraction and potential flow models
with unsteady RANSE and VOF have been investigated within the EXPRO-CFD FP5
project, and as will be discussed later, show considerable promise in this respect.
The previous state of the art review described the problem of vortex induced vibration, or
VIV, on marine risers. This is a key application of CFD and unsteady RANSE in particular.
In order to provide meaningful results, VIV simulations must be conducted with structural
responses included in order that lock- in is achieved correctly.
Calculations of vortex shedding on circular cylinders in steady flows is regarded as a standard
“benchmark” in CFD, with the objective of reproducing the Karman vortex street observed in
experiments. It is remarkable nevertheless that many commercial CFD codes are unable to
adequately model this basic phenomenon unless very great care is taken with the numerical
discretisation scheme and near wall and turbulence modelling (Younis et. al. 1996).
Nevertheless, it can be demonstrated that, if the MARNET-CFD best practice guidelines are
followed, commonly used commercial CD codes can have some success in modelling this
phenomenon.



WS Atkins Consultants – Co-ordinators                                                    Page 35
MARNET-CFD Final Report and State
Of the Art Review

The previous state of the art review cited a number of references for fluid structure
interaction models, in which the CFD calculation is coupled to a simple ordinary differential
equation describing the motion of the cylinder. For example, Dalheim (1997), Jensen and
Shultz (1998), and Kallenderis (1998), have demonstrated that large amplitude structural
responses can be simulated successfully with 2D unsteady Navier-Stokes calculations.
Since that review, further work by Dalheim (2000, 2001) using vortex in cell methods in 2D,
and Hansen & Meyer (2001) using unsteady RANSE methods, have been able to add to the
literature. Similarly, Oliveira et al. using a finite element method, and Maeda et al (2001)
using a finite difference method, have also been able to simulate VIV using 2D approaches.
It would appear that the main challenge remains to produce a convincing and efficient 3D
calculation in CFD. The main difficulty in this respect appears to be that coherent 3D vortex
structures are very difficult to simulate with current turbulence models and computing
resources. Grid resolution along the length of the riser may need to be of order that in the
circumferential direction, which, for any common length of riser (many hundreds of metres)
would appear at present to be impractical. Practical 3D models of VIV using CFD for design
are still some way off, but research into the relevant phenomena remains important in order
that empirical models such as SHEAR7 (Vandiver & Li, 1997), might be improved.
On a larger scale, vortex induced motions on SPAR platforms are, in certain current
conditions, a significant problem. It has become apparent since the previous state of the art
review both that CFD might be an appropriate modelling tool and that, other than the problem
of the higher Reynolds numbers at full scale, may be less demanding in 3D with respect to
computational resources (owing to the much lower length to diameter ratios, of order 10,
rather than many hundreds in the case of risers). Work carried out in the EXPRO-CFD
project may confirm this conjecture in the near future.
The previous state of the art review discussed a number of aspects of extreme wave loading
on SPAR buoys. Since that review, little further work outside of the EXPRO-CFD project
has been reported. Within EXPRO-CFD, considerable progress has been made, particularly
with the accuracy and validation of CFD RANSE models for wave-cylinder interaction using
VOF methods. These results have been reported (Gallagher, OTC 2003 & Vissoneau, ISOPE
2003).
Green water loading problems have been the focus of much attention since the last state of
the art review, both owing to incidents on FPSOs and conventional ships. The SAFEFLOW
FP5 funded project has made significant progress in developing tools, experimental and full
scale data, along with structural assessment methods. Within SAFEFLOW, a CFD model
(COMFLOW) has been developed using VOF methods, and this has been shown to give
quite realistic simulations of the green water phenomenon.
Finally, as noted in the previous review, for FPSO type vessels, one key problem is that of
slow drift motions, and in particular the low frequency damping of these motions. The
interaction between potential flow and viscous effects was, and still is to a large extent,
poorly understood. Little appears to have been added to the literature in this respect, but it is
hoped that some of the research carried out in EXPRO-CFD will be able to contribute to
reducing some of these uncertainties.

4.5    MARNET-CFD Questionnaire Results and Review.

During the first year of the MARNET-CFD Thematic Network, and extensive review of the
participants views and experience of CFD was carried out in the form of questionnaires


WS Atkins Consultants – Co-ordinators                                                    Page 36
MARNET-CFD Final Report and State
Of the Art Review

within each Thematic Area Group. The results were reviewed and collated across the TAs,
and certain clear patterns emerged which were discussed in the first annual report.
During the final year of MARNET-CFD, a questionnaire was again circulated. This time a
single coordinated approach was adopted, in which only the primary conclusions from the
results of 3 years earlier were re-examined. The intention of this second review was therefore
to identify changes in attitudes and experience in the industry over the period of the Network
that had arisen either as the result of MARNET-CFD, or due to general advances and
developments in technology.
Appendix II contains a summary of the questions and results. The following analysis of these
results was reported at the final Network Workshop, with the aim of stimulating debate on the
way forward. The main highlights of the review presented relate to:
   1. The frequency of CFD usage
   2. The distribution of CFD tools used.
   3. Confidence in the use of CFD in TAs 2,3 and 4
   4. Changes in views of CFD accuracy in TA1
   5. Changes in views on CFD development needs
Figure 4.1 below shows the results from the first of these questions:


                                               10
                                                9
                         Number of responses




                                                8
                                                7
                                                6
                                                5
                                                4
                                                3
                                                2
                                                1
                                                0
                                                    0-20%   21-40%   41-60%   61-80%   81-100%
                                                                                 frequency of use


                      Figure 4.1 Frequency of use of CFD among all responses


It is fairly clear from this that there are two types of users among the maritime community.
We have frequent CFD practitioners, who on closer examination, consist of mainly of
Universities and developers of CFD tools. The less frequent users, who in total are the
majority, use CFD within other tasks in design or analysis. This is entirely to be expected,
and indicates that CFD is being taken up within the design process. It is hoped that the use of
the best practice guidelines and activities of the Network are helping to support the success of
CFD in this respect.
The next question relates to the types of CFD tools being used. Figure 4.2 below shows the
distribution of software among the participants.




WS Atkins Consultants – Co-ordinators                                                               Page 37
MARNET-CFD Final Report and State
Of the Art Review



                                                      USE OF CFD CODES

                         20

                         18
                         16
       NUMBER OF USERS




                         14
                         12

                         10
                         8
                         6
                         4

                         2
                         0
                              CFX4/5     COMET       FLUENT      FINFLO    SHIPFLOW    STAR-CD   OTHER/OWN
                                                                                                   CODES
                                                              CODE NAMES


                         Figure 4.2 – Distribution of CFD software among participants: All Thematic Areas

Analysis of these results in the light of the first survey reveals some interesting trends. First,
the “other/own codes” category relates primarily to software used in sea-keeping and
manoeuvring, and developed for specific hydrodynamic problems (e.g. wave diffraction).
The debate as to whether this is in fact CFD or alternatively, Computational Hydrodynamics,
continues. Next, it is clear that there has been an overall increase in the number of RANSE
code being used. SHIPFLOW, the only dedicated CFD code for steady ship hydrodynamics,
has seen an increase in usage since the last survey. However, there has been a significant
increase in the number of commercial CFD codes (CFX4/5, COMET, FLUENT and STAR-
CD) in use. It is expected that this trend will continue, and that the growth in use of CFD in
the maritime community will mainly be served by these tools.

The next set of questions were directed at Thematic Areas 2, 3 and 4, and were aimed at
establishing current levels of confidence with the Network in using CFD. The participants
were asked to rate their level of confidence across a number of applications if CFD according
to the following scale:

                              4 = Perfect Predictions
                              3 = Quantitative confidence if supported by testing
                              2 = Qualitative confidence
                              1 = Useless

The range of applications of CFD in each Thematic Area, and their scores were as follows:




WS Atkins Consultants – Co-ordinators                                                                       Page 38
MARNET-CFD Final Report and State
Of the Art Review



             TA2 - Hydrodynamic loads and responses                            Score
             global loads on ship in waves                                      2.8
             local external loads on ship: slamming                             2.3
             local external loads on ship: green water                          2.1
             local internal loads: sloshing                                     2.8
             vibrations                                                         2.6
             fire/explosion                                                     2.5
             damaged stability                                                  3.0
             dynamic stability                                                  2.7
             comfort - operability                                              3.0
             control - stabilisation                                            2.7
             speed reduction in waves, added resistance                         2.4
             course keeping in heavy weather                                    2.4
             statistics: short term, long term, most probable critical event    2.0
             conventional manoeuvring data prediction                           2.7
             high speed manoeuvring                                             2.4
             manoeuvring in waves                                               2.3
             broaching                                                          2.4
             loads on appendages                                                2.4
             thrust sizing                                                      2.3
             slip                                                               2.8


For TA2, hydrodynamic loads and responses, all results appear to vary between confidence at
a qualitative level (perhaps suitable for design trade-off studies), and quantitatively reliable if
used with test data. No participants had sufficient confidence to use CFD in isolation. It
would appear that those areas for which the greatest confidence was expressed (2.8 and over)
were on global loads, internal loads such as sloshing, predicting comfort and operability,
damaged stability. It is interesting to note that during the course of the Network, all of these
were the subject of ongoing FP5 studies. It is not clear whether these increased confidence
levels were the result of greater familiarity or a genuine greater capability. However, In some
respects, it is little surprising that global loads on ships in waves does not score more highly.
The lowest scoring levels of confidence appear to be with events such as slamming and
green-water loading and statistical aspects. Slamming and green water involve significant
non- linearities and fluid structure interaction, so it is not surprising that these should be rated
of qualitative accuracy only.

