INTEGRITY OF OFFSHORE STRUCTURES-4

Papers presented at the Fourth International Symposium on Integrity of

Offshore Structures, 2-3 July 1990, held at the Kelvin Conference Centre,
University of Glasgow, Scotland, organised by the University of Glasgow,
Departments of Naval Architecture and Ocean Engineering and Mechanical
Engineering.

ORGANISING COMMITTEE (The University of Glasgow)

Prof. D. FAULKNER (Chairman) Dept of Naval Architecture and Ocean
Engineering
Dr M. J. CowLING Dept of Mechanical Engineering
Dr A. INCECIK Dept of Naval Architecture and Ocean
Engineering

TECHNICAL COMMITTEE
The Organising Committee (as above) and:
MrS. CLINTON BP Exploration
Dr J. A. MERCIER Conoco Inc.
Dr L. R. WOOTTON W. S. Atkins Engineering Services
 INTEGRITY OF
OFFSHORE STRUCTURES-4

Edited by

D. FAULKNER*
M. J. COWLINGt
and

A. INCECIK*
* Department of Naval Architecture and Ocean Engineering,
t Department of Mechanical Engineering,
The University, Glasgow, Scotland

f.?\ Taylor & Francis

~ Taylor & Francis Group

LONDON AND NEW YORK

 Taylor & Francis
2 Park Square, Milton Park, Abingdon, Oxfordshire OX14 4RN
711 Third Avenue, New York, NY 1001 7
First issued in paperback 2011
British Library Cataloguing in Publication Data
International Symposium on Integrity of Offshore Structures
(4th: 1990: Glasgow, Scotland)
Integrity of offshore structures-4.
I. Offshore structures. Design & construction
I. Title II. Faulkner, D. (Douglas) 1929- III. Cowling,
M.J. IV. Incecik, A.
627.98

Library of Congress CIP data applied for

ISBN 978-1-851-665 19-8 (hbk)

IS BN 978-0-4 15-5 1592-4 (pbk)
IS BN 978-0-4 15-5 1592-4 (pbk)

No responsibility is assumed by the Publisher for any injury a nd/ or damage to persons or property as
a matter of products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions or ideas contained in the material herein.

Special regulations for readers in the USA

This publication has been registered with the Copyright Clearance Center Inc. (CCC), Salem,
Massachusetts. Information can be obtained from the CCC about conditions under which
photocopies of parts of this publication may be made in the USA. All other copyright
questions, including photocopying outside the USA, sho uld be referred to the publisher.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted in a ny form or by any means, electronic, mechanical, photocopying, recording, or
otherwise, without the prior written permission of the publisher.
 v

Preface

It is once more our privilege to write this short introduction to the Proceedings of
the Fourth Symposium on Integrity of Offshore Structures. These symposia are
held every three years in the University of Glasgow and this year it coincides with
Glasgow being the European City of Culture. There will thus be plenty of
other attractions, including our own Maritime Exhibition based on the theme
of 107 years of service to the marine industries.
It is this same strong urge to see good project, consulting and research work
applied that created the lOS symposia and has been our constant driving force.
For this reason the organisers set out to do several things. We structure the
programme into four relevant sessions and then deliberately seek a strictly
limited number of quality papers from a wide cross-section of practitioners and
researchers to meet these objectives. You will see that our fourth session this year
is devoted to 'In-Service Assessment Strategies', which reflects the strong trend to
increased life now required of many platforms in the North Sea especially. A
major feature is then to devote the maximum time for discussion from the
delegates. We achieve this by sending out the preprints in advance and by inviting
a distinguished Rapporteur for each session to review the highlights and main
findings of each paper. Each session then has at least an hour for open discussion
followed by a brief response from each author. All the discussion is then edited
and published in the final book.
This of course represents a lot of work for the Organising Committee, but we
have been helped this time by a small external Technical Committee and our
personal thanks are extended to them. In turn both Committees are deeply
grateful to the eight eminent Rapporteurs and Session Chairmen and not least of
course to the authors.
We have been gratified to note the continued support for this approach. There are
28 papers from 12 different countries, including a welcome increase from Japan
this year. 54% of the authors are from industry or government departments and
46% from research laboratories and universities. We believe this all illustrates a
healthy balance.
D. FAULKNER
M. J. COWLING
A. INCECIK
 VII

Contents

Preface v

Session 1: Loading and Dynamic Response

1. Modelling of Wave Loads on Moored Systems in Spread Seas
R. EATOCK TAYLOR* and K. L. MITCHELLt (*Oxford University;
t Brown & Root Vickers Ltd)
2. Refined Modelling of Hydrodynamic Loads on Dynamically
Sensitive Structures 19
0. T. GUDMESTAD (Statoil, Norway)
3. On the Modelling of Uncertainties Associated with the Ocean
Environment and their Influence on Response Evaluations . 39
K. VENKATARAMANA,* K. KAWANo,t Y. YAMADA** and H.
IEMURA** (*Sumitomo Metal Industries; tKagoshima Uni­
versity; **Kyoto University, Japan)
4. Dynamics of Double Articulated Towers 53
J. H. HELVACIOGLU and A. INCECIK (Glasgow University)
5. Numerical and Physical Simulation of Slow Drift Motion of a
Moored Floating Structure in Waves . 85
T. KINOSHITA and K. TAKAIWA (University of Tokyo)
6. Nongaussian Nature of the Response of Compliant Skeletal
Offshore Structures Subjected to Gaussian Ocean Waves 105
G. R. HooPER* and P. SWANNELLt (*Connell Wagner Group;
t University of Queensland)

Session 2: Ultimate Strength and Design

7. Reliability of Offshore Structures under Various Assumptions of
Stochastic Waves . 121
M. ARROYo,* B. JACOB,* Y. GuENARDt and M. LEMAIRE**
(* Laboratoire Central de Ponts et Chaussees; t Societe Nationale
Elf Aquitaine; ** Universite Blaise Pascal, F'rance)
 Vlll

8. A Global Approach for Reliability Based Offshore Platform Codes 137

F. MosES (Case Western Reserve University, Ohio)

9. Application of Ultimate Strength Analysis in Design of Offshore

Structural Systems 153
H. NORDAL (Statoil, Norway)

10. Optimum Design of Stiffened Cylinders Based on Reliability 167

J.-S. LEE (University of Ulsan, Republic of Korea)

11. Probabilistic Calibration of Design Criteria for Marine Risers 179

T. MoAN and G. JIAO (University of Trondheim, Norway)

12. Application of Reliability Theory to Structural Design and

Assessment of Submarines and Other Externally Pressurised
Cylindrical Structures . 199
D. FAULKNER* and P. K. DAst (*Glasgow University; t BMT,
Wallsend)

13. Design Philosophy for Composite Construction 231

C. D. GooDE (Manchester University)

14. Probabilistic Design Aspects for Deep Water Concrete Offshore

Structures 247
D. DIAMANTIDIS, P. BAZZURRO, G. M. MANFREDINI and F.
ZucCARELLI (D'Appolonia, Italy)

Session 3: Fatigue and Fracture Modelling

15. The Fatigue Performance of Tubular Joints~An Overview of

Recent Work to Revise Department of Energy Guidance 261
A. G. REYNOLDS* and J. V. SHARPt (* BP International;
tMaTSU)

16. A Review of Stress Concentration Factors for Tubular Complex

Joints . 279
P. SMEDLEY and P. FISHER (Lloyd's Register of Shipping)

17. The Effect of Uncertainties in Fatigue Life on Design and

Inspection of Offshore Structures . 297
T. W. THORPE and J. V. SHARP (MaTSU)
 IX

18. Fatigue and Crack Growth in a Submerged Ring Stiffened Tubular

Joint 319
H. HAUGLAND and T. C. THUESTAD (Norsk Hydro, Norway)

19. Japanese Research Activities on the Structural Integrity and

Reliability of Offshore Structures Particularly for Fatigue and
Fracture 335
S. MACHIDA,* M. MATOBAt andY. HAGIWARA** (*University of
Tokyo; t Mitsubishi Heavy Industries Ltd; **Nippon Steel
Corporation)

20. Prevention of Brittle Fracture in Offshore Structures . 351

C. ARBUTHNOT* and M. SMITHt (*John Brown; t Earl & Wright)

21. Fracture Behaviour of the Local Brittle Zones of Offshore

Weldments . 365
M. K09AK,* S. YAo* and L. CHENt (*GKSS Research Centre,
FR Germany; t Shanghai Jiao Tong University)

Session 4: In-Service Assessment Strategies

22. A Defect Assessment Methodology for Offshore Jacket Structures

Including Complex Joints Subject to Complex Loading 381
J. HASWELL (British Gas pic)

23. The Implications of Non-Destructive Inspection Reliability on

Offshore Structural Integrity 405
J. C. P. KAM and W. D. DovER (University College London)

24. Probability-Based Optimization of Fatigue Design, Inspection and

Maintenance. 421
H. 0. MADSEN* and J. D. SORENSENt (*Danish Engineering
Academy, Lyngby; t Aalborg University Center)

25. Fracture Mechanics and Simplified Analyses in Probabilistic

Remaining Life Calculations for Jacket Structures 439
X. N. NIU, A. H. S. WICKHAM and P. A. FRIEZE (Advanced
Mechanics and Engineering Ltd)

26. The Role of Redundancy in Reliability Based Fatigue Life

Assessments in Offshore Structures 457
C. A. PLANE,* M. J. CowLING,* V. NwEGBut and F. M.
BuRDEKINt (*Glasgow University; t UMIST)
 X

27. An Approach to the Evaluation of Service Life . 477

J. c. BAXTER,* H. P. COJEEN,* M. BOWEN,* A. THAYAMBALLit
and R. BEA **(*US Coast Guard; t American Bureau of Shipping;
**University of California)

28. Confidence Interval in Reliability Analysis of Marine Structures. 493

J. LABEYRIE,* M. HuTHERt and G. PARMENTIERt (*IFREMER,
Brest; t Bureau Veritas, Paris)

Welcoming Address 511

Symposium Opening Address 513

Keynote Address: Unified Design Codes for Floating Systems 517

D. FAULKNER (Glasgow University)

Rapporteurs' Reports and General Discussions for Sessions 1-4 533

Closure 621

Errata 623
 1
MODELLING OF VAVB LOADS ON
MOORED SYSTEMS IN SPREAD SEAS

R.Eatock Taylor
Department of Engineering Science
University of Oxford, UK.
and
K.L Mitchell
Brown & Root Vickers Ltd
Vimbledon, London, UK

ABSTRACT
Design of the moorings for floating production and compliant systems
is strongly influenced by low frequency wave drift forces which excite
resonant responses. The paper discusses the modelling of these forces by
second order theory, involving the use of directional quadratic transfer
functions. Theoretical predictions using such functions are compared with
the results of experiments in bidirectional and fully spread seas.
Simulated force spectra are also compared with spectra estimated from
experiments in directional seas. Conclusions are drawn concerning the
influence of directionality, and uncertainties in the modelling of drift
forces.
1. INTRODUCTION

Design of the moorings for floating production and compliant systems

requires knowledge of environmental load effects due to winds, ocean
currents and waves. Critical aspects are the static offset (hence mooring
line tension) due to these effects, and dynamic responses in the low
frequency modes of surge, sway and yaw of the moored platform. These
dynamic responses may be associated with instabilities driven predominantly
by quasistatic winds and currents; or by dynamic load effects associated
with the waves. The work described here forms part of an investigation of
the latter.

DOI: 10.1201/9781003076599-1
 2

The resonant frequencies of floating systems in the horizontal modes

are generally much lower than the frequencies at which there is significant
energy in ocean waves. At full scale, typical resonant periods may lie in
the range of 1 to 5 minutes. This leads to the type of behaviour
illustrated in Figure 1, which shows time histories of wave elevation and
surge response of a turret moored tanker in spread seas (The results have
been scaled from model tests performed at a scale of 1:81 in a wave basin,
where no winds or currents were present). In this example the waves had a
characteristic period of lOs, whereas the surge response period is seen to
be about 94s. The problem for the designer, concerned with the ultimate
strength and the fatigue life of the mooring lines, is to predict the
statistics of the line tensions under these conditions. And the part of
that problem with which we are concerned here is prediction of the low
frequency wave forces causing such drift responses.

