88
HSE Health & Safety Executive Comparative evaluation of minimum structures and jackets Prepared by WS Atkins Consultants Ltd for the Health and Safety Executive OFFSHORE TECHNOLOGY REPORT 2001/062

Prepared by WS Atkins Consultants Ltd for the Health and

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Prepared by WS Atkins Consultants Ltd for the Health and

HSEHealth & Safety

Executive

Comparative evaluation of minimumstructures and jackets

Prepared byWS Atkins Consultants Ltd

for the Health and Safety Executive

OFFSHORE TECHNOLOGY REPORT

2001/062

Page 2: Prepared by WS Atkins Consultants Ltd for the Health and

HSEHealth & Safety

Executive

Comparative evaluation of minimumstructures and jackets

WS Atkins Consultants Ltd Woodcote Grove

Ashley RoadEpsom

Surrey KT18 5BWUnited Kingdom

HSE BOOKS

Page 3: Prepared by WS Atkins Consultants Ltd for the Health and

ii

© Crown copyright 2002Applications for reproduction should be made in writing to:Copyright Unit, Her Majesty’s Stationery Office,St Clements House, 2-16 Colegate, Norwich NR3 1BQ

First published 2002

ISBN 0 7176 2353 X

All rights reserved. No part of this publication may bereproduced, stored in a retrieval system, or transmittedin any form or by any means (electronic, mechanical,photocopying, recording or otherwise) without the priorwritten permission of the copyright owner.

This report is made available by the Health and SafetyExecutive as part of a series of reports of work which hasbeen supported by funds provided by Mobil North SeaLimited (UK), BP Amoco (USA), Exxon ProductionResearch (USA), Minerals Management Service (USA),Odebrecht Oil & Gas (UK), Saudi Aramco (SA) and theExecutive. Neither the Executive, nor the contractorsconcerned assume any liability for the reports nor do theynecessarily reflect the views or policy of the Executive.

Page 4: Prepared by WS Atkins Consultants Ltd for the Health and

Steering Committee Members from Sponsoring Organisations

Mr D. Galbraith (Chairman) Mobil North Sea Limited (UK)

Dr B. Stahl & Mr S. DeFranco BP Amoco (USA)

Mr W. Turner Exxon Production Research (USA)

Mr W. Jones Health & Safety Executive (UK)

Mr R.W. Smith Minerals Management Service (USA)

Mr B.L. Smith Odebrecht Oil & Gas (UK)

Dr A.M. Radwan Saudi Aramco (SA)

Steering Committee Members from Project Consultants

Dr N. Shetty WS Atkins Consultants Ltd (UK)

Dr A. Dier MSL Engineering (UK)

Mr J. Waegter Ramboll (DK)

Prof. R.G. Bea University of California, Berkeley (USA)

Contributors to the Report

Dr N. Shetty WS Atkins Consultants Ltd (UK)

Dr J.T. Gierlinski WS Atkins Consultants Ltd (UK)

Dr B. Rozmarynowski WS Atkins Consultants Ltd (UK)

Dr A. Dier MSL Engineering (UK)

Mr J. Waegter Ramboll (DK)

Mr S.T. Christoffersen Ramboll (DK

Prof. R.G. Bea University of California, Berkeley (USA)

Mr R.B. Lawson University of California, Berkeley (USA)

Mr B.L. Smith Odebrecht Oil & Gas (UK)

iii

Page 5: Prepared by WS Atkins Consultants Ltd for the Health and

iv

Page 6: Prepared by WS Atkins Consultants Ltd for the Health and

v

EXECUTIVE SUMMARY

Due to their low fabrication and installation costs, there has been a trend in recent years to useMinimum Facility Platforms (MFPs) for the “fast track” development of marginal oil and gas fields inwater depths of up to 60 m.

Compared to traditional jackets, minimum structures are characterised by a slender layout, lowstiffness, and a low level of redundancy. There are concerns that these structures may be verysensitive to damage and defects which may occur due to design, construction or operational errors. There is therefore a need to understand the performance of these structures with regard to reliability,life-cycle costs and risks, so that informed decisions can be made about their feasibility for aparticular field development.

The Joint Industry Project was set up with the overall objective of evaluating and comparing thesystem reliability levels of three minimum structures against a standard four-pile jacket under extremestorm, fatigue and ship collision conditions. The study also considered the potential for errors due tohuman and organisational factors during design, construction, and operation of minimum structures,and to quantify their effect on the reliability of these structures. The structural concepts consideredare: (i) 3-pile Monotower, (ii) Vierendeel Tower, (iii) Braced Caisson, and (iv) a conventional 4-pileJacket.

To enable a consistent comparison, the structures were designed using a common design criteria,analysis and design procedures, and for operation at the same field, and to support the sametopside. The key members were designed to have utilisation ratios close to 0.8 under the 100-yearreturn environmental loading. Welded joints were designed to have minimum fatigue lives of 5 timesthe service life (assumed = 20 years) for the three minimum structures and 3 times the service life forthe 4-pile Jacket. The in-place operational condition, vortex shedding, and on-bottom stabilityrequirements were considered, but stresses due to fabrication, transportation and installationconditions were not explicitly checked. The structures were not particularly designed for shipimpact, minor reinforcements were made to joint cans and braces to ensure that the structures fullymobilise their capacity during impact.

The reliability of the as designed structures were initially evaluated considering them to be “free”from gross defect/damage which could arise from human error. Three loading conditions: extremestorm, fatigue and ship collision were investigated.

The performance of the four structures under extreme storm conditions was studied by carrying outdeterministic pushover and system reliability analyses. The pushover analysis was performed byfactoring up the 100-year design values of wave and current forces until structural collapseoccurred. The sequence of member failures and the corresponding load levels were noted. In thecases where the foundation piles were seen to fail, their failure was suppressed by artificiallyincreasing the yield strength of pile steel and/or increasing the pile penetration. Based on the jointprobability distributions of wave height, period and current parameters and the ultimate capacity ofthe structure from the pushover analysis, and also accounting for the uncertainty in the calculated

Page 7: Prepared by WS Atkins Consultants Ltd for the Health and

vi

hydrodynamic loads and capacity of the structure, system reliability index and probability of failurewere evaluated for each structure.

The results of the above analyses show that the environmental load factors on the 100-year loadingat which collapse occurs are broadly similar for all the structures, ranging from 4.0 to 5.0, with thehighest value for the Braced Caisson structure. Similarly, the system reliability indices under extremestorm condition are broadly similar for all the structures with the annual reliability index ranging from5.58 to 6.23. Although the 4-pile Jacket has the highest ultimate strength, it also attracts 40-50%higher environmental loading for the same wave height compared to the minimum structures resultingin a collapse load factor comparable to the MFPs.

The above results correspond to the case when failure of the foundation was suppressed byartificially strengthening the piles and increasing their penetration depth. When foundation failure wasallowed, the collapse load factors reduce to between 2.5 and 3.2.

The reliability under fatigue condition was evaluated corresponding to the failure of individual jointsand sequences of two, three or four joints, assuming that initial joint failures are not detected andrepaired. The impact of fatigue failure of joints on pushover reliability was evaluated by calculatingthe conditional probability of collapse due to environmental overload given the initial failure of one ormore joints by fatigue, and multiplying this with the probability of the fatigue failure sequenceoccurring. A number of dominant fatigue failure sequences were identified and evaluated for eachstructure.

The 3-pile Monotower and Vierendeel Tower structures show a significant influence from fatigue onsystem reliability. Progressive fatigue failure of any two tubular joints in the critical braces wouldsignificantly reduce the pushover capacity of these two structures. The 4-leg Jacket shows amoderate reduction in reliability when fatigue is considered. The Braced Caisson shows no influenceof fatigue as its pushover capacity is largely derived from the piles and the caisson.

Time-domain, non-linear, ship/structure interaction analyses were carried out to study theperformance of the selected structures against collision from a supply vessel. Following the impact, apost-impact pushover analysis was carried out to determine the reduction in pushover capacity as aresult of ship impact damage. For each structure analyses were carried out for a number of vesselmass and velocity combinations in the range of 500 to 3500 tonne mass and impact velocities of upto 2.5 m/sec which were considered as credible limits for operation in the Southern and CentralNorth Sea fields.

All the structures, except for the 3-pile Monotower show adequate capacities to resist collisionsfrom vessels over the range of mass and impact velocities considered. Although significant dentingwas caused at the impact location, this did not reduce the pushover capacity of any of the structures. The 3-pile Monotower failed during the impact event for certain combinations of vessel mass andvelocity. Although this reduced the reliability of this structure, the resulting value is still comfortablyhigh. Maximum limits for this structure are considered to be vessel mass of 2000 tonne with amaximum velocity of 2.5 m/sec or 3000 tonne vessel mass with a maximum velocity of 2.0 m/sec.

Page 8: Prepared by WS Atkins Consultants Ltd for the Health and

vii

A methodology for human and organisational error analysis was developed and implemented into asoftware system called SYRAS. Based on a review of world-wide accident database for marinestructures and reported incidents of damage to offshore structures in the North Sea and other fields,five potential error scenarios for minimum facility platforms were identified. These cover errorswhich could occur during design, fabrication, installation, and operation phases of a structure. Forthe selected error scenarios, the likely damage to the structures were determined and theirreliabilities under the damaged condition were evaluated considering fatigue, extreme storm and shipcollision conditions, as relevant.

Of the five human error scenarios considered, two scenarios which affect fatigue strength showconsiderable influence on the 3-pile Monotower and the Vierendeel Tower structures. Hence thesetwo structures can be regarded as less robust against human and organisational errors.

The Braced Caisson, despite having far fewer members than the other three structures, shows aremarkably high system reliability and high robustness against Human and Organisational Errors(HOE). This is because its ultimate strength is derived primarily by the central caisson and the twopiles which are all large diameter stocky members.

Considering the overall performance of the structures under different loading conditions and errorscenarios, and also accounting for the fabrication and installation effort, the Braced Caisson may bethe best choice in the Gulf-of-Mexico where the installation costs are relatively low. The highinstallation cost and the high risk associated with the significant offshore work could make thisconcept unattractive in the North Sea. The 3-pile Monotower and the Vierendeel Tower structurescould be the preferred options for the North Sea, provided these are adequately designed for fatigueand ship collision, and effective QA/QC procedures are put in place to safeguard against human andorganisational errors. Apart from a slightly higher initial cost, the conventional 4-leg jacket can stillbe a very attractive concept for the North Sea.

Key features which influence the life-cycle reliability characteristics of each structure have beenidentified and recommendations for design have been made to improve their performance.

This JIP has clearly demonstrated that a life-cycle system reliability assessment offers considerablebenefits by providing a better insight into the performance of minimum structures. It is thereforerecommended that a system reliability assessment is performed initially at the feasibility stage of aproject to select the best concept for the particular field and operations requirements, andsubsequently during the detailed design stage to further enhance the life-cycle reliabilitycharacteristics of the selected concept.

In summary, minimum structures can be made as reliable as conventional jackets by betterengineering:

ü by designing for ship impact to mitigate the risk of damage to wells/risers considering thedynamic interaction between the vessel and the structure;

ü by designing critical welds for fatigue lives >10 times the service life;

ü by using a life-cycle system reliability-based approach during design.

Page 9: Prepared by WS Atkins Consultants Ltd for the Health and

viii

Page 10: Prepared by WS Atkins Consultants Ltd for the Health and

ix

CONTENTS

Page

1. INTRODUCTION ............................................................................................................ 11.1 Background ..................................................................................................................... 11.2 Objectives ....................................................................................................................... 21.3 Project Organisation......................................................................................................... 21.4 Method of Approach ....................................................................................................... 3

1.4.1 Structures Selected for Comparison.......................................................................... 31.4.2 Framework for Comparative Evaluation.................................................................... 91.4.3 Work Programme .................................................................................................. 10

1.5 Organisation of the Report.............................................................................................. 13

2. CONCEPTUAL DESIGN............................................................................................... 152.1 Introduction................................................................................................................... 152.2 Design Premises............................................................................................................. 15

2.2.1 Design Criteria ....................................................................................................... 152.2.2 Design Data........................................................................................................... 16

2.3 Design and Analysis Procedures..................................................................................... 172.4 Key Design Features...................................................................................................... 19

2.4.1 3-Pile Monotower.................................................................................................. 192.4.2 Vierendeel Tower .................................................................................................. 192.4.3 Braced Caisson...................................................................................................... 202.4.4 4-Pile Jacket.......................................................................................................... 202.4.5 Comparison of Key Figures.................................................................................... 21

3. RELIABILITY UNDER EXTREME STORM AND FATIGUE CONDITIONS............. 233.1 Structural and Load Modelling........................................................................................ 233.2 Deterministic Pushover Analysis...................................................................................... 233.3 Reliability Under Extreme Storm Conditions.................................................................... 243.4 Reliability Under Combined Fatigue and Extreme Storm Conditions ................................ 253.5 Results........................................................................................................................... 25

3.5.1 Extreme Storm Condition....................................................................................... 253.5.2 Combined Fatigue and Extreme Storm Condition.................................................... 283.5.3 Generalisation of Results to Other Wave Environments............................................ 29

4. RELIABILITY UNDER SHIP COLLISION CONDITIONS.......................................... 314.1 Structural and Load Modelling........................................................................................ 314.2 Ship Collision Analysis ................................................................................................... 324.3 Reliability Analysis.......................................................................................................... 334.4 Results........................................................................................................................... 34

5. HUMAN & ORGANISATIONAL ERROR ANALYSIS................................................ 375.1 Failures due to Human and Organisational Errors............................................................ 375.2 Factors Influencing Error Likelihood............................................................................... 38

Page 11: Prepared by WS Atkins Consultants Ltd for the Health and

x

5.3 Quality Assurance and Control....................................................................................... 405.4 Sources of Information for HOE Quantification............................................................... 405.5 Methodology for Quantification of Error Likelihood ........................................................ 435.6 The SYRAS Software.................................................................................................... 45

6. EVALUATION OF ERROR SCENARIOS.................................................................... 486.1 Identification of Human Error Scenarios.......................................................................... 486.2 Design HOE Scenario: Fatigue due to Pile Driving Stresses............................................. 496.3 Fabrication HOE Scenario: Fit-up and Welding Flaws .................................................... 516.4 Installation HOE Scenario: Pile Insertion Damage............................................................ 526.5 Operations HOE Scenario: Ship Collision Damage.......................................................... 536.6 Operations HOE Scenario: Dropped Object Damage ..................................................... 54

7. COMPARISON OF SELECTED STRUCTURES .......................................................... 567.1 General.......................................................................................................................... 567.2 Fabrication and Installation Effort.................................................................................... 567.3 Performance Under Extreme Storm Condition................................................................ 597.4 Performance Under Fatigue Condition............................................................................ 607.5 Performance Under Ship Collision Condition.................................................................. 617.6 Performance Under Human and Organisational Errors..................................................... 637.7 Overall Comparison....................................................................................................... 65

8. CONCLUSIONS AND RECOMMENDATIONS ......................................................... 668.1 Conclusions ................................................................................................................... 668.2 Limitations ..................................................................................................................... 698.3 Recommendations.......................................................................................................... 70

9. REFERENCES................................................................................................................ 72

Page 12: Prepared by WS Atkins Consultants Ltd for the Health and

xi

TABLES

Table 2.1: Environmental Parameters Used for Design................................................................. 17

Table 2.2: Comparison of Key Design Figures for the Selected Structures................................... 22