For TA3, the following levels of confidence were expressed with respect to the performance
of propulsor calculations.




WS Atkins Consultants – Co-ordinators                                                       Page 39
MARNET-CFD Final Report and State
Of the Art Review



             TA3 - Propulsors                                           Score
             Open water flow                                             2.9
             Propeller in inclined flow                                  2.8
             propeller in non-uniform onset flow                         2.9
             design for prescribed blade pressure distribution           2.8
             application of optimisation techniques                      2.6
             propeller interaction (with rudder, hull, shaft)            2.4
             propeller in semi-tunnel                                    2.4
             propulsor performance when manoeuvering,
             accelerating,decelerating                                   2.5
             propulsor performance in waves                              2.2
             tip vortex modelling                                        2.6
             Inception prediction of steady cavitation                   2.7
             Extent prediction of steady cavitation                      2.4
             unsteady cavitation prediction                              2.3
             propulsor induced hull pressures                            3.0
             fluctuating shaft forces                                    2.7
             scale effects                                               2.6
             propulsor design                                            2.8


Calculations on propulsors demonstrated on average, slightly higher levels of confidence,
although again, nowhere was it considered appropriate that CFD alone could be used with
qualitative confidence. The areas where greatest confidence exists is in open water flow, the
effect of ship boundary layer and inclined flow on thrust, torque and efficiency, propeller
induced hull pressures and inverse design methods (by prescribed blade surface pressures).
The lowest levels of confidence relate to caviatation. The prediction of inception for certain
types of cavitation is thought to be relatively straightforward. However, quantitative
predictions for all types of cavitation effe cts are low in confidence rating. It is fitting
therefore that projects in FP5 have been funded.

             TA 4 - Offshore Engineering                               Score
             Hydrodynamic loading on fixed structures                    3.0
             Hydrodynamic loading on floating structures                 3.0
             Hydrodynamic loading on components (e.g. riser pipes)       3.1
             Wind loading                                                2.9
             Environmental (e.g. gas dispersion of hazardous gas)        2.9
             Fire, explosion                                             2.5
             Internal flows (e.g. ventilation)                           3.3


For offshore engineering, the general calculation of hydrodynamic loads is regarded with
some confidence. This is primarily because many leading order effects can be quantified
using diffraction calculations that have been in practical use for a number of decades. It is
perhaps a shortcoming of the questionnaire that we did not ask direct questions relating to
either extreme loading in large waves (including green water), vortex induced motions or
either risers or SPAR buoys, or viscous effects in drift motions, all of which are areas of
current concern, and which in most cases would have scored 2 or below.
Taking an overall view, it would appear that across these TA participants, CFD is being used
to considerable degree to give qualitative advice on design options, or in conjunction with
testing, to give quantitative data for motions, loads and manoeuvring performance.



WS Atkins Consultants – Co-ordinators                                                 Page 40
MARNET-CFD Final Report and State
Of the Art Review

In the case of the TA1 survey, it was possible to carry out a quantitative review as above and
compare the answers with those given in the first review. The results are given in the form of
histograms for ease of comparison. In this case the scoring system was designed to fit with
the original survey as follows:
               5 = perfect predictions
               4 = quantitative confidence when supported by testing
               3 = broad qualitative agreement
               2 = only certain features or measures agree
               1 = useless




WS Atkins Consultants – Co-ordinators                                                 Page 41
                                                                                                                        RATING
                                                                       de
                                                                           ep
                                                                               wa
                                                                    sh             ter
                                                                                                                                                                                        Of the Art Review




                                                                       al               wa
                                                                          ow                  ve
                                                                              wa                  re
                                                                                  ter                sis




                                                                                                                       0.0
                                                                                                                       0.5
                                                                                                                       1.0
                                                                                                                       1.5
                                                                                                                       2.0
                                                                                                                       2.5
                                                                                                                       3.0
                                                                                                                       3.5
                                                                                                                       4.0
                                                                                                                       4.5
                                                                                                                       5.0
                                                                     de                wa                tan
                                                                        ep                   ve               ce
                                                                             wa                   re
                                                                sh               ter                sis
                                                                   al                vis                tan
                                                                      ow
                                                                           wa             ou                 ce
                                                                              ter              sr
                                                                                                                                                                                        MARNET-CFD Final Report and State




                                                                                                   es




WS Atkins Consultants – Co-ordinators
                                                                                  vis                 ist
                                                                                       co                  an
                                                                                          us                  ce
                                                                                                re
                                                                                                   sis
                                                                                                      tan
                                                                                                            ce
                                                                               de               air
                                                                                   ep               re
                                                                                                       sis
                                                                                        wa                  tan
                                                                            sh                ter               ce
                                                                               al                  wa
                                                                                  ow                   ve
                                                                       de              wa                   wa
                                                                           ep               ter                 sh
                                                                               wa                wa
                                                                    sh             ter               ve
                                                          pr           al               de                wa
                                                            op            ow                 liv              sh
                                                              ell             wa                er
                                                                 er                                ed
                                                                    ind           ter                   po
                                                                        uc             de                   we
                                                                           ed              live
                                                                                                 red            r
                                                                  bo           vib
                                                                     un            ra                 po
                                                                        da            tio                  we
                                                                           ry
                                                                                         ns
                                                                                               an             r
                                                                              lay



                                        OBJECT FUNCTION
                                                                                  er               dn
                                                                     sig             no                ois
                                                                        na               ise                e
                                                                            tur                 (tra
                                                                               es                   ns




Page
                                                                                    (p                  itio
                                                                                       re                   n)
                                                                                          ss
                                                                                               ur
                                                                                                  ev
                                                                                                      or
                                                                                                         tic
                                                                                                             es
                                                                                                sm              )
                                                                                                    ok
                                                                                                       e
                                                                                                           eff
                                                                                                              ec
                                                                                                                  ts
                                                                                                                                         TA 1 - MARINE VEHICLE PERFORMANCE PARAMETERS




                                                                                                                        02/03 RESULTS
                                                                                                                        INITIAL SURVEY




42
MARNET-CFD Final Report and State
Of the Art Review

As with the levels of confidence associated with the other thematic areas, the majority of
results are in the range between a broad qualitative agreement and quantitative confidence if
backed up with experimental data.
In all areas of hydrodynamic resistance prediction, there has been an increase in confidence
in predictions made such that CFD is now viewed as a useful tool to supplement physical
testing. Indeed it is quite widely used to explore alternative hull forms and in searches for
optimal or improved performance, reducing the amount of tank testing needed and associated
costs I tank time and physical model building. However, it would appear that no ships have
yet been built in which only CFD will have been used in resistance and powering
calculations.
Reduced confidence was expressed in the use of CFD for the prediction of noise signatures
and boundary layer transition, where even broad qualitative agreement appears difficult to
achieve, and similarly in the area of wash and wake. This may be because both areas have
been the subject of increased study over the period of the Network, and a more realistic view
is now prevalent. It is certainly the case found in other areas of CFD application that
quantitative agreement for integrated quantities such as drag or lift is achieved more easily
than the point values such as local pressures.
Overall, the same observation could be made of the responses of the other TA groups, and
that the trend is that of increasing confidence in the prediction of integrated quantities such as
global loads and resistance, but with greater uncertainty over local values, particularly where
non- linearieties (e.g. in free surface behaviour), or the details of the turbulence modelling
used (e.g. in boundary layers) will have more of an influence.
The final set of questions were aimed at establishing opinions of where further research is
needed. This question was also set in the first review of 1998/99, and so it ha been possible
to establish cha nging views. A matrix relating qualities of accuracy, applicability and user-
friendliness to certain aspects of the modelling processes, namely:
       §   CAD surface modelling
       §   Grid generation
       §   Flow solver
       §   Post-processing
was set up and participants asked to rate the ir level of relative importance for development by
allocating scores. The scores could take any value but the total of the distributed scores was
to be 100. The averaged results across all participants were as in the following tables.