The underlying hypothesis is that the loading causing large low

frequency motions and tensions depends on terms proportional to the square
of the wave elevation process. Such terms are omitted in first order
theory, which then can be developed using a frequency domain approach based
on linear transfer functions between loads and waves. The analogous
frequency domain procedure for the second order theory employs Quadratic
Transfer Functions (QTF's). It is clearly more complex than linear theory,
and there are several practical difficulties in its application; but recent
years have seen much work in this area, and a methodology for long crested
random seas is now reasonably well established. The second order frequency
domain analysis reported by Langley(l)is characteristic of the approach
that has been taken by several investigators.

Notwithstanding this development of a rational theory, several

difficulties have delayed its widespread implementation in professional
practice. These include uncertainties in the calculation of second order
hydrodynamic forces; and problems in performing appropriate experiments and
analysing the resulting data. The present investigation has been directed
at shedding some light on these two aspects; and at meeting the requirement
for extending application of the theory to study the influence of
Birectional spreading in the waves. The latter problem has also been
considered recently by Pinkster< 2>, Dalzell( 3 ), Maeda, Morooka and
Kasahora< 4 >, and Nwogu and Isaacson< 5 >.
 3

Some of the work has previously been reported by Eatock Taylor, Hung
and Hitche11< 6 >. There we summarised some of the theoretical considerations
underlying the extension to directionally spread seas, and provided
calculations illustrating the influence of various approximations in the
hydrodynamic analysis. The present paper attempts to provide experimental
evidence confirming the underlying hypothesis of the wave drift force
analysis, and offers some limited comparisons between theory and
experiments in directionally spread seas. The latter were performed at
small scale, and are subject to several sources of uncertainty - but we
believe such data to be extremely rare at the present time.

The paper is organised as follows. The next Section provides a brief

recapitulation of the theoretical background to the QTF corresponding to
unidirectional and bidirectional waves; and· describes experiments and the
resulting data leading to comparison of theoretical and experimentally
estimated QTF's. Section 3 is concerned with theoretical simulation and
experimental estimation of drift force spectra: first in crossing (ie
bidirectional) seas; and secondly in fully spread seas. The final Section
lists some brief conclusions.

2. QUADRATIC TRANSFER FUNCTIONS FOR VAVB DRIPT FORCES

2.1 Summary of Theory

It is convenient first to consider two sinusoidal waves of frequency

~i; direction of propagation ~i and phase cf>i relative to some datum; and
amplitude Ai' where i=1,2. The low frequency second order wave force on a
body in direction k may then be expressed as

(2) 1 2 1 2
f k{f) = ZA1 H11(~1'-~1) + ZA2H22(~2'-~2)

+A1A2Re [H12<~1'-~2)exp (i (<~1- ~2)t + +ccf>2)] (1)

Here Hij(~i'-~j) is the quadratic transfer function for the force due to
unit amplitude waves of frequencies ~i and ~j in directions i and j. The
force is in direction k (k=1,2 or 3 for surge, sway and yaw), but the
subscript k is here omitted from the QTF for simplicity. Pinkster< 2 >uses
an alternative notation, which may be related to the above by substitution
 4

of

H12(Cill-Cil2) = 2 (<P12et~- iQ12et~) (2)

with et and ~ denoting the wave directions instead of ~ and Ctz• The
advantage of the form given in Equation (1) is that the QTF thereby defined
is the double Fourier transform of a quadratic impulse response function,
arising in the Volterra series representation of the second order
process< 3 • 6 >.

Ye next use the OTF to describe the force in directional random

seas. Ye assume that the waves are constituted by a sum of N
unidirectional waves, having one sided spectral density functions of wave
elevation Gaai(Cil), for i=1 toN. The mean drift force in direction k is
then found to be given by

E [fk( t) ] ~=1 1 H ii(Cil, -Cil)Gaai(Cil) dCil; (3)

and the single sided drift force spectrum is

Gff(Cil) =2 ~=1 J=l J IHij(~,Cil-~)

0
12 Gaai(~) Gaaj( ICil-~1> d~ (4)

This form in terms of one sided spectra is based on the assumption

that the frequencies Cll relevant to low frequency drift forcing are much
lower than wave frequencies ~; hence there is no contribution of any
significance to the integral in Equation (4) when ~<Cil. One observes that
the drift force spectrum at frequency Cll is made up of contributions from
all pairs of wave components whose frequencies sum to Cll.

This is also observed from the Fourier component of drift force at

frequency Cll, which in spread seas can be written

F(2)(Cil) ~1= 1 ~J =1 _., H..

J"'
1J
(~,Cil-~) A.(,....)
1
A. (Cil-~)d~
J
(5)
 5

with the obvious notation that Ai(~) is the Fourier transform of the wave
elevation ai(t) in direction i. Ye may then take the inverse transform of
Equation (5) to obtain the time history of low frequency drift force in
directional seas when N individual directional components have been
distinguished as time series. The latter may be resolved using beam
forming techniques< 7 >.

Arising from Equations (4) and (5) is the concept of using sum and
difference frequencies to identify the dependence of the drift force on the
QTF. Thus it is useful to define

21=ool- 002' 22= 001+ 002 · (6)

The second order force at frequency oo is then determined by the line

22=oo in the bifrequency plane defined by axes oo1and oo2 . It is noted that
in the general formulati~n the frequencies may be both positive and
negative (c.f. Equation (5)), and in that region of the bifrequency plane
corresponding to slowly varying drift forces, one of these is usually
negative (c.f. the mean drift component in Equation (3)). The frequency
22= oo of the slowly varing component is then given by the sum of a positive
and negative QTF frequency argument. These ideas are illustrated in the
results below.

2.2 Results for a Tanker.

An extensive set of theoretical and experimental results has been
obtained< 7 •8 >, from which the following are typical. The experiments were
conducted in the 9m square directional wave basin at Heriot-Yatt
University, using a system of mooring lines attached to springs to react
the low frequency drift forces. The natural frequency of the system was
designed to lie between the frequency range of relevance to low frequency
second order forcing and that of the higher wave frequency motions.
Although the low frequency forces transmitted through the springs to force
transducers in each mooring line are then a good approximation to the wave
drift forces on a vessel responding freely at wave frequencies there are
nevertheless inherent problems with this method:
 6

(i) The measured force can contain substantial force components at

wave frequencies;
(ii) in random seas the frequency ranges of relevant drift forces
and wave frequency forces can overlap, leaving no intermediate
region in which the resonant frequency can safely lie.
These problems have been minimised by adjusting the natural frequency so
that the force component in the mooring lines due to the resonant response
is minimal; and by filtering the resulting time series to remove all but
the low frequency signals.

An additional source of complexity in these experiments is the

estimation of the QTF from the measured data. It is relatively
straightforward to estimate single points on the QTF bifrequency plane from
tests in regular and biharmonic waves, but a large number of tests would be
required to define even a small area of the QTF. The alternative explored
here is to estimate the QTF by the application of hi-spectral analysis
techniques to elevation and force time histories measured from tests in
random waves. Various methods for estimating the QTF from random wave
tests have been examined in considerable detail by Sincock< 9 >, from which
the "direct" method was adopted here. This is based on the relation

* *
Lim T E [F(ool+ oo2) Al(ool)A2(oo2)1 (7)
H12(ool, oo2)= T-+.., 4n
E[ IA£oo1 ) 12 E[ IA 2 <oo2 ) 12 1

where F and Ai are the finite Fourier transforms of the low frequency force
and wave records over an intevra1 T, * designates complex conjugate, and
the expectation operator E implies averaging across different realisations.
In practice of course this relation must be transformed in terms of
discrete Fourier transforms of blocks of N points, at a sampling interval
6t. For the following results we used 131,072 data points sampled at
intervals of 0.9s, which were split into 256 blocks of N=512 points; the
expected values were then obtained by averaging over the 256 blocks.

The experiments were performed on a 1:81 scale model of a tanker

whose properties are given in Table 1: the results, however, are quoted at
full scale. Figure 2 shows the experimentally estimated QTF for surge on
the tanker in long crested bow quartering seas. It has been obtained by
 7

filtering from the force record all frequency components above 0.167 rad/s;
and by smoothing each ordinate with its eight adjacent ordinates in the QTF
plane. Prior to smoothing the function was set to zero in the region where
the denominator of Equation (7) was less than 12% of its peak value. This
is because estimates in regions where the quadratic input is low are
inevitably unreliable due to the predominance there of the effects of noise
and statistical variability.

It is difficult to compare experimental results in the form of

Figure 2 with corresponding theoretical predictions; but by taking sections
through the QTF surface similarities and differences between theory and
experiment can be readily identified. Figures 3a and 3b show the
theoretical and experimentally estimated surge force QTF's in bow
quartering waves, along the sum frequency sections 22=0.0 and 0.109 rad/s
respectively. The corresponding sway force QTF's are shown in Figures 4a
and 4b. The theoretical results were calculated using the computer program
DYHANA, based on a comb1ned finite element/boundary element numerical
scheme< 10 )which has been extended to permit evaluation of second order
hydrodynamic forces in bichromatic waves< 6 >. Two sets of theoretical
results are shown: those corresponding to the freely floating vessel
originally intended to be tested; and results based on using an increased
roll stiffness, which takes some account of the restraint to roll motions
caused by the attachment of the mooring lines above the centre of rotation
(the first order motions in the other five rigid body modes were not
observed to be affected). The results in Figures 3a and 3b estimated from
the experimental records have not been subjected to frequency smoothing,
but otherwise they have been obtained in the same way as the data
plotted in Figure 2.

By examination of Figures 3 and 4, one can observe rapid changes in

the theoretical QTF's at the difference frequencies which require first
order data at the roll resonances (0.5 rad/s and 0.627 rad/s for the free
and restrained cases respectively). For the sum frequency sections 22= 0.0
rad/s, these difference frequencies equal twice the resonant frequencies.
For the sections 22=0.109 rad/s, the corresponding difference frequencies
are 21= 0.891 rad/s and 1.109 rad/s when the roll resonanse is at 0.5
rad/s; while 21= 1.145 rad/s and 1.363 rad/s are the difference frequencies
 8

corresponding to the restrained case roll resonant frequency of 0.627

rad/s. These irregularities in the theoretical QTF's, which are not
distinguishable in the experimental estimates, are probably due to the fact
that viscous roll damping effects were not included in the calculations.
The roll motions at resonance were consequently overpredicted, leading to
inaccuracies in the second order forces which use components at these local
frequencies.

Further details of the experiments, and an extensive commentary on

the results, have been given by Mitchell(B). Some of the sources of
discrepancy between theory and experiment can be attributed to the
aforementioned difficulties of measuring low frequency forces in random
waves; reflections in the wave basin; and the assumptions behind the theory
of ideal flow. Futhermore, the effects of the second order velocity
potential were ignored for the calculations shown in Figures 3 and 4.
Despite these various problems, the degree of agreement between these
theoretical and experimental results is thought to be most promising.

3. DRIFT FORCE SPECTRA IN DIRECTIONAL SEAS

Here we present theoretically simulated results and experimental

data for a vessel in various directional seas. The vessel is the same
tanker investigated in the previous section. The objective of these
results is specifically to shed light on the modelling of low frequency
loading in spread seas.

3.1 Crossing seas

Taking N = 2 in Equation (4), we have an expression for the spectrum

of low frequency force in crossing seas comprising a combination of
(different) long - crested random waves from two directions. One observes
that the low frequency second order excitation is larger than the sum of
the excitations due to each unidirectional wave taken independently. The
mean drift force, however, given by Equation (3) with N = 2, is exactly
equal to the sum of the mean forces in each independent unidirectional
wave.
 9

The QTF's employed in these expressions can be expensive to obtain.