Table 3.1: Probability distributions for environmental loading variables........................................ 24

Table 3.2: Probability distributions for fatigue resistance variables............................................... 25

Table 3.3: Key results for the extreme storm condition............................................................... 27

Table 3.4: Reliability indices under combined fatigue and pushover conditions ............................. 28

Table 4.1: Probability distributions for reliability analysis under ship impact condition.................... 34

Table 4.2: Results of ship impact analysis .................................................................................... 35

Table 6.1: Results for design HOE: fatigue due to pile driving stresses......................................... 50

Table 6.2: Results for fabrication HOE: fit-up and welding flaws................................................. 51

Table 6.3: Results for installation HOE: pile insertion damage...................................................... 52

Table 7.1: Comparison of design features influencing fabrication and installation effort .................. 58

Table 7.2: Comparison of performance under extreme storm condition........................................ 59

Table 7.3: Comparison of performance under fatigue condition.................................................... 61

Table 7.4: Comparison of performance under ship collision condition......................................... 62

Table 7.5: Comparison of the performance of structures under various error scenarios................. 64

Page 13: Prepared by WS Atkins Consultants Ltd for the Health and

FIGURES

Figure 1.1: Organisation of the Project.......................................................................................... 3

Figure 1.2: 3-Pile Monotower ...................................................................................................... 5

Figure 1.3: Vierendeel Tower....................................................................................................... 6

Figure 1.4: Braced Caisson.......................................................................................................... 7

Figure 1.5: 4-Pile Jacket............................................................................................................... 8

Figure 1.6: Framework for reliability evaluation considering human errors ...................................... 9

Figure 1.7: Flow-chart for Stage-I of the Project......................................................................... 11

Figure 1.8: Flow-chart for Stage-II of the Project ....................................................................... 12

Figure 3.1: Pushover collapse modes for the four structures........................................................ 26

Figure 3.2: Environmental load factor vs. horizontal deck deflection............................................ 27

Figure 3.3: Pushover system reliability indices for different offshore locations................................ 30

Figure 5.1: Factors influencing human and organisational errors................................................... 38

Figure 5.2: Nominal human task performance unreliability........................................................... 43

Figure 5.3: Generic human task error rates.................................................................................. 43

Figure 5.4: Operating team HOE causes and influencing factors................................................... 44

Figure 5.5: Quality Attribute Form.............................................................................................. 46

Figure 5.6: Quality Attribute Life-cycle Phase Form.................................................................... 46

Figure 5.7: Task Structure Form................................................................................................. 47

Figure 5.8: Task Information Form............................................................................................. 47

Figure 6.1: Error rates for design HOE: Fatigue due to pile driving stresses.................................. 49

Figure 6.2: Error sources for ship collision HOE scenario ............................................................ 53

Figure 7.1: Base shear versus wave height for the four structures ................................................. 59

xii

Page 14: Prepared by WS Atkins Consultants Ltd for the Health and

1

1. INTRODUCTION

1.1 Background

Due to their low fabrication and installation costs, Minimum Facility Platforms (MFPs) (e.g.monotowers) have become attractive within the last decade, especially for the “fast track”development of marginal oil and gas fields. This would allow marginal fields to startproducing typically at about half the cost and in half the time compared to those associatedwith standard four-pile jackets.

Compared to traditional jackets, minimum structures are characterised by a slender layout,low stiffness and a low level of redundancy. This could make these structures very sensitiveto damage and defects that may occur due to design, fabrication or operational errors. There is therefore a need to understand the performance of these structures with regard toreliability, life-cycle costs and risks, so that informed decisions can be made about theirfeasibility for a particular field development.

Minimum structures have hitherto been used as unmanned platforms in water depths of40m-60m, mainly for the development of marginal fields. However, Operators are nowconsidering the use of these structures in deeper water, to support higher topside loads andfor providing accommodation facilities as well. Such changes could considerably increasethe potential consequences of failure.

The Operators considering using minimum structures are faced with two key questions:

1. How do the reliability levels of minimum structures compare with those of standard four-pile jackets?

2. How does one choose between a jacket and a minimum structure concept for a givenfield considering life-cycle costs and risks?

The choice between a minimum structure and a jacket, and between alternative minimumstructure designs, is likely to be influenced by a number of factors such as: lead time,production revenue, service life, initial costs of fabrication and installation, in-servicemaintenance costs, probabilities of failure, consequences of failure, etc.

Experience has shown that human errors are the root causes of many failures of offshorestructures (e.g. Sleipner collapse). The likelihood of occurrence and effects of these errorsare influenced by organisations, procedures, hardware and equipment, and external andinternal environments. These errors can develop in design, construction and operation of astructure. Because of their low level of redundancy, minimum structures could be verysensitive to damage and defects arising from human errors.

Page 15: Prepared by WS Atkins Consultants Ltd for the Health and

2

1.2 Objectives

The Joint Industry Project was set up with the overall objective of evaluating and comparingthe life-cycle reliability and risk characteristics of minimum structures with those of traditionaljacket structures. More specifically, the objectives include:

• To evaluate and compare system reliability levels of three minimum structures against astandard four-pile jacket under extreme storm, fatigue and ship collision conditions.

• To develop procedures for the evaluation of potential for errors due to human andorganisational factors during design, construction, operation and maintenance of minimumstructures and to quantify their effect on the reliability of these structures.

It is not the intention of this project to “rank” the selected concepts or to recommend anyone as the best. The objective is to identify key features which influence the reliabilitycharacteristics of each structure, and if possible, to suggest how the performance of eachstructure can be improved.

The focus of the JIP is on quantifying the inherent reliability of the sub-structure (i.e. thejacket), and for this reason, the failure of the foundation, damage to conductors/risers, waveimpingement on the deck, wave breaking, fire and blast effects have specifically beenexcluded from the study.

Given the increased interest in the use of minimum structures within a “fast track” fielddevelopment programme, the results from the JIP are expected to give considerable benefitsto the industry by providing a better understanding of their behaviour and quantifying theirinherent reliability and susceptibility to gross errors. The study also aims to identify keycharacteristics that might be changed to improve the life-cycle performance of minimumstructures.

1.3 Project Organisation

The JIP was executed by four consultants with the roles defined as below:

WS Atkins Project Co-ordinatorReliability under extreme storm and fatigue conditions

Ramboll Dy. Project ManagerConceptual design of selected structuresInputs to identification of human error scenarios

MSL Engineering Dy. Project ManagerReliability under ship collision conditions

Univ. of California Dy. Project ManagerHuman and organisational error analysis

Page 16: Prepared by WS Atkins Consultants Ltd for the Health and

3

The project was sponsored by seven organisations as listed at the beginning of the report. WS Atkins acted as the main contractor with the sponsors while Ramboll, MSL and UCBwere sub-contractors to WS Atkins.

A Project Steering Committee (PSC) was constituted with one representative each from thesponsoring organisations and the consultants to monitor the project. David Galbraith fromMobil acted as the Chairman of the PSC. The organisation of the project is shown in Figure1.1.

Secretary

(N Shetty)

Project Co-ordinator

(N Shetty)

Chairman(D Galbraith)

SteeringCommittee

AtkinsTeam

UCBTeam

MSLTeam

RambollTeam

Dy. P.M.

(J Gierlinski)

Dy. P.M.

(R Bea)

Dy P.M.

(N Nichols)

Dy. P.M.

(ST Christoffersen)

Secretary

(N Shetty)

Project Co-ordinator

(N Shetty)

Chairman(D Galbraith)

SteeringCommittee

AtkinsTeam

UCBTeam

MSLTeam

RambollTeam

Dy. P.M.

(J Gierlinski)

Dy. P.M.

(R Bea)

Dy P.M.

(N Nichols)

Dy. P.M.

(ST Christoffersen)

Figure 1.1: Organisation of the Project

1.4 Method of Approach

1.4.1 Structures Selected for Comparison

Three minimum structure concepts which have been widely used in the North Sea and Gulf-of-Mexico were selected for comparison with a 4-pile jacket. The selected structures were:

1. 3-Pile Monotower

2. Vierendeel Tower

3. Braced Caisson

4. 4-Pile Jacket

Page 17: Prepared by WS Atkins Consultants Ltd for the Health and

4

To enable a consistent comparison, the selected structures were designed within the projectto a consistent design premise assuming that the structures will be operating at the sameoffshore field and will be supporting the same topside loads, see Section 2.

Schematic views of the as-designed structures are shown in Figures 1.2-1.5.

Page 18: Prepared by WS Atkins Consultants Ltd for the Health and

5

Figure 1.2: 3-Pile Monotower

Page 19: Prepared by WS Atkins Consultants Ltd for the Health and

6

Figure 1.3: Vierendeel Tower

Page 20: Prepared by WS Atkins Consultants Ltd for the Health and

7

Figure 1.4: Braced Caisson

Page 21: Prepared by WS Atkins Consultants Ltd for the Health and

8

Figure 1.5: 4-Pile Jacket

Page 22: Prepared by WS Atkins Consultants Ltd for the Health and

9

1.4.2 Framework for Comparative Evaluation

The system reliability levels of the selected structures were compared considering thefollowing three design conditions:

1. Extreme storm

2. Fatigue

3. Ship collision

For each condition, the reliabilities were evaluated in two parts and then combined:

1. Probability of failure assuming that the structure is “error-free”. This is called “intrinsic”or “natural” probability of failure, denoted Pf

I.

2. Probability of failure of the structure with one or more error scenarios which could occurduring design, construction, maintenance or operation of the structure. This is evaluatedby first identifying a number of error scenarios relevant to a structure and quantifying thelikelihood or probability of the error scenario Xi occurring, PE

Xi. This part is called

“human error analysis”. Next, the conditional probability of failure [PfE|Xi] of the

structure given a defect or damaged state arising from the error scenario Xi is evaluated. This is called “fragility analysis”. The product of the two probabilities then gives the“extrinsic” or “human” probability of failure Pf

E = [PEX

i] * [PfE|Xi]. The likelihood of

human error is influenced by “organisation”, “procedures”, “hardware” and“environment” under which an activity is carried out.

The above framework for reliability evaluation considering the potential for human andorganisational errors is shown schematically in Figure 1.6.

System Reliability

ExtremeStorm Fatigue Ship

Collision

HumanPf

ENatural

PfI

Design Construction Maintenance Operations

Human Error AnalysisPE

X Fragility Analysis

PfX |E

Figure 1.6: Framework for reliability evaluation considering human errors

Page 23: Prepared by WS Atkins Consultants Ltd for the Health and

10

1.4.3 Work Programme

The Project Work Programme was divided into two Stages, with three Tasks within eachstage as below. The Consultant responsible for each task is indicated in braces.

Stage-I: Reliability evaluation of “error-free” structures

Task I.1: Conceptual design (Ramboll)

Task I.2: Reliability under extreme storm and fatigue (WS Atkins)

Task I.3: Reliability under ship collision condition (MSL)

Stage-II: Human and organisational error analysis

Task II.1: Methodology for human error analysis (UCB)

Task II.2: Quantification of error probabilities (UCB, Ramboll)

Task II.3: Reliability analysis for error scenarios (WS Atkins, MSL)

The flow-chart of the tasks for Stage-I is shown in Figure 1.7 and for Stage-II is shown inFigure 1.8.

Page 24: Prepared by WS Atkins Consultants Ltd for the Health and

11

START

Data Collection

Task I.1Conceptual Design

Ramboll

Task I.2Reliability under extreme

Storm and Fatigue

WS Atkins

Task I.3Reliability under Ship

Collision Condition

MSL Engineering

Reporting

Input to Stage -II

3-PileMonotower

BracedCaisson

VierendeelTower

4-Pile Jacket

Figure 1.7: Flow-chart for Stage-I of the Project

Page 25: Prepared by WS Atkins Consultants Ltd for the Health and

12

START

Data Collection

Task II.1Methodology & Software

Development

UCB

Task II.2Quantification of Error

Probabilities

UCB & Ramboll

Task II.3Reliability Analysis of

Error Scenarios

WS Atkins & MSL

Reporting

Structures from Stage - I

Results from Stage - I

Figure 1.8: Flow-chart for Stage-II of the Project

Page 26: Prepared by WS Atkins Consultants Ltd for the Health and

13

1.5 Organisation of the Report

This report summarises the work carried out under the various tasks within the project andpresents and discusses the key results obtained. For detailed information reference shouldbe made to the individual task reports listed at the end of this report.

An overview of the remaining sections of this report is given below.

Section 2: Conceptual Design

This Section summarises the work carried out by Ramboll under Task I.1. The DesignPremise which forms the basis of design of the selected structures is discussed and the keydata used in design is summarised. The design and analysis procedures used to ensure thatthe structures are comparable on a consistent basis are discussed, and key design featuresof the selected structures are compared.

Section 3: Reliability Under Extreme Storm and Fatigue Conditions

The work carried out by WS Atkins under Task I.2 of the Project is summarised in thisSection. The methodology used for pushover analysis and system reliability analysis underextreme storm and fatigue conditions is outlined and the probabilistic modelling of the basicvariables is summarised. The key results obtained for the four structures are presented andcompared. Possible generalisation of the results to other geographical locations withdifferent wave climates is discussed.

Section 4: Reliability Under Ship Collision Conditions

The ship collision study carried out by MSL under Task I.3 is presented. The methodologyfor dynamic ship/structure interaction analysis is outlined and the background work forcollecting ship collision data for the North Sea is summarised. The methodology forreliability analysis is described and the results from the study are presented and discussed forthe four structures.

Section 5: Human and Organisational Error Analysis

The methodology for human and organisational error analysis and its implementation into theSYRAS software developed by UCB under Task II.1 is presented. The overall frameworkfor the identification of dominant error scenarios during design, construction, maintenanceand operation of minimum structures is presented. The influence of organisations, hardware,procedures and environments on the likelihood of human error is discussed and amethodology for reducing the error rates using appropriate QA/QC measures is described.

Section 6: Evaluation of Error Scenarios

This section summarises the work carried out by UCB and Ramboll in Task II.2 and that byWS Atkins and MSL under Task II.3. The application of the methodology for human erroranalysis to the selected structures is presented. The six dominant error scenarios identified

Page 27: Prepared by WS Atkins Consultants Ltd for the Health and

14

covering the different life-cycle phases and the quantification of their error probabilities aredescribed. The reliability analysis for the identified error scenarios considering extremestorm, fatigue and ship collision conditions is outlined and the results from this study arepresented and discussed.

Section 7: Comparison of the Selected Structures

This section compares the life-cycle reliability characteristics of the four structures in termsof their fabrication effort, reliability under extreme storm, fatigue, and ship collisionconditions, and performance under human and organisational error scenarios.

Section 8: Conclusions and Recommendations

Based on the results from the different tasks, a number of conclusions are made and theirimplications on the design of minimum structures are discussed. Recommendations aremade for improving the life-cycle reliability characteristics of minimum characteristics.

Page 28: Prepared by WS Atkins Consultants Ltd for the Health and

15

2. CONCEPTUAL DESIGN

2.1 Introduction

This section summarises the conceptual design of the three minimum structures and a 4-pilejacket carried out by Ramboll under Task I.1 of the Project. Detailed information on thedesign process and the configuration of the structures can be obtained from References 1-5.

The scope of work under Task I.1 was to carry out a conceptual design of the fourstructures which will form the basis of a comparative reliability evaluation. To enable aconsistent comparison, it is important that the structures are designed to a consistent criteriaand using the same design data on topside loading, environmental conditions, material, andsoil parameters. For this purpose, the “design premises” were established at the beginningof the project and all the structures were designed according to this.