WS Atkins Consultants – Co-ordinators                                                     Page 43
MARNET-CFD Final Report and State
Of the Art Review

                                                         Accuracy   Applicability User-friendliness   Other
CAD surface representation                                  4.5         3.5              5.5             2.1
Grid generation                                             6.0         7.1              8.2             2.1
Flow solver                                                17.7        10.3              5.7             2.2
Post-processing                                             5.0         4.4              5.7             2.2
Other                                                       2.0         2.5              2.0             1.8

DATA FROM 2002/3 SURVEY

                                                         Accuracy   Applicability User-friendliness   Other
CAD surface representation                                  1.7         2.8              3.3             0.0
Grid generation                                             5.8        13.7              7.8             3.3
Flow solver                                                25.0        18.7              3.0             3.3
Post-processing                                             0.8         3.0              3.7             0.0
Other                                                       0.8         0.0              0.0             1.7

DATA FROM 1998/99 SURVEY (MERGED)



Overall, it can be seen that the pattern of the weighting has remained much the same, with
flow solver accuracy and range of applicability remaining the most important areas for
development.
However, perhaps owing to a combination of wider use in practice and the increasing
proportion of commercial CFD software, it would appear that there is a slight increase in the
need for better CAD tools, particularly with respect to accuracy, and for post-processors of
greater accuracy, applicability ad user- friendliness. It is perhaps the case that the commercial
tools are not set up to provide commonly used hydrodynamics measures of performance, as
post-processed quantities, and hence they will be less user- friendly in the detailed design or
optimisation process.
Our overall conclusions from this review were that:
     1. There has been a significant increase in the use of “commercial” CFD codes – the
        previous survey having highlighted the majority use of in- house tools.
     2. There has been a general improvement from qualitative to quantitative (when backed
        up by testing) for resistance and powering, but:
     3. This is m atched by a drop in applications such as wake and wash, signatures and
        related more complex flow issues.
     4. For seakeeping (TA2) and propulsors (TA3) – both are mid-way between qualitative
        and quantitative, but with greater confidence for prediction of global parameters as
        found in many other branches of CFD
     5. For offshore engineering, confidence was highest across all categories, but the
        questionnaire did not provide enough categories of flow to distinguish between
        RANSE solvers and traditional potential flow methods.
     6. Solver accuracy and applicability remains as the key development need – but pre- and
        post-processing growing in importance.




WS Atkins Consultants – Co-ordinators                                                                 Page 44
MARNET-CFD Final Report and State
Of the Art Review

5.0      SURVEY OF GAPS, RES EARCH AND INDUSTRY NEEDS.

5.1      Introduction
During the final 18 months of the MARNET-CFD Thematic Network, a number of activities
have contributed to both the formulation of the foregoing state of the art and the survey of
gaps, research and industry needs. These have included the ITTC review process, the
MARNET-CFD questionnaire on the state of CFD in Europe, the revision of the Martime
R&D Master Plan, and the development of ideas among participants in MARNET-CFD for
an Integrated Project in FP6.
The following review therefore summarises for each of the MARNET-CFD Thematic Areas,
the key points that should be considered for future research to satisfy the needs of industry.

5.2      Needs in each TA

TA1 – Ship Performance
Two areas for future work stand out as being in need of urgent attention:
      1. The need for full scale validation data at realistic Froude and Reynolds numbers.
      2. The need to develop fast, multi-parameter, hull form optimisation tools linked to
         CFD.
The overwhelming consensus regarding steady ship flows remains that the largest technical
gaps lie in the area of validation. In particular, full-scale validation data is urgently required
on the turbulent flow structure around a ship at full scale Reynolds number.
A project proposal for the third call of FP5 was submitted and accepted for funding in this
area by a consortium containing members of the MARNET-CFD Network. The project
EFFORT, aims to develop techniques for full scale PIV and LDA, carry out these
measurements, and perform CFD calculations for comparison. This data will provide an
invaluable resource to all members of MARNET-CFD once made available as part of the
database or otherwise.
A further contribution to the state of the art by Ecole Centrale de Nantes during the third year
of MARNET-CFD was to transfer techniques in form optimisation (genetic algorithms and
similar), for aerospace (wing section design) to hydrodynamics. The results reported at the
Crete 2001 meeting of MARNET-CFD were very promising, but the computing power
required is clearly beyond that available to normal shipyards and designers.
There is therefore a clear need to develop efficient optimisation techniques and software
integrated within a CFD environment and suitable for use within the timescales of a typical
ship design process. Furthermore, the optimisation schemes should be multi-parameter, and
not lead to extreme optima which compromise other important design variables, such as
seakeeping, manoeuvring, noise or wake.

TA2 – Unsteady Ship Flows
Given that this thematic area covers both sea-keeping and manoeuvring, we address the most
important gap in each area.
For sea-keeping, the numerous difficulties in predicting large scale motions and loads on
ships, bow flare slamming, green water were identified in previous reports. It was also noted



WS Atkins Consultants – Co-ordinators                                                     Page 45
MARNET-CFD Final Report and State
Of the Art Review

that this need was most urgent in the area of high-speed passenger vessels. This remains the
case. In particular, and as with TA1, full scale validation of CFD models for loading and
response is required.
As the result of investigations undertaken in related FP5 projects such as EXPRO-CFD and
SAFEFLOW, both of which address aspects of non- linear wave loading and response, the
technical route forward has become clearer. It would appear that algorithms which combine
and overall non- linear potential flow modelling approach for global wave loading with
unsteady RANSE, operating on nested grids, would be optimal. Recently this has become
known as the “open ocean” approach, and requires that the RANSE equations are re-cast into
a form wherein the irrotational and rotational parts are de-coupled. It has also been
demonstrated that this is entirely compatible with current commercial RANSE codes.
In offshore engineering hydrodynamics, this is accepted as a valid methodology. It has
however yet to be applied to the case of a ship with forward speed. It is hoped therefore that
some space can be found within the future FP6 programme to exploit this type of
development.
For manoeuvring, the performance of high-speed craft was identified as an important area for
investigation in previous reports. This remains the case. No FP5 projects were proposed in
this area, but it is hoped that this may be included within a future Integrated Project in FP6.

TA3 - Propulsors
For propeller design, the prediction of cavitation volumes and cavitation inception within the
ship wake remains an important gap in hydrodynamic design. The prediction of leading edge
cavitation is an area of particular concern from the point of view of validation. The award of
the EROCAV project in the second call of FP5 was encouraging, but this does not offer the
prospect of any advances in CFD based design techniques or validation.
In the last report, MARNET-CFD recommended that there should be at least one large-scale
project within the framework programme that addresses the outstanding needs of the
propeller design community in the above areas of concern. This has been addressed with the
success of the proposal LEADING EDGE in the final call of FP5.
For the future, the integration of CFD models for propellers and propulsors with RANSE
calculations of the complete hull flow and rudders represents the greatest development need.

TA4 - Offshore
The development of projects EXPRO-CFD and SAFEFLOW within FP5 has addressed a
number of the previously identified gaps in the state of the art. The following therefore is
based on the findings of those projects and extensions to the techniques that have been
developed thus far.
The key problem areas that continue addressed are al associated with deep water and extreme
environments:
       §   Steep wave and green water loading
       §   SPAR Vortex Induced Motions (VIM)
       §   Riser Vortex Induced Vibrations (VIV)




WS Atkins Consultants – Co-ordinators                                                  Page 46
MARNET-CFD Final Report and State
Of the Art Review

The first of these have been addressed in both EXPRO-CFD and SAFEFLOW, and progress
has been made. However, there may still be a need to consolidate and exploit the systems
that have been developed. It is likely that a further EXPRO-CFD JIP will address this.
The SPAR and Riser VIM and VIV problems have not been addressed in any EU research
programmes, but are in fact major problems for ocean engineering in deep water. Both are
current induced phenomena, and in both cases, CFD can be shown to provide the potential for
suitable design solutions.
Calculations of VIM using the systems developed in EXPRO-CFD have shown that it is
possible to obtain acceptable results for lock-in phenomena at high Reynolds number. These
need to be extended to 3D, and assumptions regarding the sensitivity of the response
prediction to CFD mesh quality and structure response model confirmed. Similarly, theses
systems are capable of modelling multiple bodies and riser-riser interaction and clashing in
principle, but this needs to be proven in case studies and validated using full scale data.


5.3        Industry Requirements for Greater Exploitation
In previous reports, the following additional areas of concern to industry were identified:

      1.   CFD flow solver speed and accuracy
      2.   Interface of CFD with CAD systems.
      3.   The need for fast, automatic and high quality grid generation.
      4.   The prediction of manoeuvring to achieve IMO criteria, particularly in shallow water.
      5.   The design of propellers to minimise induced vibrations and noise.
      6.   Tools for fluid structure interaction prediction on offshore floating production systems
           in ultra-deep water.