The effort can be minimised by only evaluating them in the areas of the
bifrequency plane where there is significant input of power. The
contributions from the various terms in Equation (4) depend therefore on
the various products of the wave elevation spectra. If the spectra of the
two unidirectional waves comprising a crossing sea have little overlap on
the frequency axis (eg. in the case of swell from one direction and wind
driven waves from the other), the contribution from interaction between the
two directions will be small in the low sum frequency region of the
bifrequency plane. Under these conditions the low frequency excitation
could be taken as the sum of the drift forces due to the waves approaching
from each direction independently.

Two other factors can help to limit. the bifrequency region over
which the QTF's are required. If low frequency resonant response is
critical, then a set of sum frequency lines in the bifrequency plane which
span the natural frequency of the moored vessel needs to be considered, but
sum frequencies outside this region can be disregarded. Yhen the frequency
difference between the peaks of the two unidirectional spectra equals the
resonant frequency, the interactive terms (i#j) in Equation (4) have their
greatest influence on forcing at resonance. The second aspect to consider
is that there is little point in evaluating the QTF's for sum frequencies
higher than the lowest frequency of significant wave forcing.

These principles have been applied in the simulation of drift forces

on the tanker in a crossing sea with waves approaching from equal angles of
16° on either side of the bow. The wave components were defined by ISSC
spectra, having significant wave heights I characteristic periods of Sm/12s
from -16°, and 6m/10s from 16° respectively. The QTF's were evaluated over
the grid of points shown in Figure 5. Linear interpolation of real and
imaginary points of the QTF's was used to obtain intermediate values. The
spectra of surge and sway drift forces on the tanker, evaluated using
Equation (4), are designated closed form solutions in Figure 6. The
continuous lines, designated estimated spectra, were obtained by performing
spectral analysis on 8192 point time series simulations of drift force
(averaging 16 separate estimates and smoothing over 5 adjacent frequency
ordinates). These time series were based on simulated wave elevations
 10

generated by colouring white noise, followed by inverse transformation of

Equation (5) for the case N = 2.

It may be observed from Figure 6 that there is reasonable agreement

between the underlying trends of the closed form solutions and the spectra
estimated from the simulated drift force time series. The irregularity of
the latter is associated with the statistical variability of the estimates
based on a limited set of data. The results also demonstrate the important
contribution made by the interactive components of Equation (4) in this
case - at some frequencies these constitute up to 50% of the total low
frequency force spectra in this crossing sea state.

3.2 Fully spread seas

Ve now consider the behaviour of the tanker in multi-directional

seas where N>2. Our starting point is a continuous directional spectrum,
which for the theoretical calculations is discretised into N representative
unidirectional spectra. The number of QTF's required to evaluate the force
spectrum in each mode using Equation (4) is N(N+l)/2 (taking account of the
symmetries of the QTF's< 6 >. Ve used N = 9 for the following theoretical
results. To provide flexibility in simulating forces due to spread seas
generated in the wave tank, it is convenient to choose equally spaced
directions, and here we used four directions at 16.8° intervals on either
side of the predominant wave direction (head seas).

Figure 7 shows simulated surge force results for a series of spread

seas represented by a spreading function of the form cos 2se, over
-nl2<e<nl2. In each case the point wave spectrum was defined by the ISSC
formulation with significant wave height 6m and characteristic period lOs.
It is clear from the figure that over a substantial part of the frequency
range the low frequency surge drift force increases with increased
spreading of the seas. This is related to the fact, which we have observed
in both our experimental and theoretical data, that for waves approaching
the vessel from directions other than ahead the magnitudes of the surge
force QTF in some areas of the bifrequency plane are greater than the
corresponding head seas QTF's in the same areas.
 11

Low frequency force spectra were also estimated from experiments in

spread seas, and compared with simulations in nominally the same wave
conditions. Figure 8 shows the estimated directional spectrum generated in
the tank during one sequence of tests (spread sea A). This was estimated
using the Maximum Likelihood Method (MLM), based on wave elevation time
series from an array of seven probes. In this as in all the experiments the
waves were measured during a separate run without the vessel in the tank:
the pseudo random wave generation signals were repeatable, and by means of
a reference probe could be aligned with the time histories of forces
measured in the separate tests. All records of the wave elevations and
forces consisted of 32,768 data points sampled at an interval of 0.9s. The
cross power spectral densities required for the MLM analysis were averaged
over 64 estimates, and each spectral ordinate was averaged along the
frequency axis with nine adjacent estimates.•

Figure 9 shows the estimated surge and sway drift force spectra in
this sea state, and also in two other cases (spread sea B and a
unidirectional sea). The calculated results shown in these figures were
based on the theoretical QTF's, and discretisation of the estimated
directional wave spectra (eg. Figure 8) into nine directions.

These figures lead to several observations. Very obvious are the

peaks at about 0.1 rad/s, which are close to the lowest calculated standing
wave resonant frequency of the wave basin (0.112 rad/s). The resonant
frequencies of the moored vessel itself in surge and sway were estimated
from measurements to be 0.24 rad/s and 0.28 rad/s respectively, and it is
noticeable how the drift forces rise rapidly towards these frequencies.
For these reasons the spectra in Figure 9 which have been estimated from
the experimental measurements are not considered to give a reliable
representation of the drift forces on the freely floating vessel, in the
open sea, above a frequency of about 0.075 rad/s.

It should also be noted that the method of discretising the

directional spectra for the simulations increased the spread of the wave
energy (since the lumping occurs at the mid point between directions rather
than at the centroid of the energy lying between those directions). As a
result the calculated sway force spectrum would be expected to be slightly
 12

larger than that estimated from the experiments; and (in this case) the
calculated surge spectrum would be appear to correspondingly less.

As a final comment on the comparision of theoretical and

experimental results in spread seas, it is appropriate to emphasise the
complexity of the various analytical and numerical procedures employed in
the calculations, the difficulty in measuring drift forces on a "freely"
floating vessel; and the statistical variability inherent in the estimation
of parameters from tests in random waves.

4. CONCLUDING RBKARKS

It has been found that the non-linear interaction of waves from

different directions can have a significant influence on the low frequency
loads in mooring lines, at least for the moored tanker system considered
here. In some cases this component constituted up to 50% of the total
drift force. Futhermore, it has been observed that the surge drift forces
(and indeed the surge wave frequency forces) can either increase or
decrease as the spread of the sea state is increased. Hence the calculated
or experimentally measured mooring loads in unidirectional seas should not
be considered a conservative estimate of the loads occurring in a real sea
having the same point spectral density as the unidirectional sea.

Ye have performed a series of experiments in crossing seas and fully

spread seas which have tended to support our theoretical simulations. All
the results point to the significant influence of wave directionality. The
results presented here, however, (both theoretical and experimental)
highlight the difficulties currently associated with predicting low
frequency loads in real seas. The further stage of predicting responses has
well known additional difficulties related to the modelling of low
frequency damping. Considerable further work is required, both on
theoretical and numerical aspects, experimental techniques and methods of
data analysis. Only then will there be adequate tools, and scope for
generating sufficient statistical data, to enable soundly based reliability
techniques to be applied in the design of moored systems.
 13

ACKNOVLEDGEKENT

This work formed part of a project conducted by the authors in the

Department of Mechanical Engineering, University College London. It was
supported by the Managed Programme on Floating Production Systems (FPS),
jointly sponsored by industry and the Science and Engineering Research
Council through Marine Technology Directorate Ltd. The experiments were
carried out in the directional wave basin at Heriot-Watt University in
collaboration with another project in the FPS Programme, led by
Mr.B.T.Linfoot. His help and cooperation are much appreciated.

REFERENCES

1. LANGLEY, R.S.: 'Second order frequency domain analysis of moored

vessels', Appl. Ocean Res., vol. 9, pp. 7-18, 1987.

2. PINKSTER, J.A.: 'Drift forces in directional seas', NSMB Publication

Z50545, 1985.
3. DALZELL, J.F.:'Quadratic response to short crested seas',
Proceedings of the 16th Symposium on Naval Hydrodynamics, Berkeley,
1986.

4. MAEDA, H., MOROOKA, C.K. and KASAHORA, A.: 'Motions of floating type
offshore structures in directional waves'. In Proceedings of the
5th Offshore Mechanics and Arctic Engineering Symposium, Tokyo, vol.
1, pp. 94-101, 1986.

5. NWOGU, 0. and ISAACSON, M.: 'Drift motions of a floating barge in

regular and random multi-directional waves'. In Proceedings of the
8th Offshore Mechanics and Arctic Engineering Symposium, The Hague,
vol. 2, pp. 441-448, 1989.
6. EATOCK TAYLOR, R.,HUNG ,S.M. and MITCHELL, K.L.: 'Advances in the
prediction of low frequency drift behaviour'. In BOSS '88,
Proceedings of the International Conference on Behaviour of Offshore
Structures, ed. T. Moan, N. Janbu and o. Faltinsen, Tapir
Publishers, Trondheim, vol. 2, pp 651-666, 1988.

7. MITCHELL, K.L., KNOOP, J. and EATOCK TAYLOR, R.: 'Prediction of low

frequency responses in directional seas, Final Report on project
A2, Managed Programme on Floating Production Systems, 1989.

8. MITCHELL, K.: 'Slow drift behaviour of floating structures in multi­

directional seas', Thesis submitted for the degree of Doctor of
Philosophy in the University of London, University College London,
1989.
 14

9. SINCOCK, P.: 'Non-linear compliant systems in irregular seas',

Thesis submitted for the degree of Doctor 01 Philosophy in the
University of London, University College London, 1989.

10. EATOCK TAYLOR, R. and ZIETSMAN, J.: 'Hydrodynamic loading on multi­

component bodies', In Behaviour of Offshore Structures, Proceedings
of the 3rd BOSS Conference, ed. C. Chryssostomidis and J.J. Connor,
Hemisphere Publishing Corp., Washington, vol.1, pp. 424-446, 1982.

Table 1 Particulars of tanker vessel

Scale 1:81
Loaded displacement 109,000 Tonnes
Length L 254.0 m
Breadth B 38.4 m
Draft T 13.0 m
Block coefficient CB .765
Centre of buoyancy above keel 6.7 m
Centre of buoyancy forward of midship 6.6 m
Centre of gravity above keel 8.0 m
Radii of gyration
Transverse k 13.4 m
XX
Longitudinal k 55.8 m
yy
Vertical k 55.8 m
zz

8
::-~"~
-8 0
1798 20 2517 84 3237 48 3957.12
'
4676 76 5396 40
Time/Sec

::~~ 1?'98.20 2517.84 3237.48

Time/Sec
3957 12 4676 76 5396 40

Figure 1 Behaviour of a turret moored tanker in spread seas

a) wave elevation; b) surge response.
 15

"'
...
..

..
n,.n. Ro.ds/sec
Units of vertical axis: Njm2

Figure 2 Magnitude of estimated surge force QTF in bow

quartering waves.
~=..-------------, ~!:.,----------,

~. ..: b

l lMo 0 o0 {
0

oooOoogQ

1.00
CJ 1-<.J 2
1.25
rad.s/sec
1.50 .... 1.00 1.25 1.~

CJ 1 -<o.~ 2 rads/sec

Figure 3 Theoretical and experimental surge force QTF in bow

quartering waves a) 0 2 - 0.0 rad/s - real part;
b) 0 2 - 0.109 radjs - magnitude.

