Professional Documents
Culture Documents
MICA(P) 037/02/2011
About KOMtech
Launched in December 2007, Keppel Offshore & Marine Technology Centre (KOMtech)
is an extension and strengthening of research and development initiatives in Keppel
Offshore & Marine. Its mission is to develop competencies, promote innovation, stimulate
and carry out Application Research & Product/Process Development, and engage in
Technology Foresight to create strategic advantages for Keppel Offshore & Marine.
KOMtech complements and augments the work of the existing three design and
engineering units within Keppel Offshore & Marine - Offshore Technology Development,
Deepwater Technology Group and Marine Technology Development. Leveraging existing
and proprietary technologies, and in collaboration with universities, research institutes
and industry partners worldwide, KOMtech continues to develop innovative solutions
that are commercially viable and adaptable to the needs of the industry.
Ongoing research efforts in KOMtech include Arctic structures, drilling systems and
mini-LNG supply-chain for associated gas. Current areas of research also include
development of a Slim Drillship and Offshore Wind Energy research on self-installing
platforms for substations, wind turbine foundations, wind turbine installation vessels
and cable laying vessels.
Chairmans Message
Commercial viability has always been the key focus of Keppels technology
innovation efforts. It forms the guiding principle for KOMtech in conducting
the research and development of new solutions and enhancements to products
and processes.
Each year, a good sampling of the vast research work by KOMtech is showcased
in the Technology Review. Since the inaugural edition in 2009, KOMtech
has published 43 papers on a spectrum of concepts targeted at improving
the reliability, effectiveness and sustainability of various offshore operations.
Of these innovations, 19 provisional patents have been led and will be further
developed into concrete solutions for our customers in the near future.
Previous issues of the Review have attracted substantial interest from the
offshore and marine industry, which includes our customers, professional
institutions and academia. Several of KOMtechs papers, such as the
Conceptual Development of a Slim Drillship, have also been republished in
journals by the Society of Naval Architects and Marine Engineers Singapore.
It is with great pleasure that I present you with the much anticipated third
KOMtech publication, Technology Review 2011. This latest edition brings
KOMtechs multidisciplinary R&D expertise to bear, with a look at trends
for Arctic offshore structures, innovative vessel launching methods, as well as
new cable laying solutions for offshore wind farms.
Technology Review 2011 is another outstanding collection of rigorously
documented research produced by KOMtechs team of able researchers, led
by Centre Director, Dr Foo Kok Seng and Advisor/Director Mr Charles
Foo. The publication also brings together inputs from our research partners
and customers, who represent some of the best minds and authorities in
our industry.
I trust that you will nd this third edition of the Review equally stimulating
and path-nding.
Yours truly,
Choo Chiau Beng
Chairman
Keppel Offshore & Marine
2 KOMtech Technology Review 2011
Contents
About KOMtech
1 Chairmans Message
3 Editorial Note
5 An Integrated System for Improving Geotechnical
Performance of Jackup Rig Installation
A new instrumentation technology integrated with the Jackup to assist operators
with decision making to avoid geotechnical hazards
19 Design of a DP2 Drillship Power Plant Load analyses with different
operational modes are employed to evaluate the vessel power plant
31 Dedicated Cable Laying Vessels for Future Wind Farm
Development KOMtech has developed two types of dedicated cable laying
vessels for the demands of future wind farms
41 Numerical Simulation for Ship Side Launching
Numerical Simulation of ship side launching using a CFD software coupled
with a user dened program
47 CFD-based Resistance and Propulsion
Prediction for Wind Turbine Installation
Jackup Vessels and Semisubmersibles
The high accuracy of the present CFD simulation is
promising for the development of numerical towing tank
techniques for efcient offshore and marine applications
59 Airgap and Wave Run-up Prediction for Semisubmersibles
Using the CFD Simulation Approach An effective tool to predict
the airgap and wave run-up response in extreme sea conditions
69 Nonlinear Control of an Active Heave Compensation System
A nonlinear controller for the active heave compensation system is designed
using backstepping technology
79 Ice-Induced Vibrations of Slender Offshore Structures
A review of the current trends and challenges to understand the
mechanism of ice-induced vibrations of slender offshore structures
87 A New Pipe Handling Approach for a Deepwater
Slim Drillship A new pipe handling approach is developed
to integrate with, and enhance, the operational capabilities of
the slim drillship
97 Revisiting the Grouted Connection KOMtech has revisited
the grouted connection for use in self installing platforms, wind turbine
foundations and extendable draft semi connections
Editorial Note
This marks the third publication of our Technology Review. As with last years review, this
publication highlights some continued research as well as some new research directions of
KOMtech. Continuing areas of research include wind turbine installation vessels, the slim
drillship, LNG systems and research into arctic structures. We are also pleased to have
new papers on a diverse range of topics including cable laying vessels, control systems and
a geotechnical investigation system designed to further enhance Keppels well established
Jackup designs.
These papers present a snapshot of the ongoing work in KOMtech and are intended to
stimulate further thoughts and innovations and to showcase some of the many competencies
available and being developed in KOMtech. Every attempt has been made to ensure accurate
and factual representation of the research presented, however it is important to note that
much of this work is ongoing and represents the opinions and views of the authors and not
necessarily an ofcial position of KOMtech.
We have enhanced our collaboration with Keppel Business Units, Universities and Industry
over the past year and their contribution to our research and to the papers presented in
this publication are much appreciated. Technology Review is the result of many of hours
of hard work from the authors and editors, and I would like to thank all involved for their
hard work and support.
Michael Perry
Chairman, Editorial Committee
EDITORIAL COMMITTEE :
Advisors Mr Charles Foo and Dr Foo Kok Seng
Editorial members Dr Michael Perry, Dr Ronald Chew, Dr Henry Krisdani,
Mr Alex Tan, Mr Wayne Yap, Ms Sarah Pang and
Ms Wong Chai Yueh (Group Corporate Communications)
105 Emissions from Ships: Upcoming Regulations
and Their Implications A look at recent trends likely to affect
environmental regulations and potential mitigation measures and
technologies which may allow the industry to conform these regulations
115 LNG Regasication Selection System
LNG regasication system selection methodology for offshore
applications using the KOMtech in-house design master program
123 Re-condensing Systems for LNG Boil Off Gas
Various technical solutions for managing BOG are reviewed for the
purpose of optimal selection based on technical and commercial aspects
131 Offshore Oil & Gas Produced Water Treatment
Technologies: An Overview Strict new environmental
regulations are driving a relook at the management of produced water
141 Energy Audit An important rst step in the overall process of
reducing energy costs and carbon emissions
4 KOMtech Technology Review 2011
4 KOMtech Technology Review 2011
No part of the materials published in this journal may be reproduced, stored in a retrieval system or transmitted in any form whatsoever without the prior written permission of KOMtech
An Integrated System for Improving Geotechnical
Performance of Jackup Rig Installation 5
PREDICTABILITY AND CONSEQUENTIAL SAFETY OF JACKUP RIGS during
installation and removal remain non-trivial issues for the industry despite larger
deployment of jackup rigs and operations in emerging frontier regions with more
complex soil conditions. Jackup foundation hazards such as unpredicted leg
penetration, rapid leg penetration, punch-through, spudcan-footprint interaction
and leg extraction difculties continue to occur in spite of the industry stepping up
efforts to better manage these risks. Apart from improvements of the installation
guidelines and practices within the industry as well as implementation of proper
site specic assessments, safer performance of jackup rigs may be achieved
through advancement in jackup instrumentation technology. In this paper, a new
instrumentation technology integrated with jackup rigs is proposed to assist the
jackup operators in making decision and taking measures to prevent or mitigate
potential geotechnical hazards, particularly punch-through.
ACKNOWLEDGEMENT
The authors would like to acknowledge IADC (International Association of Drilling
Contractors) and SPE (Society of Petroleum Engineers) for the permission to
reproduce the paper, originally prepared for presentation at the IADC/SPE Asia
Pacic Drilling Technology Conference and Exhibition held in Ho Chi Minh City,
Vietnam, 13 November 2010.
In addition, the authors would like to thank Steve Nowak of Integrity Offshore
for provision of the eld data. Dr Chen Hongs work on the development of
leg punch-through simulation software is greatly appreciated. Contribution of
Dr Shazzad Hossain on discussion of the application of eld penetrometer data
for spudcan penetration prediction is gratefully acknowledged.
An Integrated System for
Improving Geotechnical
Performance of
Jackup Rig Installation
Matthew QUAH, PhD, CEng, CMarEng, FIMarEST
Henry KRISDANI, PhD, B.Eng
Okky Ahmad PURWANA, PhD, M.Eng, B.Eng
Julianto CAHYADI, MSc, B.Eng
Mark RANDOLPH, PhD, Winthrop Professor (University of Western Australia)
6 KOMtech Technology Review 2011
INTRODUCTION
A unique feature of jackup rigs is self-installing
capability which differentiates its foundation from
conventional terrestrial foundations. A jackup rig
is typically installed by pumping ballast water into
the hull until its legs carry a certain load equivalent
to the anticipated maximum footing reaction under
design storm condition plus some safety margin.
Under the imposed load, the legs (simultaneously
or sequentially) continue penetrating the seabed
until bearing capacity of the foundation soil equals
the load. This process, often called preloading,
is essentially installing and proof loading the
foundation by imposing sufciently large gravity
load. Unlike installation of conventional shallow
foundations in land where the foundation is placed
at a pre-determined embedment depth through a
ground preparation, the jackup foundation will
naturally nd its nal embedment depth after force
equilibrium in the foundation soil is satised.
Due to its nature of installation, several potential
geohazards can be present during installation of
jackup foundations. Osborne (2005)
[1]
described in
detail typical geohazards associated with jackups and
their statistics. Spudcan-footprint interaction and leg
punch-through are identied as the most common
geohazards which may lead to serious consequences.
In addition to potential structural damage to the
jackup leg and their holding systems, such incidents
may result in collision with the adjacent platform
and worse still endanger personnel on board. Punch-
through is often dened as sudden leg penetration
due to unpredictable drop in bearing capacity in
which the legs cannot be jacked fast enough to
restore the hull levelness. The old perception that
punch-through is uncontrolled leg penetration
has been changing with the advancements in jackup
technology and geotechnical knowledge.
Beyond operational procedures applied by the jackup
operator during installation, the occurrence of punch-
through may be attributed to inaccurate spudcan
penetration prediction and/or poor monitoring of
load-penetration response. The former is associated
with the quality of geotechnical survey and assessments
conducted for the intended site. Inadequate site-
specic geotechnical information, improper
geotechnical survey, poor selection/interpretation
of engineering parameters, and inappropriate use of
bearing capacity prediction methods are several factors
that may eventually result in poor leg penetration
prediction. On the other hand, provision of accurate
penetration prediction may become ineffective if
the actual load- penetration response is not tracked
during installation. Without the ability to monitor
the response, occurrence of unpredicted spudcan
behavior cannot be observed and effective preventive
measures cannot be taken in time.
CURRENT PRACTICE
Geotechnical Assessment
Prior to jackup deployment at a site, geophysical and
geotechnical surveys are normally conducted from
a dedicated vessel. The geotechnical investigation
typically consists of continuous borehole sampling
and sometimes combined with continuous
penetrometer tests. An alternate sampling and
penetrometer testing at a single borehole is also
practised in some situations. With the samples
taken, various offshore or onshore laboratory tests
are conducted from which shear strength data
and other engineering parameters are derived.
Some degree of data scatter is typically present
which requires an engineering judgment to derive
a single design shear strength prole or upper and
lower bounds on the prole. Figure 1 demonstrates
Figure 1. Undrained shear strength parameters
from eld and laboratory tests
An Integrated System for Improving Geotechnical
Performance of Jackup Rig Installation 7
an example of available shear strength information
for a jackup site in which some degree of variability
is observed. Together with interpreted layer
boundaries, the design shear strength is used in
the bearing capacity prediction model to produce
load-penetration relationships.
In general, the bearing capacity formulations used
by the assessors are the same but different assessors
have different preferred methods or empirical
adjustment factors and technique when using
certain combinations of bearing capacity equations.
Among several assessors, the predicted nal
spudcan penetration depth is normally similar
despite potentially different trends of load-
penetration response at some intermediate depths.
The variation in the response trends depends on the
complexity of soil layering and methods selected
to tackle the interpreted multi-layered proles. In
Figure 2, spudcan penetration predictions provided
by three different assessors based on the shear
strength information given in Figure 1 indicate a
wide range of predicted penetrations.
In the current practices, establishing an accurate
spudcan prediction curve requires numerous
judgments at various levels of assessment, starting
from selection of soil investigation scope, design
shear strength prole and other engineering
parameters, interpretation of soil layering condition,
as well as methods and associated detailed
implementations deemed suitable for the adopted
ground model.
In view of the heavy involvement of experienced
geotechnical engineers in typical derivations of
penetration prediction, revisiting the prediction at
the site in the event of unforeseen circumstances or
incorporating additional geotechnical information
to enhance the prediction are only possible if the
experts were directly involved in the rig move. The
traditional approach prohibits such an analysis to
be conducted if critical decisions are required when
the actual execution of rig-move differs from the
scenario envisaged during the desk assessment.
Load Penetration Monitoring
During preloading operations, the total elevated
load is commonly monitored to ensure the target
leg loads and hence the target preload base reactions
are achieved. The total elevated load is typically
derived from the weight of the hull, cantilever, drill
oor, solid variable loads, liquid variable loads, and
preload ballast, from which the distribution of leg
load among the individual legs can be computed.
The leg reaction is then calculated based on the
summation of the leg load and the total effective
weight of a single leg and spudcan. The accuracy of
the results calculated from load distribution method
using a simple spreadsheet/application is deemed
acceptable.
A potential inaccuracy in the monitoring of leg
load found in practice is typically caused by the
calculation tool that may not incorporate the effect
of hull buoyancy, especially when the rig is partially
submerged under water with some inclination.
Depending on the draft level, the hull buoyancy
can vary from little (hence negligible effect to the
leg loads) to a quite signicant amount, which
will reduce the actual leg loads signicantly. Any
reduction in the leg load will translate to a decrease
in the base reaction at the spudcan level. Therefore,
without proper consideration of the hull buoyancy
effect, it is difcult to achieve accurate monitoring
of leg load and associated leg reaction.
Figure 2. Spudcan penetration predictions and
actual data measured in the eld
8 KOMtech Technology Review 2011
Another issue typically encountered in practice is
the monitoring of leg penetration. To monitor the
leg penetration, input parameters such as leg ag
marking readings, distance from the hull base to
the leg ag marking, water depth (including tidal
variation), hull inclination, and also air gap or draft
level at each leg are required. Each of these parameters
has to be accurately measured in order to monitor the
leg penetration accurately. Typically, inaccuracy in
the monitoring of leg penetration can be attributed
to the difculty in measuring the true water depth,
tidal variation, and accurate air gap or draft level at
each leg. Without proper sensors for measuring each
of these parameters, it will be difcult to obtain an
accurate reading of leg penetration. Despite most
rigs being equipped with a single echo sounder
mounted at the hull bottom, the fact that the seabed
underneath the rig is likely not perfectly at implies
that the water depth at each leg position may vary
somewhat. However, in practice it is not uncommon
to assume a uniform water depth for all the legs.
Such an assumption potentially introduces error in
measuring the leg penetration.
Measuring accurate air gap during preloading
also poses another challenge. While measuring
the draft level is easier as indicated by draft marks
shown on the hull, it is certainly not the case for
measuring the air gap. The air gap measurement
is often just obtained from visual readings of leg
mark or observation of rack teeth with some degree
of subjectivity. Especially at night time or in low
light conditions, it may be even more difcult to
get accurate readings of the air gap. The use of laser
sensors to measure the air gap for offshore platforms
has been reported in the past
[2]
, but application in
current practice for jack up rigs is still very limited.
INDUSTRYS EFFORT TO IMPROVE
SAFETY OF JACKUP INSTALLATION
In the last two decades, the industry has been
constantly improving the guidelines for site-
specic assessment of mobile jackup units including
geotechnical analysis. Since its inception in 1994,
SNAME T&RB 5-5A
[3]
is still considered as the
main recommended practice for jackup geotechnical
assessments. For spudcan penetration analysis,
SNAME provides bearing capacity formulations
mainly for a single clay and sand layer as well as
double-layered soil. A general approach for tackling
multi-layered soil using an equivalent two-layered
system is also recommended. Since then, offshore
geotechnical engineers have based their jackup
foundation analysis on the recommended equations
in SNAME by incorporating subjective judgments
on the best methods to tackle certain situations
together with empirical factors or technique to
make the prediction more accurate.
Recently, the jackup industry collectively supported
a joint-industry project InSafeJIP with the
aim to develop improved guidelines for jackup
installation and removal
[4]
. The InSafeJIP reviewed
the current practices, explored the use of state-of-
the-art bearing capacity prediction models and
veried them with available eld data. Apart from
the detailed study on improving accuracy of the
bearing capacity prediction methods, the InSafeJIP
also highlighted the advancement and advantage of
in-situ penetrometer testing to acquire continuous
soil proles as well as the importance of using
advanced instruments to monitor spudcan load-
penetration response during preloading.
CHALLENGES IN IMPROVING
GEOTECHNICAL PERFORMANCE
OF JACKUPS
Besides improvements in the assessment guidelines,
geotechnical performance of jackups particularly
during installation can be enhanced by availability of
suitable on-site tools to perform further geotechnical
assessment, to identify potential geohazards, to
prepare preloading strategy and plans, as well
as to monitor the actual behaviour of spudcan
penetration. Provision of such an instrumentation/
device on board jackups will provide the jackup
operator with resources to detect unforeseen
situations and to help make the right decision for
mitigating any potential problems in the course of
installation. Such a system relies on an intelligent
framework for identifying proper shear strength
and implementing suitable bearing capacity model
with potential direct use of eld penetrometer
data for the penetration prediction as well as
measurement technology to give more accurate
load-penetration response. The eld penetrometer
data can also be obtained directly from a jackup as
described in Quah et al. (2008)
[5]
.
In an ideal pre rig-move assessment, accurate
geotechnical information should be available with
layer boundaries and shear strength parameters
identied. Variability in the shear strength measured
from several types of tests, with potentially
some degree of sampling disturbance, leads to
fundamental uncertainty that may in turn lead to
inaccuracies in the bearing capacity calculation
process. In the context of jackup foundation
assessment, where the best estimate of true
shear strength is crucial, signicant developments
of in-situ testing is expected to reduce reliance
on the variable shear strengths determined from
laboratory tests and minimise empirical selection of
soil strength. Shear strength directly derived from
penetrometer testing is deemed more reliable than
that obtained from laboratory tests, particularly for
sensitive soils. Penetrometer testing to some extent
can be seen as a miniature footing penetration from
which suitable parameters can be derived or directly
applied to the spudcan penetration prediction
without empirical selection of soil shear strength
and use of conventional bearing capacity models.
The use of eld penetrometer for jackup foundation
assessments including the analytical framework
of direct-correlation methods are described in
Erbrich (2005)
[6]
.
As an intermediate step before a direct-correlation
method can be fully established, a framework
can be developed to enable identication of layer
boundaries and shear strength parameters directly
from penetrometer results. The subsequent bearing
capacity calculation can be automated to produce
predicted load-penetration response. This semi-
automated process has advantages as it provides more
consistent approach in dealing with multi-layered
system and reduces subjectivity involved in the
selection of soil parameters and the implementation
of bearing capacity prediction methods by the
traditional approach.
Current practice of measuring load-penetration
response can also be further improved by
incorporating advanced measurement technology
or instrumentation systems. A simple application
to account for hull buoyancy and inclination
in the leg load distribution can be developed.
Incorporation of measurement sensors, such as
pressure transducers, laser, or sonar devices, is
also envisaged for better estimation of water
depth and draft/airgap. Integration with the
real-time data typically available on a jackup
will enable better tracking of load-penetration
response. Proper load-penetration tracking
complements the spudcan penetration prediction
and provides early warning should unexpected
behavior occur during installation. The success
of the Rack Phase Difference (RPD) system in
tackling eccentric leg reaction suggests
that the jackup industry is adaptable to
introduce practical new technology to deal
with preventive or corrective measures on
the actual site.
AN INTEGRATED SYSTEM FOR
IMPROVING GEOTECHNICAL
PERFORMANCE OF JACKUPS
With the latest advancement in geotechnical
knowledge and instrumentation technology,
jackups can be equipped with an integrated
system to identify punch-through potential and
to closely monitor progression of spudcan
installation. Such an integrated system is readily
available with some of the sub-systems having
been extensively implemented in practice. As
illustrated in Figure 3, the integrated system consists
of four sub-systems with associated software:
spudcan penetration prediction system; punch-
through simulation system; preloading planning
and stability system; and leg penetration
monitoring system. The analysis ows are briey
described below.
Geotechnical performance of jackups particularly during
installation can be enhanced by availability of suitable
on-site tools to perform further geotechnical assessment.
An Integrated System for Improving Geotechnical
Performance of Jackup Rig Installation 9
10 KOMtech Technology Review 2011
Figure 3. Schematic diagram of integrated jackup installation monitoring system
An Integrated System for Improving Geotechnical
Performance of Jackup Rig Installation 11
Sub-system 1: Spudcan Penetration
Prediction System (SPPS)
The tool automates calculation of spudcan
penetration prediction from penetrometer test results
or a given set of derived shear strength parameter.
There are two approaches that can be adopted to
derive load-penetration curves: i) direct correlation
from penetrometer result; ii) calculation based on
shear strength prole and soil layer boundaries using
conventional bearing capacity prediction formulae.
The shear strength prole in the latter approach can
be derived directly from the penetrometer results or
other available shear strength tests. With the target
preload base reaction derived from in-place analysis
conducted at the site-specic assessment stage,
potential punch-through can be identied from the
predicted load-penetration curve. An example tool
that enables semi-automation of load-penetration
prediction is illustrated in Figure 4
[7]
.
Sub-system 2: Punch-Through Simulation
System (PTSS)
Especially for sites where the seabed conditions
may potentially pose punch through/ rapid leg
penetration hazards, PTSS is instrumental to
estimate behaviour of the leg in question when
experiencing punch-through under given initial
conditions. Load- penetration response of the
leg as well as structural integrity of the leg and
holding system during a punch-through event
can be simulated with PTSS. With the predicted
penetration curve, the maximum punch-through
depth, at which the leg reaction is again equalised by
the soil bearing capacity and buoyancy of the hull,
can be estimated. If any structural over-utilisation is
indicated from the analysis, alternative preloading
strategies can be prepared to allow the installation
operation to be conducted within the rig tolerable
limits. This may include whether the preloading has
Figure 4. Example screenshot of spudcan penetration prediction system
12 KOMtech Technology Review 2011
to be executed underwater (partially submerged
with draft condition), with zero airgap, or with a
maximum allowable airgap, etc. Figure 5 illustrates
example punch-through simulation software
[8]
.
With the understanding of punch-through behavior,
the jackup operator can assess and anticipate the risk
of potential punch-through at the installation site
and if necessary consider seabed preparation when
the calculated risk is deemed unacceptable.
Sub-system 3: Preload Planning and Stability
System (PPSS)
PPSS is a system developed for the jackup operators
to plan for executing the preloading operations in
order to achieve the target preload base reaction.
The system allows various possible operation
scenarios to meet preload requirements, for example:
simultaneous preloading or sequential preloading,
staged preloading, preloading with or without
air gap, or preloading underwater. PPSS is also
developed in such a way that the jackup operator
is able to input and retrieve the preload tanks level
conditions in any given preloading plan stages.
An example screenshot of such a system is shown
in Figure 6
[9]
.
Apart from planning for the preloading operation,
PPSS can serve as a stability interface during rig
transit conditions. With this stability interface,
jackup operators can congure ballast water required
in each of the sea water tanks in order to keep the
hull stable during rig transit. A stability summary
is also provided so that the users can easily obtain
detailed stability information such as the calculated
draft level, corrected VCG, etc.
Figure 5. Example screenshot of leg punch through simulation system
An Integrated System for Improving Geotechnical
Performance of Jackup Rig Installation 13
Sub-system 4: Leg Penetration Monitoring
System (LPMS)
LPMS monitors both leg penetration and reactions
during jacking and preloading operations on a
real-time basis. This enables the jackup operator to
verify the observed leg-penetration response against
the predicted leg penetration curves, to be alerted
when the legs are approaching a critical penetration
depth at which rapid leg penetration is predicted to
occur, and to take necessary measures should any
unpredicted behavior be observed. Figure 7 shows
an example monitoring interface
[9]
.
To monitor the leg penetrations, LPMS takes
into account input parameters such as leg ag
marking readings, distance from the hull base to the
leg ag marking, water depth (including tidal
variation), and also air gap or draft level at each of
the legs. It is worth emphasising that accurate
reading of leg penetration will be hard to
achieve without proper sensors for measuring
water depth, draft, and air gap. The advanced
measurement technology has been extensively used
in the marine industry despite their limited use
for jackups.