Items 1, 2 and 3 remain as major concerns, as denoted by the results of the most recent
MARNET-CFD questionnaire.
It is expected that item 4 will be addressed in forthcoming proposals for and Integrated
Project in FP6. Items 5 and 6 have been addressed to some extent in FP5 as noted above, but
clearly more could be done.
Some new areas of concern have arisen as the result of discussions within MARNET-CFD
and with related networks. Briefly, these are:

1. In offshore, the further development of CFD fluid-structure interaction tools for deep-
   water riser systems, particularly as they relate to problems with sea-bed behaviour.
2. Sub-sea processing, i.e. applications of CFD to automatic separation of oil, gas and water
   at depth.
3. Optimisation: in its broadest sense, developing techniques appropriate to the naval
   architect for hull forms, propellers and combinations of the same, and reducing
   computing times.
4. Non-linear systems modelling: wherein CFD calculations are use in Monte-Carlo like
   simulations with irregular inputs and randomly chosen initial conditions, and combined in
   a number of applications, to provide data in a format for reliability type assessments.
   These include fluid loading problems and safety applications such as in capsize.




WS Atkins Consultants – Co-ordinators                                                      Page 47
MARNET-CFD Final Report and State
Of the Art Review

Items 3 and 4 here have the common theme of requiring very large, parallel, computing
resources at the present time, and hence research might be focused also on ways in which to
make such systems more efficient and cost-effective. Again, this is likely to be included
within the forthcoming IP proposal in FP6.




WS Atkins Consultants – Co-ordinators                                              Page 48
MARNET-CFD Final Report and State
Of the Art Review

6.0    BUDGET STATUS

The contracted total budget for the MARNET-CFD Thematic network was ,199.7 K Euros.
This was originally intended to run for the period 1/10/1998 until 30/09/2001 but the end date
was first extended to 30/09/2002, and then completed 31/03/2003.
The state of the budget was a continued concern to the Coordinator and members of the
Steering Committee throughout.
The initial payment of 40% of the total budget was paid and distributed among the
participants in proportion to their total budgets. The first year cost statements were submitted
in a reasonably timely manner (within 2 months). However, some participants failed to carry
out this simple piece of administrative work, and although the payment was eventua lly made
by the Commission, it was made some 11 months after the end of year 1. By this time the
Network participants were expressing their deep concern, particularly as the need to submit
cost statements for the second year were pressing.
The first year cost statements were not subjected to any recoveries of the initial payment, and
instructions were supplied with the payment regarding its distribution, which the coordinator
duly acted upon.
In hindsight, these initially generous payments by the Commission were mistaken, and left a
requirement to recover much greater proportions on the initial payments later in the Network.
A further difficulty started to arise during the second year, with a number of participants
reducing their activities and participation in meetings. These same participants later failed
consistently to submit cost statements despite many communications warning them of the
need to do so. As a result, the overall rate of spend dropped considerably from the second
year onward.
The second year cost statement was again submitted in good time, and was paid in January
2001. However as the result of the low spend rate and initially generous payments, the
proportion recovered of the initial payment was large and left a balance paid of only 26K
Euros. The same followed with the 3rd year payment which was only 57K Euros, although in
this case it was the result of confusion within the Commission Contracts branch, who had not
accounted for the extension of the Contract from 3 to 4 years, and duly recovered all of the
initial payment in the third year cost statement.
At this point the Coordinator pointed out these mistakes and a number of other issues of
concern, and a one off interim payment was made of 96K Euros to support some of the active
participants who were experiencing difficulty with cash flow.
At no time in all of these difficulties did it look as though the Network would exceed its
original budget however.
The following two tables provide a complete description of all payments claimed through
cost statements, and paid by the Commission.




WS Atkins Consultants – Co-ordinators                                                   Page 49
MARNET-CFD Final Report and State
Of the Art Review

   SUMMARY OF ALL COST STATEMENTS FOR MARNET CFD

                                                                              SUM OF                     NEW SUM    CONSOLIDATED              COMMENTS                  TOTAL
   Name of Partner      Original   1st yr    2nd yr    3rd year 4th year    STATEMENTS    4.5th year   TO NETWORK    STATEMENTS                                        AMOUNT
                        Budget                                              TO 4th YEAR                     END                                                        CLAIMED
   WS Atkins            280200     94165     85182      81467   67235          328049       61354         389403       389403                                           389403
   Sirehna              187125     65686     60839      47263   33074          206862       12801         219663       220313               Exchange rate ?             220313
   AESA/IZAR             31275        0        0           0       0               0          0              0                                                             0
   Flowtech              64650        0        0        49944   15259           65203       6940           72143        72143                   Correct                  72143
   VTT                   64650     20750     15092      20961    8297           65100       10291          75391       75391.2                  Correct                  75391
   Imperial college      64650        0      16210      16678 13524.46        46412.46     6995.58       53408.04      66587.12       Missing 1st year statement ?       66587
   Germanisher Lloyd     25200     10760     5483      2082.55 2994.64        21320.19     687.13        22007.32      21956.81             Exchange rate ?              22007
   HSVA                  25200      6106     1767        8425  5207.34        21505.34     7999.69       29505.03      29506.54                 Correct                  29506
                                                                                   0
   FDS                  15750         0        0        584.3       0           584.3         0           584.3                                                           584
   Odense               15750       2386       0                    0            2386         0           2386
   FSG                  15750       1855     4293       5705        0           11853      3113.77      14966.77       14964.06                 Correct                  14967
   J.L Meyer            15750         0        0                    0              0          0             0                                                              0
   Chantiers de L'A     15750         0      8662                   0            8662      1042.47       9704.47       6216.89        Adjustment due to CA errors         6217
   Fincantieri          15750         0        0                    0              0          0             0          9683.08             Only this available            9683
   Bazan                15750       3071       0                    0            3071         0           3071                                                            3071
                                                                                   0
   DMI                  15750       3472     2295       1067        0            6834         0           6834                                                           6834
   BEDC                 15750         0        0                    0              0          0             0                                                             0
   MARIN                15750        440     1823       3136        0            5399       704.81       6103.81       6105.47                  Correct                  6105
   DERA                 15750      116.22   1100.27              3373.51         4590         0           4590          4590                    Correct                  4590
                                                                                   0
   DNV                  15750       2250     2461                   0            4711        2066         6777          9351          Missing 4th year statement ?       9351
   Bureau Veritas       15750         0      3800       1839     2267.17       7906.17      992.34       8898.51       8899.12                  Correct                  8899
   RINa                 15750        519       0        1154      1043           2716          0          2716         2715.77                  Correct                  2716
                                                                                   0
   Numeca               15750       2750     6752       2610     2553.54      14665.54     1137.2       15802.74       15773.22                    OK                     15803
   IRCN                 15750       1959     2524       1357                     5840      2211.46       8051.46       9924.25        Missing 4th year statement ?         9924
   ECN                  15750       2368       0                                 2368      6857.03       9225.03                                                           9225
   TU Berlin            15750       2543     1156       1022                     4721      4651.01       9372.01                                                           9372
   IST Lisbon           15750       3844     5565       3006     3001.33      15416.33     1093.13      16509.46       16510.9                     OK                     16511
   Uni. Southampton     15750       1311     1814       9560                    12685         0           12685        12686.25                    OK                     12686
   CEAT - Poitiers      15750       2411     2542       8080      2715.9       15748.9        0          15748.9        15750                      OK                     15750
   CIMNE                15750       8018     5038       5327     6049.42      24432.42     2824.65      27257.07       27259.09                    OK                     27259
   TU Denmark           15750       3706                2027                     5733      1108.97       6841.97       6971.08              OK - some rounding             6971
   Uni. Liege           15750        212      426                                 638         0            638                                                             638
   UCL                  15750       1163       0                 5800.82       6963.82     678.81        7642.63       6479.63 Missed 1st year in Consolidated statement 7643
   TU Hamburg           15750       2014     2590       2615                     7219         0           7219                                                             7219
   TU Athens            15750       2759     2319       2641     1033.01       8752.01        0          8752.01             Missed final 6 months in Consolidated statement
                                                                                                                       9003.21                                             9003
   Uni. Glasgow         15750       1333     1366       1652     1032.55       5383.55     864.09        6247.64       6249.45                     OK                      6249
   Uni. Oxford          15750      888.57    1077      1548.08   3459.41       6973.06     440.84        7413.9         7413.6                     OK                      7414
                TOTAL   1199700    248856   242176.3   281751    177921.1   950704.09     136854.98    1087559.07    1071846.74                                       1110034



WS Atkins Consultants – Co-ordinators                                                                      Page                                                                   50
MARNET-CFD Final Report and State
Of the Art Review