The design premises are discussed in Section 2.2 followed by a summary of the design andanalysis procedures used in Section 2.3. The key design features of the four structures arecompared and discussed in Section 2.4.

2.2 Design Premises

2.2.1 Design Criteria

With the agreement of the Project Steering Committee (PSC), the Davy field in the SouthernNorth Sea was chosen as the reference site where all the structures would be located for thepurposes of design. It was also chosen to adopt standard North Sea design criteria anddesign procedures for conceptual design. Where possible, generalisation of the results toother wave environments would be made following the baseline study for North Seaconditions.

The following conditions were considered for the design of structures:

• Extreme storm (100-year return)

• Operating storm

• Fatigue

• Vortex shedding

• On-bottom stability

During the initial phase of the project, following standard North Sea practice, ship collisionfrom a 2500 tonne vessel at 2 m/sec was also used as a design condition. This, however,was seen to totally govern the section dimensions of most of the structures which made theother loading conditions insignificant. Ship collision is currently not a design requirement inthe Gulf-of-Mexico and many other offshore fields and hence it was decided by the PSC todrop this design requirement. However, for each structure after being designed for theabove listed conditions, its inherent capacity to resist a ship collision was determined asexplained further in Section 2.3.

Page 29: Prepared by WS Atkins Consultants Ltd for the Health and

16

In view of the shallow water depth, it was assumed that fabrication, load-out, transportationand installation conditions do not have a significant influence on design. In view of theconceptual nature of the design, these conditions were not assessed in detail. However, it isconsidered that the resulting designs are representative of real structures and can actually befabricated and installed, following detailed engineering.

The designs were carried out largely according to API Recommended Practice 2A-WSD,20th Edition [Ref. 6].

In order to ensure a consistent comparison of the structures, the key members of all thestructures were designed for a “utilisation ratio” of close to 0.8. In order to achieve thisrequirement, wall thickness of members were selected in increments of 1 mm. Hence it ispossible that some of the member dimensions may not follow the standard pipe sectionschedules available in practice.

The focus of the JIP was to compare the reliability characteristics of the primarysubstructure (i.e. the jacket) without being influenced by the chosen soil profile. It wastherefore important that the system reliability of the structures were not governed byfoundation failure. For this purpose, the pile axial utilisation ratios were kept well below0.60. It is considered that this has no significant influence on the primary structuredimensions and its failure behaviour.

The service life of the platforms was considered to be 20 years. The design for fatigue wasbased on the UK HSE guidelines [Ref.7]. In line with the draft ISO standard for design offixed offshore steel structures (ISO 13819-2, Fixed Steel Offshore Structures, Draft C,1997), a distinction was made between design fatigue life for the three minimum structuresand for the traditional jacket based on an assumed difference in robustness. Consequently,for the 4-Legged Jacket, a Fatigue Life Safety Factor of 3.0 was applied, while for theremaining structures a factor of 5.0 was used.

2.2.2 Design Data

All the platforms were designed to accommodate four O.D. 26 inch conductors and oneO.D. 12 inch export riser. For the 3-Pile Monotower the conductors and the riser areassumed to be located inside the central column while for the Braced Caisson theconductors are located inside the caisson and the riser is assumed clamped onto the caissonon the outside. For the Vierendeel Tower, the conductors are located inside the four cornercolumns while the riser is located within the perimeter of the Vierendeel tower. For the 4-Legged Jacket both the conductors and the riser are located centrally within the jacket toprovide maximum protection against ship impact.

The topside is assumed to have a total weight of 400 tonnes. A simplified modelling of thetopside was adopted. Additional loads due to eccentricity of the C.O.G. of the topsideloading, out-of-vertical tolerance for installation, weight of anodes and other appurtenantelements were accounted for.

Page 30: Prepared by WS Atkins Consultants Ltd for the Health and

17

As mentioned previously, the Davy field in the Southern North Sea was chosen as thereference site. Accordingly, all the environmental data and soil properties were taken fromthis site. The key environmental parameters are summarised in Table 2.1.

Table 2.1: Environmental Parameters Used for Design

Water Depth including storm surge 36.2 m

100-year return wave height 16.4 m

Period of the 100-year wave 12.6 m

Associated current speed at the surface 0.96 m/sec

Associated wind speed (1 hour mean @ 10 m above LAT) 32.2 m/sec

2.3 Design and Analysis Procedures

The environmental loading for the extreme storm and fatigue analysis were generatedfollowing API RP 2A-20th edition wave load recipe. Stream function theory was used forthe extreme wave loading and Stokes 5th Order wave theory for the fatigue loading. Awave kinematics factor of 0.9 was used and current blockage and Doppler effects weremodelled according to API. The environmental criteria given in Table 2.1 was appliedomni-directionally.

A marine growth thickness of 50 mm was assumed between LAT and -12.2 m and athickness of 25 mm from -12.2 m down to the mud line. Values of the hydrodynamiccoefficients used were: Cd = 1.05 and Cm = 1.20 for marine growth fouled members and Cd

= 0.65 and Cm = 1.60 for members not fouled by marine growth. In addition, for largediameter vertical members such as the caissons, the hydrodynamic coefficients wereevaluated as a function of the Keulegan-Carpenter number. The above values of thecoefficients were increased by 5% to account for the presence of anodes.

A non-linear soil/pile interaction analysis was used for the extreme wave and ship impactanalyses, while for fatigue analysis a linear boundary model was used for the foundationsupport. The soil spring characteristics were calculated according to API.

The dynamic response of the structure for extreme wave and fatigue analysis was modelledthrough a Dynamic Amplification Factor corresponding to the eigen period of the firstbending mode determined based on the analysis of a single degree of freedom system with2% of the critical damping.

The wind loading on the topside was determined based on a rectangular box of L = 15 m,W = 8 m, and H = 16 m). A wind shape factor of 1.5 was used.

The structures were designed to avoid the risk of vortex shedding induced vibrations and theresulting loads. This was achieved by designing all elements to be outside the locking-on

Page 31: Prepared by WS Atkins Consultants Ltd for the Health and

18

range for cross-flow and in-line excitations. The calculations followed the methods outlinedin DNV Classification Note No. 30.5 [Ref. 8]. The particle velocities used for calculation ofthe reduced velocity parameter, vr, was the resulting 100 year return period velocities at therelevant depth for the element in question after pertinent combination of wave and current.

Although the structures were not explicitly designed for ship collision condition, their inherentcapacity to resist a ship collision was determined in terms of the critical velocity of a 1000tonne vessel. Both broad and stern side impacts were analysed and the worst case from thetwo was used to define the capacity.

The ship impact capacity was determined based on a Plastic Limit State utilisation ratio limitof 1.00 (as opposed to the Elastic Limit State limit of 0.80 applied in connection with theenvironmental loads). During the performance of the design, it was further decided (with theapproval of the PSC) to re-size any of the joints which failed before the failure of thecorresponding member in the ship collision analysis. The intention was to achieve a highership impact rating if this can be achieved at a marginally extra cost by re-sizing the joints. Afull non-linear progressive collapse analysis was considered beyond the scope of aconceptual design (this was to be performed in any case by MSL under Task I.3 which ispresented in Section 4).

For the ship collision analysis of the three minimum structures, a time-domain dynamicship/structure interaction analysis was performed. The full model of the structure used in theextreme wave analysis was simplified for dynamic analysis using a modal approach, in whichthe structure is represented by the lowest bending mode shape. Based on the accelerationsand forces found in the non-linear two-degrees of freedom system, maximum d’Alembertforces (inertial forces) and direct collision forces were established. These were thentransferred back to the full structural model for which a design check was subsequentlyperformed. The impact velocity resulting in a maximum Plastic Limit State member utilisationratio of 1.0 was determined by iterative and interpolation methods.

For the 4-pile jacket on the other hand, in view of its low dynamic sensitivity, the shipimpact analysis was performed using the traditional impact energy conservation method. Theinitial kinetic energy of the drifting ship (including the effect of added mass) was assumed tobe absorbed by the following four mechanisms:

• Local plastic deformation (denting) of the impacted jacket member;

• Plastic deformation (denting) of the supply vessel;

• Development of a hinge mechanism in the impacted jacket member;

• Global elastic deformation of the jacket.

The critical impact velocity which results in a maximum Plastic Limit State member utilisationratio of 1.0 was determined by iterative and interpolation methods.

Page 32: Prepared by WS Atkins Consultants Ltd for the Health and

19

2.4 Key Design Features

The 3-D views of the as designed structures are shown in Figures 1.2-1.5. Detaileddrawings for the structures along with all relevant results such as member and joint utilisationratios, fatigue lives, etc. can be obtained from References 2-5. The governing conditions foreach structure are discussed in the following and the key design features of all the fourstructures are compared in a table at the end of this section.

2.4.1 3-Pile Monotower

The Monotower member dimensions are mainly governed by the in-place extreme stormcondition. The tubular joints as well as the lower and middle inclined braces are governed byfatigue.

Some parts of the centre column and the pile sleeves have relatively small utilisation ratios asthey are governed by the maximum allowable diameter over wall thickness ratio (= 100). Inaddition, the horizontal braces at the mud line have small utilisation ratios as their dimensionsare governed by vortex shedding considerations.

The penetration length of the piles and their axial tension and compression capacities aregoverned by the extreme storm condition.

The critical velocity for impact from a 1000 t supply boat is found to be 1.78 m/s. Thecorresponding estimated maximum collision force is 5.7 MN giving a maximum indentationof 502 mm corresponding to an indentation-to-diameter ratio of 0.21.

2.4.2 Vierendeel Tower

Above elev. (-)5.0 the joint cans and members of the Vierendeel tower are mainly governedby the in-place extreme storm condition. The upper inclined braces, the upper horizontalbraces and their joints (the cowhorn system) are governed by the in-place operational stormcondition. At elevation (-)23.0, the joints and horizontal braces are governed by fatigue.

The dimensions of the horizontal braces connecting the pile sleeves in elev. (-)23.0 werechosen to avoid vortex shedding lock-in. The diameter of the horizontal braces connectingthe pile sleeves just above mud line was chosen for geometric reasons. Consequently, theutilisation of these members in the 100-year extreme storm analysis is rather low. Similarly,the buttress braces and the lower corner columns have low utilisation ratios as they weredesigned to avoid vortex shedding lock-in.

For a few members, the minimum acceptable wall thickness was reached, and for a fewother members (e.g. the legs) the diameter-to-thickness ratio was governing.

The soil penetration length of the piles is governed by the extreme storm condition.

The critical velocity for a 1000 t supply boat was found to be 1.0 m/s. The correspondingmaximum collision force is 4.04 MN giving a maximum indentation in the corner column of

Page 33: Prepared by WS Atkins Consultants Ltd for the Health and

20

256 mm. This corresponds to an indentation-to-diameter ratio of 0.31. For this load level,the section at the impact point was found to be fully plastified. Nevertheless, the capacity ofthe dented member is shown to be sufficient, since a collision force of 5.1 MN would berequired to establish a three-hinge mechanism in the member.

The analyses for the 1.0 m/s impact velocity showed utilisation ratios of 2.08 in the memberstress check of the horizontal braces at elev. (-)5.0 and 2.39/1.47 in the punching shearcheck of the corner column cans at elev. (-)5.0 and (+)3.5, respectively. By increasing thedimensions of the horizontals at elev. (-)5.0 from Ø559x20 to Ø711x43, the corner columncans at elev. (-)5.0 from Ø854x37 to Ø960x90 and the corner column cans at elev. (+)3.5from Ø878x49 to Ø894x57, the utilisation ratios in the member stress check and punchingshear check were reduced to max. 1.0.

2.4.3 Braced Caisson

The design of the Braced Caisson is mainly governed by the in-place extreme stormcondition in the case of members and by fatigue in the case of joints.

The upper inclined braces, the upper horizontal braces and their joints (the cowhorn system)are governed by the in-place operational storm condition. The pile sleeves and the horizontalbraces at elevation (-)15.0 are governed by element fatigue. The joint fatigue governs for allthe joint cans of the pile sleeves and for the caisson sleeves at elevation (-)15.0 and atelevation (+)4.0.

The penetration lengths of the caisson and piles are governed by extreme storm.

The critical velocity for a 1000 t supply boat is found to be 1.78 m/s. The correspondingestimated maximum collision force is 5.3 MN for impact on the caisson, giving anindentation of 369 mm corresponding to an indentation-to-diameter ratio of 0.18.

For impact on the pile sleeve and pile the estimated maximum collision force is 6.8 MN.This gives a maximum indentation of 300 mm for the pile sleeve, corresponding to anindentation-to-diameter ratio of 0.19.

The analysis for the 2.0 m/s impact velocity showed a utilisation ratio of 5.93 in the punchingshear check of the caisson at elev. (+)11.6. By increasing the dimensions of the caisson canat this elevation from Ø2134x37 to Ø2134x44, the utilisation ratio in the punching shearcheck was reduced to 1.13. At elev. (+)17.5 the punching shear check also showed autilisation ratio of 1.13. By linear interpolation a velocity of 1.78 m/s is expected to give autilisation ratio of approximately 1.0 for both the joints.

2.4.4 4-Pile Jacket

The design of the 4-pile Jacket is mainly governed by the in-place extreme storm condition.The majority of the joints are governed by fatigue.

Page 34: Prepared by WS Atkins Consultants Ltd for the Health and

21

The dimensions and layout of the horizontal bracing system just above the mud line werechosen to avoid vortex shedding lock-in. Consequently, the utilisation of these members inthe 100-year extreme storm analysis is rather low. Similarly, a number of other braces havelow utilisation ratios as their diameters were chosen to avoid vortex shedding lock-in,sometimes in combination with requirements of diameter-over-thickness ratios. For a fewbraces the minimum acceptable wall thickness (fixed at 7 mm) was reached. Some parts ofthe legs have relatively small utilisation ratios as they are governed by the maximumallowable diameter-over-thickness ratio.

The penetration length of the piles is governed by the extreme storm condition.

The critical velocity for a 1000 t supply boat is found to be 1.50 m/s. For impact on a jacketleg the impact force is found to be 3.39 MN. It is found that no hinge mechanism willdevelop in the leg for this impact force. A local dent of 705 mm is, however, introduced atthe point of impact corresponding to one half the diameter of the leg.

For impact on the X-bracing the impact force is found to be 1.11 MN. It is found that ahinge mechanism will develop in the X-bracing system for this impact force. A totaldeformation of 0.877 meter at the impact point is estimated. A local dent of 46 mm is furtherinduced at the point of impact.

The analyses for the leg impact with 1.5 m/s impact velocity showed punching shearutilisation ratios of 1.46 at the X-joint in the upper bay, 1.14 at the X-joint in the middlebay, and 1.07 at the leg cans at elev. (+)17.0. Similarly, the impact on the X-bracing with1.5 m/s impact velocity showed punching shear utilisation ratios of 5.95 at the X-joint in theupper bay and 2.94 at the leg cans at elev. (+) 17.0. The wall thickness of the abovementioned joint cans were increased so as to provide maximum punching shear utilisationratios of ≈1.0.

2.4.5 Comparison of Key Figures

For comparison, a number of key figures for the four platforms are shown in Table 2.2. Thenumber of nodes and braces gives an indication of the relative effort involved in fabricationand the number of items potentially requiring offshore inspection.