To monitor the leg reactions, LPMS accounts for
distribution of total weights on the jackup which
includes the weight of the hull, cantilever, drill
oor, solid variable loads, liquid variable loads,
preload ballast, legs and spudcans, together with
their respective lateral, transverse and vertical centers
of gravity. Buoyancy effects, due to submerged
part of the leg/spudcan and partial hull submersion
under draft conditions, are also incorporated in
the system. With proper account of the buoyancy
force, both leg load and leg reaction can be
accurately monitored.
Figure 6. Example screenshot of preload planning and stability system
14 KOMtech Technology Review 2011
USE OF FIELD PENETROMETER DATA FOR
SPUDCAN PENETRATION PREDICTION
Laboratory tests are able to derive various engineering
parameters for detailed bearing capacity calculation.
However, for certain soil types, sampling techniques
and test procedures may inuence the sample
quality and accuracy of parameters derived. Sample
disturbance and different testing methods may
also result in a considerable variability of shear
strength which in turn leads to difculty deriving
true shear strength prole. If the sampling
gap is too large, layer boundaries cannot be
identied properly especially in complex multi-
layered soils.
Advancement in in-situ testing, e.g. piezocone,
T-bar and ball penetrometer testing, enables more
representative soil properties to be obtained.
Especially for soil conditions where undisturbed
sampling is very demanding and challenging, eld
penetrometer testing is seen as a better alternative
to derive in-situ shear strengths and other
parameters. In silty material, drainage characteristic
parameters can also be derived from specic
procedures, such as varying the rate of penetration in
a twitch test
[10, 11]
, with the objective of obtaining
the operational coefcient of consolidation. With
these parameters, the effect of partial drainage
on the penetration resistance can be quantied.
The bearing capacity calculation therefore can
tackle intermediate soil as opposed to idealised
clay or sand types of soil as currently practiced.
Direct use of penetrometer data for design of
foundations has been widely applied for conventional
foundations such as piles using an appropriate
averaging method
[12]
. Similar implementation
for spudcan bearing capacity is envisaged, either
Figure 7. Example screenshot of leg penetration monitoring system
An Integrated System for Improving Geotechnical
Performance of Jackup Rig Installation 15
through direct correlation or traditional bearing
capacity methods
[4]
. The latter typically consists
of a two-step approach in which shear strength
is derived from the penetrometer resistance for
use in semi-empirical bearing capacity models.
The penetrometer and spudcan are considered
as having similarities in that they both generate
full-ow mechanisms when penetrating the seabed.
The effect of different strain-rates and relative scale
between the foundation (or penetrometer) diameter
and layer thickness, are two key factors that need to be
resolved for enabling direct correlation between the
penetrometer and spudcan penetration resistance
[6]
.
Soil plug development ahead of the spudcan base,
particularly following penetration through a hard
layer into softer material, is also another area which
needs further investigation to improve penetration
prediction.
While correlations between penetrometer resistance
and soil properties in sand or clay have been
established
[13, 14]
, their relationship for calcareous
and silty material and their use for bearing capacity
prediction are still very limited. In these materials,
some degree of consolidation may occur during
penetrometer tests even though spudcan penetration
at the same location may occur under undrained
conditions. Due correction is thus necessary to
correct the penetrometer resistance to an equivalent
value suitable for spudcan penetration. In view of
limited understanding in this area, it is therefore
imperative to extend the current framework to cover
intermediate soils.
To identify the behaviour of spudcan penetration in
a wide range of soil types and layering conditions, an
experimental programme is being conducted in the
Centre for Offshore Foundation Systems (COFS)
at the University of Western Australia under COFS-
Keppel collaborative research project. The study
covers centrifuge modeling of spudcan installation in
intermediate soils such as calcareous sand, silt, clay
as well as silty sand. The behaviour of spudcans and
cone penetrometers in multi-layered soils, including
proles containing intermediate soil layers, is also
being investigated. The study is expected to extend
the use of penetrometer results to derive more
representative shear strength parameters for use
with conventional bearing capacity methods and
an analytical framework to enable direct correlation
with spudcan penetration behavior in any soil types
and layered conditions. Progressive calibration of
the method against observations made during rig
moves is also being undertaken.
An example on the use of piezocone results
for spudcan penetration prediction in clay is
demonstrated herein using the same information
obtained at the same site as given in Figure 1. In
Figure 8, the estimated bearing capacity derived
directly from the net cone resistance is presented.
Averaged net cone resistant values within 0.1 x
spudcan diameter (equivalent diameter of 14 m
in this example) above and below the depth in
question were used in the calculation. A ratio of
Nc (for spudcan) and Nkt (for cone) of 6.6/13.5
is used to correlate the cone and spudcan bearing
resistance. The Nc factor of 6.6 is regarded as an
average bearing capacity factor up to penetration
of one spudcan diameter while the Nkt factor of
13.5 is the mean cone resistance factor relative
to an average shear strength, i.e. triaxial compression,
simple shear and triaxial compression
[15]
. Spudcan
volume is accounted for and full backll is assumed.
Figure 8. Spudcan penetration prediction based
on direct correlation with cone
penetrometer result
16 KOMtech Technology Review 2011
Compared to the measured load-penetration
response, the spudcan penetration predicted directly
from the cone resistance is quite accurate.
The prediction is also carried out by the traditional
approach after deriving undrained shear strength
from the cone resistance and identifying the layer
boundaries. With the interpreted shear strength,
a two-layered approach bearing capacity model is
used. The result presented in Figure 9 shows that
the indirect prediction offers reasonable match
compared to the observed load-penetration response.
The results obtained from the two approaches
suggest that with a proper framework either method
can provide reasonable estimates of spudcan
penetration. In this case, use eld penetrometer data
and its direct use for bearing capacity prediction are
demonstrated to deliver reliable results with reduced
subjectivity throughout the course of analysis.
CONCLUSION
As jackup deployments are moving towards deeper
waters and frontier regions with more complex
soil behaviour, introducing available advanced
instrumentation technology is therefore essential
to enhance the current practice in ensuring
foundation stability. The recent safety record of
jackup installation suggests that leg installation,
as one of the critical stages in jackup operations,
imposes several challenges that the industry has yet
to resolve. Besides improvements in the guidelines,
provision of practical tools on-board jackups will
offer the jackup operator more information to help
make the right decision for mitigating any potential
problems during jackup installation.
It has been demonstrated in this paper that an
integrated system can be established for jackups to
enable spudcan penetration prediction, potential
punch-through identication, preloading plan,
and preload monitoring. Combined with existing
data acquisition systems typically available on-
board jackups, such an integrated tool is capable of
providing the necessary information at every stage
of jackup installation.
In view of the advancements of in-situ testing, use
of eld penetrometer testing is seen as a suitable
method besides the traditional approach for
predicting bearing capacity of jackup spudcans.
Recent investigation on various types of soil and
layering conditions has shown that it is possible to
establish a robust geotechnical analysis framework
for direct correlation of penetrometer and spudcan
resistance.
Combined with existing data acquisition systems typically
available on-board jackups, the integrated instrumentation
and monitoring system is capable of providing the
necessary information at every stage of jackup installation.
Figure 9. Spudcan penetration prediction based
on undrained shear strength derived
from cone penetrometer
An Integrated System for Improving Geotechnical
Performance of Jackup Rig Installation 17
AUTHORS CONTACT Matthew.Quah@keppelfels.com
REFERENCES
[1] Osborne, J.J. (2005). Jackup Drilling Option Ingredients for Success. Oil and Gas Producers OGP/CORE Workshop. National University of Singapore.
[2] Mes, M.J. (1992). Improved Accuracy of Laser-assisted Airgap Surveys Offshore. Proc. Offshore Technology Conference, Houston, OTC Paper 6946.
[3] SNAME T&RB 5-5A. (2008). Guidelines for Site Specic Assessment of Mobile Jackup Units - Rev. 3, August 2008. Society of Naval Architects and
Marine Engineers.
[4] Osborne, J.J., Houlsby, G.T., Teh, K.L., Leung, C.F., Bienen, B., Cassidy, M.J., Randolph, M.F (2009). Improved Guidelines for the Prediction of
Geotechnical Performance of Spudcan Foundations during Installation and Removal of Jackup Units InSafeJIP. Proc. Offshore Technology Conference,
Houston, OTC Paper 20291.
[5] Quah, M., Foo, K.S., Purwana, O.A., Keizer, L., Randolph, M.F. and Cassidy, M.J. (2008). An integrated in-situ soil testing device for jackup rigs. Proc.
2nd Jackup Asia Conference and Exhibition, Singapore.
[6] Erbrich, C. T. (2005). Australian frontiers: Spudcan on the edge. Proc. 1st Int. Symp. on Frontier in Offshore Geotechnics ISFOG, Perth, pp. 49-74.
[7] SMAS v3.3. (2010). Spudcan Penetration Analysis Software. Keppel Offshore & Marine Technology Centre.
[8] LePSuS v2.0. (2009). Leg Punch-Through Simulation Software. Offshore Technology Development.
[9] SeaMATE v3.0. (2009). Transit, Preload Planning and Monitoring Software. Offshore Technology Development.
[10] House, A.R., Oliveira, J.R.M.S. and Randolph, M.F. (2001). Evaluating the Coefcient of Consolidation Using Penetration Tests. Int. J. of Physical Modelling
in Geotechnics, Vol. 1(3), pp. 17-25.
[11] Randolph, M.F. and Hope, S. (2004). Effect of Cone Velocity on Cone Resistance and Excess Pore Pressure. Proc. Int. Symp. of Engineering Practice
and Performance of Soft Deposits, Osaka.
[12] Heijnan, W.J. (1974). Penetration Testing in the Netherlands, State of the Art Report. Proc. Eur. Symp. on Penetration Testing, Vol 1.
[13] Lunne, T., Robertson, P.K. and Powell, J.J.M. (1997). Cone Penetration Testing in Geotechnical Practice, Blackie Academic and Professional, London.
[14] Randolph, M.F. (2004). Characterisation of soft sediments for offshore applications, keynote lecture. Proc. 2nd Int. Conf. on Site Characterisation,
Vol. 1, Porto, Portugal, 209231.
[15] Low, H.E., Lunne, T., Andersen, K.H., Sjursen, M.A., Li, X. and Randolph, M.F. (2009). Estimation of intact and remolded undrained shear strength
from penetration tests in soft clays. (paper under review).
18 KOMtech Technology Review 2011
18 KOMtech Technology Review 2011
No part of the materials published in this journal may be reproduced, stored in a retrieval system or transmitted in any form whatsoever without the prior written permission of KOMtech
Design of a DP2 Drillship
Power Plant
BAI Songlin, MSc, B.Eng
WHEN DESIGNING A DIESEL ELECTRIC DYNAMIC POSITIONED(DP2)
DRILLSHIP, the principal challenges include determining the generator size,
the number of on-demand generators, the optimum conguration of power plant
equipment and its control system. These are crucial factors impacting the
station-keeping performance, its ability to deal with various failure scenarios,
as well as construction and operational costs. This paper focuses on the
power plant basic design by analysing various operational modes for electrical
power generation and its distribution. Load analyses with different operational
modes are employed to evaluate generator size, to further obtain engine size
as well as to determine whether a given vessel power plant design is suitable for
DP2 environmental conditions specied in the SDS7500 slim drillship technical
specication even if worstcase single failure occurs. The results from the
analysis and calculations clearly demonstrate that six generators with 3745eKW
each are able to meet the power demands of KOMtech slim drillship in different
operational modes, and the power plant and its control system with redundancy
are efcient and reliable.
Design of a DP2 Drillship
Power Plant 19
20 KOMtech Technology Review 2011
INTRODUCTION
Diesel electric systems are widely utilised in DP2
operated drillships. This is because of its high
exibility and reliability during operation. During
normal operation, the electrical power system with
power management system (PMS) supply all electrical
energy in the drillship. Therefore, the generator
sizing, power plant design and power system
conguration become highly important during the
basic design stage of a DP drillship. All of them will
signicantly affect the subsequent construction and
operational costs. Though most of the design
principles and basic conguration of the power plant
and control system for drillships are similar, the
designer has a tendency to provide larger power plant
size than required. One of the reasons is that there are
distinct differences in the selection of technology and
solutions. Another reason is that the generator sizing
and power plant design, to a great extent, depends
on the designers experience and prociency in
mechanical, electrical and control systems, as well as
the operation of the whole vessel system and mutual
coordination of different systems. In addition to the
basic philosophy for sizing and designing of power
plant, the power plant design of a DP2 drilling vessel
should meet design guidelines and rules from
Classication Societies for Dynamic Positioning
System (DPS). The subsequent section will present
how to obtain the optimal power plant size and
power plant design with redundancy by calculating
and analysing various operation and failure modes,
and also briey introduce the main components and
PMS function in the power system of KOMtech
SDS7500 drillship.
SYSTEM DESIGN AND ANALYSIS
DP capability
DP Capability is dened as the ability of the unit
to hold station in certain environmental conditions.
DP Capability Analysis is generally employed to
establish the maximum environmental conditions
in which a DP vessel can maintain its position and
heading for a proposed thruster conguration.
KOMtech SDS7500 drillship will have ve thrusters
installed in two separate rooms as follows
[1]
:
Two Azimuth propulsion thrusters , each of
3.7MW at aft
One Azimuth thruster, 3.3MW at forward
Two tunnel thrusters, each of 2.5MW
at forward
The DP sea environmental conditions for SDS7500
drillship is dened in KOMtech technology
specication
[1]
. The Kongsberg Maritime computer
program StatCap has been utilised for simulation at
different environmental conditions. The simulation
result with all thrusters intact is shown in
Figure 1a and 1b
[2]
under the specied sea
environment conditions.
Kongsbergs DP Capability Analysis is applied to
obtain thrust utilisation envelope and also clearly
indicated the following design basis (shown as
Table 1) which can be utilised for subsequent
analysis and calculation:
Load analysis
The electrical designer must have information on
the loads and operational modes before designing
the power plant system. Therefore, it is necessary
to identify the operational mode for KOMtech
SDS7500 drillship. The six operational modes in
the load analysis are as follows:
Table 1. Basic condition used in load analysis
Weather Maximum Drilling Thrusters
condition heading deviation Activities condition
Stand-by 30 Reduced All oK
Operational 30 Full All oK
Stand-by 30 Reduced 1 thruster out
Operational 30 Reduced 1 thruster out
Design of a DP2 Drillship
Power Plant 21
Figure 1b. Thrust utilization envelope used in load analysis for stand-by mode
Figure 1a. Thrust utilization envelope used in load analysis for operational mode
22 KOMtech Technology Review 2011
Figure 2. Load analysis summary
mode 1: transit
mode 2: station keeping with reduced drilling
mode 3: station keeping with full drilling ahead
mode 4: station keeping with reduced drilling
in worst case scenario
mode 5: station keeping with reduced drilling
and one thruster out
mode 6: emergency mode
In order to correctly size the generators and ensure
that there is enough power available in all operational
conditions, an electrical load analysis must be
performed. The electrical load analysis consists of
a list of all electrical loads present on the vessel and
associated power factors with dened operational
modes
[3]
. The primary load grouping for KOMtech
SDS7500 drillship includes thrusters, drilling
equipment and distribution. The 440V distribution
includes ship service and accommodates all general
use of power. The drilling category contains all the
loads associated with drilling operations such as
the hoisting high pressure unit (HPU), Ring Line
HPU, top drives, mud pumps, and rotary tables,
etc. The load analysis with specied sea environment
condition was performed under six operational
modes. Figure 2 shows a load analysis summary
without emergency mode.
The electrical consumption as shown in Figure 2
is calculated based on the estimated drilling and
station keeping loads according to each individual
operational mode. The operational mode 3 is the
governing mode, taking a total of approximately
18.4eMW.
Six main generators are intended to be used. DP2
classed ship requires that the DP capability should
not be affected in the case of a single failure. The
calculated power is about 3680eKW per generator.
In theory, the rated power of 3690eKW can be
selected as the generator size if we consider it only
from a purely mathematical viewpoint. But the
drilling equipment in KOMtech SDS7500 drillship
includes some big hoisting HPU pumps with
directly on-line (DOL) starting. These hydraulic
pumps may produce large transient load variation
at power network during full DP operation. Thus,
a slightly higher capacity of 3745eKW is selected as
the generator size.
Is the assumption of using six generators in the load
analysis
[3]
reasonable? It will be discussed through
calculations in Table 2.
The following results can be deduced from Table 2:
1. Total power for six generators with 3.7eMW
each should be the minimum power capacity.
2. The utilisation factor with 3.7eMW each under
transit mode should be at the maximum,
efcient and economical.
Based on the above two items as well as considering
installation space in the engine room, six generators
with 3.7eMW each is a suitable selection for
KOMtech SDS7500 drillship.
SYSTEM DESCRIPTION
The best power plant system in the DP2 operated
drillship should be one that will be cost effective,
safe, and adequate with redundant supply of
electrical energy to power consumption.
The section below describes the selection, design,
installation and conguration of such a system.
The power system is divided into a number of
different voltages, namely, 11KV, 970VDC,
690VAC, 440VAC and 230VAC.
There are six main diesel generators (DGs)
supplying two 11kV bus bars. Each main generator
is connected to individual section of the bus
bar separated by bus-tie breakers. The ends of
Table 2. Comparison of load analysis summary for 4, 5 and 6 generators
Mode Six generators Five generators Four generators
with 3.7MW each with 4.7MW each with 6.3MW each
% No. of generators % No. of generators % No. of generators
in operation in operation in operation
1 Transit mode 91.1 3xMain gen. 72.8 3xMain gen. 81.2 2xMain gen.
2 DP2+reduced drilling mode 96.5 5xMain gen. 96.1 4xMain gen. 95.6 3xMain gen.
3 DP2+drilling ahead mode 98.3 5xMain gen. 97.9 4xMain gen. 97.4 3xMain gen.
4 Worst case for DP2 90.5 5xMain gen. 90.2 4xMain gen. 89.7 3xMain gen.
+reduced drilling mode
5 One thruster out for DP2 93.4 5xMain gen. 93.1 4xMain gen. 92.6 3xMain gen.
+reduced drilling mode
6 Emergency mode 98.0 1xE. gen. 98.0 1xE. gen. 98.0 1xE. gen.
Total power for main Gen.(KW) 22.47 23.5 25.2
Design of a DP2 Drillship
Power Plant 23
each switchboard (SWBD) are connected via
interconnector breakers to those of the other
SWBD so as to form a ring bus.
The 970VDC SWBD shall be supplied from two
separate 11kV bus sections after transformers and
rectication. The 970VDC SWBD supplies power
to the variable speed drives of the drilling equipment
such as mud pumps and top drive.
Similarly, a 690VAC SWBD shall be supplied from
either one of the two separate sections of the 11kV
bus-bar after power transformers TRAFO5 and
TRAFO6. This ensures that the 690VAC SWBD
remains powered even if one of the supplies breaks
down. The 690VAC SWBD mainly supplies power
to the hoisting HPU and the Ring Line HPU
motors.
The 440VAC SWBDs shall be powered from
four different sections of the main 11kV bus-bar
via transformers. The 440VAC SWBD mainly
powers 440VAC motors which includes the marine
and drilling consumption and also powers other
miscellaneous 440VAC consumers.
As a part of the power system, shore power shall be
connected to 440VAC SWBDs. This will provide
an alternative to connect shore power when power
system needs to use external power.
The 230 VAC SWBDs shall be powered from two
different sections of the 440VAC bus-bars.
Lastly, the emergency power generation and
distribution consist of one emergency generator
rated 500 eKW located in the emergency generator
room together with the 440V and 230V emergency
SWBDs.
Power system protection
In the KOMtech SDS7500 drillship, system
protection comprises over current and grounding
fault protection. Over current in a power network
can occur in both normal and abnormal conditions.
The Neutral earthing for 11kV power source and
The best power plant system in the DP2 operated
drillship should be one that will be cost effective, safe,
and adequate with redundant supply of electrical energy
to power consumption.
24 KOMtech Technology Review 2011
transformers should include high resistance earthed
neutral via earthing resistors or dedicated neutral
transformers. All other systems must be isolated.
Protection is provided to ensure that a fault on
any part of the power network can be properly
isolated and permit the remainder of the network
to continue operation.
MAIN EQUIPMENT
Gen-sets
Electricity is most often generated at a power plant by
electromechanical generators, primarily driven by
combustion engine fueled either by diesel, gas, dual
fuel or heavy fuel oil. Diesel engines are often
manufactured in two stroke cycle and four stroke
cycle versions. In a compact power plant, the diesel
engines are normally four stroke cycle version with
medium or high-speed engines. Almost all the
vessels have an AC synchronous generator with
AC distribution. These synchronous generators can
operate in parallel or in stand-alone mode. Selection
of Gen-sets should consider the following items:
Fuel consumption of less than 200g/kWh at the
optimum operational point
Engine-mounted high/low temperature water
cooling pumps
Engine-mounted lubrication oil pumps
Emission compliant with IMO
The stationary voltage variation on the
generator terminals shall not exceed 2.5%
of nominal voltage
The largest transient load variation shall not
give voltage variation exceeding -15% or +20%
of the nominal voltage unless other has been
specied and accounted for in the overall
system design
Brushless excitation for reduced maintenance
and downtime
Higher sub-transient inductance is suitable if
needed to reduce S/C current
Insulation/temperature classes: F class or above
IP enclosure
Switchboards (SWBDs)
In order to obtain the redundancy requirements
for the KOMtech SDS7500 drillship, the main
SWBDs are usually distributed or split into two or
more sections according to the rules and regulations
for DP classed drillship. The SWBD shall tolerate
the consequences of one section failing due to a
short circuit (S/C).
Table 3 lists the failure result of generator and
drilling capacity with different number of main bus
sections during worst case condition.
It shows that the maximum single failure scenario
is to lose 50% of generator capacity if the SWBD
is a two-split conguration with equally shared
generator capacity and load on both sides.
Simultaneously, DP2 vessel would lose its
position, and drilling equipment would not have
enough power.
Table 3. Six generators with 3745eKW each connected at different bus section number and working
at the worst case condition
Main bus Lost generator capacity DP2 vessel Drilling capacity
section
2 50% Lost position No
3 33.3% Maintain position Reduce drilling 70%
4 33.3% or 16.6% Maintain position Reduce drilling 70% or 10-25%
5 33.3% or 16.6% Maintain position Reduce drilling 70% or 10-25%
6 Max. 16.6% Maintain position Reduce drilling 10-25%
11KV can be used to choose the main circuit breaker
(CB) with rated short-time withstanding current
at 25KA or 31.5K
[7, 8]
. There is also advantage in
using this voltage level to reduce electrical equipment
and cable size. It is possible that 6.6KV voltage level
can also be used as an alternative after further
optimal conguration.
CBs are used for connecting and disconnecting
generator or load units to the SWBDs, or different
parts of the SWBDs together. Various CB
technologies are applied as follows:
Air insulated switchgear, rarely applied today
except at low voltage levels
Gas insulated switchgear such as sulfur
hexauoride (SF6)
Vacuum breaker
The 690V/440V SWBDs shall be metal clad single
bus bar panels with withdrawable motor-starters
using molded case circuit breaker (MCCBs),
contactors and motor protection and control
device. Motor starters shall be directly on-line
(DOL) starting.
Transformers
A transformer is a static piece of apparatus used
for transferring electrical power from one circuit to
another by electromagnetic induction. The electrical
power is always transferred with frequency identical
before and after the transformation, but may involve
changes in magnitudes of voltage and current, and
sometimes for phase shift.
Cast resin dry type transformers are normally used in
the offshore industry. There are also various special
Design of a DP2 Drillship
Power Plant 25
Table 3 shows that there are three, four, ve and
six section congurations and it indicates that six
sections will be the optimum. This is because this
conguration will lose minimum power capacity
and reduce minimum drilling equipment load in
the worst case condition.
Considering system exibility and redundancy
philosophy, the 440V auxiliary SWBD is separated
into four sections. Based on this philosophy, any
electrical or auxiliary technical single failure in
the installation, excluding re and ooding, is able
to occur without reducing power below that needed
for the most critical operational situation.
Due to the limitation on the thermal and mechanical
stresses in bus bars and the switching capacity of
the switchgear, it will be advantageous or necessary
to increase the system voltage and hence reduce the
current levels.
Table 4 shows most common selection for the
main distribution system based on the IEC voltage
levels with application guidelines from IEC
[4]
and
NORSOK
[5]
:
The limitation of the electrical equipment is
determined by the load current and short current.