         SUMMARY OF ALL RECEIPTS OF PARTICIPANTS IN MARNET-CFD

                               Initial            2nd/3rd  Additional  SELECTED    TOTALS          4th Year      TOTAL       OUTSTANDING     INBALANCES
         Name of Partner      Payment    1st Yr     Year  Balances and BALANCE    RECEIVED      payment from   RECEIVED      DIFFERENCES   OWED      OWING
                                                             Returns   PAYMENTS   To 3rd Year    Commission    to 4th Year
         WS Atkins            112080     89900     84177      30290                 286157          41382        327539         61864                61864
         Sirehna               74850     65686                           8992       149528          40000        189528         30785                30785
         AESA/IZAR             12510       0                                         12510             0          12510        -12510      12510
         Flowtech              25860       0                            8992         34852          10000         44852         27291                27291
         VTT                   25860       0                            8992         34852          20000         54852         20539                20539
         Imperial college      25860       0                            8992         34852           7500         42352         24235                24235
         Germanisher Lloyd     10080     10760                                       20840                        20840          1167                1167
         HSVA                  10080      4829                                       14909          3250          18159         11347                11347
                                                                                       0                                            0
         FDS                   6300        0                                          6300                       6300           -5716      5716
         Odense                6300       2386                                        8686                       8686           -8686      8686
         FSG                   6300       1855                                        8155          2000         10155           4812                4812
         J.L Meyer             6300        0                                          6300                       6300           -6300      6300
         Chantiers de L'A      6300        0                                          6300                       6300             -83       83
         Fincantieri           6300        0                                          6300                       6300            3383                3383
         Bazan                 6300        0                                          6300                       6300           -3229      3229
                                                                                       0                                            0
         DMI                   6300       3457                                        9757                        9757          -2923      2923
         BEDC                  6300        0                                          6300                        6300          -6300      6300
         MARIN                 6300       440                                         6740                        6740           -635      635
         DERA                  6300        0                                          6300                        6300          -1710      1710
                                                                                       0                                            0
         DNV                   6300       2406                                        8706                        8706           645                  645
         Bureau Veritas        6300        0                                          6300                        6300           2599                2599
         RINa                  6300       519                                         6819                        6819          -4103      4103
                                                                                       0                                            0
         Numeca                6300       2726                                        9026          3000         12026           3777                3777
         IRCN                  6300       1959              2409.55                 5849.45                     5849.45        4074.55               4074
         ECN                   6300       2368                                        8668                       8668            557                  557
         TU Berlin             6300       2543                                        8843                       8843            529                  529
         IST Lisbon            6300       3844                                       10144          3000         13144           3367                3367
         Uni. Southampton      6300       1311                                        7611                       7611            5075                5075
         CEAT - Poitiers       6300       2263                                        8563          4000         12563           3187                3187
         CIMNE                 6300       8018                                       14318          5000         19318           7941                7941
         TU Denmark            6300       3706                                       10006                       10006          -3035      3035
         Uni. Liege            6300       212                                         6512                       6512           -5874      5874
         UCL                   6300       1256                                        7556                       7556              87                 87
         TU Hamburg            6300       2014                                        8314                       8314           -1095      1095
         TU Athens             6300       2742                                        9042                       9042             -39       39
         Uni. Glasgow          6300       1330              3269.55                 4360.45                     4360.45        1888.55              1888.55
         Uni. Oxford           6300         0                                         6300                       6300            1114                1114
                      TOTAL   479880     218530                                   812875.9         139132      952007.9       158026.1     62238   220263.55




WS Atkins Consultants – Co-ordinators                                                                   Page                                                   51
MARNET-CFD Final Report and State
Of the Art Review

The following conclusions can be drawn from these tables:

   1. Taking into account previous records and participants consolidated cost statements,
      the total amount claimed is 1,110,034 Euros against an original budget of 1,199,700
      Euros.
   2. The total amount paid out by the Commission to date is 952,007.9 Euros, taking into
      account all statements and recoveries of initial payments up to and including the end
      of the 4th year of the Network.
   3. As the result of the factors described earlier, there are a significant number of
      participants who have under-spent not only their original budgets, but also the sum of
      their initial and 1st year payments. The particular organisations are denoted in the
      second table on payments made by an imbalance on the “owed” side of the final two
      columns.
   4. The total under-spend by those participants is 62,238 Euros.
   5. The amount owed to active participants that have contributed to the work carried out
      in MARNET-CFD and to its deliverables is 220,263 Euros.
   6. The overall balance owed is 158,026 Euros.

If the Commission were to pay the remaining 158,026 Euros, the imbalance of 62,238 Euros
will have to be taken by the remainder participants. This as been foreseen as a major problem
for some time and considerable efforts have been made to recover these owed budgets
(indeed the amount owed has historically run at well over this amount and up to 126K Euros
at one time).

If the Commission were to pay only the participants currently owed, the total to be paid out
would still be les than the original budget (952,007 + 220,263 = 1,172,270).

Efforts will continue to recover the unspent funds among those organisations note above.

As the result of this experience, we would suggest that the mechanisms involved in funding
such Networks be reviewed. Although we are aware that changes were made between FP4
and FP5 the following should also be considered.

   1. The one year payment cycle is too long and causes may difficulties with cash flow
      among participants – it should be shortened.
   2. The Commission’s insistence in providing instructions to the Coordinator to pay
      participants initially and in the first year in this case caused difficulties later on.
   3. A better solution would be to provide the Coordinator with the funds in advance as
      normal, but to allow participants to invoice the Coordinator quarterly, supported by
      evidence. In this way greater control over the funds would be achieved.




WS Atkins Consultants – Co-ordinators                                                 Page 52
MARNET-CFD Final Report and State
Of the Art Review

7.0      CONCLUSIONS AND RECOMMENDATIONS

MARNET-CFD started with the idea that improved coordination, dissemination of ideas, the
sharing of knowledge and the provision of some tools (in the form of validation data and best
practice guidelines), would support and accelerate the development and exploitation of CFD
among key stakeholders in marine technology in Europe.
The Thematic Network “vehicle” that was used, being an FP4 Network, was somewhat
different from later ideas in this area, and was certainly highly novel for the marine industry.
Nevertheless, 37 participated in the initial formation of the Network, and a further 12
organisations became affiliated later, contributing considerably to the technical meetings and
similar outputs.
The key objectives of the Network have undoubtedly been met in so far as:
•     It has provided a network within the European Union for shipbuilders, naval architects,
      offshore engineers and consultants, research institutes and towing tanks, marine
      manufacturers, Classification Societies and Universities, to co-ordinate their efforts in the
      developme nt and exploitation of computational fluid dynamics for all marine
      applications.
•     It has established a shared database of experimental and computational research results
      for a variety of marine and offshore vehicle forms, generic geometries and operating
      scenarios.
•     It has coordinated efforts within the European marine and offshore industries in the
      validation of predictive techniques in hydrodynamic design.
•     It has provided best practice guidelines for the application of CFD to common marine
      problems.
•     It has provided a regular review of the state-of-the-art in Europe, including a survey of the
      CFD tools available (including interfaces to CAD) to the industry.
•     It has established an overview of the level of confidence in CFD within the industry that
      can be used to direct the development of future programmes.
•     It has stimulated Industry to pose the “grand challenge” problems, for all technical areas
      open to investigation using CFD, that generate progress in the research community, as
      evidenced in some of the papers presented at Network Workshops, and FP5 projects
      developed.
•     It has held 4 annual workshops with the purpose of comparing progress in the application
      and validation of numerical models, and subsequently produced proceedings that reflect
      well the state-of-the-art for the benefit of practising designers and hydrodynamicists.
•     It has developed a Web site on the Internet, containing all of the key deliverables and
      stimulating interest world-wide in its activities.
•     It has supported the development of new consortia for research within the Fifth
      Framework Programme to further European competitiveness in marine design and
      construction, many of which have led to projects with a direct impact on CFD validation
      and development.




WS Atkins Consultants – Co-ordinators                                                      Page 53
MARNET-CFD Final Report and State
Of the Art Review

The originally proposed 3 year programme was extended to 4.5 years both as the result of
budget under-spend, and in order to achieve all of the above over a timescale that took in the
whole of the FP5 programme.
The main achievements of the Network however can be summarised as:
•   The annual workshops, that have regularly been able to generate close to 20 good quality
    technical papers, and have supported the development of FP5 project consortia.
•   The best practice guidelines for marine CFD, developed in collaboration with the
    ERCOFTAC Industrial Advisory Committee, and available to al participants on the web.
•   To stimulate a significant increase in the amount of EU funded marine CFD research
    consortia.
•   The MARNET-CFD database of experimental and computational results.
•   Establishing a clear picture of the state of opinions and confidence in the use of CFD in
    marine applications.
The elements of the Network that did not live up to expectations were few but significant,
particularly to coordination aspects. They were that:
•   Not all of the original participants were properly involved through-out, and this has
    caused significant difficulties in the management of the funding of the Network, the rules
    for which were quite inflexible.
•   Some misunderstandings and delays with cost statements, both with participants and at
    the Commission have led to considerable problems with cash flow that clearly affected
    some participants’ level of commitment.
•   As an FP4 Network, the Contractual format did not allow the majority of participants to
    claim staff costs – this meant that many participants could not contribute as much as they
    would have liked and the full potential of the Network could not be met (Note that FP5
    Thematic Networks are not constrained in this way and are, in this author’s opinion, more
    effective as a result.)
•   The Network has so far been unable to provide a vehicle for the dissemination of FP5
    project results, other than though foreground technical papers presented at the annual
    workshops. More use could have been made of the MARNET-CFD database and web-
    site to make results available in the public domain.