~~.....-----------,

•
)f. 022
oooooooa 000
~:.,----------.
0
o
0 o2go

•
& 0o
0

a b

l:
0

1~
:.
888M&.u.••
ooooo

"'.:.g~•"'ta
a'

1.00 l.ZS 1.~ t.'oo 1~ 1.~ l:'r.l

CJ1-(ol 2 rad.sfsec CJ 1-fo.1 2 rads/sec

LECEND LEGEND
Rnl part of e111ll.mat.ed QTF Wagn~.tude of e11li.mated QTF
o DYHANA tree nuel o DYHAN4 free fl!IIUl
6 DYHANA restrained Yeuel DYHANA re1tramed Ye.. el

Figure 4 Theoretical and experimental sway force QTF in bow

quartering waves a) 0 2 - 0.0 rad/s - real part;
b) 0 2 - 0.109 rad/s - magnitude.
 16

b"!
=·~--------------------------------, ;<!::r-----------.
······... "
; '',,, + + ++
!l

. •• + +

•·..... ·
', + +
'+ +
'1 • •• + ...
il ......
'+ + +
~­
e~
••
'+,
+
+
+
•... +
5
.
~
·••
'+
+
••• +
+
+.
', + ""
"! ·•·····...
. ...
"
··....
i N

7
~
' "
o.o 0.4 0.1 0.11 1.0 1.2 Q~.-..-.--•.-.•~-.-..~-.~..~.--.~.~.---.~.•~~.-3-.~.~.3~.--~0.40
"' rads/sec rada/sec

LEGEND LEGEND
+ Poiabl calculated with DYIIANA o Unidirectional
----~P-~ . '?.1!!~~!-~~-----­
. 6 Cos20
+ CoalO
X Co16
• Coo2

Figure 5 Grid of points used for Figure 7 Spectra of surge drift

QTF calculations. force in unidirectional
and spread seas.

b~
~e,----------------------------------. ~:,-----------------------------,
'
:i" .\
/:'•
''
a
." ,.•'•'
,.
''''''
b

" ::
.~
''\f!\V'J!.... \j"\
' '
li : i !:

Vti~JuJ \{\
~
...
~:
:z;
"
oi
rl/~
··u .
~r --~

0
.." ~ ·. ' ~\-· ..\
"
"a+-~----~--~~~--~ "
0.00 0.05 0.10 0.111 0.20 o.u 0.30 0.35 .. ~.-.•-.--.~.~----.~.•-.---.-..•~--....
- --~----.-.3~.---.~.3.
....
rad.l/sec rad.l/sec

LEGEND LEGEND
~-~~~~P-~~------ ................................................................... . .
Clo1ed. form 1olutton
--~~~!1'.~------------------------------­
Cloled. form. aolution
···-~-~-~-~~-~~~~~~~-~~-~~--~~~~-~-~~-~~~~~--­ ·--·~-~-~':!~.~~~~':!~.~~.. -~~~~-~~~~~)

Figure 6 Spectra of drift forces in crossing seas a) surge; b) sway.

 •'1

--1J
Units o( vertical axis: m2 secfrads'­
-~~~~~~::::::::::::~.::::::::::::-.:~
~
-..]
0.0000.025 0.050 0.075 0.100 0.125 0.150 0.1~ 0.20(
CONTOUR PLOT ~+-------------------------------~
rads/sec
-
LEGEND a LEGEND b
c Unidirectional sea (meaaured) o Unidirectional sea (measured)
o Spread sea A (measured.) o Spre.ld sea A (meuured.)
.a. Spread. na 8 (measured) .a. Spread sea 8 (measured)

-~--l!~~~-~~-~~~
-~--~~~~-1!!~-~-(~~!~'------
-~--~~~-!!I!_J!_(~~~). ____ _

''-~R~.----"-----~~---c.~.----~~--~c.-----i
Direction degree!! Figure 9 Spectra of drift forces in spread and
unidirectional seas a) surge; b) sway.

Figure 8 Estimated directional spectrum .

 2
REFINED MODELLING OF HYDRODYNAMIC LOADS
ON DYNAMICALLY SENSITIVE STRUCTURES

OVE T. GUDMESTAD

Statoil, Box 300,

4001 Stavanger.

ABSTRACT
Methods for calculating hydrodynamic loads on dynamically sensitive
structures and the associated response of these structures are discussed.
Particular caution is raised against extrapolation of present design
practice to new types of dynamically sensitive structures. As research
is being focused on the parameters causing the largest uncertainty in the
loading, calculation of the design load must also be refined to include
dynamic and non-linear effects. Use of stochastic time domain
simulations is recommended. Examples from the calculation of loads on
dynamically sensitive jackets and jackups are given.

Notation

cd drag coefficient
Cm mass coefficient
du
dt :~ = water particle acceleration
D member diameter
d water depth
F force per unit length on member of structure
"max maximum wave height
Hs significant wave height
K Keulegan Carpenter Number
k wave number = 2 w /L
L wave length
Re Reynolds number
Tmax wave period of maximum wave
Ts significant wave period
Tz zero crossing period
u summation of water wave velocity v and current velocity
u current velocity
v water wave velocity
1.1 water surface elevation
p density of water
wave frequency
"'a.. dominating wave direction

DOI: 10.1201/9781003076599-2 19
 20

1. INTRODUCTION
Correct modelling of hydrodynamic loads on offshore structures is
required in order to ascertain their structural safety.

For '"small member'" structures traditional design codes are based on

Morison's equation, Stokes' fifth order regular wave theory, constant
hydrodynamic coefficients and deterministic analysis. As low oil prices
have necessitated the design of slimmer and more cost efficient structures
in general, and the installation of slender structures for deeper water in
particular, there is a need to accurately predict the hydrodynamic loads
on these structures. It must also be noted that these structures are
more sensitive to dynamic loads. Furthermore, there is an increased
interest in jackups for deeper water (e.g. North Sea conditions and 100 m
water depth) where dynamics are becoming important.

Through the extrapolation of existing design practice to deeper

water and/or the design of more dynamically sensitive structures, there
is, however, a danger that the inherent safety in the traditional way of
calculating the total force is lost. For these structures it is
therefore necessary to use the proper values of all parameters and to
include all important terms contributing to the total force.
Furthermore, of particular importance for these structures is the
load-history. An investigation of the safety aspects of the procedure
utilised for the calculation of hydrodynamic loads is found to be
necessary.

The calibration of the procedure must be done through careful

dynamic analysis of the relevant structures. Parameter studies aiming at
the identification of important parameters giving the largest uncertainty
in the calculation of the response are recommended. For this purpose
structural reliability analysis can be very useful. The conclusions of
the calibration and parameter studies will give guidance to the designers
and will help to define research needs.

To carry out a careful dynamic analysis of dynamic sensitive

non-linear drag dom.inated structures, i t is considered necessary to use
stochastic time domain simulations. This will provide the best possible
description of the irregular sea surface and non-linear near surface
effects, non-linear drag forcing terms and the most correct wave
kinematics model can be included. Furthermore, the time domain analysis
procedure must allow for consistent extrapolation of the non-linear
 21

non-Gaussian extreme dynamic response quantities to design values.

Special effects such as shielding(9) may be particularly important

for deep water and the effects of these should be considered where they
can be adequately documented.

2. JD)BLLIRG OF THE HYDRODYNAMIC LOAD

For the calculation of loads on slender offshore structures,
including jackets and jackups, the state of the art is represented by the
Morison formula(l). The force for unit length is given by
2 dU
F = 0.25 w D Cm dt + 0.5 D Cd u lui ( 2 .1)

Although this "established" formula has been challenged(2,3),

Morison's formula is still the only internationally recognised model for
the calculation of forces on slender offshore structures. Research
studies related to the inclusion of current velocity in the
formula ( 4 • 5 • 6) · have been carried out and this issue is not completely
solved, in particular for those cases where the Morison formula is being
linearised(7 ,8). In a time series analysis the full non-linear Morison
formula is normally being applied and assumptions related to linearisation
are not required. However, more research is required to completely
define the appropriate form of the Morison formula for the wave-current
situation. For structures with large movement the velocity u should be
replaced by the relative velocity u-x where x is the velocity of the
structure.

When calculating local forces on structures according to formula

(2.1), there are several parameters which must be determined. These are:

Wave condition represented by significant wave height Hs• wave

period Ts or Tz• and wave spectrum if a stochastic sea state
analysis is to be carried out. If a deterministic analysis is
performed the maximum wave height Hmax and associated period Tmax
must be given. For a dynamic structure the choice of dynamic load
factor must also be given.

Wave spreading models to be incorporated in the wave spectrum.

Wave kinematics models to determine water wave velocity v and

acceleration dv/dt.
 22

Value of current velocity associated with the defined wave

situation.

Wave scatter diagram representing load conditions associated with

fatigue damage. Furthermore, the wave scatter diagram is required
to determine a consistent set of (Hs, Ts) values for the extreme
wave analysis.

Marine growth thickness influencing member diameter D.

The hydrodynamic load coefficients Cd and Cm.

Other effects of importance such as shielding(9) and blocking

effects.

In estimating the total load on the platform, it is of crucial

importance to integrate local load effects over the full loaded part of
the platform, i.e. to the wave surface. This is either achieved in a
deter11inistic analysis or in a time domain stochastic analysis, while a
frequency domain analysis is restricted to include load effects below MWL
only(lO).

For the calculation of design loads, we cannot therefore give a

general recommendation to use the frequency domain analysis technique
although factors compensating for the missing inclusion of surface effects
can be introduced for static responding structures(11), e.g. by adjusting
the hydrodynamic coefficients. For dynamically sensitive structures,
however, the non-linear load caused by surface effects and also the
non-linear terms of the Morison formula (2 .1) cause resonant behaviour and
non-Gaussian response ( 12 l. It may thus be unsafe to start to introduce
compensating factors as there is no simple procedure established to
determine such compensating factors which would have had to depend upon
the degree of resonance of the platform.

Furthermore, it is considered unsafe to determine compensating

factors for one water depth, wave condition and type of structure and to
extrapolate these to other conditions. It is therefore recommended that
the calculation of the design load on dynamically sensitive structures
should use a simulated surface and solve the equation of motion in the
time domain(10,13).

In order to have a clear understanding of the non-linearities, the

dynamics and the non-Gaussian contributions, and to control the results of
 23

the time domain analysis, it is also recommended that a deterministic

analysis be carried out by stepping a deterministic design wave through
the structure.

For fatigue calculations, where the load range is of major

importance, the deficiency of using the frequency analysis technique is
not as severe as for the ultimate state analysis. Furthermore, the
non-linear resonances at multiples of the structure's frequency are not as
strong for the lower wave conditions since the load for lower sea states
generally is mass dominated rather than drag dominated. This analysis
method may therefore be utilised for fatigue analysis(14).

Having established that a stochastic time domain analysis (TDA) is

required to most accurately determine the hydrodynamic load on dynamically
sensitive structures, some discussion on the possible limitations of the
TDA is necessary:

This method is well established to treat simple 2D structural models

while a TDA analysis of large 3D models of offshore structures is
considered by many to be too costly. The argument is in our
opinion not sufficiently strong to discard the accurate
determination of the load, as parameter studies can be performed on
simple 2D models. Furthermore, efficient programming techniques
and powerful computers are becoming available to perform the
calculations.

The determination of the sea state condition for the time domain
simulation.

Specific wave kinematic models are required to handle the wave

kinematics above the MWL in a TDA ( 10), as the linear Airy theory is
not valid above the MWL for high frequency waves(15). Several wave
kinematic models have been proposed which can solve this problem.
However, none of these models satisfy the governing hydrodynamic law
and the boundary condition. Statoil has therefore, in
co-operation with Norwegian Hydrodynamic Laboratories and other oil
companies, carried out an ambitious wave kinematics measurement
programme(17). Preliminary indications are that the Wheeler wave
kinematics model(16) seems to give a good comparison with tank data
for irregular waves. Gudmestad' s model ( 15) does also seem
promising although it is more time consuming than the simpler
 24

Wheeler model given by:

H cosh k !Z'+d)
v =r-
11
. cos (kx-wt) (2. 2)
sinh kd
d
where Z' = (z-n)
d+n
for each frequency w

Further data from measurements in the open sea are, however,

required to finally determine the best wave kinematic model.