When the power system includes variable frequency
drive (VFD) loads, the actual power limitation for
each system voltage may be different with IEC and
NORSOK recommendations. These VFD loads
do not contribute to S/C power in the distribution
system, often allowing reduction of the voltage
limitation to a lower level.
Table 5 shows three phases S/C calculation for
KOMtech SDS7500 drillship based on NORSOK
guideline and IEC standard
[6]
. The S/C level for
Table 4. Voltage level recommendation in IEC and NORSOK
Voltage level Generator capacity Motor/Utility capacity
11KV >20MW >=400KW
6.6KV 4MW< and <20MW >=300KW
690V <4MW <400KW
440/230/110V Utility
26 KOMtech Technology Review 2011
purpose transformers such as Variable frequency
drive (VFD) transformers. VFD transformer with
3-windings is used to isolate between the feeding
network and the VFD. The two transformers in each
SWBD windings shall be designed to make a twenty
four pulses conguration in order to minimize the
harmonics across the line supply. Selecting and
sizing transformer for VFD should consider the
following items:
Insulation/temperature classes: F class or above
Cooling type: air force cooling or water cooling
Overloading capacity
Equipped for temperature monitoring
High-voltage tapings re-connectable on
terminals
The VFD transformer for drilling system and
propulsion shall be of twelve pulse types, two
transformers shall operate as pseudo
twenty-four pulse system on the net. This will
considerably reduced the THD on the net
Vector group
IP enclosure
Motor for propulsion and thruster
The electrical motor is accepted widely in the
offshore industry to convert electrical energy
to mechanical energy and utilised for electric
propulsion, thrusters, and most of the other loads
in ship installations such as pumps, fans, winches,
etc. In the KOMtech SDS7500 drillship, the two
main aft azimuth propulsions, two forward tunnel
thrusters and one forward azimuth thrusters are each
driven by an asynchronous motor rated 3.7MW,
2.5MW and 3.3MW separately; 4.16 kV, 720 RPM
which shall be fed from a frequency converter. The
converter makes it possible to control the speed of
the asynchronous motors.
Table 5. 3-Phase short circuit calculation summary for KOMtech SDS7500 drillship
Bus Short circuit current (kA, rms)
ID KV Sub transient Transient Steady State
230V ESB 0.230 17.25 17.128 10.098
230V D-1 0.230 18.095 17.965 10.420
230V D-2 0.230 18.108 17.965 10.420
BusA-1 11.00 15.848 9.767 3.021
BusA-2 11.00 15.848 9.767 3.021
BusA-3 11.00 15.848 9.767 3.021
BusA-4 11.00 15.848 9.767 3.021
BusA-5 11.00 15.848 9.767 3.021
BusA-6 11.00 15.848 9.767 3.021
BusB-1 0.690 67.089 60.954 34.015
BusB-2 0.690 67.105 60.969 34.015
BusC-1 0.450 57.083 53.001 43.623
BusC-2 0.450 58.604 54.493 43.682
BusC-3 0.450 59.604 54.493 43.682
BusC-4 0.450 60.609 56.459 43.756
Bus ESB 0.450 60.609 56.459 43.756
Design of a DP2 Drillship
Power Plant 27
Variable frequency drive (VFD)
Figure 3 shows a typical VFD system. It normally
includes three main parts: AC to DC rectication,
energy storage, and DC to AC switching. The
VFDs of two aft azimuth thrusters, two forward
thrusters and two drilling rectiers in KOMtech
SDS7500 drillship shall be of 12 pulse types. Two
transformers operate as a pseudo 24 pulse system
on the net. However, the 3.3MW forward thruster
VFD shall be a real 24 pulse type.
The drilling drive systems are very similar to the
thruster system. But it includes a common DC bus
conguration which consists of two sets of drilling
SWBDs. The drilling SWBDs shall power three sets
of mud pumps, and one top drive via inverters. The
power section of the drive converter shall make up
the following items:
12-pulse diode rectifer
Voltage DC-link with capacitors
Three-level or fve-level inverter with
high voltage insulated-gate bipolar
transistor(HV-IGBT) or injection enhanced
gate transistors (IEGT)power modules
Sinusoidal output flter
Water cooling
Power management system (PMS)
PMS is an automation control system designed
to provide full control over the complete power
network, allowing reliable and exible energy
management and supervision. Figure 4 shows
overall PMS and DP control system topology for
KOMtech SDS7500 drillship. The main task of an
electrical PMS is to control the generation plant and
to ensure availability of the electrical power as well
as to avoid blackouts.The PMS with redundancy
is required as a part of the DPS design. The PMS
consists of the following main functions:
1. Load dependent start/stop
2. Blackout recovery
3. Heavy consumer control
4. Blackout preventions
Blackout can be avoided by monitoring total load
compared to available power. In case of an overload,
the power system will quickly reduce load to avoid
overload on the generator. The following methods
are often used to avoid blackout in PMS:
Load shedding
Power reduction by DPS
Power limitation / load reduction
Limitation of consumers load increase
Load reduction based on average bus load
Figure 3. Typical AC VFD system
28 KOMtech Technology Review 2011
Load reduction based on individual
generator load
Load reduction based on bus frequency
Load reduction based on tripped generator,
bus tie or interconnector breaker
Split of 11kV main SWBDs
The above section describes the design of DP drillship
power system, which is part of the basic design for
KOMtech SDS7500 drillship. In addition, it also
introduces load analysis, optimum power system
and switchboard conguration with redundancy,
system protection, equipment selection and PMS.
CONCLUSION
This paper introduces the generator sizing, basic
design and the optimal conguration of power plant
system for KOMtech SDS7500 slim drillship. The
analysis and calculation show that six generators
with 3745ekW each can provide overall reliability
and efciency in the electric power distribution
even if worstcase single failure occurs. The power
plant and its control system with redundancy can
be utilised in KOMtech SDS7500 slim drillship
without sacricing DP2 capability. The basic design
of the power plant can be used further for electrical
detail design. At the same time, it also provides some
essential assistance in the selection and installation
of equipment for KOMtech SDS7500 drillship.
The power plant and its control system with redundancy
can be utilised in KOMtech SDS7500 slim drillship
without sacricing DP2 capability.
Figure 4. PMS and DP control system topology
Design of a DP2 Drillship
Power Plant 29
AUTHORS CONTACT songlin.bai@keppelom.com
REFERENCES
[1] Technology specication for PRD 7500 slim drillship, KTDE-030-01-DOC-001, July, 20, 1010.
[2] DP Capability Analysis for KOMtech 030 slim drillship, The Kongsberg Maritime AS, June, 24, 2010.
[3] Load analysis, KTDE-030-01-ELL-001, Rev.01, July, 11.2010.
[4] IEC61892-2, Mobile and xed offshore units-Electrical installations. Part 2: System design.
[5] Norsok Standard, E-001, Edition 5, July 2007.
[6] IEC61363-1, Electrical installations of ships and mobile and xed offshore units. Part 1, First edition, 1998-02.
[7] Circuit breakers, http://www.abb.com.sg/product/us/9AAC754143.aspx?country=SG
[8] Circuit breakers, http://aunz.siemens.com/Energy/Distribution/Pages/PTD_5020_MediumVoltage.aspx
30 KOMtech Technology Review 2011
Dedicated Cable Laying Vessels For Future
Offshore Wind Farm Development 31
No part of the materials published in this journal may be reproduced, stored in a retrieval system or transmitted in any form whatsoever without the prior written permission of KOMtech
OFFSHORE WIND FARMS, AN IMPORTANT RENEWABLE ALTERNATIVE TO
CONVENTIONAL GLOBAL ENERGY RESOURCES, are growing fast in recent
years in terms of the number of wind turbines, turbine sizes and capacities.
Offshore wind farms require larger scale and more complex networks of subsea
power cables to gather and deliver generated power to land based sub-stations.
The existing cable laying barges or vessels, which are not specially designed
for offshore wind farm power cable installation, are not able to deliver efcient,
quality work.
Recently, Keppel Offshore & Marine Technology Centre (KOMtech) has developed
two types of dedicated Cable Laying Vessels (CLV) for future offshore wind
farm projects. The In-eld CLV is capable of carrying 130km of power cable
with a diameter of 130mm, while the Export CLV has a capacity of 50km of
250mm cable. This high capacity, combined with the necessary equipment tted
on board, allows the vessels to complete the cable laying operation in a single
vessel transit. These vessels will be an important asset in completing large wind
farms within the aggressive time frame set by the developers and governments.
ACKNOWLEDGEMENT
The equipment arrangement for KOMtech Cable Laying Vessels presented in
this paper has been jointly developed by KOMtech, Parkburn Precision Handling
Systems Ltd (PPHS) and Soil Machine Dynamics (SMD) Pte Ltd.
Dedicated Cable Laying
Vessels For Future
Offshore Wind Farm
Development
Abul BASHAR, B. Eng
James CAO Yimin, B. Eng
TAN Cheng Hui, P. Eng, MSc
32 KOMtech Technology Review 2011
INTRODUCTION
The offshore wind energy industry is developing
faster than any time over the past decades.
According to the World Wind Energy Report-2009
published by World Wind Energy Association
(WWEA)
[1]
in March 2010, the installed wind
power capacity has more than doubled every
three years. In a press statement issued by WWEA
on 19 October 2010, it was mentioned that
16 Gigawatt of wind power has been added in rst
half of year 2010 and global capacity of wind
power will reach almost 200 Gigawatt at the end
of 2010. For the uninterrupted production and
transportation of this increased offshore power,
wind power industry needs continuous support
from offshore installation services such as cable
laying operation (CLO) and supply of various
maintenance facilities, such as cable repairing.
Considering the offshore work volume for CLO,
the water depth and the distance from shore; an
optimal solution for efcient cable laying operation
is yet to be developed. This creates an opportunity
to examine the capacity of those supports and
supplies which will be required for the future wind
power industry.
Currently, barge type vessels are used for cable
laying operation in offshore wind farms. But those
barges have limitations in their propulsion, position
keeping, operational water depth and cable laying
speed; which are considered the most important
features for an efcient Cable Laying Vessel (CLV).
At present, some of the offshore support vessels are
being converted to work as a CLV. But these vessels
do not have enough deck space to accommodate
the necessary cable laying equipment. Hence, they
must be supported by other vessels to perform their
intended cable laying jobs, resulting in higher cost
with longer time delay.
With the increase in offshore wind power demand,
more turbines have to be installed further away
from the shore and also in deeper water. More
efcient CLV will be required to meet the future
demand for cable laying operation. A CLV with self
propulsion and Dynamic Positioning (DP) system,
higher bollard pull capability and dedicated cable
laying equipment on board can perform the cable
laying operation efciently.
Keppel Offshore & Marine Technology Centre
(KOMtech) has recently designed two independent
CLVs for the wind farm industry. The rst vessel
is an In-Field CLV, which lays cable between
the wind turbines, substation and the second type
is an Export CLV, which lays cable between the
substation and shore facilities. These are dedicated
CLVs which are able to perform the cable laying
operation efciently in future projects. They are
self-propelled vessels with DP-2 facility. They are
also designed with 150 tonnes (maximum) bollard
pull capacity and about 2560 square meter of
free deck space to accommodate all the necessary
cable laying equipment to work as a dedicated and
independent cable laying vessels. The development
of these vessels further strengthens Keppels move
into offshore wind energy, supporting the work on
Wind Turbine Foundations, Turbine Installation
Vessels and Electrical Substations reported in last
years Technology Review
[3-6]
.
OFFSHORE POWER CABLES
The electrical power produced in offshore wind
turbines is transferred to the substation and then
to shore by heavy duty offshore power cables. These
cables are divided into two categories, namely,
in-eld power cables and export power cables, as
shown in Figure 1.
In-Field Power Cables: These cables are connected
between two wind turbines and also between the
turbine and the offshore substation as illustrated in
Figure 1. These cables are used to connect the wind
turbines in blocks within a wind farm. Sizes of these
cables vary with the power production capacity of a
single turbine and the inter-turbine connection
Figure 1. Offshore Power Cable and Wind Turbine
Arrangement
[2]
. (Courtesy of Barrow Offshore Wind)
may be stored on the deck in horizontal axis cable
reels or in a vertical axis cable carousel. Ineld cables
are used in short lengths, ranging from 500m to
1000m, to suit the distance between the two turbines.
Because of the short lengths, they can be stored either
in a cable carousel or in reels. In contrast, export cable
is a single length cable from the offshore sub-station
to land power grid network which is much longer,
meaning export cables need to be stored in a suitable
size of cable carousel on the deck.
Cable Load-Out
Cables, which are stored in reels or in a cable carousel,
need to be loaded out from the deck to sea bed for
laying operation. The equipment involved with the
cable load-out operation are the loading tower, cable
engine, cable tensioner, cable brake, conveyer system
and cable chute. Capacities of these equipment are
selected on the basis of cable size and the water depth.
Cable Laying Operation
The most challenging task for a cable laying vessel is
the underwater cable laying operation. It includes
seabed trenching, cable laying and cable burying.
Some CLVs undertake these tasks in several steps,
whereas, some vessels do these simultaneously. Export
cables are stiffer than in-eld cables and its laying
operation is done by a suitable plough, which is
pulled by a deck mounted winch. Depending on the
vessel design, in-eld cable laying is done either by a
plough or a Remote Operated Vehicle (ROV).
Cable Abandonment and Recovery
In the event of emergency situation, cable laying
operation has to be stopped and the loaded-out cable
needs to be disconnected from the vessel. In such a
case, the CLV should be able to release the cable safely
by cutting it at the end of the vessel and lowering it to
the seabed by an abandonment and recovery winch.
Once the emergency situation is over, the released
cable is then pulled back on board and joined to the
remaining cable so that the operation can be resumed
back to normal.
A CLV with self propulsion and Dynamic Positioning (DP)
system, higher bollard pull capability and dedicated cable
laying equipment on board can perform the cable laying
operation efciently.
arrangement. In-eld cables are 150mm maximum in
diameter and are deployed in short lengths. These
cables are stored either in a cable carousel or in reels
on an in-eld CLV.
Export Power Cables: As shown in Figure 1, export
cable is connected between the offshore sub-station
and the shore facility. The length of the export cable
depends on the distance separating these two
locations. It is the largest diameter and the longest
cable in a wind farm project. It is always stored in a
suitably sized cable carousel on an export CLV to
maintain its integrity, as in most cases, joints are not
accepted in power cables.
In a typical wind farm project, in-eld cable laying
process always needs more time than laying of export
cable as the total length of in-eld cable is more but is
laid in shorter length. In-eld cables and export cables
are signicantly different in their physical properties,
such as stiffness, cable diameter, weight and length of
the cable required for a wind farm. Because of these
differences, they need different equipment for their
handling, storage and cable laying operation, which in
turn, justies the need for two different types of cable
laying vessels.
DESCRIPTION OF A CABLE
LAYING VESSEL (CLV)
Offshore power cable laying operation involves multi-
functional tasks, including route planning and clearing,
cable lay and trenching, cable connections and nally,
the post laid cable inspection. It is not possible for a
single vessel to perform all of these tasks. The main
functions of a dedicated cable laying vessel (CLV) are
cable storage, cable load-out, seabed trenching, cable
laying, and burying of the cable into the trench.
Cable Storage
Typically, a CLV should be designed to carry the full
length of in-eld or export cable required for a wind
farm project. Depending on the type of CLV, cables
Dedicated Cable Laying Vessels For Future
Offshore Wind Farm Development 33
34 KOMtech Technology Review 2011
PRESENT CABLE LAYING VESSELS
AND THEIR LIMITATIONS
Currently, cable laying operation for offshore wind
farm is carried out using barges or converted
Offshore Support Vessels (OSV). All the necessary
equipment for cable laying operation are installed
on the existing hull of such barges or OSVs to
convert them into cable laying vessels. But, these
types of conversions do not ensure the synchronisation
of vessels characteristics with the installed cable
laying equipment on them. Hence, it is difcult for
these converted vessels to perform the cable laying
operation efciently. Modication of a vessel can
increase its functionality in completing some special
tasks. But the benets of these modications must
be weighed against the performance of an intended
job, the increased cost and high day rates of these
converted vessels.
Future wind farms will be located further away from
the shore and in deeper water. The vessel required for
cable laying operation in those future wind farms
should have a larger deck area to carry all the necessary
equipment and power cables on deck, and be self
propelled and dynamically positioned with the
required bollard pull capacity. They should be
independent and dedicated vessels specially designed
for cable laying operation.
Current cable laying barges have enough deck space
but they are not self propelled and are unable to
maintain their position by themselves. This means
that they are always dependent on other vessels for
their position, manoeuvring and pulling power
required for cable laying operation. On the other
hand, modied OSVs are self propelled and
dynamically positioned vessels but, do not have
enough deck space to carry the required length of
cable and the equipment necessary for a complete
cable laying operation. They also need supports from
other vessels for carrying extra cable and equipment
requires completing the cable laying tasks. Hence,
cable laying operation cannot be done efciently with
these converted CLVs,
Considering the above dependency on other vessels
for propulsion, manoeuvring, position keeping; lack
of pulling capability and less deck space, these
modied barges or OSVs may not be suitable for
future wind farm projects located further from shore
and in deeper water than the present projects.
KOMTECH DESIGN OF CABLE
LAYING VESSELS
Two main types of offshore power cables, in-eld cable
and export cable, are signicantly different in their
physical properties, such as stiffness, sizes, weight and
length of the cable required for a wind farm. They also
need different capacities for cable handling, storage
and burial equipment, as the physical properties of
these cables are different. Due to these differences in
power cable characteristics, a single vessel specied to
meet the cable laying requirement for both types of
cables is unlikely to be an efcient solution. It may not
be economically feasible for a single vessel or a single
equipment to handle all sizes of power cable required
for a wind farm. Because, laying of bigger cable needs
a vessel with higher pulling capacity as well as higher
engine power. This also requires higher rating of cable
laying equipment. Again, a vessel designed for laying
smaller size cables may not be capable for handling
bigger size cables due to its capacity limitation. In such
a case, the most optimal solution is to use two different
types of vessels specied for different cable sizes and to
add some extra features on them so that they can also
be used for some extra tasks, such as repair and
maintenance of offshore power cables.
KOMtech has recently designed cable laying vessels
(CLV) for the offshore wind power industry. Considering
the limitations of current CLV, the differences in offshore
power cables properties, and the impracticality of a
combined CLV as mentioned above, KOMtech has
designed two different types of vessels for the purpose of
laying two different types of cables, namely; In-eld
Cable Laying Vessel and Export Cable Laying Vessel.
KOMtech In-Field Cable Laying Vessel
General arrangement of the in-eld cable laying vessel
with the cable laying equipment is shown in Figure 2.
KOMtech has designed two different types of vessels for
the purpose of laying two different types of cables, namely;
In-eld Cable Laying Vessel and Export Cable Laying Vessel.
F
i
g
u
r
e
2
.
G
e
n
e
r
a
l
A
r
r
a
n
g
e
m
e
n
t
o
f
K
O
M
t
e
c
h
I
n
-
F
i
e
l
d
C
a
b
l
e
L
a
y
i
n
g
V
e
s
s
e
l
Dedicated Cable Laying Vessels For Future
Offshore Wind Farm Development 35
36 KOMtech Technology Review 2011
Main function of in-eld CLV is to lay in-eld power
cables within the wind farm in between the turbines.
As wind turbines are positioned close to each other
within a distance of 1km, in-eld CLV should have
a better position keeping capability to lay cable in
predetermined route and good manoeuvrability to
prevent hitting against the support structures.
The KOMtech in-eld CLV is powered by a Diesel-
Mechanical (DM) propulsion system with two
engine rooms. Forward engine room consists of three
auxiliary power generators, each with a capacity of
1000ekW; a harbour generator of capacity 750ekW
and other marine auxiliaries in combination with two
diesel engine driven retractable azimuth thrusters of
1230kW each; and an electrical driven bow tunnel
thruster of capacity 600ekW. The aft engine room
accommodates two diesel engine driven propulsion
azimuth thrusters, each with a capacity of 1230kW.
An emergency generator set of 250ekW capacity
is installed inside the accommodation block to
the forecastle deck level. The combined thruster
power of an in-eld CLV can maintain dynamic
positioning (DP-2) and good manoeuvrability of
the vessel at sea state condition four in unrestricted
service worldwide except in ice.
A six-tier accommodation block is arranged on
main deck in the front part of the vessel. The
KOMtech in-eld CLV has a large open deck area
of 1650m, designed for uniform loading of 10
tonne/m suitable for the cable laying equipment.
Equipment, such as cable tensioner, cable guide,
cable chute, etc, are arranged on deck considering
their sequential operation and ensuring that there
will be no interference among the equipment and
their working envelop. A cable carousel of 24m
outer diameter with 4000MT in-eld cable capacity
is placed at the centre of the main deck. Table 1 lists
the main specications of the cable carousel.
A work class ROV is used for cable laying operation.
The ROV with its launching A-Frame and other
accessories are placed on the main deck at forward
side. The ROV is capable of laying up-to 150mm
diameter cables and bury it into a trench of up to two
meter deep. As the ROV has trenching capability, no
extra bollard pull is required for it, which in turn,
reduces the power requirement of the vessel. The
arrangement of cable load-out system including the
loading tower, cable engine, cable tensioner, cable
brake, conveyer system and cable chute ensures the
smooth operation of power cable in a simple and
safe way without any damage. Specications of a
KOMtech designed typical in-eld CLV are given
in Table 2.
KOMtech Export Cable Laying Vessel
General arrangement of KOMtech export cable
laying vessel with the cable laying equipment is
shown in Figure 3.
The main function of export CLV is to lay the export
cable between the offshore substation and the shore
facility. The export cable is laid on a predened
Table 1. Cable carousel capacity of In-Field CLV
Items Description
Carousel Size (Do x Di x H) 24m x 3.0m x 5.0m
Carousel Capacity 4000MT of in-eld
cable (equivalent to
130km x 130mm)
Cable Size operable Up to 150mm
Table 2. KOMtech In-Field Cable Laying Vessel
Specications
Items Value
Length overall (m) 90.0
Breadth Molded (m) 28.0
Depth Molded (m) 6.5
Draft Maximum (m) 4.7
Approx Vessels speed (knots) 10.0
Accommodation (Persons) 60
Main Propulsion Engine / 2 x 1230
Thrusters (kW)
Diesel Driven Fwd Retractable 2 x 1230
Azimuth Thrusters (kW)
Bow Tunnel Thruster (CP) (ekW) 1 x 600
Auxiliary Generators (ekW) 3 x 1000
Harbour Generator (ekW) 1 x 750
Cable Carousel Capacity (Tonne) 4000
Maxm Cable Size Operable (mm) 150
F
i
g
u
r
e
3
.
G
e
n
e
r
a
l
A
r
r
a
n
g
e
m
e
n
t
o
f
K
O
M
t
e
c
h
E
x
p
o
r
t
C
a
b
l
e
L
a
y
i
n
g
V
e
s
s
e
l
Dedicated Cable Laying Vessels For Future
Offshore Wind Farm Development 37
38 KOMtech Technology Review 2011
to deploy the plough from the stern of the vessel.
Cable load-out system arrangement ensures the
smooth delivery of power cable into sea in a simple
and safe way without any damage. An ROV option
is available for cable inspection and cable repairing
services of the post laying cable. The ROV with its
launching A-Frame and other accessories are placed
on main deck at foreword side of the carousel.
Specications of a KOMtech export CLV is given
in Table 4.
Table 3. Cable carousel capacity of Export CLV
Items Description
Carousel Size (Do x Di x H) 26m x 6.8m x 7.0m
Carousel Capacity 6000MT of
export cable
(equivalent to
50km x 250mm)
Cable Size operable Up to 300mm
Table 4. KOMtech Export Cable Laying Vessel
Specications
Items Value
Length overall (m) 122.6
Breadth Molded (m) 30.0
Depth Molded (m) 12.0
Draft Maximum (m) 8.0
Approx Vessels speed (knots) 12.0
Accommodation (Persons) 60
Main Generator (ekW) 4 x 3745
Harbour Generator(ekW) 1 x 1000
Stern Azimuth Thrusters (ekW) 2 x 3500
Fwd Retractable Azimuth Thruster 1 x 2500
(ekW)
Bow Tunnel Thruster (ekW) 2 x 1000
Cable Carousel Capacity (Tonne) 6000
Maximum operable Cable Size (mm) 300
Available Bollard Pull (Tonne) 150
route using a plough which is pulled by the ship.
The export CLV must therefore be a dynamically
positioned vessel with high bollard pull capability.
KOMtech export CLV is powered by a Diesel-
Electrical (DE) propulsion system consisting of
one engine room at the foreword and two thruster
rooms, one at the aft and another one at the
foreword of the vessel. Generator sets and marine
auxiliaries are placed in the engine room. Electrical
power is generated by four propulsion generator
sets, each with a capacity of 3745ekW and a harbour
generator set of 1000ekW capacity. An emergency
generator set of 250ekW capacity is installed inside
the accommodation block to the forecastle deck
level.