The total weight (jacket + pile) (taking the pile weight of 185 t for the Monotower andignoring the weight of followers for the 4-pile Jacket) for the three minimum structures is atthe most 10% lower than the 4-pile Jacket. The fabrication effort, however, depends onseveral factors such as the number of braces and joints, the diameter and thickness ofmembers, extent of nodal construction and outdoor work, etc. Similarly the installationcosts depend on the number of piles to be driven, the use of followers, the extent andduration of offshore work, etc. A detailed comparison of the fabrication effort of the fourstructures is presented in Section 7.2.

Page 35: Prepared by WS Atkins Consultants Ltd for the Health and

22

Table 2.2: Comparison of Key Design Figures for the Selected Structures

4-Pile Jacket 3-PileMonotower

VierendeelTower

BracedCaisson

Jacket Weight:Primary steel 260 t 215 t 260 t 260 tSecondary steel 50 t 45 t 45 t 9 t

Pile Weight 189(+47)[2] t 140(185)[1] t 170 t 190 t

Caisson/Leg Dia. 1.25 m 2.4 m 0.88 m 2.1 mPile Diameter 1.2 m 1.2 m 1.2 m 1.5 mPile Penetration 40 m 34 m 32 m 41 m

No. of Braces 60 20 56 15Tubular Joints 120 36 108 26Cicumf. Welds 232 100 212 82No. of Piles 4 3 4 2

Critical Velocity(m/s)for 1000 t vessel

1.5 (0.7)[3] 1.8 1.0 (<0.5)[4] 1.8 (1.0)[5]

Dent Depth/Dia. 0.50 0.21 0.31 0.19

1. The 140 t pile weight is based on an estimated wall thickness graduation. The 185 t pile weight isbased on the constant 40 mm thickness used in the structural analysis.

2. Pile weight 189 t, followers 47 t.3. m/s is the capacity as determined from the final layout. Without minor can reinforcements the

capacity is approx. 0.7 m/s.4. m/s is the capacity as determined from the final layout. Without minor can and brace reinforcements

the capacity is estimated to be below 0.5 m/s.5. m/s is the capacity as determined from the final layout. Without minor can reinforcements the

capacity is approx. 1.0 m/s.

Since the ship impact rating has been determined on a component basis, it is somewhatunfair to compare the structures on this basis as some structures exhibit considerable reservestrength. The performance of the four structures in terms of their system reliability levelsunder extreme storm, fatigue and ship collision conditions, with and without gross errors, willbe compared in the following sections.

Page 36: Prepared by WS Atkins Consultants Ltd for the Health and

23

3. RELIABILITY UNDER EXTREME STORM AND FATIGUE CONDITIONS

This section presents the work carried out by WS Atkins under Tasks I.2 of the projectinvolving deterministic pushover analysis and system reliability analysis under extremeenvironmental and fatigue conditions, see [Ref.9]. The methodology used is summarisedand the key results from the analyses are presented and discussed.

A consistent approach was used for the analysis of all four structures in order to allowmeaningful comparisons. All the analyses were undertaken using the RASOS softwarepackage, [Ref.10], which is a specialised computer code for load generation, progressivecollapse analysis and structural system reliability analysis of offshore structures.

3.1 Structural and Load Modelling

Data for structural modelling, comprising of material parameters and geometrical dimensionswere taken from the Conceptual Design documents for individual structures, [Refs. 2, 3, 4and 5].

All the structures were modelled as space frames with each member represented by an"engineering beam/column" element. For the purpose of collapse analysis joints weremodelled as separate elements. Piles inside legs were modelled using beam elements withequivalent properties representing combined stiffness and strength of the two components. A simplified model of the deck was used with members having equivalent stiffness propertiesto simulate the actual stiffness of the deck structure.

Soil data and geometrical dimensions of plies, used to calculate the foundation responsewere taken from the Design Premises [Ref.1] document. The foundation for each structurewas modelled using pile elements supported on non-linear springs distributed along the piles. The piles themselves were modelled as tubular beam/column elements. Non-linear stiffnessof support springs (lateral p-y and axial t-z and q-z springs) were calculated from the soil properties and pile dimensions according to API [Ref.6].

Data for environmental conditions, in terms of water depth, wave and current characteristics,marine growth and hydrodynamic coefficients were taken from the Design Premises [Ref.1]. For the extreme environmental loading condition the analyses were based on a staticapproach. The environmental loading, represented by distributed forces, was calculatedusing the API RP 2A 20th Edition recipe, [Ref.6], and the Stoke's 5th order wave theorywas used for calculating particle kinematics.

The structural response under 100-year return environmental loading calculated by the threeconsultants using different software codes, namely Ramboll - ROSA, MSL - USFOS andWS Atkins - RASOS were compared for each structure. After some adjustments of theUSFOS and RASOS computer models satisfactory agreement was obtained for allstructures.

3.2 Deterministic Pushover Analysis

The pushover analysis employed for calculation of the non-linear response of a structurerequires an incremental - iterative strategy, as outlined below.

Page 37: Prepared by WS Atkins Consultants Ltd for the Health and

24

The first step in this strategy was to calculate the deterministic response under the dead loadand environmental loading for 100-year return conditions. This analysis was carried outemploying an iterative technique, in order to take into account non-linearity in the soilresponse.

The global progressive collapse analysis was carried out by factoring-up the wave andcurrent forces from their initial 100-year values until structural collapse occurred. When amember or joint "failed", by yielding or buckling, the surplus forces were redistributed to theremainder of the structure. The plastic deformation and the resulting global non-linearresponse of the structure was calculated using the Virtual Distortion Method (VDM)developed by Holnicki-Szulc and Gierlinski, [Ref.11]. The algorithm used in this methodintroduces virtual distortions into the failed locations to simulate plastic deformations thatsatisfy the constitutive law and the global equilibrium. This results in a virtual stress-strainstate of the structure. Superimposing the virtual state on the original linear-elastic stress-strain state gave the final non-linear stress-strain state of the structure with one or morecomponents failed. The key feature of the above approach is that the governing equationsare constructed for the degrees of freedom in damaged locations only. Thus, the number ofequations is considerably smaller compared to that for standard FE approach, leading to asubstantial reduction in computational effort.

3.3 Reliability Under Extreme Storm Conditions

For reliability analysis, the structure was modelled as a single component with its meanresistance represented by the ultimate base shear capacity obtained from the deterministicpushover analysis.

A number of loading, resistance and model uncertainty parameters were treated as randombasic variables described using appropriate probability distributions as summarised in Table3.1.

Table 3.1: Probability distributions for environmental loading variables

Variable Distribution Mean COV

Wave Height, H [m] Gumbel 12.6 0.10

Wave Period [sec.] Lognormal 0.432*H + 5.61 0.10

Current Speed [m/sec.] Lognormal 0.028*H + 0.48 0.15

Load Model Uncertainty Normal Bias = 1.0 0.15

Ultimate Strength Uncertainty Lognormal Bias = 1.0 0.15

The random base shear due to the applied loading was evaluated as a function of the basicvariables wave height, wave period, current speed and wave load model uncertainty. First-and Second- Order Methods (FORM/SORM) were used for calculating the probability offailure.

Page 38: Prepared by WS Atkins Consultants Ltd for the Health and

25

3.4 Reliability Under Combined Fatigue and Extreme Storm Conditions

The probabilistic model for fatigue variables is summarised in Table 3.2.

Table 3.2: Probability distributions for fatigue resistance variables

Variable Distribution Mean COV

Global analysis model uncertainty Lognormal 1.0 0.20

S.C.F. model uncertainty Lognormal 1.0 0.15

S-N curve random factor Lognormal 3.38 0.58

Miner’s rule model uncertainty Lognormal 1.0 0.25

Under combined fatigue and extreme storm conditions three types of failure sequencesleading to collapse of the structure need to be considered:

1. Sequence of member/joint static failures under an extreme storm condition,2. Sequence of fatigue failures of joints at random points in time, and3. Initial failure of one or more joints in sequence by fatigue followed by collapse of the

weakened structure during an extreme storm.

The union of all the dominant failure sequences of the above three types then gives theoverall system failure probability. In the present study, the pushover failure scenarios underextreme environmental conditions, either from an intact state of the structure (in 1 above) orfollowing initial damage by fatigue (in 3), were analysed using a simplified single resistancevariable approach as discussed in Section 3.4. Sequence of fatigue failures of joints (in 2and 3 above) were analysed using the Selective Enumeration Method, Shetty, [Ref.12].

Besides the overall system reliability levels of the candidate structures, the analysis under thecombined conditions provides useful information about the relative importance of fatigue andextreme storm conditions for the selected structures.

3.5 Results

3.5.1 Extreme Storm Condition

Deterministic pushover analyses of the four structures were carried out by factoring-up thewave and current forces from their initial 100-year values until collapse occurred. Only themost critical wave direction, selected on the basis of design calculations, was considered. Initial analyses showed that the collapse of all the structures was governed by the failure ofthe foundation system. In order to focus the comparisons to the jacket part of thestructures, the foundation failure was suppressed by either strengthening the piles or byincreasing their penetration depth. The results below correspond to the revised designs. Thecollapse modes and sequence of member failures for the four structures are shown in Figure3.1.

Page 39: Prepared by WS Atkins Consultants Ltd for the Health and

26

Figure 3.1: Pushover collapse modes for the four structures

1

2

3

4

56

7

8

9

10

6

(a) 4-legged Jacket

12

3

45

6

8

9

10

(b) 3-pile Monotower

12

4

3

5

(d) Braced Caisson

1

212

6

4

8

73

9

5

10

13

11

(c) Vierendeel Tower

Page 40: Prepared by WS Atkins Consultants Ltd for the Health and

27

The load factor on the 100-year environmental loading versus the horizontal deflection at thedeck level for the structures are plotted in Figure 3.2.

0.0

1.0

2.0

3.0

4.0

5.0

6.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Horizontal Deflection at Deck Level (m)

En

viro

nm

enta

l Lo

ad F

acto

r

Monot.

Caisson

Vierend.

Jacket

Figure 3.2: Environmental load factor vs. horizontal deck deflection

From the above figure two types of behaviour could be distinguished. The 3-pileMonotower and the Braced Caisson show a much stiffer behaviour until very close tocollapse and a rapid increase in deck displacement as the collapse load is reached. Incontrast, the 4-Legged Jacket and the Vierendeel Tower, show a local maximum at around60% to 70% of the ultimate capacity usually associated with local buckling of keycompressive braces. This results in a dynamic change of the load path with the remainingmembers of the structures supporting further increases in the load.

The key results from the deterministic pushover analyses and system reliability analyses forthe extreme storm condition are summarised in Table 3.3.

Table 3.3: Key results for the extreme storm condition

Result 4-PileJacket

3-PileMonotower

VierendeelTower

BracedCaisson

100-year design env. base shear [kN] 7,970 3,450 4,646 3,700

Ultimate env. base shear [kN] 32,600 15,180 18,440 18,500

Env. load factor at collapse 4.09 4.40 3.97 5.00

Component reliability index (annual) 4.62 4.91 4.44 5.20

System reliability index (annual) 5.73 5.91 5.58 6.23

Most likely collapse wave [m] 26.60 27.30 25.90 > 27.00

Env. load factor to collapse for theoriginal foundation design

2.52 3.12 3.20 3.12

Page 41: Prepared by WS Atkins Consultants Ltd for the Health and

28

Of the four structures, the lowest ultimate load factor was obtained for the VierendeelTower while the highest value was achieved for the Braced Caisson. Notwithstanding thesedifferences, the collapse load factors for all the four structures are relatively high.

The above trend is also reflected by the system reliability indices for the four structureswhich shows that the Braced Caisson has the highest system reliability (6.23) while theVierendeel Tower gives the lowest reliability (5.58). While the values are reasonably highfor all the structures, the difference between the highest and lowest values is equivalent toone order of magnitude in annual failure probability.

The above results can in part be explained by the amount of environmental loading thestructures attract for a given wave height. Comparing the environmental base shear for100-year return design conditions and ultimate base shears at collapse for the four structures, itcan be seen that for the three minimum structures the base shears for the same design waveheight are roughly of similar magnitude. In contrast, the jacket structure attracts about twiceas much load as the other structures. The ultimate capacities of the four structures alsocompare similarly, and it can be seen that the wave height corresponding to collapse isnearly equal for the four structures. This could explain why, despite the high ultimatecapacity of the jacket, its ultimate environmental load factor is not significantly different fromthe minimum structures.

Comparing the three minimum structures themselves, it can be seen that the VierendeelTower attracts about 20% more hydrodynamic loading than the 3-pile Monotower at thedesign level. This structure also has a 10% higher ultimate capacity than the 3-pileMonotower. However, due to differences between base shears for these two structures atthe design level, the collapse load factor for the 3-pile Monotower ends up about 10%higher than that of the Vierendeel Tower.

For the sake of comparison, the collapse load factors for the original design of thefoundation are given in the last row of Table 3.3. In this case collapse occurred due tofailure of the foundation for all structures. It can be seen that foundation failure governs theultimate capacity of all the four structures.

3.5.2 Combined Fatigue and Extreme Storm Condition

Component reliability indices for fatigue failure of joints and system reliability indices undercombined fatigue and pushover conditions are given in Table 3.4 for all the four structures. The results are for a service exposure of 20 years.

Table 3.4: Reliability indices under combined fatigue and pushover conditions

Structure First joint failure byfatigue, ββ fcf

Any joint failure byfatigue, ββ afcf

SystemReliability, ββ sys

3-pile Monotower 3.07 2.57 3.14

4-legged Jacket 2.69 1.82 5.43

Vierendeel Tower 3.23 2.70 4.10

Braced Caisson 3.21 2.85 6.23

Page 42: Prepared by WS Atkins Consultants Ltd for the Health and

29

Of the four structures, the 3-pile Monotower is seen to be the most sensitive to fatigue, witha combined fatigue and pushover reliability of 3.14 compared with 5.91 when fatigue is notconsidered. The fatigue-sensitive joints of this structure all form part of the primarymembers of the underwater truss structure. It was found that progressive fatigue failure ofany two tubular joints in sequence was sufficient to significantly reduce the pushovercapacity of the structure. The critical sequences all involved initial fatigue failure of any oneof the joints in the top compression braces in the three frames. Subsequent fatigue failure ofany one of the joints in the lower braces, or in the remaining top braces gives a probability ofcollapse under pushover loading of greater than 0.5.

The system reliability of the Vierendeel structure is also seen to be very sensitive to fatigue,with a combined fatigue and pushover reliability of 4.1 compared with 5.57 when fatigue isnot considered. The fatigue-critical joints for this structure are in the four pyramidal rakingbraces supporting the Vierendeel frame. Fatigue failure of joints in any two out of these fourbraces will lead to the platform collapsing under its self-weight. The system reliability indexis thus determined by the probability of these fatigue sequences occurring. The joints in theeight horizontal braces close to the mudline are also fatigue-sensitive. However fatiguefailure of these joints, although reducing the pushover capacity of the structure, does notinfluence the overall system reliability under combined fatigue and pushover conditions.

The 4-legged jacket is seen to be relatively less sensitive to fatigue, with the combinedfatigue and pushover reliability of 5.43 compared with 5.73 for the pure pushover condition. The fatigue-critical joints are in the vertical diagonal braces in the bottom and middle bays.Failure of the four bottom bay brace joints by fatigue is more critical than failure of themiddle bay brace joints. However, the platform still has considerable reserve of strengthfollowing the fatigue failure of these brace joints owing to the bending capacity of the mainlegs.