The determination of the design load level from a stochastic time

domain analysis is not straight forward and a sufficient number of
load maxima must be calculated to treat the maxima
statistically(l3). Of major importance is the identification of
probabilistic models to flt the maxima. Evaluation of the
stability of the design load predictions using different simulation
time lengths and different time series for the simulation must be
performed.

Several studies(10) have documented that a 3 parameter Weibull

parameter distribution could be used for estimating response maxima
and minima for drag dominated structures. Furthermore, the
stability of the method has been carefully checked and found to be
satisfactory for a jackup applied in deep water, (See Fig. 2.1).
Referring to this figure, a time series of about 3000 s seems
appropriate.

Having selected the analysis method and wave kinematics model,

noting the importance of non-linearities in the loading mechanisms (free
surface effects and non-linear Morison force), there are still important
parameters to be determined prior to calculating the hydrodynamic load.
These are the more tradi tiona! parameters which also must be determined
for a deterministic analysis.

The wave condition represents a most important contribution to the

uncertainty in the load estimate ( 18). Although some part of this
uncertainty relates to the inherent variability of the waves,
uncertainties due to lack of data and refined models for prediction
of extremes, etc. can be reduced through future research. The
potential for improvement should be quantified.

The wave spreading model is to be given attention and conservatism

 25

~0
0.010
•• LS

[
~
>
0.058-1
~.
~ t.Oj
j
.­
1,...,. .
I I ell I '
/ ........ _·-·-·----... -1--c:
.......... ,._
I
I
'
I
,...... _____
0.056-1 {
I
\ a.S I

--------­
I

r
I

o.~ 0.0
0 I
2000 I
~ I
6000 ,_,'
11000 a zOOo ~000 6000 11000

·r'L
0.08 8
§
-·-·-..
~
: ...­ -...__________ -tl--· •0
~ ~ / ...... ____ ...._.... _...
0.07-1 "' .. A­

i
I
I ' "
i
I
I
o.orH: ~ I

:
I
I
I

~ O.OSi------;;------r-----,,-----~ 24-----r---~r---~----~
0 2000 ~000 6000 ( _,
8000 a 2000 ~000 6000 ,_,
11000
r:..,.,. T~-

Deck Displacement

i. tOE~ .Y..-__Jl..-_ _ L_ _ .L__-+

•• -0.2

i
-' I
~ \ ,-*--it_...~\. I
-D.4
-~· IS£~ I 'f, ,X' 'x--~--1<
ill
I
I
\ I ''~<
-0.6
\ )f.-M--><-->e--M-~-*-.-1(
\ ,'
I ,
I ,
-t .2lE-o4 \ I \ ~X
II -o.8 II ''

* -t.O
1(
-1~~-fr-----r----~-----,----~
0 2000 ~ 6000 ,_,
11000 0 2000 ~ 6000 ,_,
11000

c5.0E~
~ I 'y'~ ....... _'"'*'_M-...1('-~--M l< •j
84-----~--~----~----~

..- I !
' ~
D I
...> 6 ll.,
'2.5E~ :
'
" I
:{ 'x.. ...,.._"M-_"it... ..oc--)(·-·-·-~
11 ' 4

.L/--.-----r---::::r:~
I

t
I
I
I

U:z.OE~ 24-----r---~r---~----~
0 2000 ~ 6000 11000 0 2000 4000 6000 ,_,
IDIID
r:... l'­

Overturning moment

Figure 2.1 Stability of statistical parameters with simulation

length of time series.
 26

is urged pending good data from 3D measurements in open sea. Lack

of sufficient data on spreading may be a source for over design, and
the Petroleum Directorate (19) suggest that the following spreading
function can be used for sea states with a significant wave height
less than 10 m:

D (9m - 9) = C Cos 0 (9m - 9) (2.3)

where 9m is the dominating wave direction and n is chosen between 2
and 8 in the most unfavourable way for the structure.

Careful definition of the current velocity to be combined with the

design sea state condition is required for drag dominated structures
where current plus wave loading is being squared(20). Of
particular concern is the recurrence level for the current to be
associated with the design wave. NPD(19) recommends that a 10 year
current is utilised together with a 100 year design wave condition.
This combination may be relatively conservative and statistical
analysis of wave-current data should be performed to determine the
appropriate combination of wave and current(18).

The assumed marine growth thickness is a critical parameter since

marine growth will significantly increase the member dimension and
the associated hydrodynamic forces on slender offshore structures
(e.g. 0.1 m marine growth will increase the effective diameter of a
0.8 m member by 25%). The cost of specifying large marine growth
thickness must be carefully weighed against the cost of cleaning
excessive marine growth. For the Veslefrikk platform, Statoil has
introduced an antifouling coating(21) to limit the marine growth and
the hydrodynamic load. For jackup structures it may be advisable
to clean the members during the regular inspection intervals.

Determination of hydrodynamic drag and mass coefficient has been

given much attention(1,22) over many years. We have focused on the
measurement of wave kinematics(17) and consider it essential to link
estimates of hydrodynamic coefficients to measured values of forces
and kinematics. Since conservative wave kinematics models are
traditionally used in deterministic analysis(23,24), there might
have been a tendency to use too low values of Cd in such analysis.
As compensating factors(11) cannot be extrapolated to slender
 27

structures and deep water, the most accurate models and values must
be used for both kinematics and hydrodynamic coefficientsl22).
Although a Cd value of 0.7 to 0.75 traditionally has been used for
the calculation of design forces on slightly fouled members(22),
indications are that the most proper value is closer to 0.9(11,22),
for those values of Keulegan - Carpenter number K and Reynolds
number Re relevant for the design wave situation. This value
should therefore be used in a TDA. It should be noted that this
value is also equal to the adjusted value proposed by Olsen and
Torseth(11) for the calculation of load using frequency domain
analysis.

For the mass coefficient, an opposite tendency to allow reduction

compared to the traditional value of 2.0 exists and a value of 1.8
for non-linear stochastic analysis is suggested, Fig. 2.2 from ref.
22). Although these values may well be discussed, and although
sufficient research is lacking to fully document the values, we find
it necessary to select these most proper values while supporting
continued research.

The drag coefficient is uncertain due both to lack of representative

measurements and to the fact that the drag force is being produced
when wakes are released behind structural members making the drag
forcing a stochastic process. The variability of the drag
coefficient has been reviewed by several authorsl22,25). If a
probabilistic analysis be carried out, it is suggested that a normal
distribution for the drag coefficient be utilised having a mean of
0.90 and a standard deviation of 0.15.

It has been suggested that the drag and inertia coefficients be

varied with the flow conditions. i.e. that a different Cd and Cm be
used, as the Reynolds and Keulegan Carpenter numbers have different
values in the flow( 26). Although this might be possible in a
deterministic sea condition, the lack of consistent measurements and
good data and the uncertainties involved in determining the
hydrodynamic coefficients has led us to conclude that it is
unrealistic to introduce variable hydrodynamic coefficients.
Furthermore, as the definition of the Reynolds number and Keulegan
Carpenter number are linked to deterministic wave conditions, the
definition of these numbers for a stochastic wave condition is not
 28

0
~

.....'·
1.4
1.2
.1.1
Cd
<> 00
:::r~oo
1.0 0000
1.0
Re: 150000

.... 1.1
1.•

1.2
<>
oo<>
'[ 00 1.0

.
0
·' ~~0

~· ..
K K

~-~-::--::---:o~--':o-o~o~o"-=o~o~o~o-.-o
= : : s ~ i ~~
1
.
.,. N ., ot ill • ,._ e •2 • • ~ =: ~

Fig. 2.2 Drag & inter5ia coeff for smooth cylinders. Re

(Fig. 9 2.11-2.42<10 .
of ref. 21)
Full curves: Results given in Fig. 13 of Ref. 21.

Diamonds : CBT results. NOTE:

A strong cross current is present and
data may not be relevant for offshore
structure.
Circles OTS results.

The Veslefrikk Jacket Structure Arrangement of deck

Figure 3.1.
The Veslefrikk Jacket
 29

straight forward. The use of coefficients varying continuously in

a stochastic sea may appear particularly unrealistic as the size of
the wake behind a member may not be linked to the full frequency
band of the wave condi lion, For a stochastic fatigue analysis,
however, values of Cd = 1. 0 and Cm = 2. 0 may be more adequate (lower
Re and K numbers) than those recommended for the ultimate design
analysis.

In conclusion, it is recommended that time-domain stochastic

analysis (TDA) be used for the calculation of loads on slender offshore
structures. The traditional design parameters must be selected in a
consistent manner to reflect the best estimate for each of the parameters.

Having adapted this approach to estimating the hydrodynamic loading,

it is recommended that other features which can be documented to give a
more correct estimate of the loading be introduced. Of particular im­
portance are shielding effects I 19 I which can give considerable reduction
in loadingl21l. However, it must be noted that the distorted flow
pattern around conductors will change the flow around the platform and
increase the load on some members in the structure. It is of major
importance to treat the complete platform-riser-conductor system
consistently when calculating shielding effect.

3. EXAMPLE CALCULATION
a) Calculation of the environmental load on the relatively dynamic
sensitive Veslefrikk jacket structure (Fig. 3.1) has been presented
elsewherel10,13,25). This platforml2l I was installed by Statoil
during summer of 1989 in a 175 m water depth in the Northern North
Sea. The first natural period is estimated as 3.3 s. For base shear
and overturning moment an 18% dynamic amplification was found using
a TDA. Furthermore, different wave kinematics models gave a differ­
ence of about 20% in base shear and OTM with the constant horizon­
tal velocity value above mean water level (Vertically Extrapolated
Airy Model) (Fig. 3.2) being more conservative over the stretched
linear velocity profile (the Wheeler model(16) and the Gudmestad< 15 1
model). A remarkable increase in design load incorporating the
design current at the Veslefrikk field should be particularly noted.
Further research into application of direct summation of wave
particle velocity and current in a time domain stochastic dynamic
response analysis (TDA) is considered necessary. (See also Ref. 10)
 30

TT~,., Extension of linear

Airy wave up to the
free surface

B) Stretching of the
L____f:{j'i '>"<: 1 7 MSL linear velocity
profile to the free
surface {velocity at
bottom equal to the
velocity found by the
linear theory}
Wheeler (16).

TT .
.
'
MSL C) Constant horizontal
velocity above the
MSL .

Illustration of the horizontal fluid velocity near the

mean sea level for different representations of the
field.
Velocity profile according to the Airy theory.

Fig. 3.2 Vertically Stretched Airy Wave Kinematics (B) and

Vertically Extrapolated Airy Wave Kinematics (C)

WATER I:EI'TH 110m.

Fig. 3.3. General view of TPG 500 Jackup

 31

b) Some parametric studies reflecting the hydrodynamic load on a

dynamically sensitive jackup structure will be discussed (see also
Ref. 27). A design for relatively deep water (the TPG 500) has
been taken as an example (Fig. 3.3). The two first natural periods
of this jackup in 110 m water depth have been estimated as 5.16 s.
The damping has been set equal to 2% Rayleigh damping plus
hydrodynamic damping caused by relative velocity.

The dynamic response is clearly non-Gaussian (Fig. 3.4) with higher

extremes than in a Gaussian process. The sources of this high
non-linearity are drag forces and the surface effects (integration
to actual surface). The wave and current induced response is thus
strongly influenced by the dynamic resonance effects caused by
amplified superharmonic load componentsl12). Figure 3.5 gives the
response spectra for the wave alone case.

The results show that the variation in loading when varying the
different parameters changes for different parts of the jackup legs
as the loads locally are dependent upon the degree of local effects
of dynamics.

Variation of the spectral peak parameter by ~ 2 s around the base

case value of 14.5 s results in a load varying between 80 and 110%.

Increasing the significant wave height by 1 m from the base case

value of 14.5 m causes an increase in extreme response between 0 and
14% (Table 3.1).