Aft propulsion thruster room accommodates two
electrical motor driven azimuth thrusters with a
capacity of 3500ekW each. The forward thruster
room accommodates one electrical motor driven
retractable azimuth thruster of 2500ekW capacity
and two tunnel thrusters of 1000ekW capacity each.
All the thrusters are driven by frequency controlled
electrical motor. The combined thruster power of
the export CLV is strong enough for its propulsion,
position keeping (DP-2) and manoeuvrability of
the vessel at sea state condition four in unrestricted
service worldwide except in ice and is capable to
produce a 150-tonne bollard pull which is sufcient
to pull a plough during the trenching operation.
KOMtech export CLV also has a six-tier
accommodation block on forward part of the main
deck. It has a large open deck area of 2560m for
cable laying equipment with a uniform deck loading
of 10 tonne/m. Similar to the in-eld CLV, cable
laying equipment for the export CLV are arranged
on deck considering their sequential operation and
avoiding working interference. A cable carousel of
26m outer diameter with 6000MT export cable
capacity is placed at the centre of the main deck.
Specications of the cable carousel are given in
Table 3.
The operation of laying export cable is done with
a plough. It is capable of laying up to a 300mm
export power cable at three meter trenching depth.
A hydraulically operated stern A-Frame is used
CONCLUSION
Recent developments of cable laying equipment
have increased their functionality and hence,
improved the cable laying operation. KOMtech
CLVs have been designed in consideration of the
recent developments of cable laying equipment
from major manufacturers. In addition, opinions
of several wind farm developers on their likely
vessel requirements for future offshore wind
farm have been considered in the design process.
The vessels hulls and their marine equipment
are designed in such a way that they have the
capability to synchronise the vessels characteristics
with the functional capability of these cable laying
equipment. Following are some important features
of KOMtech CLVs-
Better manoeuvrability of the vessel.
Improved stability of the vessel to work in
specied environment condition.
Large deck space to accommodate the necessary
cable laying equipment.
Bigger cable carousel to carry maximum amount
of cable required for a wind farm project.
Available bollard pull capacity to tow a cable
laying device as required.
Cable engine/cable tensioner with wide range of
operation.
Conveyor belts for smooth running of cable
and to prevent the cable damage.
Stern chute with roller type system makes cable
load-out simple and safe.
An ROV option for cable laying and repairing
function.
Combination of these vessels properties and the
cable laying equipment can work efciently as a
future CLV. These types of vessels can be considered
as a Dedicated CLV for Future Offshore
Wind Farm Development. A dedicated cable
laying vessel is inherently safer than a temporarily
converted vessel.
AUTHORS CONTACT Abul.Bashar@keppelom.comYan.Wang@keppelom.com
REFERENCES
[1] World Wind Energy Association (http://www.wwindea.org/)
[2] Barrow Offshore Wind (BOW) (http://www.bowind.co.uk/project.shtml )
[3] K.S. Foo, A. Mortensen, T.T. Wong and Y.J. King, Offshore Wind Turbine Installer, KOMtech Technology Review (2010), pp 5-17
[4] M. Perry and H. Krisdani, Foundation Structures for Offshore Wind Turbines, KOMtech Technology Review (2010), pp 19-26
[5] M.H. Chia, Electrical Power Transmission Schemes for Offshore wind Farms, , KOMtech Technology Review (2010), pp 27-33
[6] W. Henstra, Self-Installing Substation for Offshore Wind Farms, KOMtech Technology Review (2010), pp 35-42
Dedicated Cable Laying Vessels For Future
Offshore Wind Farm Development 39
40 KOMtech Technology Review 2011
Lorem Dolor
Lorem dolor ipsum 41
No part of the materials published in this journal may be reproduced, stored in a retrieval system or transmitted in any form whatsoever without the prior written permission of KOMtech
SHIP SIDE SLIPWAY LAUNCHING INVOLVES THREE PHASES, sliding in the
quay, tilting but still in contact with the quay and fully-oating in the water. From a
hydrodynamic point of view, there might be some risks during these phases, such
as touching the canal bottom, hitting the canal quay, or worst of all, capsizing. In
order to manage the risks for ship side launching, shipyard needs a tool to assess
the hydrodynamic behaviour of ship side launching. This paper presents numerical
simulation of ship side launching using a Computational Fluid Dynamics (CFD)
software, StarCCM+, coupled with a user dened program. A series of numerical
simulations for a 92m platform supply vessel side launching have been performed
successfully. The water ows around the vessel and the motion of the vessel has
also been evaluated. The results show the ability of CFD to predict the dynamic
behaviors that need to be considered for ship side launching.
Key words: Numerical simulation, Computational Fluid Dynamics (CFD), Ship side launching
ACKNOWLEDGEMENT
The authors would like to thank Forrukh Ahmed and Yujie Cong from Keppel
Singmarine for providing empirical calculation results and AutoCAD drawings,
Mr Peter Ewing from CD-adapco Australia for providing invaluable advice.
Numerical Simulation for
Ship Side Launching
LIANG Jun, M.Eng, B.Eng
James CAO Yimin, B.Eng
TAN Chenghui, C.Eng, P.Eng, MSc, BSc
Numerial Simulation for
Ship Side Launching 41
42 KOMtech Technology Review 2011
INTRODUCTION
Generally, there are three methods for ship
launching, which are gravitational method, oating
out method and ship lift method. The gravitational
method is divided into longitudinal slipway
launching and side slipway launching. Before
launching, coatings such as wax and grease should
be spread on the slipway. As these coatings are
very sensitive to temperature, some shipyards use
intensive steel-rollers instead of wax and grease.
The oating out method is adopted for ship that
is built in drydock. After construction, water is
allowed to ow into drydock and subsequently
the ship is oated out. This method is very
reliable and relatively simple but the initial
investment is high. Ship lift method is normally
used for small and special ships such as sailing
boat and catamaran. This method requires
complicated equipment and also high maintenance
fee. Besides all the above mentioned method, there
are some innovative methods, such as using high
pressure inatable airbags
[1]
.
In 2010 Keppel Singmarine Pte Ltd acquired
a shipyard in Brazil. Because of its location and
land / waterfront limitations, side launching becomes
the only option. As a result, quick and reliable
assessment of hydrodynamic behaviour of ship side
launching is vital to every successful operation.
As shown in Figure 1, the side launching of a ship is
divided into the following three phases
[2]
. Phase 1,
ship is sliding on the quay, Phase 2, ship is tilting
but still in contact with the quay, and Phase 3, ship
is oating in the water.
In this paper, we investigate the hydrodynamic
behaviour of ship side launching such as the
calculation of roll angles and trajectory of chines.
Using user dened programs and commercial
software StarCCM+
[3]
, which is a general-purpose
Computational Fluid Dynamics (CFD) code based on
the nite volume method, side launching of a multi-
purpose Platform Supply Vessel (PSV) constructed
by Keppel Singmarine (Figure 2) has been studied.
Figure 1. Three phases of side launching a ship
Figure 2. Sketch of platform supply vessel
Figure 3. Coordinate system
Numerial Simulation for
Ship Side Launching 43
NUMERICAL SIMULATION
The global coordinate origin is dened at the point
on the edge of the quay. The x-coordinate direction is
dened along this edge. The y-coordinate is dened
going portside, while z-coordinate is dened going
upwards (Figure 3).
This paper presents two cases to evaluate the
sensitivity of vertical centre of gravity (VCG)
locations during launching. The VCG used in the
study are 7.86m for case 1 and 9.50m for case 2.
GEOMETRY AND MESH
The vessel for side launching numerical study is a
multi-purpose PSV. Its main parameters, as well as
the canal and launch ways parameters are presented
in Table 1. The sketch of numerical simulation of
side launching is shown in Figure 3.
In the numerical model, the 3D hull with a skeg has
a total of 300,000 trimmed cells. The mesh around
the water surface has been rened locally. A morpher
and dynamic uid body interaction technique has
been selected for the mesh deformation due to rigid
body motion. Because of the large scale motion,
negative volume mesh cannot be avoided. Hence,
the mesh should be re-meshed manually several
times during the simulation. The section view of the
whole model is presented in Figure 4.
Table 1. Main Parameters
Parameters Symbol Unit Value
Length of canal L m 200.00
Width of canal W m 250.00
Canal Height of left quay H_PS m 30.00
Height of right quay H_SB m 30.00
Water depth d m 28.00
Density of water ton/m
3
1.025
Slipway slope xita deg 5.73
Launch way Height of launch way, including cradle h m 2.50
Friction coefcient of slipway Cf 0.08
Launch sliding distance D m 12.50
Initial speed at Phase 2 V m/s 3.67
Length over all Loa m 93.60
Breadth of the ship B m 19.70
Vessel Depth of the ship T m 7.85
Launch weight ton 4000.00
Longitudinal Center of Gravity from AP LCG m 49.78
Transversal Center of Gravity from CL TCG m 0.00
Radius of gyration (Kxx=0.34B) kxx m 6.70
Figure 4. Section view of domain mesh
44 KOMtech Technology Review 2011
ANALYSIS METHODOLOGY
There are some simplications that need to be
adopted during CFD simulation. Actual ship
launching should consider the deformation of the
slipway and support structures. Since our aim is
to study the hydrodynamic behavior of launching,
one simplication is that the slipway and ship are
considered as rigid bodies without any deformation.
In addition, the CFD simulation starts from
Phase 2 and the aerodynamic force is ignored.
The normal force from the quay is difcult to obtain,
but it could be derived from the relation when
the cradles bottom is in contact with the slipway,
as long as the vessel is still tilting (phase 2). The
detailed derivation was performed by Bobby Hak
[2]
.
In order to calculate the force from the quay, a user
dened program has been coded and coupled with
CFD solver at each time step. The normal force N
from the quay is as follows:
(1)
where, P is the gravitational force (N); R
z
, R
y
,
M
RX
, is the uid force (N) in each direction and its
moment (Nm) along x axis; m is the mass of vessel
(kg); I
x
is the rolling inertial moment (kgm2). a, b,
d,e, f are dened as:
where, is the friction coefcient; k
G
is the angle
of cradle bottom regarding to the horizontal plane
(rad);
G
is the angle from vertical plane (rad); h
G
is
the distance from CG to vessel bottom including
the cradle (m); is the initial angle of vessel from
horizontal plane (rad).
The uid force is computed by the numerical
method solving the uid mass continuity equation
and the incompressible Reynolds-Average Navier-
Stokes (RANS) equations with a Standard K-Epsilon
model. The free surface is captured using Volume-
of-Fluid (VOF) method. The time-dependant
pressure distributions around the hull are generated
at each time step. 6-DOF of rigid body motion has
been added to the simulation.
RESULTS AND DISCUSSIONS
A full scale simulation was carried out in order to
enable direct comparison with empirical analysis
performed by Keppel Singmarine
[4]
based on
SNAME
[5]
. As shown in Table 2, the empirical
results of case 1 are in good agreement with those
from CFD. Referring to the positions of chine,
both starboard and portside chine exceed 7m below
the water level. That means if the water depth is
less than 7m, the ship will hit the canal bottom.
Although the VCG has been changed, the position
traces of chine do not change much. The reasons
are that the small change of VCG has little effect
on rolling moment and the radius of gyration is
assumed the same for both cases.
N =
(P+R
z
)(R
y
+M
Rx
me
+
md
)
I
x
b+af+(az
g
+by
g
)
me
I
x
a = sin(k
G
) cos (k
G
)
b = cos (k
G
) sin (k
G
)
d =
e =
t = tan (k
G
)
2y
G
G
cos(k
G
)+2y
G
G
2
sin(k
G
)
+
1+sin
2
(k
G
)
cos
3
(k
G
)
cos
3
(k
G
)
h
g
cos()G
2
y
G
+
sin
2
(k
G
)
cos
2
(k
G
)
h
g
cos()G
2
cos
2
(k
G
)
(2)
(3)
(4)
(5)
(6)
Numerial Simulation for
Ship Side Launching 45
Figures 5 to 7 show the rolling and translations of
the vessel. For CFD case 1, the largest rolling angle
is 38.6 degrees at 4.3s. When the vessel rolls back,
the largest rolling angle is 32.4 degrees at 10.2s. For
CFD case 2, the largest rolling angle is 39.4 degrees
at 4.3s. When the vessel rolls back, the largest rolling
angle is 32.4 degrees at 10.3s. The largest rolling
angle does not decrease too much because the
bilge keels and other appendages are not included
in the numerical analysis. In addition, according
to Hak, 2005
[2]
, large radius of gyration (0.34B in
our case) might cause the rolling to decay slowly.
From Figures 5 and 6, it can be deduced that the
vessel moves away from the quay very quickly,
around 25m from the quay after 14s. This condition
is very favorable since it will avoid collision with
canal quay. The positions of the starboard and
portside chine are calculated based on the positions
of CG. Figure 7 shows the trace of portside chine
during launching. From the position of the portside
chine, the required minimum canal water depth
can be determined.
Table 2. Comparison between Numerical and Empirical results
Parameters CFD Empirical
case 1 case 2
1 VCG (m) 7.86 9.50 7.86
2 Max rolling angle (deg) 38.6 39.4 36.0
3 Max CG distance from the quay (m) 25.8 25.4 N/A
4 Max Chine_ps below the water level (m) 7.6 7.1 7.4
5 Max Chine_sb below the water level (m) 7.8 7.2 7.9
Figure 5. Comparison between two cases and
empirical formulae
CONCLUSIONS AND REMARKS
Numerical simulation of ship side launching has been
presented in this paper. The morpher and 6 DOF
rigid body motion techniques have been applied for
large displacement motion. A user dened program
is successfully applied for calculating the normal
force from the quay at each time step. The CFD
result indicated that the change of VCG does not
affect the maximum rolling angle of the ship. The
ship will roll back shortly after dropping in the canal,
which means that the ships stability is sufcient and
the side launching operation is safe.
Figure 6. Rolling angle
Figure 7. Position trace of portside chine
46 KOMtech Technology Review 2011
AUTHORS CONTACT Jun.Liang@keppelom.com
REFERENCES
[1] Qindao evergreen shipping supplies co.ltd, Introduction of ship launching airbags, 2009.
[2] Bobby Hak, Numerical simulation of the side launching of a ship, University of Groningen, August 2005
[3] CD-adapco, StarCCM+ User Guide, 2008.
[4] Forrukh Ahmed and YJ Cong, Side launching calculation, Keppel Singamarine, July 2010
[5] SNAME, Principles of Naval Architecture, 1967
No part of the materials published in this journal may be reproduced, stored in a retrieval system or transmitted in any form whatsoever without the prior written permission of KOMtech
THE CFD (COMPUTATIONAL FLUID DYNAMICS) SIMULATION APPROACH
is presented as an effective and suitable method to predict the resistance
and propulsion of non-streamlined wind turbine installation jackup vessels and
semisubmersibles. In this approach, a general-purpose commercial CFD solver is
adopted to achieve the solution accuracy, while the offshore & marine engineering
related issues such as the free surface, turbulence effect and wave reection
are specially addressed. Furthermore, a practical propulsion prediction method
is proposed to tackle the hull-propeller interaction and the latest HPC
(High Performance Computing) techniques are used to speed up the CFD
computation. As a result, this approach can achieve a good compromise
between accuracy and efciency. The present approach is nally applied to
resistance and propulsion prediction for wind turbine installation jackup vessels
and semisubmersibles. The corresponding CFD results are in good agreement
with the available experimental results. The signicant effect of the vessel
bottom openings on resistance and propulsion due to water oscillations is
also highlighted. It is suggested that the CFD simulation approach for resistance
and propulsion prediction should gain wider acceptance for initial practical
ship design.
ACKNOWLEDGEMENT
This work was partially supported by Offshore Technology Development (OTD)
and Deepwater Technology Group (DTG). We are grateful to Margarita Ivanova
Georgieva and Wang Chao in OTD as well as Edwin Nah Hock Choon and Zhang
Yanting in DTG for sharing their insights and expertise. We also would like to
express our gratitude to Peter Ewing from CD-adapco Australia for his helpful
technical support of the CFD software.
CFD-based Resistance and
Propulsion Prediction for
Wind Turbine Installation
Jackup Vessels and
Semisubmersibles
WANG Shengyin, PhD, M.Eng, B.Eng
Matthew QUAH Chin Kau, PhD, CEng, CMarEng, FIMarEST
CFD-based Resistance and Propulsion
Prediction for WInd Turbine Installation
Jackup Vessels and Semisubmersibles 47
48 KOMtech Technology Review 2011
INTRODUCTION
In ship hydrodynamics, resistance (or drag) is the
horizontal component of the force opposing the
forward motion of a ship
[1]
while propulsion is the
act of providing the thrust as the driving force to
overcome the resistance to push forward the ship
with the aid of a propulsive device such as a
propeller
[2-4]
. It should be noted that the term
resistance is preferred in ship hydrodynamics, while
the term drag is generally used in aerodynamics and
for submerged bodies. Conventionally, resistance
analysis is the starting point for propulsion
prediction
[2,3]
, though in reality the two parts are
closely inter-dependent due to the interaction
between ship hull and propeller. It has been well
accepted that accurate prediction of resistance and
propulsion is a critically important but always
challenging task.
There are several options available for resistance and
propulsion prediction, ranging from the traditional
methods through to the more advanced CFD
methods
[2,4]
. The basic classes of approach
[4]
are the
traditional and standard series methods, the
statistical regression based procedures, the direct
model test methods, and the CFD simulation
approach. Unlike the CFD and direct model test
approaches, the former two are based on the
traditional naval architectural parameters of hull
form. When the hull forms become more complex
and the requirements become stricter, these few
traditional parameters tend to be unable to reect
the growth of the boundary layer, as well as wave
making and eddy making components. As a result,
the actual resistance of modern ship hulls may often
be underestimated
[2]
. The direct model test approach
is still seen as an essential part in the design of a ship
to predict or validate the power requirements in
calm water
[2]
. Nevertheless, the direct model test
approach can be costly and time consuming with a
limited number of data points and poor repeatability
and arguable accountability to convert the model
test results into the prediction of a full scale ship.
CFD simulation is emerging as a more suitable
approach due to the rapid developments in computer
technology and numerical methods
[5-7]
. The CFD
analysis can be done repeatedly in design loops to
provide useful insights into the phenomenological
behavior where classical extrapolation techniques in
the direct model test approach are not applicable.
However, the CFD simulation approach for
resistance and propulsion prediction has not gained
wide acceptance yet due to the past difculties in
reaching a solution within a reasonable time,
modeling the critically important free surface
accurately, generating quality volume mesh for
complex geometries with localised grid renement,
achieving non-reecting boundary conditions and
simulating the ship hull-propeller interaction in the
high Reynolds number turbulent ow accurately.
The objective of the present work is to further
develop the CFD simulation approach for accurate
and efcient resistance and propulsion prediction
for non-streamlined ships using state-of-the-art
CFD methods and HPC techniques. The HPC
techniques based on parallel computation are used
to obtain a much faster CFD solution speed while
the lter scheme approach in
[8]
is further developed
to achieve robust non-reecting boundary conditions
to damp out the non-physical wave reection from
the articial boundaries. To facilitate the ship
hull-propeller interaction analysis with sufcient
engineering accuracy and to enable reliable
propulsion prediction, the factors and efciencies
derived from vendors open water propeller data and
the CFD simulation results are utilised. Hence, the
traditional difculties and technical barriers are
resolved and more accurate and efcient prediction
results can be expected.
CFD-BASED RESISTANCE AND
PROPULSION PREDICTION
Effective CFD Methods for Resistance
Prediction
In the present CFD simulation approach, the CFD
methods were used to predict the calm water
resistance of a steadily advancing vessel in bare hull
form without a propulsive system.
In order to perform the CFD analysis, a sufciently
large computational domain enclosing the vessel
was established. Inside the CFD computational
domain, both the air and water exist and thus a
multiphase ow model was adopted to obtain the
critically important free surface between the air and
water. The uid ow with air and water was modeled
as a two-phase ow and the VOF (Volume Of Fluid)
method
[9]
was used to capture the free surface, which
was once a major obstacle for applying viscous CFD
CFD simulation is emerging as a more suitable approach
due to the rapid developments in computer technology
and numerical methods.
methods to offshore & marine engineering
[6,7]
. The
steady forward motion of the vessel was represented
by an equivalent at VOF wave
[7]
with a constant
speed of water and air at the far eld while the vessel
is initially xed at zero sinkage and trim. It was
assumed that the vessels Froude number was low
enough within the present vessel speed range of
interest so that the effect of dynamic sinkage and
trim would be insignicant
[10]
. Furthermore, the
two-phase ow was assumed to be unsteady to
achieve a robust solution convergence and the
impact of initial conditions was alleviated by using
a ramp function to apply the boundary conditions
gradually. Due to the full-scale vessel size in the
CFD simulation, the unsteady VOF ow is turbulent
with a high Reynolds number, which was proven to
be difcult for the direct model test approach
[2]
. In
the present CFD simulation approach, the RANS
(Reynolds-averaged Navier-Stokes) equations
[2]
,
time-averaged equations of motion for uid ow,
were adopted to deal with the highly turbulent ow
and thus appropriate choice of a turbulence model
becomes essential. Menters SST k-omega turbulence
model
[11]
, was adopted as a much better choice than
Wilcoxs k-omega model or the popular standard
k-epsilon model to capture the nominal turbulent
wake and estimate the turbulence effects around the
hull. In order to recover the behaviours of the wall
treatment in the limit of very ne or very coarse
meshes for a turbulence model, the All y+ Wall
Treatment in
[7]
was employed. It is a hybrid approach
to emulate the High y+ Treatment for coarse meshes
and Low y+ Treatment for ne meshes and may even
give reasonable results for intermediate meshes
where the wall-cell centroids fall within the buffer
region of the boundary layer. The control volume-
based nite volume method was adopted in the
present CFD solver to guarantee the solution
accuracy and mass conservation and a nite-element
type mesh can be used to achieve the efciency. To
generate a high quality volume mesh with an
appropriate boundary layer resolution, the trimmer
meshing model together with a prism layer mesher
in
[7]
was used. The resulting trimmed mesh was
dominantly composed of hexahedral cells to reduce
the total number of unknowns as a good compromise
to achieve the computational efciency and accuracy.
Furthermore, the local mesh renement allows for
high quality resolution of the free surface and ow
separation around the hull. In order to damp out
the unphysical wave reection in the vicinity of the
articial boundaries at the far eld, a dissipation
zone with extra active damping effect was predened
and the corresponding wave reection damping
lter was fullled by adding a user dened
momentum source in the RANS solver
[7]
, similar to
but more robust than the lter scheme approach to
achieving non-reecting boundary conditions
[8]
.
To implement the present CFD simulation
approach for resistance prediction, the next
generation general-purpose control-volume based
commercial CFD software tool STAR-CCM+
v4.04 was utilised, which includes the modern
software development technology, state-of-the-art
computational continuum mechanics algorithms
[7]
,
and excellent user environment design.
Practical Propulsion Prediction Method
Propulsion prediction during the design process is
usually difcult due to the lack of knowledge of the
wake eld in which the propeller is to operate
[2-4]
.
Usually, a model nominal wake eld is measured
during the ship model test prior to construction to
allow the designer to understand in a qualitative sense
the characteristics of the wake eld. However, the
designer needs to transform the model nominal wake
eld into a ship effective velocity distribution before
it can be used for quantitative design purposes. This
transformation is far from clearly dened within the
current state of knowledge
[4]
. In the case where the
ship has not been model tested, the designer must
rely on his knowledge of other similar ships and the
way they performed to make empirically based
estimates. Furthermore, application of CFD to the
analysis and design of marine propeller is still at a
relatively early stage of its development for practical
propeller computations
[2,4]
.
CFD-based Resistance and Propulsion
Prediction for WInd Turbine Installation
Jackup Vessels and Semisubmersibles 49
50 KOMtech Technology Review 2011
of the cluster and complexity of the problem and is
difcult to state in advance. It is recommended that
the minimum number of mesh cells placed on each
cluster node to be at 100,000.
RESULTS AND DISCUSSION
CFD Validation for Resistance of the Ship
Model DTMB 5415
In general, the use of an advanced CFD tool as
well as the application of the viscous CFD methods
can be depended for simulation-based design only
if the numerical simulation results have been veried
and validated. The well-known surface combatant
ship model David Taylor Model Basin (DTMB)
5415
[5]
was selected for the present CFD validation
in resistance analysis. The ship model DTMB
5415
[5]
is a towing tank model representing a
modern US naval combatant. The model geometry,
which includes both bulbous bow/a sonar dome
and a transom stern as shown in Figure 1, has been
adopted by ITTC as a recommended benchmark
for CFD validation in resistance and propulsion
[5]
.