The pushover capacity of the braced caisson is seen to be insensitive to the effects of fatigueof failure of the joints in the cross braces which connect the caisson and piles below themean sea level. The reliability against combined fatigue and pushover is thus the same as thereliability under pushover condition with a reliability index of 6.23.

Note that all the fatigue reliability results are for an exposure period of 20 years (assumedservice life) and assume no inspection during service.

3.5.3 Generalisation of Results to Other Wave Environments

The results for the extreme storm condition can be generalised to other offshoreenvironments under the following assumptions:

1. Structures are designed to the same premises;

2. They are designed to 100-year return conditions of the offshore field considered;

Page 43: Prepared by WS Atkins Consultants Ltd for the Health and

30

3. The ultimate environmental load factors are assumed to remain the same as those given inTable 3.3 above.

Under the above assumptions, the system reliability indices under extreme storm conditionfor the different offshore fields differ as a function of the different coefficient of variation(COV) of the wave height for these locations.

The system reliability indices for different offshore locations are compared in Figure 3.3assuming the COV of wave height as: North Sea - 0.10, Other location - 0.15, and Gulf-of-Mexico - 0.20.

0

1

2

3

4

5

6

7

Jacket Monotower Vierendeel Caisson

Sy

ste

m R

eli

ab

ilit

y I

nd

ex

North Sea

Other

G-o-M

Figure 3.3: Pushover system reliability indices for different offshore locations

Page 44: Prepared by WS Atkins Consultants Ltd for the Health and

31

4. RELIABILITY UNDER SHIP COLLISION CONDITIONS

The ship collision study carried out by MSL under Task I.3 is presented in this Section. Themethodology for dynamic ship/structure interaction analysis and reliability analysis is outlined.The main results from the study are presented and discussed for the four structures. Furtherdetails can be obtained from the MSL report [Ref.13].

A consistent approach to analyses was applied to all structures so that valid comparisonscould be made between their respective behaviours.

4.1 Structural and Load Modelling

Structural and load modelling and subsequent ship impact and pushover analyses werecarried out using the USFOS computer program [Ref. 14].

The first step was to create the structural models including boat landings, appurtenances andmarine growth within USFOS. The structural details were taken from Ramboll reports[Refs. 2-5] for individual structures. Subsequent changes to pile dimensions and materialproperties provided by WS Atkins, as discussed in Section 3, were incorporated before thefinal analyses were carried out. Soil springs were generated using an EXCEL spreadsheetbased on API RP 2A [Ref. 6] formulations.

The loading on the structures included gravity loads from the self weight of the structure andtopside equipment, wind loading on deck, buoyancy, and wave and current loads. Theeffects of eccentricity of C.O.G. of topside loading and out-of-verticality tolerance duringinstallation were simulated using additional horizontal loads at the deck level.

The substructure steel density was increased by 5% to account for the weight of sacrificialanodes. Wind loads were simulated by point loads distributed over the topsides structure. Stream Function theory was used to calculate wave loads for the 100-year return designstorm conditions. The hydrodynamic coefficients were taken from Ramboll reports.

The mean value of yield strength of steel was taken to be 15% higher than the specifiedminimum value used by Ramboll in design.

The structural and load models were checked by comparing base shears, over-turningmoments and axial loads in selected members to values obtained by Ramboll and WSAtkins. Satisfactory agreement was obtained.

All bracing members were automatically given an initial bow and assigned certain plasticityfactors such that the buckling strength estimated by USFOS match the API RP2A strengths(without the safety factors).

Baseline pushover analyses of all the structures in their intact condition were carried out byfactoring up only the wave and current loads from their initial 100-year return values untilstructural collapse occurred. The results of collapse base shear and RSR were compared

Page 45: Prepared by WS Atkins Consultants Ltd for the Health and

32

with the corresponding results from WS Atkins and satisfactory agreement was found, seeSection 3.

4.2 Ship Collision Analysis

In a separate study carried out by MSL for HSE, one of the Sponsors of the JIP, the HSE’scollision database pertaining to UK sector of the North Sea was examined. From thisstudy, credible ranges of ship mass and velocity were established. Based on this, the shipimpact analyses for the four structures considered a range of ship mass between 500 and3500 tonnes and impact velocities up to 2.5 m/sec.

A non-linear, time-domain, dynamic ship/structure interaction analysis was used for all of thestructures. Still water conditions and gravity loads were considered at the time of impact.

Two non-linear springs were inserted in series into the model at the impact location. Onespring represented the member denting process for which the P-δ non-linear stiffnessrelationship was obtained from the work of Pettersen and Johnsen [Ref. 15]. The otherspring simulated the deformation characteristics of the vessel. The P-δ relationship specifiedby DNV [Ref. 16] was used for this spring.

After applying gravity, buoyancy and other ‘static’ loads, a mass representing the vessel andassociated added mass was given the initial velocity and applied to the end of the (ship) non-linear spring. Appropriate levels of damping were used during this phase of the analysis. Normally, following a short period, the response of the structure and vessel was such thatseparation occurred.

After separation, the damping levels were increased to damp out structural vibrationsquickly, in preparation for the quasi-static post-impact pushover analysis under extremeenvironmental conditions. The impact damage was modelled in terms of the residual dentand bow damage of the impacted member and any residual plastic deformations elsewherein the structure.

For each structure, only one impact location was selected and a broadside impact wasassumed as this proved to have the worst effect on the structure. From the knowledge ofthe critical members for the pushover condition obtained from the initial baseline analyses,the direction of the ship was chosen such that the impact will cause maximum damage(yielding/buckling) to these critical members.

A maximum limit of 0.7 was set for the dent depth to member diameter ratio. The impactanalysis was stopped when either this limit was reached or the structure failed during theimpact itself. In reality, the well conductors and risers within the impacted member may beseverely damaged before the limiting dent depth is reached. In order to capture this effectcorrectly a detailed modelling of the conductor package would be necessary. There is alsoa danger that following this approach would make the results specific to the system used. Therefore, it was agreed by the Project Steering Committee that the impact analysis wouldignore the presence of conductors/risers and aim to determine the maximum capacity of the

Page 46: Prepared by WS Atkins Consultants Ltd for the Health and

33

structures to withstand ship impact. If necessary, the Operators can set a lower limit takinginto account the exact configuration of the conductor/riser system.

4.3 Reliability Analysis

For reliability analysis, the structure was modelled as a single component with its resistancerepresenting the ultimate strength of the structure under pushover condition. The reductionin pushover capacity due to ship impact damage was modelled using a reduction factor f(M,v) which is a function of the ship mass and velocity of impact.

The safety margin, Z, for collapse of the structure under extreme environmental loadingfollowing ship impact is expressed as

Z = Xmodel . Rinit (1 – f (M, v)) – Xhydro (a . H b) – Xmwind . Fwind (4.1)

where:

Xmodel = random factor for uncertainty in ship impact and pushover capacity

Xhydro = random factor for uncertainty in base shear calculations

H = annual maximum wave height (random)

Xmwind = random factor for uncertainty in wind force calculations

Fwind = Base shear due to associated wind loading on the deck (random)

Rinit = ultimate capacity of the structure in terms of base shear at collapse

f(M,v) = function to account for degradation of system strength due to shipimpact (see below)

a, b = structure dependent parameters, fitted from analysis results, to relatebase shear to wave height.

The function f(M,v) depends on the mass (M) and velocity (v) of the ship. The results of theanalyses described in Section 4.4 next suggest that f(M,v) = 0 for all structures except the3-pile Monotower.

The Monotower structure failed during the ship impact for certain combinations of mass andvelocity. The safety margin for failure during ship impact for this structure is expressed as

Z = [411.5 v2 - 3971 v + 9560] - M (4.2)

The term in the [ ] braces represents the capacity of the structure against ship impact andwas obtained by fitting a function to those values of mass and velocity which resulted in thefailure of the structure during impact.

Page 47: Prepared by WS Atkins Consultants Ltd for the Health and

34

The vessel sizes which could visit a structure were modelled using a uniform (rectangular)distribution between 500 and 3500 tonnes. (This does not include the added mass whichwas taken into account during analyses.) For a given vessel, the uncertainty in its mass wasmodelled using a normal distribution with a coefficient of variation of 0.15. The velocity ofimpact was taken to be exponentially distributed with a mean of 0.3 m/s and a standarddeviation of 0.3 m/s.

The distributions for the environmental parameters and model uncertainty factors are thesame as those discussed under Section 3, and are summarised in Table 4.1.

Table 4.1: Probability distributions for reliability analysis under ship impact condition

Variable Distribution Statistics

Xmodel

(Modelling uncertainty)Normal Mean = 1.0

COV = 0.15

Xhydro

(Base shear uncertainty)Normal Mean = 1.0

COV = 0.15

H(Annual max. wave height)

Gumbel Mean = 12.55mCOV = 0.097

Xmwind

(Wind force model uncertainty)Lognormal Mean = 1.0

COV = 0.15

Xwind

(Wind force)Lognormal Mean = varies

COV = 0.20

M(Ship mass)

Rectangular 500 to 3500 t

Xmship

(Ship mass uncertainty)Normal Mean = 1.0

COV = 0.15

v(Ship velocity)

Exponential Mean = 0.3 m/sStd. dev. = 0.3 m/s

4.4 Results

A series of dynamic ship impact and post-impact pushover analyses were conducted toestablish the degradation of system strength, if any, due to the impact. The results arepresented in Table 4.2. For the range of vessel mass and velocities considered (500 to3500 tonnes and up to 2.5 m/s) it can be seen that only the 3-pile Monotower’s systemstrength was affected.

Page 48: Prepared by WS Atkins Consultants Ltd for the Health and

35

Although the dent must have a weakening effect on the impacted member, the member didnot participate in the collapse mechanism under pushover conditions. Therefore, providedthe structure survived the ship impact itself, the reserve strength under subsequent pushoverconditions remained unaffected.

Table 4.2: Results of ship impact analysis

For certain impact cases, very high dent depths (up to 0.7 of the member diameter) wereobtained without global collapse and without a significant influence on the pushover capacity. In reality, the well conductors and risers within the impacted member may be severelydamaged before these dent depths are reached.

For the 4-pile Jacket, Vierendeel Tower and the Braced Caisson, the system reliability istherefore governed by the pushover condition and the results will be the same as those givenin Section 3 for the intact structure without any damage due to ship impact.

Structure and ship impact location

Ship Mass

(Tonnes)

Ship Velocity

(m/s)

Energy (MJ) Dent (m) d/D Post Impact Pushover Strength

λλ

3P Monopod 1000 1.78 2.275 0.200 0.083 3.71Caisson φ=2.400m 1000 2.00 2.870 0.270 0.113 ''+4.0m 1000 2.50 4.488 0.370 0.150 ''

1000 3.25 7.585 0.600 0.250 3.651000 3.30 7.821 - - Failed during impact2500 1.78 5.687 0.420 0.180 3.712500 2.00 7.180 0.550 0.230 ''2500 2.35 9.915 0.600 0.250 3.4452500 2.50 11.219 - - Failed during impact3500 1.50 5.654 0.440 0.180 3.713500 1.78 7.962 0.580 0.240 ''3500 1.90 9.074 0.600 0.250 3.653500 2.00 10.052 - - Failed during impact

4 Legged Jacket 1000 1.50 1.800 0.580 0.410 3.05Leg φ=1.422m 1000 2.00 3.200 0.870 0.610 ''-2.0m 1000 2.50 5.000 1.000 0.700 ''

2500 1.25 3.125 0.870 0.610 ''2500 1.50 4.500 1.000 0.700 ''3500 1.10 3.388 0.870 0.610 ''3500 1.25 4.375 1.000 0.700 ''3500 2.50 17.500 1.000 0.700 ''

Vierendeel 1000 1.00 0.800 0.070 0.083 2.76Column φ=0.840m 1000 2.00 3.200 0.375 0.446 ''-2.0m 3500 2.00 11.200 0.588 0.700 ''

3500 2.50 17.500 0.588 0.700 "Braced Caisson 1000 1.00 0.719 0.058 0.030 3.76Column φ=2.134m 1000 2.00 2.876 0.315 0.150 "+2.0m 3500 2.00 7.876 0.800 0.370 "

3500 2.50 12.306 1.000 0.470 "

Notes: 1) For each structure limit of denting is d/D=0.700 2) For 4 legged jacket removal of X-bracing for sternside. X bracing impact had no effect on reducing pushover strength.

Page 49: Prepared by WS Atkins Consultants Ltd for the Health and

36

The 3-pile Monotower structure failed during the ship impact for certain combinations ofmass and velocity. In the other cases, the pushover capacity was only marginally affectedfor values of mass and velocity very close to those which caused failure under ship impactitself. Therefore the reliability of this structure for failure during ship impact was evaluatedusing the safety margin given in Eq. (4.2).

The probability of failure during impact, using the First Order Reliability Method (FORM),was calculated as 6.36E-04. In Task II.2, the probability of impact was established as9.0E-03 per annum based on historic collision data for North Sea structures. Combiningthese two probabilities gives an overall probability of failure due to ship impact of 5.72E-06equivalent to a reliability index of 4.39. This is lower than the reliability index of 5.13 underpushover condition obtained using the safety margin Eq. (4.1) with f (M, v) set to zero.

In summary, all structures except perhaps the 3-pile Monotower have proven to be robustagainst impacts from vessels up to 3500 tonnes and velocities up to 2.5 m/sec. Even for the3-pile Monotower, the reliability index remains at a comfortably high level. These results,however, ignore any damage to conductors and riser within the impacted member(caisson/leg) which in practice will limit the size of the vessel and impact velocities to muchlower values.

For the 3-pile Monotower, Vierendeel tower and the Braced Caisson, the values given inTable 4.2 should therefore be considered as upper bounds to the impact capacities of thesestructures, while those given in Table 2.2, which are based on component failure, should betaken as lower bound values. For the 4-pile jacket, in which the conductors and the riserare located centrally within the jacket, the ship impact capacity will correspond to the valuesgiven in Table 4.2.

Page 50: Prepared by WS Atkins Consultants Ltd for the Health and

37

5. HUMAN & ORGANISATIONAL ERROR ANALYSIS

This section describes the work carried out by University of California, Berkeley under TaskII.1 of the project. A general framework for the identification of human errors andquantification of their likelihood was developed, [Ref.17]. The methodology wasimplemented into the SYRAS software to facilitate the error evaluation, [Ref.18].

5.1 Failures due to Human and Organisational Errors

Experience has shown human errors to be the basic cause of failures of many engineeredsystems. In almost all cases, the initiating event can be traced back to a catastrophiccompounding of human and organisational errors, [Ref.17]. A careful examination andevaluation of past incidents of major failures in different industrial sectors has provided abetter insight into the causes of human errors.

Failure can be defined as any undesirable or unanticipated state or poor performance whichmakes a structure unfit for its intended purpose. Failure could occur, for example, due toloss of safety, serviceability, durability, or other performance requirements such as budget,project schedule, aesthetics, etc.

Failures could occur due to:

(i) “intrinsic” (natural, inherent) causes - those that could have been or were anticipated, i.e.predictable causes, and

(ii) “extrinsic” (human error related) causes - those that could not have been or were notanticipated, i.e. unpredictable causes.

The causes of failure due to human and organisational errors (HOE) can be organised intothree categories:

1. the initiating actions which are the direct cause of failures,

2. the contributory factors that underlie the above actions, and

3. the compounding or propagating actions.