Different wave kinematics models give even larger variations (Table

3. 2). The vertically extrapolated Airy wave model gives the most
non--Gaussian response and the highest extreme between 30% and 55%
higher than for the Wheelerl16) model. The Gudmestad model (15)
gives a slightly lower extreme response than the Wheeler model.

Different wave-current interaction models for the wave crest I 28 l

have been found to give relatively small changes in the loading.
This confirms the findings of ref. (29).

The limited parameter study of forces on a dynamic sensitive jackup

has confirmed the importance of introducing non-linear forcing
terms, resonance response terms and treating the response as
non-Gaussian in a stochastic time domain analysis. Of particular
importance for the results is the indentification of wave climate
 32

1:: WW WWPcr..,t. ....-~..­ W WWW Pa-enl ..,..~,_

I
J
--......,,~
--GcMAeLon
- - - aa..-1.01'1 fLl - - - ac....I.Cri
f1..l

0.9999 J •
0,9999
o.ssoo • 0.9990~
• •• •
0.9900 ••••••• 0.""""~ ••
0.9000 0.90001
0.5000
0.5000

0,\000 o.moo

0.0100 0.0100
f1.00\0 0.00\0
Cl' JOOI 0.0001
-<t -3 -2 -1 t2!"t56l
-3 -2 -t Ot23"t56
ST. DEV. CF PIIRENT SERIES
ST. OEV. OF PARENT SERIES

Parent distribution Parent distribution

1::

I.....
• • • •tbc..-a ...."" ~ • • • ll'l'bc....a ..,..""
- ­ .......<4> J

I
- - - Vel.bull-1 f1..l
- - ....... <4>
- - - 'lle-.b.JLL-1 f Ll

··­ ··­
0.8889

D.!ISDD -· D.!ISDD -•-·

o.aaaa
....... ...... /

,,""·
JtE
o.taao 0,6000 /

0.1000 I.

-· 2 a " s a
ST • DEV , CF PARENT SEA I ES
1
D.OtDD 'f'--,--,---y-----,,---.---f
a .,. s
ST. lEV. CF PARENT SERIES
s

Maxima distribution
Maxima distribution
Base Shear Overturning Moment

Fig. 3.4 Sample probability distributions

.
m+----~----~---~ ~~
~~ 2IDDJ

1!­
~­ ~­
~i
~~ 2ll

•• tDDDDD

......
tO

o.t o.z
...........,....
... ... ...
AEIIEICY
Base Shear
Overturninq MomPnt

Fig. 3.5. T 17.5 sec - Dynamic response spectra

p
 33

Table 3.1 Effect on the jacK-up of variation in significant

wave height (no current) •

Mean (~) St. Dev. (0) Ext. Reap.

Response
Quantities Base Hs Base Hs Base Hs
14.5 15.5 14.5 15.5 14.5 15.5

Deck displacement 1.000 1. 059 1. 000 1.100 1. 000 1. 076

Base Shear 1. 000 1. 055 1. 000 1.106 1.000 1.106

Overturning moment 1 .000 1. 063 1. 000 1. 092 1. 000 1. 040

Top of leg
Shear 1. 000 1. 098 1. nun 1. 062 1. uon 1. 005
Bending moment 1. 000 1. 064 1 nno 1 .089 1. 000 1. 046

A.t.....ll!iL
Shear 1. 000 1. 109 I. 000 1. 072 I. 000 1. 032
Bending moment 1 .000 1. 061 1. 000 1. 110 1. 000 1.140
-

The result~ are nor•alized with the base case results

as basis

Table 3.2 Effect on the jack-up of variation in wave model

(base case seastate, no current)
VEA = Vertically extrapolated Airy model
VSA =Vertically stretched Airy model - Wheeler(16)
Gudmestad = Gudmestad model (15)

Mean (p) Standard dev. (0) Extrem~ Response

Response
Quantities Base V.E.A Gudme Base V.E.A Gud11e Base V.E.A Gudme
V.S.A stad V.S.A stad V.S.A stad

Deck displacement 1.000 1. 214 1. 093 1. 000 1. 350 0. 997 1.000 I . 4111 0. 935

Base Shear 1.000 1. 140 1.073 1. 000 1. 217 0. 995 1.000 1 . 3 25 0. 905

Overturning moment 1. 000 1. 187 1. 099 1. 000 1. 396 0. 993 1. 000 1. 411 0. 934

Top gf leg
Shear 1.000 1. 863 1.111), 1. 000 1. 370 0. 990 1. 000 1. 37 i 0 9 JS
Bending moment 1.000 1. 246 I .llQ9 I .000 1. 388 0. 993 1. 000 I . 4 4I 0. 9 JO

A.t.....ll!iL
Shear 1.000 1. 390 1. 107 1.000 1. 401 0. 985 1.000 I. 551 0. 9?.5
Bending moaent 1.000 1. 208 1. 099 1.000 1.311 1.003 1 .000 1. 363 0. 94.3

The results are normalized with the base case results as basis
 34

and selection of proper wave kinematics model.

To systematise further the information about uncertainties, it is

recommended that structural reliability methods be applied(18).

•. DISCUSSION
Through a careful discussion of non-linear loading terms and
resonant response, and through a brief summary of example calculations it
has been demonstrated that these effects play a major role in determining
the load on dynamically sensitive offshore structures. It has
furthermore been stressed that the load is non-Gaussian. A time domain
stochastic analysis (TDA) load is non-Gaussian. A time domain
stochastic analysis (TDA) is required to cover all these aspects of
loading, response and estimate of extreme values.

It has furthermore been demonstrated that it is very important to

select a correct wave kinematics model in a TDA. The process of
estimating design extremes in a TDA is found to be stable.

The wave parameters (height/period) and the combined wave-current

situation must be defined most carefully as the uncertainty in these
parameters contributes most significantly(18).

The traditional hydrodynamic parameters must be selected

consistently and it is furthermore discussed in the paper that the
extrapolation of present procedures to dynamically sensitive structures
must not take place. Each parameter must be appropriately selected to
arrive at a safe structural design.

A similar review of the effect of damping should be carried out,

both the structural, geotechnical and hydrodynamical damping must be
critically assessed.

5. ACKNOWLEDGEMENT
The author acknowledges Statoil for permission to publish this
paper. Of particular assistance has been the valuable discussion with H.
Nordal and s. Haver of Statoil. The research co-operation with Sintef's
N. Spidsoe and D. Karunakaran is particularly appreciated.
 35

REFERENCES

I. SARPKAY A, T. and ISAACSON, M.: 'Mechanics of Wave Forces on

Offshore Structures', van Nostrand Reinhold Company, 1981.

2. BENDAT, J.S. and PIERSOL, A.G.: 'Spectral Analysis of Non-linear Systems

Involving Square-law Operations', J. Sound Vibr, 81 (2) 199-213, 1982.

3. HORTON, T.E. and FEIFAREK, MJ.: 'The Inertial Pressure Concept for
Determining the Wave Forces on Submerged Bodies', J. of Energy
Resources Technology Vol. 104 pp 47-52, 1982.

4. VERLY, R. and MOE, G.: The Forces on a Cylinder Oscillating in a

Current', River and Harbour Laboratory, The Norw. Institute of
Technology, Report No. STF 60A 79061, 1979.

5. LAY A, E.J., CONNOR, J.J. and SHY AM SUNDER, S.: 'Hydrodynamic

Forces on Flexible Offshore Structures', ASCE Journal of Engineering
Mechanics, Vol. 110, No. 3, pp 433-448, March 1984.

6. American Petroleum Institute: Recommended Practice for Planning,

Designing and Constructing Fixed Offshore Platforms. API RP2A, 17.
Edition, April I, 1989.

7. GUDMESTAD, O.T. and CONNOR, J.J.: 'Linearisation Methods and the

Influence of Current on the Non-linear Hydrodynamic Drag Force',
Applied Ocean Research, Vol. 5, No. 4 pp 184-194, 1983.

8a. LIPSETT, A.W.: 'A Pertubation Solution for Non-linear Structural Responses
to Oscillatory Flow', Applied Ocean Research, Vol. 8, No. 4, 1987.

8b. GUDMESTAD, O.T.: 'Effective Damping Applying a Morison Type of

Forcing Function', Applied Ocean Research, Vol. 10, No. I, 1989 (in
response to ref (8a)).

9. JUSTESEN, P., OTTESEN HANSEN, N.E. and LYNGBERG, B.: 'The Role
of Hydro Elastic Vibrations in Marine Risers', Proc. Offshore Mechanics
and Arctic Engineering, OMAE pp 417-425, Tokyo, 1986.

10. GUDMESTAD, O.T., SPIDSOE, N. and KARUNAKARAN, D.: 'Wave

Loading on Dynamic Sensitive Offshore Structures', OMAE 90-315. To
appear in Proc. Offshore Mechanics and Arctic Engineering, Houston, Feb.
1990.

II. OLSEN, O.A. and TORSET, O.P.: 'Hydrodynamic Loading on Jackets.

Application of Linear Stochastic Methods', pp 19-37 in 'Integrity of
Offshore Structures' ed. by Faulkner, Cowling and Frieze. Applied Science
Publishers, 1982.

12. SPIDSOE, N. and KARUNAKARAN, D.: 'Effects of Superharmonic

Excitation to the Dynamic Response of Offshore Platforms', E&P Forum
Workshop on Wave and Current Kinematics and Loading, IFP, Paris, Oct.
25-26, 1989.

13. GUDMESTAD, O.T. and SPIDSOE, N.: 'Deepwater Wave Kinematics Models
for Deterministic and Stochastic Analysis of Drag Dominated Structures',
Proc. NATO-ARW Water Wave Kinematics, Molde, May 1989. Ed. A Torum
and O.T. Gudmestad. To be published by Kluver Academics, 1990.
 36

14. MOMMAS, C.J. and DEDDEN, W.: 'The Development of a Stochastic

Non-linear and Dynamic Jack-up Design and Analysis Method', Proc. 2nd
Inti. Conf. 'The Jack-up Drilling Platform', London, Sept. 1989.

15. GUDMESTAD, O.T.: 'A New Approach for Estimating Irregular Deep Water
Wave Kinematics'. To be published in Applied Ocean Research, Jan. 1990.

16. WHEELER, J.D.: 'Method for Calculating Forces Produced by Irregular

Waves', Journal of Petroleum Technology, pp 359-367, March 1970.

17. SKJELBREIA, J., TORUM, A., BEREK, E., GUDMESTAD, O.T.,

HEIDEMANN, J. and SPIDSOE, N.: 'Laboratory Measurements of Regular
and Irregular Wave Kinematics', E&P Forum Workshop on Wave and
Current Kinematics and Loading, IFP, Paris, Oct. 25-26, 1989.

18. WINTERSTEIN, S.R. and HAVER, S.: 'Statistical Uncertainties in Wave

Heights and Combined Loads on Offshore Structures', Paper OMAE 90-554,
to appear in Proc. Offshore Mechanics and Arctic Engineering, Houston,
Feb. 1990.

19. Norwegian Petroleum Directorate: Guidelines for the determination of load

and load effects, NPD, Stavanger, Norway, 1989.

20. VY AS, Y.K., LOHRMANN, A., HEIDEMANN, J.C., DAHL, F.E. and
NERMERSCH, J.A.: 'Storm Driven Current Profiles for Design of Offshore
Platforms', Proc. Behaviour of Offshore Structures, pp 519-534, Tapir,
Trondheim, 1988.

21. BAERHEIM, M and FOSSAN, T.l.: 'Weight Optimization of the Veslefrikk

Jacket', OTC 6189, pp 689-700, Proc. Offshore Technology Conference,
Houston, May 1989.

22. MOE, G. and OVERVIK, T.: 'The Use of the Morison Equation - A
Review of Field Measurements', Proc. E&P Forum Workshop on Wave and
Current Kinematics and Loading, IFP, Paris, Oct. 25-26, 1989.