The principal particulars of this model are shown in
Table 1. In the present CFD validation, only the
bare hull of DTMB 5415 with a scale factor of
1:24.824 and static sinkage and trim
[12]
was chosen.
The present propulsion prediction method was
based on the thrust identity approach
[2-4]
, similar
to the well-known ITTC 1978 speed-power
performance prediction method
[2]
. However, the
resistance and nominal wake eld were directly
obtained from the CFD simulation for resistance
prediction and only the open water propeller
performance data were obtained from the vendors
test results. Hence, further time-consuming CFD
analysis can be avoided and a reliable propulsion
prediction would become possible.
Efcient HPC Techniques
Although the term HPC-High Performance
Computing, is more commonly associated with
computing for scientic research or computational
science, it is closely related to the engineering
applications of cluster-based computing for CFD
(Computational Fluid Dynamics) in this work.
The parallel computing operating system Microsoft
HPC Server 2008 was used to manage a cluster of
Windows 2008 64 bit machines comprising of
two HP Proliant Z6000 servers with a total of
4 quad-core processors (16 cores) and 128 GB RAM
on both enterprise and private networks with a
bandwidth of 1 Gigabits per second (Gbps)
respectively. The two techniques of HPC multiple
cores and multiple machines were combined to
execute jobs consisting of individual tasks to achieve
an even more powerful HPC performance. Under
the Windows HPC Server 2008 environment, the
Message Passing Interface (MPI) based domain
decomposition technique was used for efcient
parallel computation and the CFD jobs were
submitted from our CFD software tool
STAR-CCM+ to the Windows CCS job scheduler
using MS-MPI either in batch mode or
interactively
[7]
. It should be noted that in CFD, it is
highly desirable to compute solutions in parallel to
solve problems faster and increase the total amount
of memory available for a given simulation
[7]
because of the intensive computational work
required to achieve converged solutions.
It is important to scale the simulation size to be
in line the number of cluster nodes being used
[7]
.
The optimal scaling is determined from a ratio of
the time to compute and the time taken to exchange
data between cluster nodes. The optimal number of
cluster nodes to use is dependent upon the hardware
Figure 1. Bare Hull Geometry of the Ship Model
DTMB 5415
Table 1. Principal Particulars of the Ship Model
DTMB 5415
Particular Full Scale Model Scale
Length L (m) 142 5.72
Breadth B (m) 17.973 0.724
Draught T (m) 6.15 0.248
Wetted 2972.6 4.861
Surface Area S (m
2
)
Speed U (m/s
2
) 9.252 2.097
Reynolds Number Re 1.40E9 1.26E7
Froude Number Fr 0.248 0.28
A comparison of the present resistance components
and those available in the literature
[12]
is shown in
Table 2, in which is the relative error of the
total resistance with respect to the EFD result, as
dened in
[3]
. It can be seen that the total resistance
with a relative error of 0.9% predicted by the
present CFD code in STAR-CCM+ v4.04 is almost
identical to the EFD result
[12,14]
given by IIHR
(Iowa Institute of Hydraulic Research). All the other
CFD codes listed in Table 2 provided fairly good
solutions, but cannot be as accurate as the present
code due to the signicant differences of the
CFD methods implemented. It should be noted
that all the CFD codes in
[3]
were developed at least
10 years ago while the present CFD code
STAR-CCM+ v4.04 was released in 2009 only.
Due to its newly developed robust and efcient
algorithms
[7]
, the present CFD code can be the
most accurate among all the listed CFD codes in
Table 2.
Figure 2. Grid Convergence of Resistance Components Figure 3. Trimmed Mesh with Predominantly
Hexahedral Cells
Figure 4. Predicted Wave Pattern Composed of
Primary and Secondary Wave Systems
The grid convergence of the resistance components
was rst investigated. Figure 2 shows the grid
convergence results, in which C
T
( , where
R
T
is the total resistance) is the total resistance
coefcient, C
F
( , where R
F
is the friction
resistance) the friction resistance coefcient,
C
P
( , where R
P
is the pressure resistance)
the pressure resistance coefcient, and the relative
error is with respect to the corresponding EFD
(Experimental Fluid Dynamics) result, as dened
in
[12]
. It can be seen that the grid convergence rates
were fast for all the resistance components. The
spectral convergence shown in Figure 2 is due to
the present predominantly hexahedral trimmed
mesh, which is irregular and unstructured around
the hull surface, as demonstrated in Figure 3 with
a cell count of 3,258,319. The pressure resistance
converges slowly the most due to its heavy
dependence on the high pressure gradients at the
bow and the stern
[12]
. The relative error of the total
resistance based on the coarsest mesh with a cell
count of 364,305 can still be small (below 4%) even
though the relative error of the pressure resistance
(nearly 12%) is much higher. The reason could be
that the total resistance is dominated by the friction
resistance rather than the pressure resistance due
to the present streamlined hull form
[13]
. Here,
the CFD results based on the present nest mesh
with a cell count of 3,258,319 were taken as mesh
independent solutions for the present CFD
validation. The corresponding wave pattern is
shown in Figure 4. It can be seen that it agrees
well with the well-known Kelvin wave pattern
[2]
comprising of a distinct primary wave system and
secondary divergent and transverse wave systems.
RT
0.5
SU
2
=
RF
0.5
SU
2
=
RP
0.5
SU
2
=
CFD-based Resistance and Propulsion
Prediction for WInd Turbine Installation
Jackup Vessels and Semisubmersibles 51
52 KOMtech Technology Review 2011
Figure 5 shows the relationship between the relative
speed-up ratio and the number of cores. It is shown
that the relationship was almost linear ideally and
thus the high efciency of the present HPC
techniques based on multiple cores and multiple
machines can be illustrated. The ideally linear or
super speedup cannot be achieved since the present
multiple cores cannot be as fast as multiple CPUs
to allow for massive amounts of data through the
processor continuously
[7]
.
Resistance and Propulsion Prediction for a
Jackup Vessel
A wind turbine installation jack-up vessel with four
spudcan recesses was used to demonstrate the
accuracy of the present CFD-based resistance and
propulsion prediction. The bare hull form of this
wind turbine installation jack-up vessel is shown in
Figure 6 and its principal particulars are shown in
Table 3. To perform the CFD analysis, a trimmed
mesh composed predominantly of hexahedral cells
with trimmed cells next to the surface
[7]
was
generated as shown in Figure 7 and a dissipation
zone with extra active damping was pre-dened to
damp out the non-physical wave reection from the
articial boundaries at the far eld as shown in
Figure 8.
Table 2. Comparison of Resistance Components
Organization Code S/L
2
C
T
(*1000) C
F
(*1000) C
P
(*1000) | E | (%)
IIHR Experiment 0.1490 4.23 2.88 1.35 -
KOMtech STAR-CCM+ 0.1467 4.26 2.86 1.40 0.9
IIHR CFDSHIP 0.1550 4.36 3.06 1.30 3.1
INSEAN MGSHIP 0.1485 4.47 3.23 1.24 5.7
DERA CFX 0.1517 3.94 2.95 0.98 6.9
Figure 6. Geometry Model of the Bare Hull of the
Jackup Vessel with Four Spudcan Recesses
Table 3. Principal Particulars of the Wind Turbine
Installation Jackup Vessel
Particular Value
Length (m) 115.214
Breadth (m) 49.987
Depth (m) 9.754
Draught (m) 5.121
Thruster Model WFSD400-3000
Thruster Motor Power (kW) 3000
Figure 5. Nearly Ideal Linear Relationship between the
Speed-up Ratio and the Number of Cores
Using the Multiple Cores and Multiple
Machines HPC Techniques
As a result, Figure 9 displays a CFD predicted wave
pattern generated by the steadily advancing hull,
which consists of a primary wave system due to the
high pressure eld at the bow and stern and a
secondary divergent wave system due to the moving
high pressure points
[2]
. It can be seen that the
secondary stern transverse wave system was almost
damped out due to the thick boundary layer and
severe ow separation at the wake. This is a typical
phenomenon for ships with a relatively low Froude
number
[12,15]
. A comparison of the resistance
predictions using both the direct model test and the
present CFD simulation approach is numerically
shown in Table 4 and graphically shown in Figure
10. It can be seen that the present CFD solutions
are in good agreement with the model test results
[16]
by MARIN with sufcient engineering accuracy.
The CFD predicted resistance can even be almost
identical to the corresponding model test result for
Figure 7. Trimmed Mesh for Resistance and Propulsion
Prediction of the Barge
Figure 8. A Dissipative Zone for the Wave Reection
Damping Filter
Table 4. Numerical Comparison of the Calm Water Resistance Predictions
Vessel Speed (kn) Model Test (kN) CFD Simulation (kN) Relative Error
8 873 861 -1.36%
9 1136 1096 -3.49%
10 1397 1354 -3.11%
11 1722 1739 0.97%
12 2085 2069 -0.75%
13 2527 2492 -1.37%
14 3113 2891 -7.14%
Figure 10. Graphical Comparison of the Calm Water
Resistance Predictions
Figure 9. A Wave Pattern Consisting of the Primary
and Secondary Divergent Wave Systems
CFD-based Resistance and Propulsion
Prediction for WInd Turbine Installation
Jackup Vessels and Semisubmersibles 53
54 KOMtech Technology Review 2011
the barge with a speed of 12 knots. The maximum
error of 7.14% for the highest vessel speed of
14 knots may even be reduced if a ner mesh is
used. In this work, only one mesh was adopted to
deal with all these speed variations to save the
computational time, as shown in Figure 7. Table 5
shows a comparison of the propulsion prediction
results. It can be seen that the speed prediction given
by the present practical propulsion prediction
method matches well with the prediction provided
by MARINs model test method
[15]
. The high
accuracy of the present CFD simulation approach
in resistance and propulsion prediction demonstrates
the promising prospects of developing the numerical
(virtual) towing tank techniques for efcient
offshore and marine applications
[2-4]
and may even
promote wider acceptance of the CFD simulation
approach in offshore and marine engineering. The
decades-long dream
[4]
of replacing the traditional
towing tank with the numerical towing tank for
resistance and propulsion prediction may be nally
made to come true.
Resistance and Propulsion Prediction for
a Semisubmersible
A self-propelled wind turbine installation
semisubmersible with eight thrusters was used to
study the impact of the bottom openings to sea on
resistance and propulsion. The principal particulars
of this semisubmersible are listed in Table 6 and
its bare hull form with six bottom openings to
sea (sea chests) is shown in Figure 11.
The high accuracy of the present CFD simulation
approach in resistance and propulsion prediction
demonstrates the promising prospects of developing
the numerical (virtual) towing tank techniques for
efcient offshore and marine applications.
Table 5. Numerical Comparison of the Speed Predictions
Parameter Model Test CFD Simulation Relative Error
Predicted Vessel Speed (kn) 10.1 10.28 1.78%
Rotation Rate of Fore Thrusters (RPM) 178.2 175.9 -1.29%
Rotation Rate of Aft Thrusters (RPM) 156.3 160.3 2.56%
Figure 11. A Wind Turbine Installation Semisubmersible
with Six Bottom Openings to Sea
Table 6. Principal Particulars of the Wind Turbine
Installation Semisubmersible
Particular Value
Length of main deck (m) 93.88
Breadth of main deck (m) 54.27
Length of pontoons (m) 108.3
Breadth of pontoons (m) 15.2
Height of pontoons (m) 8.5
Static Draught (m) 8
Thruster Motor Power (kW) 2500
Number of Thrusters 8
Propeller Diameter (m) 3.4
Propeller Type Fixed Pitch
increase of up to 30% for streamlined drillships,
at a moderate forward speed in
[18]
due to the
signicant hull form difference. Several ways to
reduce the water oscillations in a moonpool are
available in the literature
[17]
such as using wedges,
cut-out parts, grid of aps, single ap, vertical
bulkhead, converging openings, hysteresis effects,
damping champers, anges. However, they are
either inefcient or expensive or both to apply to
the present semisubmersible during huge water
oscillations. In this work, a more effective method
was proposed.
The present CFD simulation approach was applied
to predict the resistance and propulsion of this
semisubmersible with and without openings and
the predicted signicant effect of the bottom
openings is shown in Table 7. It can be seen that
at a constant forward speed of 10 knots the calm
water resistance of the semisubmersible with
openings to sea would be 4.22 times the resistance
of the semisubmersible without openings. The
huge added resistance is due to the large water
oscillations inside the sea chests around the openings,
as shown in Figures 12 and 13. The large water
oscillations were excited by ow separation and
vortex shedding at the leading bottom edges of
the openings as shown in Figure 12, similar to the
water oscillations in the moonpool of a drillship in
calm water at forward speed
[17,18]
. The ow separation
of the semisubmersible can be clearly indentied
from the skin friction lines as shown in Figure 14.
The added resistance increase of about 322% for
the present non-streamlined semisubmersible with
openings is much larger than the added resistance
Table 7. Effect of the Bottom Openings on Resistance and Propulsion of the Semisubmersible
Parameter Without Openings With 6 Openings Ratio
Transit Sinkage (m) -0.64 2.63 -4.10
Transit Trim (deg) -1.51 -4.03 2.68
Resistance at 10 kn (kN) 1146 4834 4.22
Predicted Transit Speed (kn) 13.05 7.56 0.58
Predicted Propulsive Efciency 65.53% 53.96% 0.82
Figure 12. Velocity Field around the Openings of the
Semisubmersible Free to Sink and Trim
Figure 13. Water Oscillations around the
Semisubmersible Free to Sink and Trim
Figure 14. Skin Friction Lines of the Semisubmersible
Free to Sink and Trim
CFD-based Resistance and Propulsion
Prediction for WInd Turbine Installation
Jackup Vessels and Semisubmersibles 55
56 KOMtech Technology Review 2011
The present method to signicantly reduce the water
oscillations inside sea chests of the semisubmersible is
to use an active feedback control system to keep the
semisubmersible xed at zero sinkage and trim during
the transit. Since the water amount inside the ballast
tanks of a semisubmersible can be actively controlled,
such a feedback control system should be able to
implement. It is well known that the vessel resistance
increases signicantly with the transit sinkage and
trim
[14]
. Hence, the resistance of the semisubmersible
would be decreased if the semisubmersible is always
xed at zero trim and sinkage during the transit. As
shown in Figures 15 and 16, by using the present
method, the water oscillations were drastically
reduced since the relative levels of the sloshing water
in the middle and aft sea chests were signicantly
dropped, though water oscillations still exist. The
signicant effect of the transit sinkage and trim on
resistance and propulsion of the semisubmersible is
numerically shown in Table 8. It can be seen that at
a constant forward speed of 10 knots, the calm
water resistance of the semisubmersible xed at zero
sinkage and trim would only be 31% that of the
semisubmersible free to sink and trim. This is indeed
a huge resistance decrease and thus the corresponding
propulsive performance can be greatly improved, ie.,
the transit speed can be 58% faster and the propulsive
efciency 23% higher. As a result, the predicted vessel
speed of 11.94 knots and propulsive efciency of
66.42% for the semisubmersible with six openings
xed at zero sinkage and trim can be comparable
to the predicted vessel speed of 13.05 knots and
the propulsive efciency of 65.53% for the
semisubmersible without openings. Hence, the
huge impact of bottom openings to sea can almost
be eliminated and the present method to reduce
the water oscillations can be quite attractive for initial
practical design.
Figure 15. Velocity Field around the Semisubmersible
Fixed at Zero Sinkage and Trim
Figure 16. Water Oscillations around the
Semisubmersible Fixed at Zero Sinkage
and Trim
Table 8. Effect of the Transit Sinkage and Trim on Resistance and Propulsion of the Semisubmersible with
6 Bottom Openings to Sea
Fixed at Zero Sinkage
Parameter Free to Sink and Trim and Trim Ratio
Resistance at 10 kn (kN) 4834 1516 0.31
Predicted Transit Speed (kn) 7.56 11.94 1.58
Predicted Propulsive Efciency 53.96% 66.42% 1.23
present CFD simulation approach for resistance
and propulsion prediction can overcome the
traditional difculties of applying the CFD
techniques and even outperform the direct model
test approach. The present approach was nally
applied to investigate the signicant impact on
resistance and propulsion of bottom openings of
a wind turbine installation semisubmersible. It
was illustrated that the zero sinkage and trim
imposed by an active control system can alleviate
the water oscillations in the sea chests around
the bottom openings so that the calm water
resistance can be greatly reduced and the propulsion
performance can be signicantly improved. It was
suggested that by replacing the actual towing tanks
with the present CFD simulations for resistance
and propulsion prediction could be promising for
practical design purposes.
CONCLUSION
The CFD simulation approach has been further
developed to predict the resistance and propulsion
performance of the wind turbine installation jackup
vessels and semisubmersibles more accurately
and efciently by using state-of-the-art CFD
methods and HPC techniques. Due to the use of a
trimmer meshing technique, a wave reection
ltering scheme, a VOF free surface capturing
method, a propulsion prediction method and
multi-core and multi-machine HPC techniques,
the challenging problem of simulating a high
Reynolds number turbulent ow around a hull
was effectively resolved. The high accuracy of
the present CFD simulation approach can be
validated by the model test results of both a
streamlined surface ship and a non-streamlined
wind turbine installation jackup vessel. Hence, the
AUTHORS CONTACT ShengYin.Wang@keppelom.com
REFERENCES
[1] Resistance Test, ITTC QM 7.5-02-02-01, 23rd International Towing Tank Conference, 2002.
[2] V. Bertram, Practical Ship Hydrodynamics, Butterworth-Heinemann, Oxford, 2000.
[3] E. C. Tupper, Introduction to Naval Architecture, 3rd Edition, Butterworth Heinnemann, Oxford, 1996.
[4] J. Carlton, Marine Propellers and Propulsion, 2nd Edition, Butterworth Heinnemann, Oxford, 2007.
[5] Benchmark Database for CFD Validation for Resistance and Propulsion, ITTC QM 7.5-03-02-02, 22nd International Towing Tank Conference, 1999.
[6] B. Bucan, M. P. Buca, S. Ruzic, and D. Matic, Numerical Modeling of the Flow Around a Tanker Hull, in Marine Special Report, S. Ferguson, Editor,
CD-adapco, 2009.
[7] CD-adapco, User Guide: STAR-CCM+ Version 4.04.011, 2009.
[8] K. M. T. Kleefsman, Water Impact Loading on Offshore Structures - A Numerical Study, Dissertation, University of Groningen, Netherlands, 2005.
[9] C. W. Hirt, and B. D. Nichols, Volume of Fluid (VOF) Method for the Dynamics of Free Boundaries, Journal of Computational Physics, vol. 39, pp. 201225,
1981.
[10] A. Olivieri, F. Pistani, A. Avanzini, F. Stern, and R. Penna, Towing Tank Experiments of Resistance, Sinkage and Trim, Boundary Layer, Wake, and Free
Surface Flow Around a Naval Combatant INSEAN 2340 Model, Iowa Institute of Hydraulic Research, The University of Iowa, IIHR Report No. 421, 2001.
[11] F. R. Menter, Two Equation Eddy Viscosity Turbulence Models for Engineering Applications, AIAA Journal, vol. 32, no. 8, pp. 1598-1605, 1994.
[12] L. Larsson, F. Stern, and V. Bertram, Benchmarking of Computational Fluid Mechanics for Ship Flows: The Gothenburg 2000 Workshop, Journal of Ship
Research, vol. 47, no. 1, pp. 63-81, 2003.
[13] W. S. Hughes, J. A. Brighton, Schaums Outline of Theory and Problems of Fluid Dynamics, McGraw Hill, 1999.
[14] A. Olivieri, F. Pistani, A. Avanzini, F. Stern, and R. Penna, Towing Tank Experiments of Resistance, Sinkage and Trim, Boundary Layer, Wake, and Free
Surface Flow Around a Naval Combatant INSEAN 2340 Model, Iowa Institute of Hydraulic Research, The University of Iowa, IIHR Report No. 421, 2001.
[15] S. Das, O. P. Sha, and S. C. Misra, An Investigation into Resistance Characteristics of Rectangular Barges, IE (I) Journal, vol. 86, pp. 14-18, 2005.
[16] J. Dang and R. Kerkhof, Multipurpose Service Vessel; Calm Water Model Tests, Report No. 24470-1-DT, MARIN, The Netherlands, 2010.
[17] G. Gaillarde and A. Coteleer, Water Motion in Moonpools - Empirical and Numerical Approach, ATMA - Association Technique Maritime et Aeronautique,
Paris, France, 2004.
[18] R. V. Veer and H. J. Tholen, Added Resistance of Moonpools in Calm Water, Proceedings of the ASME 27th International Conference on Offshore
Mechanics and Arctic Engineering, OMAE2008, Estoril, Portugal, 2008.
CFD-based Resistance and Propulsion
Prediction for WInd Turbine Installation
Jackup Vessels and Semisubmersibles 57
58 KOMtech Technology Review 2011
No part of the materials published in this journal may be reproduced, stored in a retrieval system or transmitted in any form whatsoever without the prior written permission of KOMtech
Airgap and Wave Run-up Prediction
for Semisubmersibles Using
the CFD Simulation Approach 59
IN THIS PAPER, THE COMPUTATIONAL FLUID DYNAMICS (CFD) SIMULATION
APPROACH IS ADOPTED AS AN EFFECTIVE TOOL to predict the airgap
and wave run-up response of a semisubmersible in extreme sea waves. In
the present CFD simulation approach, the commercial CFD solver is used to
guarantee the solution accuracy while the offshore and marine engineering-related
issues are addressed by dening user eld functions for the solver. Furthermore,
High Performance Computing (HPC) techniques are utilised to speed up the
fully nonlinear CFD solution. A good compromise between accuracy and efciency
is thus achieved. It is further demonstrated that the deterministic design wave
method would under-predict the airgap and wave run-up response and the
stochastic wave method would be more suitable to predict the severe dynamic
response. The present CFD-based prediction that the semisubmersible in irregular
waves under the operating conditions with a one year storm return period would
experience extreme wave events with negative airgap can be generally consistent
with the available experimental results. It is suggested that the CFD simulation
approach is a more suitable tool than the traditional methods to predict the dynamic
response of large volume oating structures.
ACKNOWLEDGEMENT
This work was partially supported by Deepwater Technology Group (DTG). We
are grateful to Edwin Nah Hock Choon and Zhang Yanting in DTG for sharing
their insights and expertise. We would also like to express our appreciation to
Peter Ewing, director of CD-adapco Australia, for his invaluable support of our
CFD software tool, STAR-CCM+.
Airgap and Wave
Run-up Prediction
for Semisubmersibles
Using the CFD
Simulation Approach
WANG Shengyin, PhD, M. Eng, B. Eng
ZHENG Xiangyuan, PhD, M. Eng, B. Eng
60 KOMtech Technology Review 2011
INTRODUCTION
A semisubmersible (semi-sub, or semi), is a specialised
large volume oating structure
[1]
consisting of
ballasted pontoons located below the ocean surface
and connected by horizontal bracing members, an
operating deck located high above the sea level, and
structural columns to connect the pontoons and
the operating deck. Due to their distinct stability
and seakeeping characteristics, semisubmersibles
have been successfully applied as deepwater mobile
offshore drilling units, crane vessels, safety vessels,
offshore support vessels, and offshore production
platforms. The critical design decision for a
semisubmersible includes the initial airgap between
the still water surface and the underside of the
operating deck
[1-3]
, which can be determined by
the required minimum airgap in extreme design
conditions
[3]
. Since a semisubmersible may operate in
a wide variety of sea states including severe incident
waves with high wave crests, the incident waves
can be signicantly amplied
[1,2]
under the deck to
generate the undesirable deck overtopping and wave
slamming on the underside of the deck
[4]
. Hence,
accurate prediction of airgap and wave run-up is of
critical importance. Nevertheless, accurate airgap and
wave run-up prediction for the large volume oating
semisubmersibles is particularly challenging.
There are several engineering analysis tools available
for predicting airgap and wave run-up of large volume
oating structures such as semisubmersibles
[1-4]
.