A detailed study of the case histories of failure of marine structures indicates that while thedirect causes of failure can be attributed to the acts of individuals, the dominant contributingand compounding causes are fundamentally "organisational" - erroneous actions by groupsof individuals that influence the direct cause of failure and exacerbate or escalate itsdevelopment through compounded errors. Of the individual errors, the majority of errorsare errors of commission (80%), i.e. what was performed was erroneous and purposefullyexecuted. Errors of omission or what was performed was not intentional account for aminority of the causes (20%). Often, the direct initiating actions are identified and the moreimportant underlying and compounding actions are ignored. This has been an importantdeficiency in most accident databases.

Human and organisational errors can occur during the design, construction, and operationphases of the life-cycle of a structure.

Page 51: Prepared by WS Atkins Consultants Ltd for the Health and

38

5.2 Factors Influencing Error Likelihood

Any activity that involves human intervention is subject to the results of human influences thathave both positive and negative impacts of varying degrees. These influences can becategorised into six groups as depicted in Figure 5.1.

OperatingTeamsFactors

OrganizationalFactors

ProceduralFactors

HardwareFactors

StructuralFactors

InterfacesFactors

EnvironmentalFactors

Figure 5.1: Factors influencing human and organisational errors

Operators. The Operator or Individual malfunctions might best be described as actions andin-actions that result in lower than acceptable performance. Operator malfunctions can becategorized by types of error mechanisms (Reason, 1990). These include slips or lapses,mistakes, and circumvention. Slips and lapses cause low quality actions when the outcomeof the action is or was not the intended outcome. Frequently, the significance of this type ofmalfunction is small because these actions are not easily recognized by the person involvedand in most cases are easily corrected. Mistakes can develop when the action is intended,but the intention is wrong. Mistakes are perhaps the most significant because theperpetrator has limited clues to indicate a problem. Circumvention (violations, intentionalshort cuts) develop when a person decides to break a rule for what seems to be a good (orbenign) reason in order to simplify or avoid a task. Often, it takes an outsider to thesituation to identify this.

Organisations. An organisational malfunction is defined as a departure from acceptable ordesirable practice on the part of a group of individuals that produces unacceptable orundesirable results.

Analysis of the history of failures of offshore platforms and other marine systems providesmany examples in which organisational malfunctions have been primarily responsible forfailures. The goals promulgated by an organisation may induce operators to conduct theirwork in a manner that management would approve. Excessive risk-taking problems are verycommon in marine systems. Frequently, the organisation develops high rewards formaintaining and increasing production while hoping for safety. The formal and informalrewards and incentives provided by an organisation have a major influence on theperformance of operators.

One of the most pervasive problems resulting in failures of offshore platforms involvesorganisational communications. In the case of the Piper Alpha platform, the break down in

Page 52: Prepared by WS Atkins Consultants Ltd for the Health and

39

organisation was found in the failure of the permit to work system, the loss of commandcontrol, and the organisation’s ignoring early warning signals issued by the field operatingpersonnel [Ref.19]. Due to incentives provided by the organisation, there were tendencies tofilter information, making things seem better than they were.

Experience indicates that one of the major factors in organisational malfunctions is the cultureof the organisation. Organisational culture is reflected in the following: views on action,change, and innovation; the degree of external focus contrasted with internal focus;incentives provided for risk taking; other rewards and incentives; the degree of lateral andvertical integration of the organisation; the effectiveness and honesty of communications;autonomy, responsibility, authority and decision making; rewards and; and the orientationtoward the quality of performance contrasted with the quantity of production. The culture ofan organisation is embedded in its history.

Procedures. These malfunctions can be embedded in engineering design guidelines andcomputer programs, construction specifications, and operations manuals. They can beembedded in how people are taught to do things. With the advent of computers and theirintegration into many aspects of the design, construction, and operation of marine structures,software errors are of particular concern because computers can only function within thelimitations of their designers.

Software errors in which incorrect and inaccurate algorithms were coded into computerprograms have been at the root cause of several major failures of marine structures.Guidelines have been developed to address the quality of computer software for theperformance of finite element analyses. Extensive software testing is required to assure thatthe software performs as it should and that the documentation is sufficient. Of particularimportance is the provision of independent checking procedures that can be used to validatethe results from analyses. High quality procedures need to be verifiable based on firstprinciples, results from testing, and field experience.

Given the rapid pace at which significant industrial and technical developments have beentaking place, there has been a tendency to make design guidelines, constructionspecifications, and operating manuals more and more complex. In many cases, poororganisation and documentation of software and procedures has exacerbated the tendenciesfor humans to make errors. Simplicity, clarity, completeness, accuracy, and organisation aredesirable attributes in procedures developed for the design, construction, and operation ofmarine structures.

Hardware. Human malfunctions can be initiated by or exacerbated by poorly engineeredsystems and procedures that invite. Such systems (hardware and/or structure) are difficult toconstruct, operate, and maintain.

New technologies compound the problems of latent system flaws. Complex design, closecoupling (failure of one component leads to failure of other components) and severeperformance demands on systems increase the difficulty in controlling the impact of humanmalfunctions even in well operated systems.

Page 53: Prepared by WS Atkins Consultants Ltd for the Health and

40

The issues of system robustness (defect or damage tolerance), design for constructibility,and design for IMR (Inspection, Maintenance, Repair) are critical aspects of engineeringoffshore platforms that will be able to deliver acceptable performance.

Environment. Environmental influences can have important effects on the performancecharacteristics of individuals, organisations, hardware, and software. Environmentalinfluences include external (e.g. wind, temperature, rain, fog, time of day), internal (lighting,ventilation, noise, motions) and sociological factors (e.g. values, beliefs, etc.).

5.3 Quality Assurance and Control

Quality Assurance (QA) is composed of those practices and procedures designed to helpassure that an acceptable degree of quality or performance is obtained. QA is focused onpreventing of malfunctions. Quality Control (QC) is associated with the implementation andverification of the QA practices and procedures. Quality control is intended to assure thatthe desired level of quality is actually achieved. Quality control is focused on reaction,identification of malfunctions, rectification, and correction.

Achieving quality goals is primarily dependent on people. QA / QC efforts are directedfundamentally at assuring that human and system performance is developed and maintainedat acceptable levels. Strategies for implementing QA /QC measures include those put inplace before the activity (prevention), during the activity (checking), after the activity(inspection), after the manufacture or construction (testing), and after the structure has beenput in service (detection). The earlier QA / QC measures are able to detect the lack ofacceptable quality, then the more effective the remediation.

It is desirable that QA / QC are very stringent for the error intolerant elements that comprisea structure system. Also, it is desirable to configure or design the element or component sothat it can be “error tolerant” for the highly likely types of design, construction, and / oroperations malfunctions. The design of damage or defect tolerant (robust) structures is veryimportant. The sensitivities of various parts of a particular structure and various parts of aparticular design process can be studied beforehand through “fragility analysis” to determinethe most error intolerant parts. Re-design and QA / QC efforts can thus be directed atthose elements and aspects with the highest criticality. Constant attention needs to be givento these elements during construction, operation, and maintenance. Inspections can helpconfirm the quality and condition of the elements most important to the integrity of theplatform and most intolerant of low quality factors.

5.4 Sources of Information for HOE Quantification

A study of the present databases on marine and offshore systems in which there has beenunacceptable levels of quality indicates that they are deficient in their ability to accuratelydefine the key initiating, contributing, and compounding factors that lead to compromises ofoperating quality. There has not been any common classification or definition of causes ofmarine accidents. There has been a dearth of well trained investigators. Investigationsgenerally have focused on the immediate causes of quality problems, not the underlying

Page 54: Prepared by WS Atkins Consultants Ltd for the Health and

41

factors that lead to these causes. Investigations have frequently been focused on placingblame rather than on determining the underlying, direct, and contributing factors. Organisational factors have largely been ignored. Due to legal action concerns, there is nota single generally available database that addresses violations or intentional circumventionrelated causes of low quality in marine systems.

In all parts of the quality improvement process, data on HOE causes and effects is sadlylacking. There has not been a common vocabulary to describe direct, contributing, andcompounding causes. There is little definitive information on the rates and effects of humanerrors and their interactions with organisations, environments, hardware, and software. There is even less definitive information on how contributing factors influence the rates ofhuman errors.

Given the requirement to improve the quality of marine structures and a need to implementalternative QA / QC strategies in design, construction, and operation of marine structures,there is a pressing need to begin gathering, archiving and analysing high quality data on HOEincidence, causes, and effects. Some organisations have begun such developments. Theseefforts need to be encouraged and extended.

Given the dearth of reliable quantitative information that is presently available on HOE indesign and construction of marine systems, the analysts are left with four primary sources ofinformation to perform evaluations:

• judgment,

• simulations,

• field, laboratory, and office experiments, and

• process reviews, accident and near-miss investigations.

All of these sources represent viable means of providing quantitative evaluations. It is rare tofind a structured and consistent use of these four approaches in HOE assessments. Giventhe lack of definitive quantitative information on which to base objective quantitativeevaluations, one must rely, at least in the near future, primarily on judgement. As adequatelystructured databases are developed and implemented for HOE evaluations, then in thefuture, more reliance can be placed on objective data and evaluations based on acombination of data and judgement. Adequately qualified and unbiased judgement will beessential to produce meaningful results.

A number of researchers have published useful summaries that provide quantifiedinformation on human errors, see [Ref.17] for sources. This information has beendeveloped primarily for evaluation of HOE effects in the operations of nuclear power plants. The information was developed primarily from experiments and simulations concerninggeneral categories of human task reliability, and are shown in figures below.

Generic human error rates are assigned to general types of tasks performed under generaltypes of influences and impediments. The range of error probabilities are intended to beassociated with the potential ranges in the influences and impediments. If the influences and

Page 55: Prepared by WS Atkins Consultants Ltd for the Health and

42

impediments are intense, then the error probabilities will be toward the upper portion of therange and vice versa.

It is important to note that the severity of the error is not captured in any of the availablequantitative information. Errors are either major and significant or minor or not significant. Itis noted that minor or not significant errors are generally caught by the individual or individuals and corrected; hence their lack of importance in the assessment of humanreliability.

1 E-6

1 E-5

1 E-4

1 E-3

1 E-2

1 E-1

1unfamilar taskperformed withspeed

simple taskperformd withspeed

change systemwith procedureswith checking

routine taskstrained, motivated

respond to system commandswith supervisory system

change system statewithout procedureswithout checking

routine taskperformed withspeed or divertedattention

Figure 5.2: Nominal human task performance unreliability(o Mean; • - 1 Standard Deviation)

10-1

10-2

10-3

10-4

10-5

1new or rarely performed task

extreme stress, very little timesevere distractions & imparements

highly complex taskconsiderable stress, little time

moderate distractions & imparements

complex or unfamililar taskmoderate stress, moderate timelittle distractions & imparements

difficult but famillar tasklittle stress, sufficient time

very little distractions or imparements

simple, frequently, skilled taskno stress, no time limits

no distractions or imparementsME

AN

PR

OB

AB

ILIT

Y O

F H

UM

AN

ER

RO

RO

R F

AIL

UR

E P

ER

TA

SK

Figure 5.3: Generic human task error rates

Page 56: Prepared by WS Atkins Consultants Ltd for the Health and

43

Information also has been developed on human error performance shaping factors. Theseperformance shaping factors are influences that can result in an increase in the mean rates ofhuman errors. Simulations, experiments, and information gathered on plant operations haveprovided this information.

These “performance shaping factors” are extremely useful in helping develop quantificationof the potential effects of changes in organisation, hardware, procedures, and environmentson the base rates of human errors. In [Ref.17] the performance shaping factors arepresented in the form of “influence scales” that give the multipliers on the base rate or

.

5.5 Methodology for Quantification of Error Likelihood

The methodology used in the JIP for the quantification of the likelihood of HOE scenarios issummarised below. The methodology attempts to account for all the major human andorganisational factors which influence errors as discussed in the previous sections.

A number of potential human and organisational error scenarios are first identified based onjudgement combined with information from historical incident databases. The scenariosshould cover the design, fabrication, installation, operation and maintenance phases of astructure’s life-cycle. The likelihood of error is evaluated separately for each scenario whichis then multiplied with the conditional probability of structural failure given the error. Theprobabilities due to different error scenarios and for different failure modes are thencombined to evaluate the overall system failure probability as shown in Figure 1.6. SpecificHOE scenarios considered in this project and the results of their evaluation are presented inSection 6.

Each error scenario is broken down into the primary functions or “tasks” involved in theactivity. If desirable, a task can be decomposed into sub-tasks. As discussed in Section5.2, in each task, errors due to the Operating teams can develop due to eight differentcauses (or sources) as given in Figure 5.4. The error rates for these eight sources can beestimated from the generic error rates given in Figure 5.3 combined with judgement.

Performance shaping factors or influence scales given in [Ref.17] are used to modify thebase operator error rates to recognise the influences of Organisations, Procedures,Hardware and Environments (external, internal, social) as discussed in Section 5.2.

Page 57: Prepared by WS Atkins Consultants Ltd for the Health and

44

.

OPERATIN G TEA M

Co m m u n ic a t io n s

Se le ct io n &T ra in in g

Plann ing &P re p ar a t io n s

Lim it a t io n s &

Im p ai rm e n t s

V i o la t io n s

S l ip s

Ig n o r an c e

Mi s t a k e s

P RO CED URES

ORGA N IZA TION S

HA RD WA RE

EN V IRON MEN TS

PE =

infl

uen

ces

.

OPERATIN G TEA MOPERATIN G TEA M

Co m m u n ic a t io n s

Se le ct io n &T ra in in g

Plann ing &P re p ar a t io n s

Lim it a t io n s &

Im p ai rm e n t s

V i o la t io n s

S l ip s

Ig n o r an c e

Mi s t a k e s

P RO CED URES

ORGA N IZA TION S

HA RD WA RE

EN V IRON MEN TS

P RO CED URES

ORGA N IZA TION S

HA RD WA RE

EN V IRON MEN TS

PE =

infl

uen

ces

Figure 5.4: Operating team HOE causes and influencing factors

In determining the overall probability of error, the correlation between the above eightsources should be considered. High positive correlation in the sources could be developedby human factors such as a consistent set of high quality individuals (human), organisation,hardware, and procedures that are allowed to permeate the entire design process. Organisation culture is one of the most important of the correlating processes. Forsimplicity, full correlation can be assumed to obtain a lower-bound to the error probability,while zero correlation will give an upper-bound.

The influence of different QA / QC alternatives on the error probability can then bedetermined with a view to select the most effective option. The attention here is focused onthe major malfunctions which can be detected and corrected by the QA / QC system. Theprobability of error detection, P(D), and error correction, P(C), play an important role inreducing the likelihood of human error. The P(D) and P(C) depend on the effectiveness andintensity of the QA / QC measure used. The probability due to each error source is thenmultiplied by [1- P(D) • P(C)] which represents the probability that the error is not detectedand not corrected.

Page 58: Prepared by WS Atkins Consultants Ltd for the Health and

45

5.6 The SYRAS Software

The SYRAS software was developed within the project to facilitate the evaluation humanand organisational errors implementing the methodology described above. The errorevaluation is achieved by progressing through a number of forms or windows, each windowperforming a specific task.

After entering relevant information about platform name, location, water depth, etc. the useris presented with the Quality Attribute Form shown in Figure 5.5. The overall systemreliability can be evaluated in terms of four quality attributes: Serviceability, Safety,Durability, and Compatibility (i.e. ability to meet budget and schedule requirements). Theresults can be obtained with and without accounting for the influence of QA / QC measures.