The following recommendations are made in light of the experience gained in MARNET-
CFD.
•   That the Thematic Network can be a strong mechanism for R&D support in this area but
    that a more flexible funding mechanism is required.
•   The state of the art and surveys of opinion within the industry suggest the following key
    priorities for future support in marine CFD:
           o The integration of tools and techniques across all aspects of the analysis
             process (potential flow and RANSE, rigid body and structural dynamics)
           o The improvement of solver speed and accuracy
           o Full scale validation of CFD through improved measurement techniques and
             equipment.



WS Atkins Consultants – Co-ordinators                                                 Page 54
MARNET-CFD Final Report and State
Of the Art Review

           o The introduction of tools for optimisation of form and performance.
•   It is unlikely that a single CFD tool suitable for all applications will be developed, but
    there are many synergies that could be better exploited.
•   The user remains the main area of uncertainty in application of marine CFD, and
    inconsistent practices the chief source of error and inconsistency. It is likely therefore
    that CFD will remain a specialist discipline in design analysis rather than become a
    mainstream process for naval architects.
•   That a review of the means of funding thematic networks be carried out in order that the
    problems experienced in MARNET-CFD be avoided in the future.

Finally, we would recommend that consideration be given to the further support of marine
CFD in Europe through a second Thematic Network in marine CFD at some point in the
future. However, its modus operandi would need to be significantly different in order that
the full potential could be achieved.




WS Atkins Consultants – Co-ordinators                                                 Page 55
MARNET-CFD Final Report and State
Of the Art Review

REFERENCES

Abdel-Maksoud, M., 2003. "Numerical and Experimental Study of Cavitation Behaviour of a
Propeller". Sprechtag Kavitation, Hamburg, 30 January 2003, 12 pp.

Abdul-Maksoud, M., Rieck, K., Menter, F., 2000. "Unsteady Numerical Investigation of the
Turbulent Flow around the Container Ship Model (KCS) with and without Propeller". A
Workshop on Numerical Ship Hydrodynamics, Gothenburg, 14-16 September 2000. 6 pp.

Ahuja, V., Hosagadi, A., Arunajatesan, S., 2001. "Simulations of Cavitating Flows Using
Hybrid Unstructured Meshes". Journal of Fluids Engineering, Vol. 123, No. 2, pp. 331-340.

Calcagno, G., Di Felice, F., Felli, M., Franchi, S., Pereira, F., Salvatore, F., 2003. "The
INSEAN E779a Propeller Test Case: a Database for CFD Validation". MARNET-CFD, Final
Annual Workshop, Haslar, 20-21 March 2003. 5 pp.

Chen, B., Stern, F., 1999. "Computational Fluid Dynamics of Four-Quadrant Marine
Propulsor Flow". Journal of Ship Research, Vol. 43, No. 4, pp. 218-228.

Coutier-Delgosha, O., Reboud, J.L., Fortes-Patella, R., 2001. "Numerical Study of the Effect
of the Leading Edge Shape on Cavitation around Inducer Blade Sections".              Fourth
International Symposium on Cavitation, Pasadena, 20-23 June 2001. 8 pp.

El Moctar, O., 1999. "Numerical Investigation of Propeller-Rudder Interaction". NuTTS'99,
Second Numerical Towing Tank Symposium at INSEAN, Rome, 2-3 August 1999, 4 pp.

Farrell, K.J., 2001. "Eulerian/Lagrangian Analysis for the Prediction of Cavitation Inception".
Fourth International Symposium on Cavitation, Pasadena, 20-23 June 2001. 8 pp.

Gindroz, B., Hoshino, T., Pylkkänen, J.V., eds., 1998. “22nd ITTC Propulsion Committee
Propeller RANS/PANEL Method Workshop Proceedings”, Grenoble, France, 5-6 April 1998.
Privately Printed, Val de Reuil.

Han, J.-M., Kong, D.-S., Song, I.-H., Lee, C.-S., 2001. "Analysis of the Cavitating Flow
Around the Horn-Type Rudder in the Race of Propeller". Fourth International Symposium
on Cavitation, Pasadena, 20-23 June 2001. 8 pp.

Hsiao, C.-T., Chahine, G.L., 2001. "Numerical Simulation of Bubble Dynamics in a Vortex
Flow Using Navier-Stokes Computations and Moving Chimera Grid Scheme". Fourth
International Symposium on Cavitation, Pasadena, 20-23 June 2001. 10 pp.

Hsiao, C.-T., Chahine, G.L., 2002. "Prediction of Vortex Cavitation Inception Using Coupled
Spherical Models and UnRANS Computations". 24th SNH, Fukuoka, 8-13 July 2002.

Hsiao, C.-T., Pauley, L.L., 1999. "Numerical Computation of Tip Vortex Flow Generated by
a Marine Propeller". Journal of Fluids Engineering, Vol. 121, No. 3, pp. 638-645.

Hu, P., Zangeneh, M., 2001. "CFD Calculation of the Flow Through a Water-Jet Pump".
International Conference Waterjet Propulsion III, Gothenburg, 20-21 February 2001. 10 pp.


WS Atkins Consultants – Co-ordinators                                                  Page 56
MARNET-CFD Final Report and State
Of the Art Review



Huntsman, I., Hothersall, R., 2001. "Development of Quasi 3D Design Methods and 3D Flow
Solvers for Waterjet Development". International Conference Waterjet Propulsion III,
Gothenburg, 20-21 February 2001.

ITTC, 1996. “Report of the Cavitation Committee". In 21st International Towing Tank
Conference Proceedings, Vols. I & II. Trondheim, 15-21 September 1996. Vols. I & II.

ITTC, 1996. “Report of the Propulsor Committee". In 21st International Towing Tank
Conference Proceedings. Trondheim, 15-21 September 1996. Vols. I & II.

ITTC, 1999. “Report of the Propulsion Committee". In 22nd International Towing Tank
Conference Proceedings, Seoul & Shanghai, 5-11 September 1999. Vols. I & III.

ITTC, 2002. “Report of the Propulsion Committee". In 23rd International Towing Tank
Conference Proceedings, Venice, 8-14 September 2002. Vol. I.

ITTC, 1999. “Report of the Resistance Committee". In 22nd International Towing Tank
Conference Proceedings, Seoul & Shanghai, 5-11 September 1999. Vols. I & III.

ITTC, 1999. “Report of the Specialist Committee on Computational Methods for Propeller
Cavitation". In 22nd International Towing Tank Conference Proceedings, Seoul & Shanghai,
5-11 September 1999. Vols. II & III.

ITTC, 1999. “Report of the Specialist Committee on Cavitation Induced Pressure
Fluctuations". In 22nd International Towing Tank Conference Proceedings, Seoul &
Shanghai, 5-11 September 1999. Vols. II & III.

ITTC, 2002. “Report of the Specialist Committee on Cavitation Induced Pressures". In 23rd
International Towing Tank Conference Proceedings, Venice, 8-14 September 2002. Vol. II.

ITTC, 1999. “Report of the Specialist Committee on Unconventional Propulsors". In 22nd
International Towing Tank Conference Proceedings, Seoul & Shanghai, 5-11 September
1999. Vols. II & III.

ITTC, 1999. “Report of the Specialist Committee on Waterjets". In 22nd International
Towing Tank Conference Proceedings, Seoul & Shanghai, 5-11 September 1999. Vo ls. II &
III.

Kaarlonen, K., 2002. "Numerical Investigation of the Flow Around Rounded Delta Wing" (in
Finnish). Espoo, Helsinki Univ of Tech, Laboratory of Aerodynamics. Report B-53. 85 pp.

Koyama, K., 1993, “Comparative Calculations of Propellers by Surface Panel Method –
Workshop Organized by 20th ITTC Propulsor Committee“. Papers of Ship Research Institute,
Supplement No. 5.