Standard analysis tools are based on semi emperical
linear or second-order hydrodynamics, as well as
direct model tests
[1,2]
. Since the prediction of negative
airgap and run-up on the operating platform would
involve strongly nonlinear wave interactions around
the hull
[4]
, these standard engineering tools may
become either inaccurate or inefcient. Linear
diffraction theory-based tools are still popular for
airgap and wave run-up predictions because of their
simplicity and ease of use
[3-6]
. The wave amplication
and modication of the kinematics caused by the
structures may be readily estimated by linear radiation
and diffraction theory
[4]
. However, linear theory-
based tools would often underestimate the maximum
wave elevations as well as the minimum airgap
under extreme sea waves
[3-8]
. Hence, the risk of wave
slamming damage to platform equipment may be
increased due to the under-predictions. Second-
order diffraction-radiation theory may represent a
signicant improvement relative to linear theory,
producing more realistic spatial patterns of minimum
airgap and maximum wave run-up
[3]
. However, as
noted by Stansberg
[3]
, second-order nonlinear analysis
tools such as WAMIT
[9]
may not be sufcient to
model the extreme wave amplication under a
semisubmersible in random seas and that fully
nonlinear wave models are required to reproduce
the extreme wave crest elevations. The physical
model tests are usually essential for the initial airgap
design. Nevertheless, the direct model tests are
expensive and conversion of the model test results
into the prediction of a full scale ship can be empirically
based and practically arguable due to the difculty in
keeping the Reynolds similarity to address the actual
wave-induced motion in the high Reynolds number
turbulent ow region
[10]
. Fully nonlinear CFD
simulations may capture the highly nonlinear
phenomena such as jet-like run-up and wave slamming
on the underside of the deck. The CFD simulations
can be done repeatedly in design loops to provide
useful insights into the phenomenological behavior
where classical extrapolation techniques in the direct
model test approach are not applicable. Hence, the
CFD simulation approach may potentially be more
suitable than the direct model test approach for airgap
and wave run-up prediction. However, the CFD
simulation approach has not yet gained wide recognition
due to some technical barriers such as the coarse free
surface resolution and prohibitive computational time,
which may have been gradually overcome with the
rapid developments in computer technology and
numerical methods during the last decade
[11,12]
.
The objective of the present work is to apply
the CFD simulation approach for accurate and
efcient airgap and wave run-up prediction for
semisubmersibles in extreme sea waves using state-
of-the-art CFD methods and HPC techniques. The
HPC techniques were used to speed up the CFD
solution while the latest CFD methods were adopted
to improve the solution accuracy. The Volume Of
Fluid (VOF) method
[10,14]
was utilised to capture the
free surface between the air and water accurately so
that the major obstacle to applying CFD methods to
offshore and marine engineering
[11]
can be resolved.
Hence, the technical barriers for applying the
CFD simulation approach are overcome and more
accurate and efcient CFD-based airgap and wave
run-up prediction can be expected.
Airgap and Wave Run-up Prediction
for Semisubmersibles Using
the CFD Simulation Approach 61
CFD SIMULATION APPROACH FOR
AIRGAP AND WAVE RUN-UP PREDICTION
Denitions of Airgap and Run-up
According to the recommendation by DNV
[1]
, the
initial airgap is the difference in elevation between
the bottom of the deck and the mean water level
and the instantaneous airgap is dened by Eq 1.
a(x, y, t) = a
0
+ z(x, y, t) - (x, y, t) (1)
where a(x, y, t) is the instantaneous airgap, a
0
the
initial airgap, z(x, y, t) the vertical displacement
of the structure at the position(x, y) and (x, y, t)
the instantaneous surface elevation at the same
horizontal position. Negative airgap, a(x, y, t) < 0,
means that there is impact between the wave and
the structure. Parts of the structure that are near
the water surface are susceptible to forces caused
by wave slamming when the structural part is
being submerged. Wave Slamming is due to sudden
retardation of a volume of sea water, generating a
considerable force on the structure
[1]
.
The surface elevation includes global upwelling due
to diffraction of incoming waves with the structure
and local run-up in the form of jets and other
strongly nonlinear effects
[1]
. Similar to the airgap
denition, local run-up r(x, y, t) may be dened by
Eq 2.
r(x, y, t) = (x, y, t) - z(x, y, t) (2)
In the present work, the global co-ordinate system
was based on the Centre of Gravity (CG) of the
semisubmersible rather than the mean water level,
as shown in Figure 1. Furthermore, only the
instantaneous heave and pitch motions were
considered for the semisubmersibles in head waves.
Hence, the instantaneous airgap can be redened
by Eq 3. where w(x, y, t) is the instantaneous heave
motion, (x, y, t) the instantaneous pitch angle,
H the distance between the bottom of the deck
and the CG of the semisubmersible in still water.
Similarly, the local wave run-up can be redened
by Eq 4.
Wave Conditions
For design purposes, wave conditions can be
described either by deterministic design wave
methods or by stochastic methods
[1]
. The
deterministic regular waves characterised by
wavelength and corresponding wave period, wave
height and crest height can be obtained by statistical
methods and may be sufcient for quasistatic
response prediction. However, according to the
recommendation by DNV
[1]
, signicant dynamic
response caused by severe wave slamming or
signicant wave amplication requires stochastic
modeling of the sea surface and its kinematics by
time series. A real sea state can be best described
by a random wave model and specied by a wave
frequency spectrum with a given signicant wave
height, a representative frequency or peak period,
a mean propagation direction and a spreading
function. To simplify the analysis, only a linear
random wave model was adopted, which was a sum
of linear wave components with different amplitude,
frequency, direction and phases. The sea state was
assumed to be a stationary random process. Usually,
the period of stationarity can range from 30 minutes
to 10 hours, though three hours is a standard time
between registrations of sea states when measuring
waves
[1]
. In this work, the period of stationarity
was assumed to be ve minutes only to avoid
excessive CFD computation time and the JONSWAP
wave spectrum was adopted to describe the short-
term wave conditions for CFD simulations of wave
elevation and kinematics.
The existence of unexpectedly large and/or steep
waves, so called freak or rogue waves, has been
generally accepted. However, no consensus has been
concluded on the denition of a freak event or about
the probability in occurrence of freak waves
[1]
. The
denition suggested by Haver
[15]
with a threshold
value of 2.45 times the signicant wave height was
adopted. For comparison, the CFD-based airgap and
wave run-up prediction was performed for a deepwater
semisubmersible in the following scenarios: regular
waves, irregular waves and freak waves.
a(x, y, t) = w(x, y, t) + H cos (x, y, t) tan (x, y, t) (x H sin (x, y, t)) (x, y, t) (3)
r(x, y, t) = (x, y, t) w(x, y, t) H cos (x, y, t) + tan (x, y, t) (x H sin (x, y, t)) + H (4)
62 KOMtech Technology Review 2011
meshes for a turbulence model, the All y+ Wall
Treatment
[12]
was employed. To generate a high
quality volume mesh with an appropriate boundary
layer resolution, the trimmer meshing model together
with a prism layer mesher in
[12]
was utilised. The
resulting trimmed mesh was dominantly composed
of hexahedral cells as a good compromise to achieve
the computational efciency and accuracy. In order
to damp out the unphysical wave reection in the
vicinity of the articial boundaries at the far eld,
a dissipation zone with extra active damping effect
was pre-dened and the corresponding wave-
reection damping lter was fullled
[12]
, similar to
but more robust than the lter scheme approach to
achieve non-reecting boundary conditions in
[13]
.
To implement the present CFD simulation approach,
the next generation general-purpose control-volume
based commercial CFD software tool STAR-CCM+
v4.04 was utilised.
Efcient HPC Techniques
The parallel computing operating system Microsoft
HPC Server 2008 was used to manage a cluster of
Windows 2008 64 bit machines comprising of
two HP Proliant Z6000 servers with a total of four
quad-core processors (16 cores) and 128 GB RAM
on both enterprise and private networks with a
bandwidth of 1Gbps respectively. The multiple
cores and multiple machines HPC techniques
were combined to execute jobs consisting of
individual tasks to achieve even more powerful
HPC performance. It should be noted that in CFD,
it is highly desirable to compute solutions in parallel
Effective Implementation of the CFD
Simulation Approach
As recommended by DNV
[1]
, the CFD simulation
approach can be used to predict airgap and local
wave run-up of semisubmersibles in waves since it
allows for wave break-up and changing topology
of the sea water domain
[1]
. In the present work,
the CFD simulation approach was effectively
implemented for airgap and wave run-up prediction
for semisubmersibles in extreme sea waves.
In the CFD analysis, a sufciently large three
dimensional (3D) computational domain enclosing
the semisubmersible is needed. Inside the CFD
computational domain, both the air and water exist
and thus a multiphase ow model was adopted to
obtain the free surface between the air and water.
The uid ow with air and water was modeled
as a two-phase ow and the VOF method
[14]
was
used to capture the free surface, which was once a
major obstacle to applying viscous CFD methods to
offshore & marine engineering
[11,12]
. The incoming
waves were applied at the inow boundary of this
CFD domain. Due to the full-scale size in the
present CFD simulation, the unsteady VOF ow
is turbulent with a high Reynolds number. In the
present CFD simulation approach, the RANS
(Reynolds-averaged Navier-Stokes) equations
[12]
were adopted. Menters SST k-omega turbulence
model
[16]
, which blends a k-epsilon model in the
far-eld with Wilcoxs k-omega model near the
wall, was used to estimate the signicant turbulence
effects. In order to recover the behaviors of the wall
treatment in the limit of very ne or very coarse
Figure 1. Airgap Denition
z
H
W (x, y)
x
to solve problems faster and increase the total
amount of memory available for a given simulation
[12]
because of the computationally intensive work
required to acquire converged CFD solutions.
To apply the HPC techniques efciently, it is
important to scale the simulation size with the
number of cluster nodes being used
[12]
. The optimal
number of cluster nodes to use is dependent upon
the hardware of the cluster and complexity of the
problem and is difcult to state a priori. In this
work, the number of mesh cells placed on each
cluster node was around 100,000.
RESULTS AND DISCUSSION
Wave Run-up Prediction for a Box-shaped
Barge in Regular Waves
A box-shaped barge in the open sea under the
action of a plane wave
[17]
was chosen as a benchmark
example for the present CFD analysis. The principal
particulars of this barge are shown in Table 1
and its 3D solid geometry model is shown in
Figure 2. It was assumed that the barge had zero
forward velocity and the incident plane wave was a
regular sinusoidal wave. As shown in Figure 3, the
wave run-up was measured by the relative vertical
motion of the water particle along the barge surface
with respect to the body-tted coordinate system
with an origin at the CG of the barge
[12]
. The trimmed
volume mesh for the present CFD analysis is shown
in Figure 4, which is dominated by hexahedral cells
to improve the accuracy and efciency.
Table 1. Principal Particulars of the Box-shaped Barge
Particular Value
Length L (m) 200
Breadth B (m) 30
Draught T (m) 15
Wavelength (m) 300
Wave height (m) 20
Figure 2. Geometry Model of the Box-shaped Barge
Figure 3. Wave Run-up around the Box-shaped Barge
Figure 4. Trimmed Volume Mesh for CFD Simulation of
the Barge Motion in Waves
Complicated hydrodynamic problems of large oating
offshore structures can be solved by full-scale CFD simulations
with a good compromise between accuracy and efciency.
Airgap and Wave Run-up Prediction
for Semisubmersibles Using
the CFD Simulation Approach 63
64 KOMtech Technology Review 2011
predicted by the present CFD simulation approach
agrees quite well with the analytical solution of the
strip theory
[17]
. Hence, the accuracy of the present
CFD simulation approach as well as the strip theory
to predict the pitch motion less sensitive to the
local effects can be veried. It can also be noted
that the heave motion predictions can be quite
different. The reason can be that the linear wave
theory assumes a very small wave amplitude rather
than nite wave amplitude so that the wave surface
prole can be sinusoidal and the wave crest height
can be equal to the wave trough height
[1]
, as shown
in Figure 7. For the present regular waves with wave
amplitude of 10m, this assumption cannot hold
true. It should also be noted that the CFD predicted
response can never be exactly sinusoidal since the
CFD approach is fully nonlinear and the randomly
uctuating turbulence effect due to existence of
ow separation for the present barge in the turbulent
ow with a high Reynolds number. As shown in
Figure 8, signicant ow separation can be identied
by the convergence of the skin friction lines of
the barge hull surface onto a line or point.
Figure 5. A Comparison of Wave Run-ups of the Barge
in Regular Waves
Figure 6. Pitch Angle of the Barge in Regular Waves
Figure 7. Heave Motion of the Barge in Regular Waves
Figure 8. Instantaneous Skin Friction Lines of the
Barge in Regular Waves (t=300s)
Figure 5 displays a comparison of the run-ups
predicted by the strip theory
[17]
and the present
CFD simulation approach. The wave run-up
predicted by the strip theory
[17]
was represented
by the run-up at the bow, where the extreme wave
run-ups are located for the present head waves. The
run-ups predicted by CFD are the instantaneous
extreme run-ups of the whole barge. It can be seen
that the potential ow-based strip theory would
underestimate the maximum wave run-up by up
to 20%. This observation is quite consistent with
the general recognition
[3-8]
that the linear theory-
based analysis would often underestimate the
maximum wave elevations signicantly. Hence, the
risk of green water on deck for this case would be
under-predicted by the strip theory solution
[17]
.
According to the present CFD analysis, the freeboard
must be at least 15m in order to avoid water on
deck. Furthermore, it can also be seen that the
barge bottom would go out of the water since the
minimum wave run-up is less than -15m. The regular
wave-induced pitch and heave motions are shown in
Figures 6 and 7. It can be seen that the pitch motion
Airgap Prediction for a Semisubmersible
in Waves
A semisubmersible accommodation unit
[18]
was used
to demonstrate the airgap predictions for the
semisubmersible in regular waves, irregular waves
and even freak waves using the present CFD
simulation approach. The principal particulars of
this semisubmersible are listed in Table 2 and its
geometry model in the bare hull form is shown in
Figure 9. Only the semisubmersible in head waves
under the operation conditions with a one year
storm return period was considered due to its
relatively small initial airgap of 6.71m, which is
more susceptible to wave slamming damage,
according to the model test results in
[18]
. Figure 10
displays the trimmed volume mesh for CFD analysis
and Figure 11 shows a dissipation zone denition
for an active damping lter to achieve the non-
reecting boundary conditions.
Table 2. Principal Particulars of the Semisubmersible
in Operation
Particular Value
Length of pontoon (m) 93.88
Breadth of pontoon (m) 11.58
Height of pontoon (m) 6.1
Length of central column (m) 12.19
Breadth of central column (m) 9.14
Length of aft & fwd columns (m) 13.07
Breadth of aft & fwd columns (m) 9.14
Column height (m) 12.8
Breadth outside pontoons (m) 45.2
Bracing diameter (m) 1.8
Draught (m) 12.2
Weight displacement (tonnes) 17912
VCG (m) 12.34
Kxx (m) 17.176
Kyy (m) 29.103
Kzz (m) 28.164
Figure 9. Geometry Model of the Semisubmersible
Figure 11. Active Damping Filter for Non-reecting
Boundary Conditions
Figure 10. Trimmed Volume Mesh with a Cell Count
of 536,726
Airgap and Wave Run-up Prediction
for Semisubmersibles Using
the CFD Simulation Approach 65
66 KOMtech Technology Review 2011
for practical offshore and marine design using the
popular potential ow theory-based hydrodynamic
codes such as WAMIT and MORA to achieve both
the accuracy and efciency.
After being veried by the natural periods of the
semisubmersible, the present CFD simulation
approach was applied to airgap prediction for the
semisubmersible in regular, irregular and freak waves
scenarios. The ratio between the maximum wave
height and signicant wave height is 1.8 for irregular
waves and 2.45 for freak waves
[15]
. We designed the
random waves whose maximum height occurred
after the body motions were fully developed. Figure
14 shows the instantaneous minimum airgap of the
semisubmersible in regular waves with the same
wave height of 9.2m and two different wave-lengths
(147m and 105m). Only the rst-order regular
waves
[1]
were considered. According to the present
quasistatic response prediction, the instantaneous
minimum airgap can always be positive so that the
initial airgap design of the semisubmersible may
be regarded as sufciently safe. Furthermore, the
deepwater regular waves with a shorter wavelength
of 105m would be more severely amplied under
the deck. The reason can be that the water particles
would have a much higher velocity and acceleration
at the still water level due to a shorter wave period of
8.2s resulting from the shorter wavelength
[1]
.
Figure 12. CFD-based Heave Decay Test Result
Figure 13. CFD-based Pitch Decay Test Result
Table 3. Comparison of the Natural Periods and Damping Ratios
Particular CFD by KOMtech MORA by DTG Difference
Heave natural period (s) 15.9 15.9 0
Pitch natural period (s) 17.3 17.3 0
Heave damping ratio 0.2 0.25 -25%
Pitch damping ratio 0.27 0.25 7.4%
The CFD-based heave and pitch decay tests were
performed rst. The CFD analysis results are shown
in Figures 12 and 13 and the corresponding numerical
comparison of the natural periods and damping
ratios is shown in Table 3. It can be seen that the
natural heave and pitch periods predicted by the
present CFD simulation approach are identical to
those predicted by a potential ow-based program
package MORA
[18]
. Theoretically, the natural periods
are independent of the ow theory and viscous
damping. Hence, the accuracy of the present fully
nonlinear CFD simulation approach can be veried.
It can also been seen that there is a signicant
difference of 25% in the heave damping ratios.
Since the damping ratio of 0.25 for all the motion
modes used by DTG
[18]
for the program MORA
was empirically based and may be unrealistic,
it would be more accurate and reasonable to choose
the damping ratios predicted by the present CFD
simulation approach. It should be noted that the
heave motion would be more severely damped out
and the airgap prediction would become less risky
due to this higher heave damping ratio of 0.25
[18]
.
Since the present CFD simulation approach for
numerical decay tests is less time consuming due to
the rapid convergence of the CFD solutions for the
motion decay problems, it is recommended that
offshore engineers should rely on the CFD-based
ratios rather than empirically-based damping ratios
Figure 15 shows the instantaneous minimum airgap
of the semisubmersible in the irregular waves with a
maximum wave height of 9.2m and in the freak
waves with a maximum wave height of 12.5m.
It can be seen that in both cases the semisubmersible
was under the impact of negative airgap and the
impact would be more severe when the freak waves
were involved
[1]
. Hence, the stochastic wave methods
would be more suitable than the deterministic
design wave methods to predict the severe dynamic
response caused by the signicant wave amplication,
according to the practice recommended by DNV
[1]
.
The time history of the wave elevation at the bow of
the semisubmersible is shown in Figure 16. As
expected, the freak waves would cause a much larger
local wave elevation at the bow. At the solution time
of 148s, the semisubmersible in the freak waves had
a negative airgap. The corresponding wave elevation
contour is shown in Figure 17. It can be seen that
the negative airgap is due to the high wave elevation
at the bow. Figure 18 shows that the negative airgap
was caused by thin run-up jetting along the forward
columns of the semisubmersible via local wave
amplication around columns. Hence, the negative
airgap would not cause a global deck impact resulting
from the wave slamming damage and thus the
impact of the negative airgap would be minor. It
should be noted that similar wave events were
reported
[18,19]
based on the numerical predictions
given by the potential ow codes MORA and
WAMIT, and the physical model testing.
Figure 14. Instantaneous Minimum Airgap of the
Semisubmersible in the Regular Waves
Figure 15. Instantaneous Minimum Airgap of the
Semisubmersible in Irregular and
Freak Waves
Figure 16. Time History of the Wave Elevation at the
Bow of the Semisubmersible in the Irregular
and Freak Waves
Figure 17. Instantaneous Wave Elevation Contour of
the Semisubmersible in the Freak Waves
Figure 18. Instantaneous Negative Airgap of
the Semisubmersible
Airgap and Wave Run-up Prediction
for Semisubmersibles Using
the CFD Simulation Approach 67
68 KOMtech Technology Review 2011
numerical decay tests for a semisubmersible. Excellent
natural periods prediction may indicate its high
accuracy while the more realistic prediction of the
damping ratios may establish its suitability to replace
the empirically-based methods in hydrodynamic
applications. It was highlighted that the linear wave
theory-based deterministic design wave method
would under-predict the airgap and wave run-up
response signicantly. The stochastic wave method
based on a linear random wave model was shown to
be more suitable in predicting the severe dynamic
response caused by the signicant wave amplication.
The CFD prediction for the semisubmersible in
irregular waves under operation conditions may
experience extreme wave events with negative airgap
caused by thin wave run-up jetting along columns.
The prediction is generally consistent with the
available experimental results. It is suggested that the
CFD simulation approach be more widely used in
predicting airgap and wave run-up response of large
volume oating structures in extreme sea waves.
AUTHORS CONTACT ShengYin.Wang@keppelom.com
REFERENCES
[1] DNV-RP-C205, Det Norske Veritas (DNV), Environmental Conditions and Environmental Loads, Recommended Practice 2007.
[2] D. Kriebel, Wave Amplication and Air Gap Tests on a MOB Module, Proceedings of the 24th US-Japan Marine Facilities Panel, US-Japan Cooperative
Program in Natural Resources, Honolulu, HI, 2001.
[3] C. T. Stansberg et al., Extreme Wave Amplication and Impact Loads on Offshore Structures, OTC 17487, Offshore Technology Conference, 2005.
[4] F. G. Nielsen, Comparative Study on Airgap under Floating Platforms and Run-up along Platform Columns, Marine Structures, vol. 16, pp. 97-134, 2003.
[5] D. G. Danmeier et al., Validation of Wave Run-up Calculation Methods for a Gravity Based Structure, OMAE2008-57625, Proceedings of the ASME 27th
International Conference on Offshore Mechanics and Arctic Engineering, Estoril, Portugal, 2008.
[6] B. Sweetman, S. R. Winterstein, C. A. Cornell, Airgap Analysis of Floating Structures: First- and Sec-ond-order Transfer Functions form System
Identication, Applied Ocean Research, vol. 24, pp. 107-118, 2002.
[7] C. T. Stansber, and T. Kristiansen, Non-linear Scattering of Steep Surface Waves around Vertical Columns, Applied Ocean Research, vol. 27, pp. 65-80,
2006.
[8] J. N. Niedzwecki, and A. S. Duggal, Wave Runup and Forces on Cylinders in Regular and Random Waves, Journal of Waterway, Port, Coastal and Ocean
Engineering, vol. 118, pp. 651-634, 1992.
[9] C. H. Lee, and F. T. Korsmeyer, WAMIT User Manual, Department of Ocean Engineering, Massachusetts Institute of Technology, 1999.
[10] V. Bertram, Practical Ship Hydrodynamics, Butterworth-Heinemann, Oxford, 2000.
[11] B. Bucan, M. P. Buca, S. Ruzic, and D. Matic, Numerical Modeling of the Flow around a Tanker Hull, in Marine Special Report, S. Ferguson, Editor, CD-
adapco, 2009.
[12] User Guide: STAR-CCM+ Version 4.04.011, CD-adapco, 2009.
[13] K. M. T. Kleefsman, Water Impact Loading on Offshore Structures - A Numerical Study, Dissertation, University of Groningen, Netherlands, 2005.
[14] C. W. Hirt, and B. D. Nichols, Volume of Fluid (VOF) Method for the Dynamics of Free Boundaries, Journal of Computational Physics, vol. 39, pp. 201-225,
1981.
[15] S. Haver, Freak Waves: A Suggested Denition and Possible Consequences for Marine Structures, Rogue Waves 2004 Workshop, Brest, 2004.
[16] F. R. Menter, Two Equation Eddy Viscosity Turbulence Models for Engineering Applications, AIAA Journal, vol. 32, no. 8, pp. 1598-1605, 1994.
[17] O. M. Faltinsen, Sea Loads on Ship and Offshore Structures, Cambridge Ocean Technology Series, Cambridge University Press, 1990.
[18] KFELS SSAU3600-GOM-C42 (MSSAU-1) Motion and Airgap Analysis, Document D031, Keppel FELS, 2004.
[19] Modelling of Wave Run-up, Air Gap, and Hydrodynamic Loads of Semi-submersible Structures, 2nd Progress Report, Maritime Research Centre and
School of Civil and Environmental Engineering, Nanyang Technological University, 2008.
CONCLUSION
The CFD simulation approach was presented as an
effective tool to predict the airgap and wave run-up
response of semisubmersibles in extreme sea waves
condition. The commercial control volume-based
CFD solver was adopted so that the solution accuracy
can be guaranteed. The offshore and marine related
issues such as wave boundary conditions can be
effectively addressed by dening appropriate user
eld functions. The trimmer meshing technique can
be adopted to obtain a quality mesh while the lter
scheme approach may be used to achieve the non-
reecting boundary conditions. The VOF method
can be utilised to capture the free surface and to
overcome the major obstacle in applying CFD
methods to offshore and marine engineering. The
multiple cores and multiple machines HPC
techniques were used to signicantly speed up the
CFD solution. As a result, a good compromise
between accuracy and efciency was achieved. The
present CFD approach was further validated by the
DURING HARSH WEATHER, THE HEAVE MOTION OF VESSELS greatly
affect the normal operating, safe launching and recovery of subsea modules for
drillship and installation vessel. A heave compensation system can decouple the
movement between installation ship and subsea modules. An active heave
compensation system which includes a measurement unit, a controlling unit
and an electric winch is investigated in this paper. A mathematical model for
the subsea module landing operation was established to attain the efciency
of active heave compensation system. Based on the mathematical model,
a nonlinear controller for the active heave compensation system is designed
by using backstepping technology. The effectiveness of the active heave
compensation controller is veried by computer simulation.