After selecting a quality attribute, the Life-cycle Phase Form is presented as shown in Figure5.6. The results for each life-cycle phase and for each quality attribute are presented bothwith and without QA / QC. The total life-cycle probability combining all the phases is alsopresented.

By selecting one of the life-cycle phases, the user is presented with a Task Structure Form,Figure 5.7. This allows the user to input the sequences, correlation, and organisationalrelationships between tasks. The open format is composed of a six by six grid of individualtask openings to depict different series and parallel relationships. Task information is enteredinto the structure by double clicking on the task location (represented by an empty box). Double clicking the task box allows the user to either: (a) Designate the task as asupervisory task (whose Pf will be calculated using the Pf’s of sub-tasks), or (b) Designatethe tasks as a sub-task. Task relationships are defined using the horizontal and verticalcheck boxes. Horizontal boxes designate correlation, and vertical boxes designate results offragility analyses.

Information for sub-tasks is input using the Task Information Form, Figure 5.8. After doubleclicking on the desired task location for a sub-task, the user is prompted for information onbase error rate, description, and mean for the task. The user then identifies the impacts ofthe six influence factors: the Operator, the Organisation, the Procedures, the Hardware, theEnvironment, and the resulting Interfaces. Using the multipliers inherent to the influences, thenew task error rate is calculated and displayed. When the user exits the form, the data isstored. The life-cycle phase Pf is updated upon returning to the task structure form.

After exiting the task structure form, the user is taken back to the life-cycle form to selectthe next life-cycle phase. After completing all the life-cycle phases for a quality attribute, thecontrol returns to the quality attribute form to select the next quality attribute. The evaluationis completed after finishing all the quality attributes and exiting from the quality attribute form.

The data and results from SYRAS can be exported into an Excel spreadsheet forpresentation in a tabular format.

Page 59: Prepared by WS Atkins Consultants Ltd for the Health and

46

Figure 5.5: Quality Attribute Form

Figure 5.6: Quality Attribute Life-cycle Phase Form

Page 60: Prepared by WS Atkins Consultants Ltd for the Health and

47

Figure 5.7: Task Structure Form

Figure 5.8: Task Information Form

Page 61: Prepared by WS Atkins Consultants Ltd for the Health and

48

6. EVALUATION OF ERROR SCENARIOS

This section presents the work carried out by UCB and Ramboll in Task II.2 and that byWS Atkins and MSL under Task II.3. The application of the methodology for human erroranalysis to the selected structures is presented. Six dominant error scenarios identifiedcovering the different life-cycle phases and the quantification of their error probabilities aredescribed. The reliability analysis for the identified error scenarios considering extremestorm, fatigue and ship collision conditions is outlined and the results from this study arepresented and discussed. Further details can be obtained from [Refs. 17, 9, 13].

6.1 Identification of Human Error Scenarios

High consequence accidents resulting from HOE can be differentiated into those that occurin design, construction and operation phases of a marine system's life cycle. Experience hasshown that the majority of the errors occur or manifest during the operating phase, however,many of these may have root causes founded in design and / or construction errors.

Based on the information available from the World Offshore Accident Databank (WOAD)[Ref.19], the principal cause of accidents to fixed offshore platforms are seen as blowouts,collisions, fires and explosions. Of all the causes of accidents, only 6% to 9% are attributedto structurally related causes.

Examination of the major causes for damage and repair of offshore structures in the NorthSea, [Ref.20], indicates that the leading causes are fatigue (frequency of 10-3 per annum)and collision (9×10-3 p.a.). The other sources of damage included dropped objects,fabrication and installation faults, corrosion, design and operating errors and the probabilityof damage to the platforms from these sources ranges from 3 E-3 pa to 5 E-3 pa. Currentexperience also indicates that the majority of damage that is associated with accidents(collisions, dropped objects) is discovered after the incident occurs, [Ref.20]. About 60%of fatigue and corrosion damage is detected during routine inspections. However, thebalance of 40% is discovered accidentally or during non-routine inspections.

Some experience has also been available specific to minimum structures in the Gulf-of-Mexico. This suggests that the likelihood of error for these structures are typically muchhigher than the conventional jacket structures which have reached a level of ‘maturity’ in thedesign and construction.

Based on the wider experience available from the above sources combined with the specificexperience of the JIP participants, the following five HOE scenarios were identified as themost significant for the type of structures considered in this study:

1. Design: Fatigue due to pile driving stresses (Durability)

2. Fabrication: Welding & fit-up flaws (Durability)

3. Installation: Pile insertion damage (Safety)

4. Operations: Dropped production processing package (Safety)

5. Operations: Supply / work boat collision (Safety)

Page 62: Prepared by WS Atkins Consultants Ltd for the Health and

49

The evaluation of each of the above scenarios is discussed in the following sub-sections. TheSYRAS software was used for the evaluation of the error likelihood [Ref.17], while theconditional probabilities of structural collapse given damage due to each HOE scenario wereevaluated by repeating the analyses presented in Sections 3 and 4 considering extremestorm, fatigue and ship collision conditions, [Refs. 9 and 13].

6.2 Design HOE Scenario: Fatigue due to Pile Driving Stresses

This HOE scenario relates to a design error which involves “failure to design the structurefor pile driving stresses and not making provisions to allow more precise alignment of thepiles during driving”. This error could result in significant stresses transmitted to the pilesleeves and into the rest of the structure due to pile misalignment during driving. This resultsin significant fatigue cracking in all the joints of the braces connecting the pile sleeves/guidesto the primary structure. This scenario is not likely to influence the 4-pile Jacket structure asthe piles are driven through the legs from above the water level thus allowing a betteralignment of the piles.

The methodology and the SYSRAS software described in Section 5 was used for thequantification of likelihood of this error scenario. The base rates due to the eight primarycauses of error were obtained as shown in Figure 6.1.

These probabilities should reflect theinfluence of the Organisations,Procedures, Hardware, andEnvironments. These influences can bemodified by the influence scales toreflect the practices within a specificorganisation.

Combining the error rates from the eightcauses, the overall probability of thisHOE is obtained as 8.9E-3. Thedominant sources of this HOE areignorance (56%) and selection andtraining of the designers (22%). Theignorance cause is primarily attributed tothe lack of organisational communicationand defined design procedures toaddress this problem.

There can be ‘correlations’ between theeight sources of errors due toorganisational influences that impart aspecific culture within an organisationresulting in ‘group think’ biases. Suchcorrelations could be modelled through

Figure 6.1: Error rates for design HOE: Fatiguedue to pile driving stresses

. OPERATING TEAM

Communications

Selection &Training

Planning &Preparations

Limitations &Imparements

Violations

Slips

Ignorance

Mistakes

PROCEDURES

ORGANIZATIONS

HARDWARE

ENVIRONMENTS

Pe = 5 E-4

Pe = 6 E-4

PE= 8.9 E-3

Pe = 1 E-3

Pe = 1 E-4

Pe = 2 E-3

Pe = 5 E-3

Pe = 1 E-4

Pe = 5 E-4

influ

ence

s. OPERATING TEAMOPERATING TEAM

Communications

Selection &Training

Planning &Preparations

Limitations &Imparements

Violations

Slips

Ignorance

Mistakes

PROCEDURES

ORGANIZATIONS

HARDWARE

ENVIRONMENTS

Pe = 5 E-4

Pe = 6 E-4

PE= 8.9 E-3

Pe = 1 E-3

Pe = 1 E-4

Pe = 2 E-3

Pe = 5 E-3

Pe = 1 E-4

Pe = 5 E-4

influ

ence

s

Page 63: Prepared by WS Atkins Consultants Ltd for the Health and

50

SYRAS, and assuming perfect positive correlation in this case would reduce the errorprobability from 8.9E-3 to 5.0E-3.

The error likelihood can be significantly reduced by effective QA/QC procedures. With theconventional checking procedures during design the error PE can only be reduced slightly toa range 4.6E-3 ~ 8.2E-3. On the other hand using an independent ‘third party’ verificationby an experienced engineer could reduce the PE to 1.4E-3 ~ 2.5E-3.

In order to determine the effect of this HOE scenario on the reliability of the consideredstructures, the fatigue damage due to pile driving stresses was considered to be such that itreduces the original (error-free) calculated fatigue lives by a factor of 1/10 for all the joints inbraces connecting the pile sleeves/guides to the primary structure. System reliability analysisunder combined fatigue and pushover conditions was carried out with the reduced fatiguelives for the three minimum structures.

The results of error likelihood, conditional probability of structural collapse under combinedfatigue and pushover loading and the overall probability of failure considering human errorsare summarised in Table 6.1.

Table 6.1: Results for design HOE: fatigue due to pile driving stresses

StructureError Prob.

(no QA/QC)PE

Conditional Prob.of Failure

Pf|E

Total Prob.of Failure

PE×Pf|E

Error-freeProb.

Pf

4-leg Jacket - - - 3.0E-8

3-Pile Monotower 5.0E-3 ~ 8.9E-3 0.37 1.8E-3 ~ 3.3E-3 8.0E-4

Vierendeel Tower 5.0E-3 ~ 8.9E-3 0.097 4.8E-4 ~ 8.6E-4 2.0E-5

Braced Caisson 5.0E-3 ~ 8.9E-3 2.5E-10 < 1.0E-11 2.5E-10

It can be seen that this error scenario has a significant influence on the 3-pile Monotowerand Vierendeel Tower structures and practically no influence on the Braced Caisson. Giventhe damage caused by human error, probabilities of fatigue failure of affected joints andsequence of fatigue failures increase quite considerably but the pushover reliability followingfatigue failures is unaffected. As a result, the overall system probability of failure given anerror scenario is increased considerably from the error-free case for the 3-pile Monotowerand Vierendeel Tower structures. This shows that these two structures are less robustagainst the human error scenarios considered.

Page 64: Prepared by WS Atkins Consultants Ltd for the Health and

51

6.3 Fabrication HOE Scenario: Fit-up and Welding Flaws

One of the potential HOE scenario that could occur during fabrication relates to “lack ofadequate fit-up resulting in the misalignment of the surfaces to be welded”. This increasesthe local stresses at the joints significantly. Misalignment also makes welding difficultresulting in defects such as incomplete root penetration, inadequate profiling of welds, etc.The combined effect of increased local stresses and welding flaws is to reduce the fatiguelives of joints considerably. This HOE scenario is particularly crucial for minimum structuressince very small tolerances are involved and much of the work must be done above theground.

Based on a careful analysis of the tasks involved during fabrication, the base rates for theeight error sources were determined. These probabilities reflect the influence oforganisations (low bid, fast-track contracting) and procedures (difficult to achievetolerances). Hardware and environments do not have a significant influence. The dominantcauses of the HOE are selection and training of the fabrication team (less experiencedwelders selected for speed of welding and low hourly rate) (33 %), and limitations andimpairments associated with the difficulty to fit-up and align the joints (24 %). The violationsource error was influenced primarily by lack of organisational incentives to “report badnews” (difficult alignment and fit-up that will delay the schedule) and procedures to rectifythe problem.

The overall probability of this HOE scenario is determined as 2.1E-3 (for no correlation)and 0.7E-3 (for full correlation). This probability can be reduced to 1.0E-4 ~ 2.8E-4 usingappropriate QA/QC measures that involve proper training and qualification of welders,encouraging staff to report to the management any potential problems.

The effect of the HOE scenario was modelled by reducing the original (error-free) fatiguelives by a factor of 1/10 for all the joints in braces connecting the pile sleeves/guides to theprimary structure. System reliability analysis under combined fatigue and pushoverconditions was carried out with the reduced fatigue lives for the all four structures. Theresults are summarised in Table 6.2.

Table 6.2: Results for fabrication HOE: fit-up and welding flaws

StructureError Prob.

(no QA/QC)PE

Conditional Prob.of Failure

Pf|E

Total Prob.of Failure

PE×Pf|E

Error-freeProb.

Pf

4-leg Jacket 0.7E-3 ~ 2.1E-3 1.7E-4 1.2E-7 ~ 3.6E-7 3.0E-8

3-Pile Monotower 0.7E-3 ~ 2.1E-3 0.37 2.6E-4 ~ 7.8E-4 8.0E-4

Vierendeel Tower 0.7E-3 ~ 2.1E-3 0.097 6.8E-5 ~ 2.0E-4 2.0E-5

Braced Caisson 0.7E-3 ~ 2.1E-3 2.5E-10 Negligible 2.5E-10

It can be seen that this error scenario has a significant influence on the 3-pile Monotowerand Vierendeel Tower structures and practically no influence on the Braced Caisson. For

Page 65: Prepared by WS Atkins Consultants Ltd for the Health and

52

the 4-leg Jacket, although there is a significant reduction in reliability due to the errordamage, the resulting system reliability is still considered to be acceptable.

6.4 Installation HOE Scenario: Pile Insertion Damage

A potential HOE scenario during the installation phase relates to “a pile swinging away fromits guides during stabbing and damaging a brace member”. Rough weather conditions andpoor visibility makes stabbing into the underwater pile guides difficult for the minimumstructures. This could result in a damage to one of the braces in 3-pile Monotower andVierendeel Tower structures, while in the case of the Braced Caisson this could result in adamage to one of the piles while the second one is being inserted. This scenario is not likelyfor the 4-leg Jacket as the piles are driven through the legs from above the water level whichenables a greater control of the pile during stabbing.

Combining the probabilities for each of the eight error causes, the overall probability of theerror scenario was obtained as 3.7E-3. The dominant sources are limitations andimpairments (pile stabbing in difficult operating weather conditions, limited visibility of thepile guides) (54 %), and slips associated with the difficulties to align the piles andmanoeuvering them into the pile guides (27 %). These probabilities reflect influences fromthe organisations (low bid, fast-track contracting), procedures (difficult underwater pilestabbing operations), hardware (no significant influences), and environments (significantweather influences). The base rates of errors can be further modified by the influence scalesprovided in [Ref.17].

Considering correlation due to organisational influences a lower-bound of 2.0E-3 (for fullcorrelation) and an upper-bound of 3.7E-3 (for no correlation) can be obtained for the errorprobability. Adequate QA/QC would have suggested installing an additional above waterguide which would reduce the error probability to 2.0E-4 to 3.7E-4.

The damage to the structures due to this HOE scenario was modelled by selecting one ofthe critical braces in each structure and reducing its axial capacity by 25% to account for thedent/bow effects. Pushover reliability analysis (excluding fatigue deterioration) was carriedout for all the structures and the results are summarised in Table 6.3.

Table 6.3: Results for installation HOE: pile insertion damage

StructureError Prob.

(no QA/QC)PE

Conditional Prob.of Failure

Pf|E

Total Prob.of Failure

PE×Pf|E

Error-freeProb.

Pf

4-leg Jacket - 2.0E-8 - 5.0E-9

3-Pile Monotower 2.0E-3 ~ 3.7E-3 3.0E-9 negligible 2.0E-9

Vierendeel Tower 2.0E-3 ~ 3.7E-3 2.0E-8 negligible 1.0E-8

Braced Caisson 2.0E-3 ~ 3.7E-3 3.0E-9 negligible 2.5E-10

It can be seen that the probability of failure increases only marginally as a result of the pileinsertion damage compared to the “error-free” failure probability. Taking the probability of

Page 66: Prepared by WS Atkins Consultants Ltd for the Health and

53

the error scenario into account would reduce the overall probabilities of failure as a result ofthis HOE to negligible levels for all structures.