Kubota, A., Kato, H., Yamaguchi, H., 1992. "A New Modelling of Cavitating Flows: A
Numerical Study of Unsteady Cavitation on a Hydrofoil Section". Journal of Fluid
Mechanics, Vol. 240, pp. 59-96.


WS Atkins Consultants – Co-ordinators                                             Page 57
MARNET-CFD Final Report and State
Of the Art Review



Laurens, J.-M., Grosjean, F., 2002. "Numerical Simulations of Propeller-Rudder Interface".
Cavitating Hydrofoils". . NuTTS'99, 5th Numerical Towing Tank Symposium, Pornichet, 29
Sep-1 October 2002, 5 pp.

Lee, H., Kinnas, S.A., 2001. "Modelling of Unsteady Blade Sheet and Developed Tip Vortex
Cavitation". Fourth International Symposium on Cavitation, Pasadena, 20-23 June 2001. 12
pp.

Lindau, J.W., Kunz, R.F., Boger, D.A., Stinebring, D.R., Gibeling, H.J., 2002. "High
Reynolds Number, Unsteady, Multiphase CFD Modeling of Cavitating Flows". Journal of
Fluids Engineering, Vol. 124, No. 3, pp. 607-616.

Lindenau, O., Streckwall, H., Bertram, V., 2002. "RANSE Simulations for Cavitating
Hydrofoils". NuTTS'99, 5th Numerical Towing Tank Symposium, Pornichet, 29 Sep-1
October 2002, 4 pp.

Lobachev, M.P., Tchitcherine, I. A., 2001. "The Full-Scale Resistance for Podded Propulsion
System by RANS Method". SP 2001: Lavrentiev Lectures, St. Petersburg, 19-21 June 2001.
Pp. 39-44.

McDonald, H., Whitfield, D., 1996. "Self-Propelled Manoeuvering Underwater Vehicles".
21st SNH, Trondheim, 24-28 June 1996. 12 pp.

Ohashi, K., Hirata, N., Hino, T., 2002. "Numerical Simulation of Ship Flows with Contra-
Rotating Propeller Effects". Transactions of the West-Japan Society of Naval Architects, No.
104, pp. 15-24.

Sánchez-Caja, A., Ory, E., Salminen, E., Pylkkanen, J.V., Siikonen, T., 2003. "Simulation of
Incompressible Viscous Flow Around a Tractor Thruster in Model and Full Scale". Abstract
accepted for presentation at 8th NSH, Busan, 2003. 11 pp.

Sánchez-Caja, A., Rautaheimo, P., Siikonen, T., 2000. "Simulation of the Incompressible
Viscous Flow Around a Ducted Propeller Using a RANS Equation Solver". 23rd Symposium
on Naval Hydrodynamics, Val de Reuil, 17-22 September 2000.

Seil, G.J., 2001. "The Effect of scale and Shaft Rotation on the Flow in a Waterjet Inlet".
International Conference Waterjet Propulsion III, Gothenburg, 20-21 February 2001.

Song, C.C.S., He, J., 1998. "Numerical Simulation of Cavitating Flows by Single-Phase Flow
Approach". Third International Symposium on Cavitation, Grenoble, 7-10 April 1998. Pp
295-300.

Song, C.C.S., Qin, Q., 2001. "Numerical Simulation of Unsteady Cavitating Flows". Fourth
International Symposium on Cavitation, Pasadena, 20-23 June 2001. 8 pp.

Streckwall, H., 2003. "Numerical Models for Cavitation and Propeller Induced Pressure
Fluctuations". STG Sprechtag Kavitation, Hamburg, 30 January 2003, 8 pp.




WS Atkins Consultants – Co-ordinators                                               Page 58
MARNET-CFD Final Report and State
Of the Art Review

Streckwall, H., 1999. "Numerical Techniques for Propeller Design". Wegemt, Hamburg,
1999. 16 pp.

Szantyr, J.A., 2000. "Analytical Methods for Prediction of Propeller Cavitation and Its
Consequencies". NCT'50 International Conference on Propeller Cavitation, University of
Newcastle, 3-5 April 2000, pp. 37-56.

Szantyr, J.A., 2003. "A Surface Panel Method for Hydrodynamic Analysis of Pod
Propulsors". RINA Paper, pp. 37-56.

Yakushi, R., Yamaguchi, H., Kawamura, T., Maeda, M., Sakota, M., 2001. "Investigation for
Unsteady Cavitation and Re-entrant Jet on a Foil Section - Approach by Experiments and
CFD" (in Japanese). Journal of the Society of Naval Architects of Japan, Vol. 190, pp. 61-
74.

Young, Y.L., Kinnas, S.A., 2001. "A BEM for the Prediction of Unsteady Midchord Face
and/or Back Propeller Cavitation". Journal of Fluids Engineering, Vol. 123, No. 2, pp. 311-
319.

Young, Y.L., Kinnas, S.A., 2000. "Prediction of Unsteady Performance of Surface-Piercing
Propellers". Propellers/Shafting Symposium 2000, Virginia Beach, 20-21 September 2000.
Pp. 7-1...7-9.




WS Atkins Consultants – Co-ordinators                                              Page 59
MARNET-CFD Final Report and State
Of the Art Review




                                        APPENDIX I



                          Best Practice Guidelines for
                                  Marine CFD

                                        Final Version




WS Atkins Consultants – Co-ordinators                    Page 60
MARNET-CFD Final Report and State
Of the Art Review




WS Atkins Consultants – Co-ordinators   Page 61
MARNET-CFD Final Report and State
Of the Art Review




                                        APPENDIX II



                              FINAL STATE-OF-THE-
                              ART QUESTIONNAIRE




WS Atkins Consultants – Co-ordinators                 Page 62
MARNET-CFD Final Report and State
Of the Art Review


        MARNET-CFD Questionnaire - 2002

       Name:
       Company/Organisation:
       Email:


       1. Which of the following describes the capabilities, facilities, and abilities of your organisation?


              Towing Tank                             Wave Basin                           Cavitation Tunnel
              CFD user                                CFD developer                        CFD software supplier
              Ship design                             Ship building                        Marine consultancy
              Propeller design                        Propeller manufacture                Waterjet systems
              Rudders/stabilisers etc.                Ship machinery                       Cargo handling
              Research                                Education                            Classification/approval
              Naval architecture                      Offshore engineering                 Marine engineering
              Naval vessels                           Merchant vessels                     Passenger vessels
              Yachts/small craft                      Fast ferries                         Offshore prod. systems


       2. How much of your work involves CFD?

              0-20%              21-40%           41-60%             61-80%     81-100%


       3. Which CFD tools do you use?


              CFX4          CFX5          COMET      FLUENT          FINFLO   SHIPFLOW       STAR-CD        Other - please specify:

       4. Which areas of application do you use CFD for?
       Please give your opinion about the accuracy of CFD for these applications

       Accuracy scoring scale: 1: useless, 2: qualitatively correct, 3: quantitatively reliable when supported by
       empirical/experimental input, 4: perfect predictions

       TA 1 - Marine vehicle performance                                                                    Score
       deep water wave resistance
       shallow water wave resistance
       deep water visous resistance
       shallow water viscous resistance
       air resistance
       deep water wave wash
       shallow water wave wash
       deep water delivered power
       shallow water delivered power
       propeller induced vibrations and noise
       boundary layer noise (transition)
       signatures (pressure vortices)
       smoke effects

       TA2 - Hydrodynamic loads and responses                                                               Score
       global loads on ship in waves
       local external loads on ship: slamming
       local external loads on ship: green water
       local internal loads: sloshing
       vibrations
       fire/explosion
       damaged stability
       dynamic stability
       comfort - operability
       control - stabilisation
       speed reduction in waves, added resistance
       course keeping in heavy weather
       statistics: short term, long term, most probable critical event
       conventional manoeuvring data prediction
       high speed manoeuvring
       manoeuvring in waves
       broaching
       loads on appendages
       thrust sizing
       slip




WS Atkins Consultants – Co-ordinators                                                                                                 Page 63
MARNET-CFD Final Report and State
Of the Art Review


        TA3 - Propulsors                                                                                               Score
        Open water flow
        Propeller in inclined flow
        propeller in non-uniform onset flow
        design for prescribed blade pressure distribution
        application of optimisation techniques
        propeller interaction (with rudder, hull, shaft)
        propeller in semi-tunnel
        propulsor performance when manoeuvering, accelerating,decelerating
        propulsor performance in waves
        tip vortex modelling
        Inception prediction of steady cavitation
        Extent prediction of steady cavitation
        unsteady cavitation prediction
        propulsor induced hull pressures
        fluctuating shaft forces
        scale effects
        propulsor design


        TA 4 - Offshore Engineering                                                                                    Score
        Hydrodynamic loading on fixed structures
        Hydrodynamic loading on floating structures
        Hydrodynamic loading on components (e.g. riser pipes)
        Wind loading
        Environmental (e.g. gas dispersion of hazardous gas)
        Fire, explosion
        Internal flows (e.g. ventilation)