Nonlinear Control of
An Active Heave
Compensation System
Daniel LIU Shuyong, PhD, B.Eng
Nonlinear Control of An Active
Heave Compensation System 69
No part of the materials published in this journal may be reproduced, stored in a retrieval system or transmitted in any form whatsoever without the prior written permission of KOMtech
70 KOMtech Technology Review 2010
INTRODUCTION
With increasing energy demand, offshore drilling
goes from shallow water to deep or ultra deep water
in search of more offshore oil elds. In recent years,
the market has seen that the offshore oil elds
have been developed with processing equipment
on the seabed
[1]
. These subsea equipment need
high operability for maintenance and repairs to
reduce malfunction risks. As drilling operation
goes to deeper waters, harsh weather can lead to an
expensive operation downtime for offshore drilling
and installation vessels. Also, the harsh weather
has a great threat to the safety of vessel, crew, and
the components to be handled. Under these severe
weather conditions, it is a challenge for offshore
crane handling and installation operations. In
such situations, the crane/winch operation may
result in the loss of subsea modules. Hence, the
involved crane/winch system must satisfy rigorous
requirements in terms of safety and efciency to
deal with such installations. This challenge can be
met by using heave compensation technology.
Heave compensation can be achieved by passive or
active methods. The passive compensation system is
achieved by the pneumohydraulic components like
cylinder, accumulator, and air pressure vessel (APV).
Passive devices most often store energy in one
direction and release the stored energy in the other
by using a moving piston and rod system, which
separates two dissimilar uid as shock absorbers.
This provides a natural centralizing tendency and
requires no external energy input relative to this
central point. Passive systems, do not rely on an
input signal and there is no real time control action.
On the other hand, active heave compensation
(AHC) systems, take motion reference unit (MRU)
input as the instantaneous state of the system,
and actively add or remove energy to bring the
measured and the desired states into a closer accord
by using closed loop control. Compared to passive
compensation system, AHC denitely provides
superior performance on the accuracy and stability.
In the AHC application, the crane/winch tip motion
is measured via the motion reference unit (MRU)
sensors, and then hydraulic/electric actuators are
controlled to pull in or pay out the cable to overcome
the crane/vessel motion such that the load remains
stationary. AHC technology has been investigated
for quite a few years. Several commercial products
have been offered on the market by TTS Marine,
National Oilwell Varco (NOV), Hydramarine
(MacGregor), and Bosch Roxroth etc. These AHC
systems can handle water depth up to 3000m and
loads of up to 500 tonnes.
Although research has been conducted continuously
on improving compensation accuracy, almost all
AHC solutions focus on compensating vessel motion
only
[2]
. Only a few researchers considered the
dynamics of lengthy rope which is critical for subsea
module landing operation. Neupert et al.
[3]
presented
an inverse based control strategy using the linearization
method, with the estimate and prediction of heave
motion using wave observer. The prediction method
did not need the ships property. Johansen et al.
[4]
investigated the case of the subsea module entering
the water surface and proceeding the splash zone, in
which the crane dynamics were taken into account,
and feedforward control strategy was employed and
evaluated. Korde
[5]
used the coupled mechanical
oscillators in drill-ships to achieve heave compensation
over the frequency band covered by irregular waves.
However, the complex dynamics introduced by the
long cable and subsea module are usually neglected.
Without considering these dynamics, it cannot
guarantee reliable and robust operation of heave
compensation system.
In this paper, we will rst introduce the detailed
dynamic model for subsea landing operation,
while considering the cable dynamics and subsea
module dynamics. Based on this model, a nonlinear
controller is formulated to compensate the heave
motion by using backstepping technology
[6]
.
Stability of the system is proved by Lyapunov's
theory
[7]
. The controller is then veried through
a modeling framework, which consists of various
perspectives in simulating an AHC solution, such
as wave generator, ship motion simulator and cable
dynamics. The actuator response delay has also
been included to demonstrate the efciency of the
nonlinear controller.
MATHEMATICAL MODELING FOR
SUBSEA MODULE LANDING OPERATION
To design the active heave compensation control
system, we need to investigate the operation mode
and mathematical modeling. For the ship loading
operation, it is relatively straightforward to design
the control system. For the subsea module landing
operation, the immersed weight of long length cable
often results in very high tension. Vessel motion, due
to wave action, introduces additional cyclic loads
which can, and often do cause, cable deterioration
due to exure fatigue.
Kinks can occur in the cable during zero load
condition as shown in Figure 1. The zero load
condition has also been called slack cable condition.
This slack cable condition may result in a subsequent
snap load. When lowering is stopped and the payload
impacts the cable, it often resulted in the loss of
electrical signal due to short or broken circuits. In
such scenarios, complete failure or breakage of the
cable and total loss of the subsea module package
may happen.
While hanging free from the ship, the tension in the
cable is the sum of the static load, which is due to
the cable and instrument immersed weight, and the
dynamic load which is due to cable and attached
equipment inertia and hydrodynamic resistance.
Static and dynamic effects occur simultaneously
most of the time.
In the following section, we will describe the
dynamic model of the cable tension and establish
the dynamic equation of the load motion. The
nonlinear controller will be designed for the active
heave compensation system for subsea module
landing.
Mathematical model of cable/payload system
In order to develop a heave compensator system
for subsea module landing operation, mathematical
model of a normal winch system in an uncompensated
mode is described in this section. The model is
mainly for the cable/payload system. The formulae
and certain simple analytical methods will be
presented which permit a reasonable prediction
of both the static and dynamic cable loads. The
dynamic equation for load motion and snap loads
will be used to design the nonlinear controller for
the heave compensation function.
All forces considered for steady state peak tensions
When hauling and lowering operation is carried
out during rough sea environment, the tension is
time variant, its inertia and drag forces must be
considered. Under these conditions, it is practical
to assume a zero hauling speed (winch secured) to
nd the tension in the cable at the shipboard end.
Assuming the travel path of the payload and the cable
to be vertical (or nearly so), the dynamic tension
T(s,t) at the head sheave can be evaluated with the
help of Morissons equation in one direction
[8, 9]
:
Equation (1) implicitly stipulates that cable and
payload rigidly follow the head sheave motion. In
other words, it treats the cable as a rigid bar. Despite
this oversimplication, equation (1) can be used
to calculate the maxima of expected cable tension
other then snap load. Once these calculations have
been performed for a specic vessel, the results
obtained using this technique should be condensed
and presented for ease of readability.
Figure 1. Kinks due to the zero load condition
T(s, t) = Wp + WLs + 1/2 (Cc Dcs + CpAp)V|V| + (m + m1)
dV
dt
Wp = immersed weight of the payload (lbs)
WL = immersed weight of the cable per unit (lbs/ft)
Cc = cable longitudinal drag coeffcient
Dc = diameter of cable (ft)
Cp = payload normal drag coeffcient
Ap = payload normal cross section (ft
2
)
V = constant cable speed (ft/sec), hauling being
positive and lowering being negative
|V| = absolute value of V
s = length of cable paid out
m = mass of the cable and payload
m1 = added mass of the cable and payload
= the vertical acceleration of the head sheave at
time t
dV
dt
(1)
Where:
Nonlinear Control of An Active
Heave Compensation System 71
Dynamic equation of the load motion and snap
loads
The payload free body diagram is shown in Figure
2. The schematic diagram of cable/payload spring
mass model is shown in Figure 3.
72 KOMtech Technology Review 2010
Figure 2. Schematic diagrams for free payload body
Figure 3. Schematic diagram for cable/payload spring
mass model
A simple spring mass model (see Figure 3) can be
used to predict the occurrence of snap loads and
compute the ensuing cable tensions. In this model,
the following assumptions were made:
The motion of the payload is entirely vertical
(one degree of freedom system).
The mass of the cable is assumed to be a small
fraction of the equipment mass. This would be
the case for rather short lengths of cable
(hundreds of meters instead of thousands),
or if the cable is light (Kevlar line for example),
or if the payload entrains a lot of water.
The cable acts as a linear spring, the tension
T being directly proportional to the cable
elongation L i.e.
T = kL (2)
In the elastic range of cable elongation the spring
constant k is given by:
k =
EA
(3)
L
D = 1/2 C
p
A
p
V|V|
Where E is the cable modulus of elasticity (psi),
A is the cable metallic area (in
2
) and L is the cable
un-stretched length (ft). Then k is expressed in lbs/ft.
Equipment drag is non linear and in the form of:
(4)
Equipment drag is assumed much larger than
the cable drag
The vertical displacement of the cable upper
end can be described by an explicit function of
time (such as a sinusoid).
Applying Newtons law to the payload mass mvp
yields the following equation of motion:
(5)
Where
k = cable spring constant (lb/ft)
X1 = displacement of cable upper end (ft)
X2 = displacement of cable lower end (ft)
Ls = cable elongation under pure stative loading (ft)
Wp = immersed weight of payload (lbs)
mvp = virtual mass of payload (ft/s)
X2 = immersed speed of payload (ft/s)
X2 = immersed acceleration of payload (ft/s
2
)
k(X1 X2 + Ls) Wp 1/2CpApX2|X2| = (mvpX2)
Noting that kL
s
= W
p
, this equation of motion
reduces to:
(6)
The instantaneous cable tension at the payload end
is given by:
T = Wp
+
k(X1 X2 ) (7)
The motion of the payload mass is governed by
equation (6) as long as T>0.
If T, as given by equation (7) equals zero, then the
payload is no longer pulled by the cable and a new
equation of motion will prevail. Applying Newtons
law to the payload in free ight yields:
(8)
The system can be assumed at rest initially. At time
t = 0, the upper end of cable starts moving upwards.
The motion of payload is calculated by integrating
equation (6) using numerical integration method.
We will illustrate the payload motion without active
heave compensation control. We will consider the
response of a particular payload/cable system with
the following characteristics:
k(X1 X2 ) 1/2CpApX2|X2| = (mvpX2)
Wp
1/2CpApX2|X2| = mvp
Cable characteristics
Type 319 wire rope
Size 0.375 inch
Length 3000 ft
Immersed weight 30000.191=573 lbs
Modulus of elasticity 18,000,000 psi
Metallic area 0.1 sq. in
Strength 14,800 lbs
Payload characteristics
Type heavy instrument package
Weight in air 4200 lbs
Weight in water 2200 lbs
displaced
Added mass 12 slugs
Normal area 3.14 sq. ft
Drag coefcient 1.0
Figure 4. Cable lower end displacement and tension as
function of cable upper end displacement
X
1
= 7 sin (
2
t
) (9)
4
The vertical displacement of cable upper end is
assumed to be given by:
The displacement is 7ft, and period is 4 seconds.
Using these parameters, we can generate gure 4
using MATLAB simulink:
It can be seen that the cable lower end heave
displacement is even bigger than the upper end
due to the hydrodynamics. We can also meet the
zero tension situations at cable lower end. The
mean value of the cable tension at cable lower end
is around 2200lbs, which is the payload weight in
water.
NONLINEAR CONTROLLER DESIGN OF
ACTIVE HEAVE COMPENSATION SYSTEM
Based on the dynamic model of the payload motion
equation, we can design a nonlinear controller to
make the payload motionless. The whole system
including the controller can be expressed as:
(10)
Xship XCon X1 is the displacement of cable upper end
as shown in equation (6). So with the Xship signal
from MRU and X1 from winch encoder, we can
control the winch to pull in and pay out the cable
to keep the payload movement X2 stationary. The
X2 can be measured by the attached heave sensor on
the payload.
k(Xship XCon X2) 1/2CpApX2|X2| = mvpX2
Nonlinear Control of An Active
Heave Compensation System 73
74 KOMtech Technology Review 2010
V(z
1
) =
1
z
2
(12)
2
1
V(z
1
) = z1 ( X2d X2) (13)
= z
1
z
2
a
1
z
2
1
V(z
1
) = z1 ( X2d z2 a1 z1 X2d) (15)
Here we can choose the differential of virtual control
input as
As before, the time differential of equation (16)
becomes:
According to Lyapunov theory, equation (18) is a
negative denite function, which means that the
winch system is stable and the tracking error will
converge to zero exponentially with the controller
chosen in equation (17).
2
1
2
2
V(z
1
z
2
) =
1
z
2
+
1
z
2
(16)
From the previous denition of the virtual control
input X2, we can build the second Lyapunov function
for z1 and z2 . The expression, z2 = X2 a1 z1 X2d is the
control variable related to the real control output in
equation (10).
The stabilization of z1 can be obtained by introducing
a virtual control input X2
And the equation (13) becomes:
Figure 5. Control diagram for subsea module landing operation
X
2
= z1 a2 z2 + a2 z1
+ X
2d
(17)
= mvp (k(Xship XControl X2) 1/2CpApX2|X2| )
-1
V(z
1
V(z
1,
z
2
) = a1 z
2
a2 z
2
(18)
1 2
X2 = z2 + 1 = z2 + a1 z1 + X2d (14)
X2d = X2d = X2d = 0
Nonlinear backstepping controller design
Due to the nature of nonlinear dynamics of payload,
the traditional PID controller could not be used for
the active heave compensation operation to guarantee
the stability of the whole system. The linearization of
nonlinear equations and designing a PID controller
will degrade the performance of the whole system.
The global stability of the whole system also cannot
be guaranteed by using PID controller. Next, we
will introduce the nonlinear backstepping controller
design to tackle this problem to keep the payload
stationary. Currently the nonlinear backstepping
controller is a very popular tool to handle many
critical nonlinear dynamic systems. It has often
been used for automatic navigation system, such
as trajectory tracking, path following system and
dynamic positioning (DP) system.
Using the backstepping approach, one can synthesize
the control law thus forcing the system to follow
the desired trajectory. First, we will consider the
tracking error. In our application, the control target
is . Although our control trajectory
is stationary, it can be easily changed to time varying
trajectory, such as lowering the payload at a constant
speed towards the seabed. The tracking error in our
application is:
z1 = X2d X2 (11)
Based on Lyapunov theorem, we choose the following
Lyapunov function candidate:
The time differential equation of (12) is:
. . .
Figure 6. Simulation results of tracking error of payload
displacement with nonlinear backstepping
controller
Figure 7. Simulation results of ship heave displacement
and winch output displacement
We will nd that the tracking error converge to zero
exponentially from Figure 6. The controller output
is trying to follow the ship heave displacement as
shown in Figure 7. This nonlinear controller has
no requirement for the initial condition of payload.
To highlight this characteristic, we will illustrate
another simulation result of a simple controller, in
which the winch control output would try to follow
the same signal output as the heave motion sensor
on the ship. It has the same initial condition as the
previous output with nonlinear controller.
Using the backstepping approach, one can synthesize
the control law thus forcing the system to follow the
desired trajectory.
X2d = X2d = X2d = 0
. . .
In our application, the control target is ,so the controller law will be simplied to the
following equation:
(19)
a1 and a2 are the positive controller gains which determine the convergence speed of the tracking error.
The simulation results for this nonlinear controller will be provided and compared with traditional controller
in the next section. The sensor measuring delay issue will also be considered.
XControl = Xship X2 k
-1
1/2CpApX2|X2| + k
-1
mvp( X2 + a2 X2 + a1 a2X2 + a1 X2)
SIMULATION RESULTS FOR ACTIVE
HEAVE COMPENSATION WINCH
Using the parameters in previous section, we can get
the following simulation results shown in Figure 5
which illustrate the control diagram for the subsea
module landing operation.
In our application, the initial state of the payload
system is . The ship heave movement
is described with eight seconds period and eight
feet height.
X2 = 5, X2 = 1
.
Nonlinear Control of An Active
Heave Compensation System 75
76 KOMtech Technology Review 2010
CONCLUSION
In this paper, we have investigated the active heave
compensation crane/winch application. We have
analyzed the full detailed nonlinear mathematical
model of subsea module landing operation. Based on
the nonlinear model, we have designed a nonlinear
backstepping controller to handle the nonlinear
dynamics. Simulation results have also been presented
taking into consideration of the actuator response
delay. This nonlinear backstepping controller result
has also been compared with a simple controller
results. It was shown very clearly that the simple
controller could not handle the typical actuator delay.
The nonlinear backstepping controller can achieve a
better control result.
Figure 8. Simulation results of tracking error of payload
displacement with simple controller
Figure 9. Simulation results of tracking error of payload
displacement with simple controller including
actuator response delay
Figure 10. Simulation results of tracking error of payload
displacement with nonlinear backstepping
controller including actuator response delay
The result of the tracking error of the payload
with simple controller is shown in the Figure 8.
Although we are trying to compensate the ship
heave motion, we can nd that the convergence
of the tracking error is not behaving like the
nonlinear backstepping controller. This is due to
the initial condition of the payload. The initial
moment of the payload makes it move but the
movement could calm down due to the damping
element in the dynamic model. From Figure 8,
it is also obvious that the tension of the lower
cable end varies greatly in the beginning phase
and converges to 2200lbs in the end. In this
simulation, we did not consider the actuator
response delay.
By adding a 0.2 second actuator response delay
in the simulation, we can nd that the tracking
error of payload displacement will never converge
to zero from Figure 9. The 0.2 second actuator
response delay makes the tracking error magnitude
closer to the ship heave movement. This is due to
the introduction of exciting forces by the actuator
control error into the system.
The control efciency of the nonlinear backstepping
controller can be checked by considering actuator
response delay. From Figure 10, we can nd that
the control accuracy is very good compared with
using the simple controller.
AUTHORS CONTACT Shuyong.Liu@keppelom.com
REFERENCES
[1] S. Sagatun, Active control of underwater installation, IEEE Tran. On Control Systems Technology, 10: 5 (2002) pp. 743-748.
[2] P. Heney, (2008) Fluid power negates the seas pitch-and-roll, http://www.hydraulicspneumatics.com/200/Issue/Article/False/79782/Issue ,
Accessed 3rd December 2010
[3] J. Neupert, T. Mahl, B. Haessig, O. Sawodny, and K. Schneider,A heave compensation approach for offshore cranes, The 2008 American Control
Conference, Seattle, Washington, USA, 2008.
[4] Johansen, T.A., Fossen, T.I., Sagatun, S.I., Nielsen, F.G., 2003. Wave synchronizing crane control during water entry in offshore moonpool operations
experimental results, IEEE Journal of Oceanic Engineering, 28:4 (2003) pp. 720-728.
[5] U. A. Korde, Active heave compensation on drill-ships in deep water, Ocean Engineering, 25:7 (1998) pp. 541-561.
[6] Krstic, M., Kanellakopoulos, I., Kokotovic, P.V. Nonlinear and Adaptive Control Design. Wiley, New York, 1995.
[7] Slotine, J.J.E., and Li, W., Applied Nonlinear Control, Prentice-Hall,1991.
[8] J. Newman, Marine Hydrodynamics. Cambridge,MA: MIT Press, 1977.
[9] DNV-RP-H103 Modelling and Analysis of Marine Operations. April 2010.
Nonlinear Control of An Active
Heave Compensation System 77
78 KOMtech Technology Review 2011
Ice-Induced Vibrations of
Slender Offshore Structures 79
No part of the materials published in this journal may be reproduced, stored in a retrieval system or transmitted in any form whatsoever without the prior written permission of KOMtech
SLENDER OFFSHORE PLATFORMS IN THE ARCTIC AND SUB-ARCTIC ARE
PRONE TO ICE-INDUCED VIBRATIONS. When the ice pushes past the structure,
the ice fracture and clearing mechanism can set the structure into oscillation, in a
similar way to the process of vortex shedding due to uid ows that can afict
structures in the sea. The ice-induced vibrations can lead to fatigue damage,
damage to facilities, and can cause people on the platform to feel so uncomfortable
that their productivity is reduced. Recently available data suggests that a prediction
of when ice-induced vibrations will occur may be possible. In addition, it is possible
to design structures in such a way as to reduce or eliminate the failure mechanisms
that cause the vibrations to occur. A review of the current trends and challenges
to understand the mechanism of ice-induced vibrations of slender offshore
structures is presented in this paper.
ACKNOWLEDGEMENT
The Authors thank Dalian University of Technology for making available data and
photographic materials.
Ice-Induced Vibrations
of Slender Offshore
Structures
Andrew PALMER, FRS, FREng, FICE, PhD, DSc, Keppel Professor
Cynthia WANG, PhD, MSc, BSc
Michael PERRY, PhD, CEng
80 KOMtech Technology Review 2011
Figure 1. A Bohai Sea Jacket
The ability to understand ice-induced vibrations and
when they may occur is essential, as we move on to
further explore the opportunities that exist in the
ice-covered regions of the world.
INTRODUCTION
Marine structures can be damaged by vibrations
induced by the ow of water past them, and those
vibrations have been the subject of much research,
started many years ago by Strouhal and von Krmn.
These vortex induced vibrations are reasonably
well understood. Marine structures in ice-covered
waters can, in addition, be damaged by vibrations
created by the ow of ice and that is much less well
understood.
Structures in ice are not particularly new. The
offshore petroleum industry built several production
platforms in the Cook Inlet on the south coast
of Alaska in the 1960s, where the tidal currents
are extremely high. Some of them suffered from
vibrations. Many lighthouses were built earlier in the
twentieth century in the Gulf of Bothnia between
Finland and Sweden. Some of the lighthouses were
rather slender and exible. At least one lighthouse
was destroyed by ice-induced vibration.
The same problem occurred in offshore platforms in
the Bohai Gulf in China. Figure 1 shows a Bohai
Sea jacket. The structures are conventional jackets,
and the winter ice is not particularly thick, typically
0.25m thick and without the thick pressure ridges
that are found north of Alaska and in the northern
Caspian Sea. Heavy vibrations occurred, and were
large enough to cause fatigue fractures in topsides
piping, and at the same time were severely disturbing
to people working on the platforms, to the extent
that their sleep was interrupted and their productivity
reduced. A limited number of full-scale observations
have been made, though it has not always been
possible to measure and record critical parameters
such as the ice thickness and the ice velocity. Bohai
Gulf experience has recently become more broadly
available. Interest in these developments resulted in a
successful workshop on ice vibrations at the National
University of Singapore in January 2011, and some
of these results are also included in this paper.
The ability to understand ice-induced vibrations
and when they may occur is essential, as we move
on to further explore the opportunities that exist in
the ice-covered regions of the world. In this paper,
the mechanics behind ice-induced vibrations,
testing to determine the extent of vibrations and the
possibility of designing structures with vibrations in
mind are examined.
THEORETICAL BACKGROUND
Ice-induced vibrations arise from the dynamic
nature of ice interaction with a structure. As an ice
oe passes a structure, it rst breaks on the edge that
is in contact with the structure. The broken pieces
of ice then clear away, leaving a gap of open water,
before the next intact ice edge progresses towards the
edge of the structure and fails against it. The repeated
process of ice failing and clearing induces dynamic
force on the structure. The crushing mechanism is
Ice-Induced Vibrations of
Slender Offshore Structures 81
Figure 2. A simple mass-spring-damper system
Figure 3. Dynamic Amplication Factor for a case of
10% damping
not fully understood, and is the subject of much
controversy, only incompletely resolved by the ISO
19906 standard
[1]
that has recently been released.
The mechanism is dominated by repeated fracture.
That takes it some way beyond conventional
fracture mechanics, which is principally concerned
with the initiation and growth of a single crack,
rather than with the myriad cracks that occur when
ice crushes against a structure. Fracture mechanics
does, however, account for the observed size effect
in which the force per unit area at which the ice
breaks is not independent of area (as it would be
if the ice could be idealised as a perfectly-plastic
material), but instead decreases with increasing area.
This effect has profound implications
[2-8]
.
Quantifying dynamic ice-structure interaction is
a complex task that requires input from both the
ice behaviour as well as the structural response.
Firstly, the failure of ice against the structure, while
following certain statistical probabilities, is not easy
to predict. In addition, due to the magnitude of
the structural motions in relation to the ice failure
lengths, the structural response itself can affect the
ice failure process, meaning that not only does the
response depend on the force, but the force in turn
can depend on the response.
Theoretical attempts have been made to simplify
the coupled problem by treating the ice force and
the structural response separately. This simplied
approach has been employed in other elds such
as hydrodynamics. In addition, the system may be
modelled using a mass-spring-damper, representing
the fundamental mode of vibration, as shown in
Figure 2.
The equation of motion for such a system can be
represented using equation 1, where m is the mass of
the system, c is the damping, k is the spring stiffness
and F is the time varying force. The response
of the system is given by its displacement, x, and
the time derivatives of displacement, velocity and
acceleration which are denoted in equation 1 using
a dot and double dot respectively.
The natural frequency of such a system is computed
as
n
=(k/m) and the damping ratio as =c/(2m
n
).