23. GUDMESTAD, O.T., JOHNSEN, J.M, SKJELBREIA, J. and TORUM, A.:

'Regular Water Wave Kinematics', Proc. Int. Conf. on Behaviour of
Offshore Structures, Vol. 2, pp 789-804, Trondheim, June 1988.

24. SOBEY, R.J.: 'Wave Theory Predictions of Crest Kinematics', Proc.

NATO-ARW Water Wave Kinematics, Molde, May 1989. Ed. A. Torum and
O.T. Gudmestad. To be published by Kluver Academics, 1990.

25. OLUFSEN, A.: 'Uncertainty and Reliability Analysis of Fixed Offshore

Structures', PhD report, Trondheim, Norway, 1989.

26. SHY AM SUNDER, S. and CONNOR, J.J.: 'Sensitivity Analysis of Steel

Jacket Offshore Platforms', Applied Ocean Research, Vol. 3, No. I, 1981.

27. KJEOY, H., BOE, N.G. and HYSING, T.: 'Extreme Response Analysis of
Jack-up Platforms', Proc. 2nd Int. Conference 'The Jack-up Drilling
Platform', London, Sept. 1989.

28. BARLTROP, N.D.P., MITCHELL, G.M. and ATKINS, J.B.: 'Fluid Loading
on Fixed Offshore Structures', Department of Energy, October, 1987.
 37

29. DALRYMPLE, R.A. and HEIDEMANN, J.C.: 'Non-linear Water Waves on a

Vertically Sheared Current', E&P Forum Workshop on Wave and Current
Kinematics and Loading, IFP, Paris, Oct. 25-26, 1989.
 3
ON THE MODELLING OF UNCERTAINTIES
ASSOCIATED WITH THE OCEAN ENVIRONMENT
AND THEIR INFLUENCE ON RESPONSE EVALUATIONS

K. VENKATARAMANA
Sumitomo Metal Industries,
Hasakimachi, Kashima, Ibarakiken 314-02, Japan

K.KAWANO
Dept. of Ocean Civil Eng., Kagoshima University,
Korimoto 1-21-40, Kagoshima 890, Japan

Y. YAMADA and H. IEMURA

Dept. of Civil Eng., Kyoto University,
Kyoto City 606, Japan

ABSTRACT

The exploration and exploitation of resources from sea beds at ever-increasing

depths have necessitated the re-evaluation of the design techniques for offshore structures.
The mechanisms involved in the loading process in the ocean environment are extremely
complicated and difficult for exact formulation. Even those which can be described
precisely present numerical difficulties in their computer implementation. To minimize
the computer time and to obtain a reasonably accurate result, a simplified description of
the complex problems is usually assumed. This paper is devoted to investigations on the
effects of those simplified descriptions on the response evaluations. Parametric studies
based on the perturbation method are presented to assess the influence of uncertainties
associated with the various assumptions and approximations. It is found that among
all parameters considered in the present study, variations in the wave height provide the
most significant contributions to the dynamic response. The random variable effects of
the inertia coefficient are larger for smaller wave heights and shorter wave periods whereas
those of the drag coefficient become dominant for larger wave heights and longer wave
periods. The effects of the randomness of the shea.r wave velocity in the soil are larger
when the mean wave period approaches the fundamental period of the structure due to
the increased soil-structure interaction effects.

DOI: 10.1201/9781003076599-3 39
 40

1. INTRODUCTION

The analysis and design of offshore structures is a complicated process and

requires several assumptions and approximations. Empirical models such as the Morison
equation for the wave force or the expressions for the wave energy such as the
Bretschneider's spectrum involve several coefficients which have to be determined by
experiments or field observations and have to be extrapolated to a particular problem.
The hydrodynamic interaction effect between the legs of the platforms is another
area of uncertainty and depends on multiple factors such as the characteristics of the
structures, water depth, distance between structures and the direction of ground motion,
and so on. The methods of response analysis, linear or nonlinear are also a source
of considerable debate. Linear methods are efficient and yield better results from a
probabilistic sense. On the other hand, fatigue damage and life of a structure are much
dependent on the nonlinearities in the wave, structure and soil. In practical design
methods, based on frequency-domain analysis, the nonlinear forces are replaced by their
linearized equivalents. When the structural velocities are larger, responses based on
these linearization methods may be appreciably smaller than their true values(!) and
both fatigue damage and extreme response can significantly exceed those predicted by
linear methods. Nonlinear dynamic analysis of flexible structures using perturbation
formulation< 2 ) or iterative approach< 3 ) are found to be quite efficient but their performance
for soil-structure interaction situation is not yet clear.

Some of the above problems are difficult for exact formulation. Those
problems which can be described precisely present numerical difficulties in their computer
implementation. Therefore some researchers do not recommend exact formulation of all
connected problems. Exact expression of some problems and approximate description
of the others may cause the structural responses to be approximate. To minimize the
computer time and to obtain a reasonably accurate result, a simplified description of
the complex problems mentioned above is usually assumed. This paper is devoted to
investigations on the effects of those simplified descriptions on the response evaluations.
Parametric studies based on the perturbation technique are presented to assess the
influence of various uncertainties associated with various assumptions and formulations of
dynamic response analysis method. In particular, random variable effects of the following
parameters are considered for the study:
• Inertia and drag coefficients of the Morison equation: These coefficients usually
determined by laboratory experiments or field observations exhibit large scatter
and it is practically impossible to predict the exact values of these coefficients for
design purposes.
 41

• Wave height distributions: By just looking at the real sea surface, one sees that the
surface consists of a variety of waves moving in different directions with different
frequencies. While designing a structure, it is required to represent these waves using
an equivalent wave height such as significant wave height or mean wave height or rms
wave height. However, all these representations are only approximate because the
wave height distributions vary from place to place and from time to time depending
on fetch, wind velocity, duration of the wind flow etc.

• Properties of the subsoil: A structure excited by dynamic loads interacts with

the surrounding soil. The response of the structure is strongly influenced by the
dynamic properties of the subsoil. The subsoil properties are random in nature and
are functions of multiple factors such as the type of soil and the conditions of the
site, thickness of layers in the soil strata, lateral and axial loading characteristics of
offshore environment and so on.

To perform the parametric studies, the above parameters as well as the structural
velocities are expanded in a power series about their mean values. A set of dynamic
equations are formulated in terms of these random variables and are solved to obtain
response spectral densities and rms responses. Numerical simulations are carried out to
assess the suitability of the proposed formulations.

2. THE GOVERNING EQUATION OF MOTION

Fig.l shows an offshore structure model considered for study. The structure is
discretized using the finite element method and the pile-soil foundation is modelled using
impedance functions. The governing equation of motion is obtained by the substructure
method. Using the linearized Morison equation for the wave forcing funtion, the equation
of motion in the general form may be expressed as:

([MJ + [I<mlJ {ii} + [[c] +[I< DJ] {u} +[I<]{ u} = [I<M]{ v} +[I< n]{ v} (1)
where

[I<m] ['p(CM- l)V,J [Kn] = ['~pCn4crr'\,]

[J<M] ['PCMV'.] r = v- u
in which [M] is the lumped mass matrix, [CJ is the structural damping matrix, [I<] is
the stiffness matrix, { u} is the displacement vector, { v} and { v} are the horizontal water
partcile velocity and acceleration respectively at the undeflected structure coordinate
locations, pis the mass density of water, V is the enclosed volume with respect to flow,
 42

A is the area projected in the direction of flow, CM is the inertia coefficient and Cn is the
drag coefficient.

The equation of motion of the total system can be obtained by the substructure
method in which the structure-pile-soil system is separated into two substructures:
the structure and the pile-soil foundation. The structural displacements consist of
dynamic displacements of the structure subsystem and the quasi-static displacements
associated with the interaction displacements at the structure-soil interface. The
impedance functions for the pile-soil foundation system are determined separately and
incorporated into the governing equations of motion for the total system. The dynamic
displacements of the structure are treated as the linear combination of the first few
vibration modes for the rigidly supported base condition which have significant effects
on the response. The quasi-static displacements are expressed with the superposition of
a few generalized displacements. Thus, dividing the structural displacements into two
parts, the substructure mode synthesis is possible and the computational time is greatly
reduced.

l l1f l
The governing equations of motion of the total system are finally expressed as< 4 l:

[
(I] (~ap]l f {ij} +[ ['\.2,8jjWJj"\,] 0 {q}
[Mpa] [Mp] 1{up} o (cp] {up}

+ [ ('wJJ''J ~ l J l ]l J l
{q} = [ (P.. {ti.. } (2)
0 [Kp] 1 1
{up} [Pb] {v.. }

where

[
(Pa]l
[Pb]
[
(<I>jT(KM]

[GjT[L]T[KM] [GjT[LJT(Kn]
[<I>jT(I< n] l
[M] = [M] + [Km], {u~} = (<I>]{q}

in which suffix a denotes the unrestrained nodal points above the base and suffix b denotes
the restrained nodal point at the base, suffix p denotes the pile-soil foundation, (I] is the
unit matrix, (L] is the quasi-static transformation matrix, (G] is the matrix connecting the
displacements at the base nodes and at the center of gravity of pile-soil foundation, {u~}
is the dynamic displacement of the structure for the rigidly supported base condition,
 43

[<I>] is the undamped eigen vector and {q} is the corresponding generalized displacement
vector for the rigidly supported base condition, Wfi is the natural frequency and f3!i
is the corresponding damping ratio which includes both the structural damping and
the hydrodynamic damping for the rigidly supported base condition, and superscript
T denotes the transpose of a matrix. (In the above equation, offdiagonal terms of the
nonproportional damping matrix are ommitted assuming that the lower vibration modes
have well-separated natural frequencies< 5 l.)

3. BASIC FORMULATIONS FOR REPRESENTING THE RANDOM

VARIABLE PARAMETERS

(i) Hydrodynamic coefficients of the Morison equation CM and Cv: Using perturbation
technique and assuming that each coefficient matrix can be separated into one part
containing constant coefficients and a second part with randomly varying coefficients
with zero-mean:

[I<m] [K~'I + "[K~'I )

[J<M] [I<~)]+ <t[I<~)l (3)

[Kv] [1?}~)] + <2[~)]

with,

[J<~l] = ['-p(C~l- I)V,] [J<~l] f1KM._

acM = [' Pv,]
[I<~)] ['-pc~lv ,] [I<~)] ['-pV,]

[1?10)] ['-j!pc_g>l tcr,,] [I<11)] ['-/!pfcr,,]

and E1 and E2 are defined such that:

E[Et] = 0' E[<?J = <6M

E[<2] = 0' E[<iJ = <6D

Further, each matrix of ' 0 ' suffix stands for the mean value. For example, [I<,}.0 l] = E[I<m]·

(ii) Mean wave height H: When the wave energy spectrum is expressed using the
Bretschneider's formula, the water particle velocity spectrum has the form:
 44

2
s.,.,(w) IFi(z, w) I2 S~~(w) = H2 )
IF1(z, w)l 2 a1 ( gT
2
1 )
g 2 exp { -b1 ( Tw
w
4
}

F,(z, w)H 2 (4)

Using perturbation method,

s.,.,(w) = s~~~,(w) + E3s~:~,(w) = s~~~,(w) + 2E3F,(z,w)H

2
S(O)
VJVJ
.(w) + E3 =
H S(O)
tiJVJ
.(w) (5)

where
E[<3] = 0, E[<i] =E.:..
H

(iii) Shear wave velocity in the soil Vs: Expanding the impedance functions of the pile-soil
foundation system using perturbation method, the following expressions may be obtained:

[I<p] [J<P(O)l + E4 [J<P(l)l )

(6)
(CvJ [cp<oJ] + <4 [cv<ll]

l l
where

[J<p(l)l = 1_ [ 2J<x~) (0)

]{x8
[cv<ll] = _1_ [ 2Cx~l cx8(0)
y(O)
s ]{ (0) v:s(0) c<o) lc<oJ
8x V<8~o) 8x 2 88

E[<4] = 0, E [< 42] = <Js

Thus the variations in the wave force parameters are expressed using the
perturbation technique. The coefficient matrices of the Equations of Motion (Eq.(2)) are
reformed to include these expressions. The structural responses and the wave kinematics
are also expanded using the perturbation technique as follows:

q } { q(O) } { q(l) } { q(2) } { q(3) } { q(4) }

{
Up = U~O) + El U~l) + E2 U~2) + E3 U~3) + E4 U~4) (7)

The wave kinematics are expanded a.s:

{ ~: } = { v<o)
v~o)
} + E3 { vPl}
v (1) (8)
a a
 45

Substituting the above results in Eq.(2) and neglecting higher order terms such as those
containing ci, E1 E3 etc., the governing equation of motion for each component is finally
obtained. The responses may be determined in the frequency domain by the Fourier
transformation of these equations.