        5. Do you tend to use CFD results


           Independent to other info (i.e. use CFD results unmodified)

           Supported by experience (i.e. results from CFD modified by experience)

           Supported by scaled physical model data

           Supported by full scale information



        6. CFD development needs (to make CFD a useful tool in marine applications)
        Give a score for each of the areas in the table below, making sure that the sum of all the scores in the boxes is 100
        (please modify the example below, which gives the same weighting to all areas)


                                                     Accuracy        Applicability   User-friendliness         Other
        CAD surface representation                          5              5                    5                       5
        Grid generation                                     5              5                    5                       5
        Flow solver                                         5              5                    5                       5
        Post-processing                                     5              5                    5                       5
        Other                                               5              5                    5                       5
                                                                                                               sum                      100



        Thank you for filling in this questionnaire
        Please save this file and send it to         Athena.Scaperdas@atkinsglobal.com

        The results of the survey will be published on the MARNET-CFD Website                    http://pronet.wsatkins.co.uk/marnet/
        If you have any queries about this questionnaire, please contact Athena Scaperdas (Atkins) - use email above




WS Atkins Consultants – Co-ordinators                                                                                                         Page 64
MARNET-CFD Final Report and State
Of the Art Review




                                        APPENDIX III



                Proceedings of the Final MARNET-CFD
                         Technical Meeting




WS Atkins Consultants – Co-ordinators                  Page 65
MARNET-CFD Final Report and State
Of the Art Review




WS Atkins Consultants – Co-ordinators   Page 66
MARNET-CFD Final Report and State
Of the Art Review




                                        APPENDIX IV



                        Description and Management
                          Structure of the Network




WS Atkins Consultants – Co-ordinators                 Page 67
MARNET-CFD Final Report and State
Of the Art Review

APPENDIX IV            Description and Management Structure of the Network

IV.1   Description of the Network - Aims and Objectives.
MARNET-CFD is a Thematic Network, which has been set up to support the needs of the
European Shipbuilding and Offshore Industries in computational fluid dynamics (CFD). The
Network was started in November 1998, and has been running for 2 years.
The main objectives of MARNET-CFD are:
•   To provide a network within the European Union for shipbuilders, naval architects,
    offshore engineers and consultants, research institutes and towing tanks, marine
    manufacturers, Classification Societies and Universities, to co-ordinate their efforts in the
    development and exploitation of computational fluid dynamics for all marine
    applications.
•   To establish, maintain and develop a shared database of experimental and computational
    research results for a variety of marine and offshore vehicle forms, generic geometries
    and operating scenarios.
•   To establish a common and concerted approach within the European marine and offshore
    industries to the validation of predictive techniques in hydrodynamic design, and provide
    guidance/codes of practice for application.
•   To provide a regular review of the state-of-the-art in Europe, including a survey of the
    CFD tools available (including interfaces to CAD) to the industry.
•   To stimulate Industry to pose the “grand challenge” problems, for all technical areas open
    to investigation using CFD, that generate progress in the research community.
•   To hold annual workshops with the purpose of comparing progress in the application and
    validation of numerical models, and subsequently produce summaries of the state-of-the-
    art for the benefit of practising designers and hydrodynamicists.
•   To publish a regular newsletter, supported by a Web page on the Internet, containing
    notification of research activities, latest publications, database updates and Network
    activities for all participants.
•   To stimulate the development of new consortia for research within the Fifth Framework
    Programme to further European competitiveness in marine design and construction.


IV.2   Participants
The following is the list of participating organisations within MARNET-CFD. There are a
total of 37 organisations participating as full members of the Thematic Network. A
consortium agreement has been signed, formalising the obligations of the members of the
Network. In addition, other organisations unable for various reasons to join in the first wave
have contacted the co-ordinator and have been able to participate in some of the activities in
the Network without drawing on the funds supplied by the Commission. These additional
organisations have been regarded as “Affiliated Members” of the network and are subject to a
slightly altered form of Consortium agreement.




WS Atkins Consultants – Co-ordinators                                                    Page 68
MARNET-CFD Final Report and State
Of the Art Review



  Participant   Organisation Name              Organisation Type            Member State
  No.
       1        WS Atkins Consultants          Consultants                  UK
        2       SIREHNA                        Consultants                  France
        3       Astelleros Espanoles           Shipbuilders                 Spain
        4       FLOWTECH                       Software/Consultants         Sweden
        5       VTT                            Research Organisation        Finland
        6       Imperial College of Science    University                   UK
        7       Germanischer Lloyd             Classification Society       Germany
        8       HSVA                           Research Organisation        Germany
        9       FDS                            Research Organisation        Germany
       10       Odense                         Shipyard                     Denmark
       11       Flensburger                    Shipyard                     Germany
       12       J.L. Meyer GmbH                Shipyard                     Germany
       13       Chantiers de L-Atlantique      Shipyard                     France
       14       Fincantieri                    Shipyard                     Italy
       15       Bazan                          Shipyard                     Spain
       16       DMI                            Research Organisation        Denmark
       17       Bassin D’Essais de Carenes     Research Organisation        France
       18       MARIN                          Research Organisation        Netherlands
       19       DERA                           Research Organisation        UK
       20       DNV                            Classification Society       Norway
       21       Bureau Veritas                 Classification Society       France
       22       NUMECA                         Software/Consultants         Belgium
       23       IRCN                           Research Organisation        France
       24       Ecole Centrale de Nantes       University                   France
       25       TU Berlin                      University                   Germany
       26       IST – Lisbon                   University                   Portugal
       27       University of Southampton      University                   UK
       28       CEAT – Poitiers                University                   France
       29       CINME                          Research Organisation        Spain
       30       TU Denmark                     University                   Denmark
       31       University of Liege            University                   Belgium
       32       University College London      University                   UK
       33       TU Hamburg                     University                   Germany
       34       TU Athens                      University                   Greece
       35       Registro Italiano Navale       Classification Society       Italy
       36       University of Glasgow          University                   UK
       37       University of Oxford           University                   UK


                          Full List of the MARNET-CFD Membership


IV.3   Management and Structure
The Network reflects the wide ranging interests of the industry, and the versatility of CFD in
solving practical engineering problems. In keeping with the now well established structures
used in Thematic Networks, MARNET-CFD is organised around 4 Thematic Areas, each
with its own technical spheres of interest and TA Co-ordinator. The table on the following
page provides an outline of the subject areas covered by each Thematic Area.




WS Atkins Consultants – Co-ordinators                                                     Page 69
MARNET-CFD Final Report and State
Of the Art Review


 TA No.      Technical Area           Examples of Interests Covered
 1           Marine Vehicle           Steady ship waves, boundary layers & separation, cross-flow,
             Performance              transition, appendage flows, hydro-aerodynamics, performance of
                                      new concepts.
 2           Hydrodynamic loads       Seakeeping, manoeuvring, slamming and deck-wetting, added
             and responses            resistance, internal fluid loads, high speed craft.
 3           Propulsors               Propeller flows, water-jets, cavitation & ventilation, thruster
                                      performance, duct flows.
 4           Offshore Engineering     Wave-loading on structures, breaking wave kinematics, drift forces,
                                      higher order loads and motions, offshore safety etc.


                    Technical areas covered in the MARNET Thematic Network.



The MARNET-CFD Thematic Network is managed through a Network Steering Committee
(NSC), which is made up of the following organisations and personnel:

     Role/Title                     Organisation                 Contact Person
     Co-ordinator                   WS Atkins Consultants        Paul Gallagher
     Assistant Co-ordinator         SIREHNA                      Jean-Paul Borleteau
     Chairman                       HSVA                         Gerhard Jensen
     Shipyard Co-ordinator          Astilleros Espanoles         Eduardo Minguito
     TA 1 Co-ordinator              FLOWTECH Ltd.                Lars Larsson
     TA 2 Co-ordinator              SIREHNA                      Jean-Jacques Maisonneuve
     TA 3 Co-ordinator              V.T.T                        Jaakko Pylkkanen
     TA 4 Co-ordinator              Imperial College             Mike Graham
     RCDG representative            Germanischer Lloyd           Carsten Ostergaard


The Co-ordinator and Assistant Co-ordinator are responsible for the day-to-day running of
the Network. The Thematic Area Co-ordinators have the role of organising the technical
activities within each of their spheres of interest. The shipyard co-ordinator provides a voice
for the industrial membership of the Network, and the RCDG representative provides contact
with the Research Committee of COREDES.

The NSC meets twice per year, usually at the same time as Thematic Area meetings of the
Annual MARNET-CFD Workshop.




WS Atkins Consultants – Co-ordinators                                                             Page 70

				
DOCUMENT INFO