The extent to which the structure responds to a
given input force depends on the magnitude of the
force, the frequency content of the force compared
to the structural frequency and the magnitude of
the damping. For a given input force, the structural
response can therefore be determined. For example,
if the force is idealised in the form F=Asin(
f
t ),
then the maximum dynamic response can be simply
computed in terms of the static response, F/k, and a
dynamic amplication factor (DAF) depending on
the ratio of the frequency of the force to the natural
frequency as well as on the damping level. An
example of the DAF for a damping level of 10% is
shown in Figure 3. This approach, while providing
a good feel for the problem, is however overly
simplied for a realistic ice-structure interaction
scenario where the force is not sinusoidal and can in
fact also be a function of the response.
mx + cx + kx = F
. . .
(1)
82 KOMtech Technology Review 2011
Observations suggest that the interaction force
is dominated by ice crushing failure and that the
resulting interaction force can be classied into
several distinct modes. ISO 19906
[1]
provides the
classication of time varying actions due to ice
crushing. These are:
1. Intermittent ice crushing, occurring at low
ice speed
2. Continuous brittle crushing, occurring at
high ice speed (typically greater than 100mm/s)
3. Frequency lock-in occurring at intermediate ice
speed (typically between 40mm/s to 100mm/s)
Note that these modes are detailed in terms of ice
failures in the next section on empirical work.
Examples of such theoretical solutions for time-
varying ice loads are presented by Krn & Turunen
[9]
and Eranti
[10]
where ice is represented by non-linear
springs that depict the assumed three-layer failure
zones in the ice sheet. However, these failure zones
are difcult to model. The difculty of modelling
the failure zones in ice will be discussed in the next
section.
Among the time varying ice actions, the frequency
lock-in or ice-induced vibration poses the greatest
challenge for slender offshore structures. This
condition arises from the phenomenon where the
time variation of the ice force is close to a natural
frequency of the structure at the waterline, resulting
in amplied responses such as those suggested by
the peak of the DAF in Figure 3. The steady state
vibrations induced by this phenomenon have been
known to cause fatigue damage to the structure.
EMPIRICAL WORK
It is possible to reproduce the same vibrations
discussed above in the laboratory, by pushing sheets
of oating ice against exible model platforms.
There have been many ice model tests carried out;
some using sea ice, some using fresh-water ice, and
some using ice with properties modied by dopants,
which are supposed to better represent sea ice on a
model scale. Despite this large number of tests, there
is ongoing controversy about how best to model sea
ice: that is a parallel subject of research
[11]
which we
will not address in detail here.
Because the mechanics of ice-induced vibration are
not well understood, it is unclear how to transfer
information between the model scale and the full
scale. If one researcher observes vibrations in a
model test in 10mm thick doped ice moving at
80mm/s, and another researcher observes vibrations
of a full-scale offshore structure in 500mm thick
sea ice moving at 10mm/s, is the rst test in some
sense a model of the second, or does it represent
something completely different? In particular, can
it be assumed that vibrations will always occur
when the ice velocity reaches 80mm/s or 10mm/s,
whatever the size of the structure and whatever the
thickness of the ice? If a third researcher observes
oscillations of another full-scale structure instead
to start at 100mm/s, is something different
happening? Model tests are far cheaper and easier
to accomplish than measurements on full-scale
structures, but they are of no use at all unless the
scaling laws are understood.
In other branches of mechanics, questions of this
kind are resolved by the powerful techniques of
dimensional analysis and similarity
[12]
, expressed in
the Vaschy-Buckingham Pi theorem and succinctly
summarised in a famous statement by Richardson;
...phenomena go their way independently of
the units whereby we measure them
In ship hydrodynamics, for instance, Froude
showed that wave-making resistance is correctly
modelled if Froude number U/(g) is made the
same in a model as in the prototype ship; U is
the ship velocity, g the gravitational acceleration
and l a length that characterises the ship. If
we model a ship 150m long sailing at 10m/s
by a geometrically similar model 3m long in a
towing tank, the model ought to be towed at
10/(150/3) = 1.41m/s. We can then be sure that
a non-dimensional resistance will be the same
in the prototype as in the model and we can use
measurements of the resistance of the model to
condently predict the resistance of the full-
scale ship. That technique is routinely applied in
offshore engineering. In ice mechanics, however,
the question is still open. That limits the usefulness
Ice-Induced Vibrations of
Slender Offshore Structures 83
A loose analogy may be made with vibrations
induced by vortex shedding in water or wind.
of model tests, and makes it uncertain how model
tests can be applied, or indeed if they can be
applied at all.
A loose analogy may be made with vibrations
induced by vortex shedding in water or wind.
It is known that the parameter that primarily
determines whether or not there will be vibrations,
what kind of vibrations and how large they will
be, is a dimensionless group called the reduced
velocity, U/NL, where U is the velocity of the
uid, N the natural frequency in Hertz and
L is a length that characterizes the system.
In oscillations of a vertical riser coming up
through the water surface, for instance, U is taken
as the water speed and L as the diameter of the riser.
Intuition suggests that we might apply the same
idea to ice-induced vibrations. Identifying the ice
speed and the natural frequency is straightforward.
It is not so clear what the appropriate choice of
length L ought to be. When ice is driven against
a structure, the ice breaks up into fragments, but
in a highly irregular way, so that there are some
large fragments and many small ones. (The reader
can demonstrate this by modelling a structure by
an upturned teacup resting on a table, and pushing
across the table against the teacup a sheet of brittle
material, such as an ice-cream wafer or a brittle
Swedish ryebread). Ideally, we would be able to
predict the lengths of the fragments, and to take L
as the characteristic length we need, but we lack the
knowledge to predict the length. Instead, we might
choose L as something better dened, for example
the ice thickness or the breadth of contact between
the ice and the structure.
It was not obvious whether the ice thickness or
the contact breadth would be the better choice,
and so both have been tried. Working with
researchers from Dalian University of Technology
in north China
[14]
, 11 data sets were brought
together, six from model tests and ve from
full-scale observations (two from the Bohai Gulf,
and one each from the Beaufort Sea, the Gulf of
Bothnia and the Athabaska River in Canada)
[12,13]
.
The structure diameters range from 50mm to 89m,
and the ice thickness from a 24mm to 5m. One
of the reasons for the usefulness of dimensional
analysis is that it can make sense of data from hugely
different scales.
A difculty is that there is no uniformity in
reporting; what one researcher calls a sawtooth
mode of vibration another calls intermittent
crushing and a third calls quasi-static. The divisions
between different modes are not completely clear-
cut, but broadly speaking, can be divided into four
modes, sketched schematically in Figure 4. If the ice
is moving very slowly, it creeps around a structure,
rather like ice owing down a valley glacier in
Switzerland, and the force is nearly constant. If
the ice is moving slightly faster, the force slowly
builds up, and then the ice breaks, the force falls
back and remains at a low level for some time, and
then the cycle repeats. As a result of this, the peak
load from one cycle is not affected by the dynamics
from the previous cycle and the load is therefore
referred to as quasi-static. If the ice is moving faster,
within a certain range, the structure goes into a
steady oscillation: referred to as locked-in (because
of the analogy with other kinds of vibrations in
uid dynamics and mechanics). Finally, if the ice is
moving fast, there is an irregular random oscillation.
The measurements of the vibration amplitude are
often poor or completely missing.
The data illustrated in Figure 4 rst appeared
in
[14]
. The gure plots the kind of oscillations
that occur against a reduced velocity U/Nt, the
ice velocity U divided by the lowest natural
frequency N and by the ice thickness t. The
pattern of behaviour is somewhat consistent. It
appears that locked-in oscillations occur when
the group U/Nt lies between about 0.03 and 0.3.
A plot that takes the characteristic length as the
contact breadth rather than the ice thickness
is much less consistent.
84 KOMtech Technology Review 2011
DESIGN OF STRUCTURES FOR
ICE INDUCED VIBRATIONS
It is desirable to minimise the breadth exposed to the
advancing ice, because that helps to reduce the ice
force that might break or overturn the structure, but
doing so reduces the exural stiffness of the structure
column and tends to lower the natural frequency.
It may not be practical to design an offshore
structure to be stiff enough for oscillations not to
occur at all. If the vibrations are so small that they
do not induce fatigue damage and do not create
discomfort to people, they probably do not matter.
Indeed, they might actually be advantageous if
they can reduce the maximum ice force on the
structure. The next and more challenging task is to
be able to predict the amplitude of the vibrations.
Another possibility is to modify the interaction
between the ice and a structure by altering the
geometry of the interaction. If a structure has a
sloping face, the fracture mechanism changes
completely. If the face slopes inward with
increasing height, the ice rides up the face and
fractures by repeated bending. The ice forces
are often signicantly smaller than in vertical-
sided structures, the fragments are larger, and
it is easier for the ice to clear by pushing round
the structure. Indeed, it can be argued that
structures in ice, where possible, ought always to
have sloping sides, and that ice researchers focus
on vertical-sided structures has been mistaken.
Vibrations are not often observed in structures
with sloping sides.
Figure 4. Relationship between structural oscillation and reduced velocity U/Nt
the horizontal and the higher at 78 starting 2.6 m
above the highest water level.
The success of modifying the interaction geometry
encourages us to think that the problem of ice-induced
vibration can be eliminated. Curiously, the design
problem of eliminating the vibrations has been much
less studied than the analysis problem of determining
when oscillations occur and how large they are. We
are continuing research on this question.
CONCLUSION
Ice-induced vibration phenomenon is one of the
important challenges for slender offshore structures
operating in ice. The most challenging aspect
from the problem is that little is understood of the
interaction between the ice and the structure as the
latter affects the failure of the former. Theoretical
works have thus far been limited in their ability to
explain the process of failure. Although ice-induced
vibrations can be captured in ice model test, this is
still dependent on the method of ice modelling and
is a subject of much debate.
However, observations of structures in Bohai Bay
have suggested that we can reasonably predict
when vibrations are likely to occur and have proved
that ice-induced vibrations can be mitigated by
modifying the structure such as adding a cone
around the waterline area. The ability to understand
ice-induced vibrations and when they may occur is
essential, as Keppel moves on to further explore the
opportunities that exist in the ice-covered regions
of the world.
Ice-Induced Vibrations of
Slender Offshore Structures 85
Figure 5. Conical piers of Confederation Bridge
This design principle has been applied to many
structures, among them the Confederation Bridge
between Prince Edward Island and mainland Canada,
light piers in the Gulf of St. Lawrence, ice barriers in the
Kashagan project in the northern Caspian, the Kemi
lighthouse in the Gulf of Bothnia, the Prirazlomnoye
platform in the Barents Sea, and numerous structures
in the Bohai Gulf. Figure 5 is a photograph of one
of the piers of the Confederation Bridge, where the
geometry near the waterline consists of two conical
sections, the lower at the waterline inclined at 52 to
86 KOMtech Technology Review 2011
AUTHORS CONTACT Cythia.Wang@keppelom.com
REFERENCES
[1] ISO 19906, Petroleum and natural gas industries Arctic offshore structures, International Standard, International Organisation for Standardisation, (2010).
[2] Palmer, A.C., Goodman, D.J., Ashby, M.F., Evans, A.G., Hutchinson, J.W. and Ponter, A.R.S. Fracture and its role in determining ice forces on offshore
structures, Annals of Glaciology, 4, 216-221 (1983).
[3] Palmer, A.C. and Sanderson, T.J.O. Fractal crushing of ice and brittle solids, Proceedings of the Royal Society, series A, 433, 469-477 (1991).
[4] Evans, A.G., Palmer, A.C., Goodman, D.J., Ashby, M.F., Hutchinson, J.W., Ponter, A.R.S and Williams, G.J. Indentation spalling of edge-loaded ice sheets,
Proceedings, International Association for Hydraulic Research Ice Symposium, Hamburg, 1, 113-121, (1984).
[5] Ashby, M.F., Palmer, A.C., Thouless, M., Howard, M., Hallam, S.D., Murrell, S.A.F., Jones, N., Sanderson, T.J.O and Ponter, A.R.S. Nonsimultaneous failure
and ice loads on Arctic structures, Proceedings, Eighteenth Annual Offshore Technology Conference, Houston, OTC 5127, 1, 399-404, (1986).
[6] Dempsey, J.P, Palmer, A.C. and SodhiI, D.S. High pressure zone formation during compressive ice failure. Engineering Fracture Mechanics, 68, 1961-1974,
(2001).
[7] Palmer, A.C. Ice forces the current state of knowledge. Proceedings, Arctic Oil and Gas Conference, Oslo (2008).
[8] Palmer, A.C., Dempsey, J.P. and Masterson, D.M. A revised pressure-area curve and a fracture mechanics explanation. Cold Regions Science and
Technology, 56, 73-76, (2009).
[9] Krn, T. and Turunen, R, Dynamic response of narrow structures to ice crushing, Cold Regions Sci. & Tech., vol. 17, 173-187, (1989).
[10] Eranti, E., Ice Forces in Dynamic Ice Structure Interaction, International Journal of Offshore and Polar Engineering, Vol.2(3), 204-211, (1992).
[11] Palmer, A.C. and Dempsey, J.P. Model tests in ice. . Proceedings, Twentieth International Conference on Port and Ocean Engineering under Arctic
Conditions, Lule, POAC09-40, (2009).
[12] Palmer, A.C. Dimensional analysis and intelligent experimentation. World Scientic, Singapore, (2008).
[13] Xu, N., Yue, Q., Qu, Y., Krn, T. and Palmer, A.C. Main factors of ice sheet inclined structure interaction process based on eld monitoring. Proceedings,
ASME 2009 28th International Conference on Offshore Mechanics and Arctic Engineering, Honolulu, paper OMAE2009-79848, (2009).
[14] Palmer, A.C., Yue, Q., and Guo, F. Ice-induced vibrations and scaling. Cold Regions Science and Technology. (60), 189-192, (2010).
ABOUT THE KEPPEL PROFESSORSHIP
The Keppel Professorship was launched in 2002. During the same year,
Keppel Shipyard, Keppel FELS and Keppel Singmarine, were also integrated
to become the Keppel Offshore & Marine group.
Working closely with the National University of Singapore, the Keppel
Professorship was established to build a repository of knowledge and to
enhance Singapores position as an international maritime centre. Also, it
aims to promote greater interest in the research on offshore and marine
technology which in turn serves to enhance the growth of Keppel Offshore &
Marine group as a global leader.
Since its inception, the Keppel Professorship has stimulated intellectual
discussion and broadened the knowledge base for engineers.
The current Keppel Chair Professor is Professor
Andrew Palmer. An authority on marine pipelines,
Professor Palmer also has extensive knowledge on
engineering, mechanics and construction on ice.
He is the second Keppel Chair, following on from
Norwegian marine structures expert, Professor
Torgeir Moan, who held the rst Keppel Chair
between 2002 and 2006.
No part of the materials published in this journal may be reproduced, stored in a retrieval system or transmitted in any form whatsoever without the prior written permission of KOMtech
A New Pipe Handling Approach for a
Deepwater Slim Drillship 87
A New Pipe Handling
Approach for a Deepwater
Slim Drillship
Alex TAN Kah Keong, B.Eng
Asbjorn MORTENSEN, Dr.Ing, M.Eng
Sudhan VINCENT, B.Eng
THIS PAPER OUTLINES A PROPOSAL ON THE NEW PIPE HANDLING
APPROACH FOR A DEEPWATER SLIM DRILLSHIP OPTIMISED FOR
RE-ENTRY TO EXISTING SUBSEA WELLS. Non Productive Time (NPT) is
inevitable in offshore operations and it increases operational costs. The proposed
slim drillship is designed with a new pipe handling solution to reduce downtime for
pipe handling. It revolves around a new vertical pipe racking machine with racking
speed of about 50-60 stands an hour compared to 30-40 stands an hour, which
is common for most racking machines in the market. This new pipe handling
approach is also tailored to facilitate various specialised well intervention
operations, which results in switch over time between different well intervention
operations being reduced signicantly. The ship with the new pipe handling
solution is designed to cut down considerable NPT while drilling wells and
performing subsea and well intervention operations up to 22500ft (6858m) below
the drilloor at water depth up to 7500ft (2286m). This results in an efcient slim
drilling and intervention vessel suitable for operation in the North Sea, Mexican
Gulf, West African, Asian and Brazilian deepwaters.
88 KOMtech Technology Review 2011
of the drillship is also maximised by this new pipe
handling solution to generate extended areas for
third party well intervention equipments enabling
both drilling and intervention packages to t into
the slim compact vessel. Furthermore, the optimised
layout offers the equipment direct access to the well
center ensuring rapid turnover between various
tools going into the well.
As emphasis is shifting towards technically
demanding segments of the offshore drilling
business, a new generation of high specication
deepwater drilling and well intervention units
which are versatile and efcient are required to
meet the continued growth of the business. This
new approach for the pipe handler conceptually
developed around a slim hull drillship delivers the
qualities of the next generation drilling unit.
INTRODUCTION
Driven by increasing world demand for energy, the
offshore exploration and production business is
moving from shallow waters towards deeper waters.
Oil drilling activities in water depth up to 550ft
(170m) can be carried out by jackups and xed
platform rigs while drilling in deeper waters require
oaters such as drillships or semisubmersibles.
Deepwater offshore oil production involves drilling
subsea wells, which is an area of the oil business with
rapid development of new technology. Emerging oil
demand has pushed oil companies to keep pace and
maintain or increase production levels from existing
elds. Subsea well intervention is a necessary
work operation in order to avoid declining output
from a subsea well. There will be more than 5500
subsea wells by the end of 2010
[1]
and the number
is still likely to accelerate in years to come. Most
deepwater drillships and semisubmersibles are made
for exploration drilling, and need a lot of setup
time when changing between various intervention
modes required for going into a pressurised well.
Efcient deepwater well intervention vessels are
therefore necessary with the purpose of reducing
Non Productive Time (NPT) and enhance cost-
effectiveness for well intervention.
In general, since intervention operations are less
time consuming than drilling a whole well, the vessel
should be able to move quickly between locations
to enhance efciency. Therefore, a slim ship-shaped
hull which is smaller and lighter than a full size
drilling vessel, is an obvious preference. The future
intervention vessel should also be able to minimise
downtime when arranging intervention and drilling
equipment, which is often the main challenge in the
vessel design of such class. For this proposed slim
drillship design, the prime technology driver is a
new conceptual pipe handling approach. This
new approach enables the slim drillship to reduce
downtime signicantly for tubular handling during
drilling operations via faster racking speed and
ofine stand building of tubulars. The new pipe
handling approach with dual horizontal catwalks
from forward and aft direction is also aimed at
handling third party well intervention equipment
skids in order to reduce turnaround time between
drilling and intervention. In addition, the layout
PIPE HANDLING CONCEPT
The new pipe handling solution revolves around
a new KOMtech design automated vertical pipe
handling machine, the Rotaset Racker
[2]
, housed on
a stand building deck which is a lowered deck at
4.5m below the drilloor to carry out ofine stand
building via powered mouseholes. An A-shaped box
type telescopic mast is developed to complement
the Rotaset Racker for tubular transfer to the well
center. Forward open deck area on the starboard
side of the vessel and under deck section are set aside
Figure 1. Overview of the slim drillship
Table 1. Tubular storage capacity
Tubular Storage Location Tubular Size No. of Joints Total Length
Main Deck 3 Drill Pipes (5 TJ) 1560 45200 ft (13777 m)
2 x (W=4.5m x H=3 m) or 5 Drill pipes (7 TJ) 780 22600 ft (6888 m)
or 9-5/8 Casing 400 16800 ft (5120 m)
or 13-3/8 Casing 270 11200 ft (3413 m)
Pipe Hold 3 Drill Pipes (5 TJ) 1560 45200 ft (13777 m)
(W=14 m x H=2.5 m) or 5 Drill pipes (7 TJ) 780 22600 ft (6888 m)
for tubular storage while the aft poop deck area is
used for riser storage. Horizontal tubular feeding
machines for moving tubulars and risers from their
storage location at forward and aft of the vessel
respectively towards the drilloor are also integrated
so that the complete pipe handling system can
work seamlessly.
Horizontal Pipe Handling
Horizontal tubular feeding is by means of a forward
drill pipe catwalk machine for transportation of drill
pipes/casings and an aft riser catwalk machine for
transportation of risers from storage location to the
well center and vice versa. The lower elevation of the
setback and the corresponding stand building deck
allows a more efcient horizontal pipe handling,
compared to catwalk decks on the same level as the
drilloor. The use of slanting catwalk chutes is not
considered an option, due to its track record with
accidents and dropped objects, and lower efciency.
The catwalk machine is also optimised for the
transportation of well intervention equipment
which will be discussed in the next section.
Tubular Storage
Drill pipes and casings storage are located forward
of the moonpool on the main deck starboard side
of the vessel with combined capacity for 400 casing
joints of 9-5/8 giving a total length of 16800ft
(5120m) or 1560 slim size drill pipe joints of
3-1/2 adding up to 45200ft (13777m). Table 1
shows the capacity of the tubular storage for different
tubular size. Additional storage area for drill pipes
is allocated in the pipe hold below the main deck
with capacity to hold up to 780 standard 5 size
drill pipe joints total length 22600ft (6888m).
This arrangement for tubular storage with pipe
hold below deck not only enhances tubular payload,
which makes the slim drillship less dependent
on tubular supply vessels/infrastructure reducing
downtime, but it also maximises the limited deck
space on the slim drillship creating available large
deck space on the forward port side of the vessel
for a lay down area for cargo as well as a ROV
hangar from which ROVs are activated for subsea
operations.
Pipe Craneage
The vessel comes with three major knuckle boom
cranes and a gantry crane. A tubular handling
overhead crane is also installed inside the pipe
hold for handling single tubular via two vertical
scissor lift tables to be raised onto the main deck.
Manual guiding of tubular through the pipe hatch
by operators done traditionally with crane hoisting
is therefore eliminated. This improves speed and
safety of the procedure. As illustrated in Figure 2,
the tubulars are transferred to the drill pipe catwalk
machine by a knuckle boom crane with bundle
pipe gripper for fast feeding of tubular to the
drilloor. This crane is located forward end of the
catwalk machine on top of the ROV hangar so as
to also access tubulars from the pipe hold hatch.
A gantry crane installed over the riser storage area
is designated to handle the riser onto the riser
catwalk machine at the aft of the moonpool.
Two other knuckle boom pedestal type deck cranes
are also installed at mid ship and ship aft for
transferring drill pipes/casings and risers respectively
to/from quay or tubular supply vessel.
A New Pipe Handling Approach for a
Deepwater Slim Drillship 89
90 KOMtech Technology Review 2011
A V-door arm, referred to as the V-door machine,
picks up single tubulars from the forward catwalk
and raises them to vertical orientation, in order
to put them into one of two vertical mouseholes.
The catwalk machine/ conveyor have a tail-in arm
to assist the lower end of the tubular, in order to
transfer tubular safely into the mousehole.
Vertical Pipe Handling Machine
The vertical pipe handling machine, KOMtechs
Rotaset Racker, is a conceptual racking machine
designed to be one of the worlds fastest racking
systems
[2]
. It achieves a racking speed of about
50-60 stands an hour as given by the thread make and
break cycle time and the drawworks hoisting speed,
meeting the industrys expectation and overcoming
the shortcoming of most vertical racking machine
able to rack about 30-40 stands per hour. The
Rotaset Racker is a fully automated vertical racking
system with two rotating setbacks and a separate
two arm column racker. Its tripping sequence is
divided into two parallel activities to improve
racking speed: The rst activity for internal handling
inside setback and to drilloor, and a second activity
is to handle the stand from setback vicinity to its
destination. The setback area is located on the
lowered stand building deck. Its storage capacity
includes sections with belly board for slim pipes
(3 ), sections for double casing stand building,
as well as normal triple stands, heavyweights and
collars. Casing stand building is performed in
parallel with running casing doubles into the well,
in order to reduce the size of the setback. This
not only allows the slim drillship to gain an
advantage with a lower center of gravity (COG)
and hence improves stability, but also maximises the
available space on the drilloor. The lower elevation
of the stand building deck not only permits the
forward knuckle boom crane to reach the main
deck tubular storage with ease, but also improves
safety and workability during ofine stand building
due to its relative far distance from the well center
(compared to traditional mono derricks with ofine
stand building).
The ngerboards have a double barrier giving it
redundancy to prevent stands from falling out.
This cylindrical and rotating ngerboard is placed
inside a stationary ring to prevent the stand escaping
from the ngerboard by accident.
Figure 2. Tubular storage on open deck with tubular craneage and horizontal tubular feeding machine
Knuckle boom crane with pipe gripper
Drill pipe catwalk machine
Tubular storage
Pipe hold hatch
Stand building deck
Drilloor
Tail-in arm