4. NUMERICAL RESULTS AND DISCUSSIONS

The dynamic response analysis is carried out for the offshore structure model
shown in Fig.l. The depth of water is lOOm from mean sea level. The main members
have an outer diameter of 2.8m and a thicknes of 27mm. Each leg of the tower rests on
a pile-soil foundation. The structural members as well as the piles in the foundation are
made of steel. Each pile-soil foundation is assumed to consist of 10 piles and each pile
has a diameter of 1.2m. The structure is discretized by lumping masses at selected nodal
points. There are 34 nodes and 51 members. Each node except at the base has three
degrees offreedom: horizontal displacement (in the x-direction), vertical displacement (in
the z-direction, and rotation (about they-direction). The ba.se nodes are restrained from
vertical movement. The mean values of the shear wave velocity in the soil is assumed to
be lOOm/ sec. The natural frequencies and the vibrational mode shapes are computed by
eigenvalue analysis, firstly for the rigidly supported base condition, and then for the soil­
structure interaction condition. The values of the natural frequencies are shown in Table
1 for up to the third mode of vibration. The structural damping for the first vibration
mode of the structure subsystem is assumed to be 2%. When the soil-structure interaction
is considered, by incorporating the pile-soil foundation in the offshore structure model,
the structural and radiation damping is found to be 1.35%.

The Morison equation is used to compute the hydrodynamic forces due to the
surrounding water and the sea waves. The nonlinear drag term in this equation is
linearized in the classical manner assuming that the probability density function of the
relative particle velocity distribution is Gaussian< 6 l. The linearized drag term is calculated
by a cyclic procedure, the essence of which is to alter the damping coefficients in an optimal
manner. In the present study, reasonable convergence was obtained in about three cycles
of iteration.

Random variable effects of wave force parameters on response evaluation are

investigated stochastically. The following wave force parameters are considered for the
study: (i) Inertia coefficient CM, (ii) Drag coefficient Cn (iii) Mean wave height H and
(iv) Shear wave velocity in the soil V5 . Using perturbation technique, each of the wave
force parameters is separated into two parts, one containing a constant value and a second
part with a randomly varying value. The constant value is equal to the mean value of the
 46

parameter whereas the randomly varying part has a zero-mean.

The dynamic response analysis is carried out using the frequency-domain random
vibration approach firstly using constant values of the wave force parameters and then
using the randomly varying wave force parameters. The results are expressed using rnlS
responses. Fig.2 shows the rms response displacement at node 1 for constant values of
the wave force parameters with CM = 2, Cn = 1, H =3m or 7m, Vs =lOOm/sec. The
inertia and drag components of this response are shown in Fig.3.

Fig.4 shows the contributions of the random variable effects of the mean wave
height on the response. The ratio of the response variation to the total response is
plotted against the mean wave period. The variations in the mean wave height has
increasing effects on the response when the mean wave period becomes longer. Fig.5
shows the contributions of the random variable effects of the inertia coefficient on
the dynamic response. The influence on the response by the variation in the inertia
coefficient decreases for the wave force when the mean wave period becomes longer.
Fig.6 shows the contributions of the random variable effects of the drag coefficient on
the dynamic response. Generally, the influence on the response by the variation in the
drag coefficient increases for the wave force for which the mean wave period become longer.
Thus variations in inertia and drag coefficients give different contributions to the
response. Fig. 7 shows the contributions of the random variable effects of the shear wave
velocity in the soil. The random variable effects are higher for small wave periods and
decrease rapidly with the increase in wave period and approach zero irrespective of the
size of the randomness. This result is in confirmity with our earlier observationC7l that the
dynamic soil-structure interaction effects become smaller when the the mean wave period
is away from the fundamental period of the structure.

Fig.8 shows the contributions of the random variable effects of the wave force
parameters on the total response. The horizontal axis represents the variance of the
randomly varying part and the vertical a.xis represents the amount of variation in the
response. In general, the variations in the response increase with the increase in the
randomness of wave force parameters, although, the rate of increase tends to diminish
when the variance of the randomly varying part becomes larger. It is shown that,
among all the wave force parameters considered in the present study, the randomness
of the mean wave height has the most significant contributions on the dynamic response.
The contributions of the random variable effects of the inertia coefficient are larger for
smaller wave heights and shorter wave periods whereas random variable effects of the drag
coefficient are dominant for larger wave heights and longer wave periods.
 47

5. CONCLUSIONS

Dynamic analysis of offshore structures is carried out taking into account the
vanous uncertainties associated with the characterization of the ocean environment.
A dynamic response analysis method utilizing the perturbation technique is proposed
to effectively quantify the randomness of wave force parameters. Parametric studies
are presented to assess the influence of the random variable effects of hydrodynamic
coefficients of the wave forcing function, wave height distributions and the properties of
the subsoil.

It is shown tha.t, among all the wave force parameters considered in the present
study variations in the mean wave height have the most significant effect on
the response evaluations. The random variable effects of the inertia coefficient are larger
for smaller wave heights and shorter wave periods whereas random variable effects of the
drag coefficient become dominant for larger wave heights and longer wave periods.

The random variable effects of shear wave velocity in the soil are larger when the
mean wave period is nearer to the fundamental period of the structure. Sensitivity studies
indicate that when the the mean wave period and the fundamental period of the structure
are well-separated, the structural responses are less influenced by the fluctuations in the
shear wave velocity as the soil-structure interaction effects are minimum.

For rational design of offshore structures it is necessary to clarify the reliability of

various assumptions and approximations involved in response evaluations. It is shown that
the proposed method involving perturbation technique is efficient to assess the degree of
uncertainty allowable in assuming simplified descriptions of some of the complex problems
encountered in a real sea.

REFERENCES

1. Brouwers, J.J.H and Verbeek, P.H.J.: 'Expected fatigue damage and extreme
response for Morison-type wave loading', Applied Ocean Research, Vol.5, No.3,
pp.129-133, 1983.

2. Lipset, A.W.: 'Nonlinear structural response in random waves', J. Struct. Eng.

Div., ASCE, Vol.112, No.ll, pp.2416-2429, 1986.

3. Jain, A.K. and Dutta, T.K.: 'Nonlinear dynamic analysis of offshore towers in
frequency domain', J. Eng. Mech. Div., ASCE, Vol.113, No.4, pp.610-625, 1987.
4. Venkataramana, K., et al.: 'Stochastic response of offshore structures to sea wave
and earthquake excitations with fluid-structure-soil interaction', Research Report,
No.89-ST-01, Kyoto University, 1989.
 48

5. Yamada, Y. and Kawano, K.: 'Seismic response analysis of nonproportional

damping system due to response spectrum method', J. Struct. Eng/Earthquake
Eng., JSCE, Vol.4, No.1, pp.213-222, 1987.
6. Malhotra, A.K. and Penzien, J.: 'Nondeterministic analysis of offshore structures',
J. Eng. Mech. Div., ASCE, Vol.96, No.EM6, pp.985-1003, 1970.

7. Venkataramana, K., et al.: 'Dynamic soil-structure interaction effects of the random

response of offshore platforms', Proc. 8th Int. Conf. Offshore Mech. and Arctic
Eng., ASME, pp.565-572, 1989.

Table 1: Natural frequencies of structure-pile-soil system

Vibration mode Rigidly supported Soil-structure

base interaction
first 2.24 1.61
second 11.88 8.87
third 27.49 25.14

60m

30m

--
11
1
_;:::! lOrn
--=
" 2
-:"

110m

6
z
:i4Y
;'
7 X

\l 7 \l _, j
_'/ \.'
11 ~ / \ l~'k
\ ,:;
''./ pile-soil
foundation

Fig.l: Analytical model of structure-pile-soil system

 49

CM=2 H=7m
CM=2
5 C0=1 0.10 C =1
0.10 5 0
'"'
~

~II) '"'
~

<.) ~
"'
II)
...... <.)
0.
....."' "'
......
0. 0.05
"" ....."'
~ ""
~
0~--~--~----~--~ 0~------~--------~
5 10 15 5 10 15
Mean wave period (sec) Hean wave period (sec)

Fig.2: RMS displacement at node 1 Fig.3: Inertia and drag components

of RMS response at node 1

bas = random response contribution

Us total response

1.0 -1
H=7m
CM=2
Cn=1
V5 =100m/s
I

€if =0.1
0.5 f
~ 0.05

.r--
bas
Us
0.01

5
'
10
' I
15
Hean wave period (sec)

Fig.4: Random variable effects of H

 50

0.3 H=7m
CM=2
CD=l
\1s=100m/s

0.2

6u"'
(J"'

0.1

0~----~----~------~-----J
5 10 15
Hcan wave period (sec)

Fig.5: Random variable effects of CM

0. 3 ..---------:;--~-~
H=7m
CM=2
CD=l
\1s=100m/s

0.2

6u"'
(J"'

0.1

0 •
5 10 15
Nean wave period (sec)

Fig.6: Random variable effects of CD

 51

0.2r-------------------------­
H=7m
CM=2
Cv=l
Vs=lOOm/s

OU:z:

Uz

5 10 15
Nean wave period (sec)

Fig.7: Random variable effects of Vs

1.0 - ­
H =3m, T = 7s H = 7m, T= lls

II

0.5

OU:r;

Uz

0 0.05 0.10 0 0.05 0.10

fir = (~M = (~D = f~,

Fig.S: Random variable effects of wave force parameters

 4
DYNAMICS OF DOUBLE ARTICULATED TOWERS

I. H. HELVACIOGLU and A. INCECIK

Department of Naval Architecture and Ocean Engineering

University of Glasgow, U.K.

ABSTRACT
This paper describes the dynamic motions of a double articulated tower (either coupled
with a floating structure or single) under wave excitation. Structural response predictions to
determine the axial and shear forces and the bending moments along the tower length and the
axial yoke forces (in the case of a coupled tanker-tower system) are summarised. The results
of correlation studies to validate theoretical predictions with experimental measurements are
discussed.

In the second part of the paper the results of various parametric studies to illustrate the
sensitivity of motion and structural load predictions to the wave coefficients, to the geometrical
configuration of the tower (i.e. location of the second articulation, size of buoyancy chamber,
etc.), and to water depth will be discussed.

Finally, the motion and structural performance of a single articulated tower-ship

system will be compared with that of a double articulated tower-ship system.

1. INTRODUCTION
The double articulated tower concept was first introduced by BNOC (British National
Oil Corporation) and installed in 1977 to provide export facilities for the Thistle field (see Fig.
1, duplicated from Goodfellows Ltd.( 1)). The double articulated tower configuration coupled
with either a permanently floating structure or with a shuttle transport tanker provide a cost
effective alternative to fixed platforms and pipelines in bringing oil to shore via either floating
surface or subsea structures. A typical double articulated tower configuration consists of three
parts; a deok, upper column and lower column. The lower column is connected to the sea bed
by means of a universal joint allowing angular rotations. The bottom part of the lower column
is usually ballasted with water so that the correct restoring moment can be obtained and that the
ballast water prevents the column from buckling under the external hydrostatic pressure. The
upper column provides the main buoyancy forces and its design is important if optimum

DOI: 10.1201/9781003076599-4 53