6.5 Operations HOE Scenario: Ship Collision Damage

Ship collision on offshore structures is a major cause of damage and repair in the North Seawhich has prompted the introduction of detailed risk analysis and explicit design to protectthe structures against collision. Most collisions are caused by supply boats manoeuveringnear the platform in severe seas. For the 3-pile Monotower a collision could severely dentthe central column while for the Vierendeel Tower one of the legs of the tower could bedamaged. Collision could damage the central caisson in a Braced Caisson structure, whilein a 4-leg Jacket the impact could be critical on a leg, a brace member or an intersectingnode of the X-braces.

The factors that influence the operator error include training (inexperienced boat operator),and limitations (fatigue). The organisational factors included culture, and monitoring /controlling. The hardware factors include less than desired power in the supply vessel andinsufficient fendering on the platform. The procedure factors include incomplete andinaccurate instructions to the master in conducting supply operations in severe seaconditions. The environmental factors are the severe sea conditions, and unfavourable windand current directions. There are no significant interface breakdown factors. Lack ofchecks to allow detection of the hazardous situation and correction before the accidentoccurred also influence the likelihood of this HOE.

The probabilities of each of the eightpotential causes of this are summarised inFigure 6.2.

The overall probability of this HOE scenariowas obtained as 8.7E-3. The dominantsources are selection and training of theoperating team (captain and crew) (35 %),and limitations and impairments (due tocrew and captain fatigue) (35 %). Lack ofproper planning and preparations (trying tocome along side in severe seas) and slips(using the throttles to manoeuver in sternseas) each account for 12 % of the HOElikelihood.

Considering the correlation in the errorsources, a lower-bound of 3.0E-3 (for fullcorrelation) and an upper-bound of 8.7E-3(for no correlation) can be obtained. QA/QC procedures involving hiringexperienced boat crew and captain and

.

O PE R A T IN G T E A M

C o m m u n i c a t i o n s

S e le c t io n &

T r a i n i n g

P la n n in g &

P r e p a r a t i o n s

L im i t a t io n s &

Im p a i r m e n t s

V i o l a t i o n s

S l i p s

I g n o r a n c e

M i s t a k e s

P R O C E D U RE S

O R G A N IZ A T IO N S

H A R D W A R E

E N V IR O N M E N T S

i n f l u e n c e s

PE = 8.7 E-3

P e = 3 E-3

P e = 3 E-3

P e = 1 E -3

Pe = 5 E -4

Pe = 1 E- 4

P e = 1 E-3

Pe = 1 E-5

Pe = 1 E- 4

Figure 6.2: Error sources for ship collisionHOE scenario

Page 67: Prepared by WS Atkins Consultants Ltd for the Health and

54

training them in severe sea conditions could reduce the error probability to 0.6E-3 to 1.7E-3. A second alternative of change in the regulations and procedures requiring sufficient restby boat masters and crews could reduce the error probability to 0.9E-3 to 2.6E-3.

The results of ship impact analyses for the four structures have been presented in Section 4,see Table 4.2. From this it can be seen that, within the range of ship mass and velocitiesconsidered credible in the North Sea, the ship impact damage has a significant influence ononly the 3-pile Monotower which failed during the impact event for certain combinations ofship mass and velocity. The other three structures survived the ship collision incident withoutcollapse and the resulting damage did not reduce their pushover capacity appreciably.

Reliability analysis of the 3-pile Monotower for the ship collision condition gave a probabilityof system collapse of 6.36E-4. Multiplying this with the probability of impact gives an overallprobability of failure due to ship impact of 1.9E-6 to 5.7E-06.

6.6 Operations HOE Scenario: Dropped Object Damage

Dropped object damage is a common cause of repair to North Sea platforms. This HOEscenario models “dropping of a production package during transfer from a barge to the topdeck”. The package drops through the water and strikes one of the critical vertical diagonalbraces of the structure.

The dropped object HOE could result from a series of errors committed by the riggingcrew, the work crew supervisor, the derrick operator, and the barge superintendent. Lackof experienced rigging crew (low bid, down-sized operating crew), under-sized shacklesused to attach the slings to the production package, inability of the supervisors to detect andcorrect the evolving problem, and when the package started to break loose from the slings(it had been lifted clear of the transportation barge) the barge operator could slip and put animpact loading on to the package. The package could then drop from the elevation of theupper deck and into the sea.

The base error rates for each of the above contributors to the scenario were determined andthe overall error probability of 4.1E-3 was obtained. This takes into account theorganisational (low bid, fast-track contracting) and procedural influences, while thehardware and environmental influences are considered to be less significant. The primarysources of the HOE are selection and training of the operating team (less experienced riggingteam) (24 %), lack of proper planning preparations (sling shackles not laid out and checkedfor use by rigging team) (24 %), and the slip by the barge crane operator (24 %). Ignorance and mistakes (on the part of the rigging crew) and communications breakdowns(crew supervisor and barge superintendent) were responsible for the majority of theremaining causes.

Accounting for correlation in the different error sources due to organisational influences givesa lower-bound of 3.1E-3 (for full correlation) and an upper-bound of 4.1E-3 (for nocorrelation) can be obtained for the error probability. Adequate QA/QC measures

Page 68: Prepared by WS Atkins Consultants Ltd for the Health and

55

involving the use of experienced rigging crew, close communication and supervision by thebarge operations managers could reduce the error probability to 1.4E-4 to 5.7E-4.

The dent and bow damage which could be caused by a dropped object in the selectedcritical braces of each structure was evaluated considering two impact energies of 0.5MJand 1.0MJ. To put a perspective on these energy levels, 0.5 MJ corresponds to an objectof 10 tonnes travelling at 10 m/s or 65 tonnes (eg. pile) at 4 m/s, [Ref.13].

The denting process was modelled by a non-linear P-δd characteristic obtained from thework of Pettersen and Johnsen [Ref.15]. Non-linear analyses were conducted with a pointload of increasing magnitude applied at the end of the dent spring. This allows the non-linearbowing response of member to be accurately established. The maximum and residual valuesof dent and bow were obtained by post-processing the USFOS results.

The dent and bow levels obtained were compared with the range of values recorded inpractice for North Sea structures from a database prepared by MSL for HSE and a goodcorrelation was found. The residual bow and dent values pertaining to E = 0.5MJ wereused to set the damage levels for subsequent analyses.

With the pre-existing damage due to the dropped object, the structures were subjected todynamic ship impact analyses and subsequent pushover analyses. The ship impact andpushover analyses were conducted in a similar manner to that used for intact structuresdescribed in Section 4. The results showed that only the 3-pile Monotower is significantlyaffected by the dent/bow damage. Relatively minor damage from the dropped object wassustained by the Vierendeel and Braced Caisson structures and hence their ship impact andpushover capacities remained unaffected. The jacket member, on the other hand, sufferedquite severe damage both in bow and dent but this was not seen to affect the ultimatestrength of the jacket structure.

The 3-pile Monotower failed during the impact event for certain combinations of ship massand velocity. Reliability analysis for the ship collision condition gave probability of failuredue to impact of 3.7E-3. Multiplying this with the ship impact probability of 3E-3 ~ 9E-3and dropped object error probability of 3.1E-3 ~ 4.1E-3 gives a overall probability offailure due to ship collision following a dropped object damage of 3.4E-8 to 0.13E-8.

Although the effect of dropped object damage on pushover reliability was not investigated,the results of conditional probabilities of failure given in Table 6.3 can be taken asreasonable upper-bound values. This would make the overall probability of pushover failureas a result of this HOE scenario would be negligible for all structures.

Page 69: Prepared by WS Atkins Consultants Ltd for the Health and

56

7. COMPARISON OF SELECTED STRUCTURES

7.1 General

This section compares the life-cycle reliability characteristics of the four structures using anumber of performance indicators. The key results from all the tasks which are presented inthe previous sections are synthesised and discussed. The objective of this comparison is toidentify key features which influence the reliability characteristics of each structure, and ifpossible to suggest how the performance of each structure can be improved. It is not theintention of this JIP to rank the selected concepts or to recommend any one concept as thebest.

The comparison is presented initially in terms of the following:

• Fabrication and installation effort

• Performance under extreme storm condition

• Performance under fatigue condition

• Performance under ship collision condition

• Robustness against human and organisational errors

Following this, an overall comparison is made considering all the above factors.

7.2 Fabrication and Installation Effort

The effort required in the fabrication and installation of a structure directly influences theinitial costs and the lead time from design to commissioning of the platform. Although thefabrication and installation costs were not evaluated within the project, a good indication ofthis can be obtained from the comparison of the key design features presented in Table 7.1for the four structures.

Fabrication effort/cost is influenced by factors such as:

• The total weight of jacket and pile steel

• Thickness and type of material used

• No. of tubular joints to be welded

• No. of circumferential and longitudinal butt welds in large diameter tubulars

• Split of work between covered and uncovered assembly areas

• Quantity of work carried out at height and ease of access to the work place

• Quantity of inspection required

• Size and ease of handling fabricated components

Other considerations are that fabrication in northern European yards will be more influencedby weather considerations than those in the Gulf-of-Mexico (GoM), and that different yardswill have plant and experience more suited to certain types of construction than others. It istherefore not possible to give definitive guidance on which of the substructure options it maybe cheapest to construct. After all it must be remembered that many of the MFP solutions

Page 70: Prepared by WS Atkins Consultants Ltd for the Health and

57

on the market were developed around the specific capabilities of the companies promotingthem.

The 3-pile Monotower uses a very large diameter column and relatively stocky tubulars formost of the braces. These would have to be made from rolled cans of 3 to 4 m longsections requiring a number of circumferential welds. The Braced Caisson also uses largediameter rolled cans for the caisson and the piles. The 4-pile Jacket and the VierendeelTower, on the other hand, except for the legs use small diameter members which can bemade from seamless tubes which reduces the no. of circumferencial joints and expedites thefabrication process.

Both the 4-pile Jacket and the Vierendeel Tower structures require a larger number oftubular joints to be welded but because of small diameter tubes point-to-point constructioncan be used throughout. The 3-pile Monotower and Braced Caisson need nodalconstruction and PWHT although they have far fewer nodes to be fabricated. The maindisadvantage of the 3-pile Monotower is the need to work in confined space within thecentral column to attach the various guides for the conductors and risers. The BracedCaisson has far fewer braces and joints to make but a significant amount of the constructionis carried out at the offshore site.

Installation cost depends on the type of offshore equipment used and the duration of it’s use.The GoM has a greater range and supply of installation vessels than are to be found inEuropean waters. This in turn leads to far more competitive hire rates existing in the GoMthan in Europe, where, therefore, installation cost represents a much greater proportion ofthe overall project cost than in the GoM. Typically, for a Minimum Facility Platform installedby Heavy Lift Vessel in the southern North Sea the installation cost will represent about30% of the total facility cost (topsides plus substructure). This proportion will be between15 - 20% for the GoM. It is therefore more cost effective to minimise offshore work and itsduration in the North Sea than it is in GoM. If it is assumed that all the structures are to beinstalled by heavy lift crane vessel then the probable offshore work-scope for eachsubstructure can be used as a measure of comparing the relative installation costs.

The 4-pile Jacket would be installed by driving the piles using followers. The need to insert,remove and change follower at each pile will increase the installation time compared withthat for a single one piece pile installation. For the 3-pile Monotower and the VierendeelTower the piles can be driven in a single operation thus significantly reducing the installationtime. The Braced Caisson requires a significant amount of offshore construction inconnecting the two sections of the caisson and completing the field weld at the topconnecting the caisson with the piles. The inclination of the piles may necessitate the use ofmultiple sections during driving.

Page 71: Prepared by WS Atkins Consultants Ltd for the Health and

58

Table 7.1: Comparison of design features influencing fabrication and installation effort

Item 4-Pile Jacket 3-Pile Monotower Vierendeel Tower Braced Caisson

Jacket weight: 310 t 260 t 305 t 269 tPile weight 189 t 185 t 170 t 190 t

Caisson/Leg diameter 1.25 m 2.4 m 0.88 m 2.1 m

No. of braces 64 20 60 15No. of seamless tube members 48 4 32 8No. of circumferential welds 232 100 212 82No. of tubular joints 120 36 108 26Amount of nodal construction None Considerable None ConsiderableWork at height Minimal Minimal Minimal MinimalWork in confined space None Considerable Minimal ConsiderableWork in the open Maximum Minimal Minimal SignificantCraneage required Light Heavy Heavy Heavy

Piles/caisson to be driven 4 3 4 2+1Inclination of piles 9.3° Vertical Vertical 1-Vertical; 2 @ 14°Pile Diameter 1.2 m 1.2 m 1.2 m 1.5 mPile Penetration 40 m 34 m 32 m Caisson: 36m; Piles: 41mOffshore fabrication/assembly None None None ConsiderableHammer position Above water Under water Under water Above waterPile driving method Multiple sections using

followersSingle section Single section May need multiple sections

1. Tubular sections < 610 mm diameter are assumed to be seamless tubes, the rest are rolled members2. Circumferential welds are assumed at thickness transitions and @ every 4m lengths of rolled members3. All structures are assumed to be installed using a heavy lift crane vessel

Page 72: Prepared by WS Atkins Consultants Ltd for the Health and

59

7.3 Performance Under Extreme Storm Condition

The performance of the four structures under extreme storm conditions was studied bycarrying out deterministic pushover and system reliability analyses as discussed in Section 3. The results from these analyses are summarised in Table 7.2.

Table 7.2: Comparison of performance under extreme storm condition

Item 4-PileJacket

3-PileMonotower

VierendeelTower

BracedCaisson

100-yr Wave + Current Base Shear 7.97 MN 3.40 MN 4.64 MN 3.7 MN

Load Factor for Collapse 4.09 4.40 3.97 5.00

Base Shear at Collapse 32.59 MN 15.18 MN 18.4 MN 18.5 MN

Component Reliability Index 4.62 4.91 4.44 5.2

System Reliability Index 5.73 5.91 5.58 6.23

System Failure Probability 5.0E-9 2.0E-9 1.0E-8 2.5E-10

(Note: The Load Factor above is a factor on the 100-year wave + current loading)

From the above table it can be seen that the ultimate capacities of the minimum structures,measured in terms of the wave + current base shear at collapse, are about half of that for the4-pile Jacket. However, it is interesting to note that, in terms of the Load Factor tocollapse, which measures the system reserve against environmental loading, the differencesare not that large. This is because the Jacket structure attracts relatively high environmentalloading compared to the other structures since it has many of its structural members withinthe wave zone close to the mean sea level.

0

5

10

15

20

25

10 12 14 16 18 20 22 24 26 28 30 32

Wave Height (m)

Bas

e S

hea

r (M

N)

4-Pile Jacket

3-Pile Monotower

Vierendeel Tower

Braced Caisson

Figure 7.1: Base shear versus wave height for the four structures

Page 73: Prepared by WS Atkins Consultants Ltd for the Health and

60

This is evident from the base shear versus wave height relationships for the four structuresshown in Figure 7.1. For the Jacket structure, the base shear increases much more rapidlywith wave height compared to the other structures.

For the 4-pile Jacket, 3-pile Monotower and Vierendeel Tower structures, the pushovercapacity is governed by the two vertical framing levels close to the sea bed which are in linewith the wave approach. For the Braced Caisson, the ultimate capacity is derived largely bythe axial capacity of one pile and bending capacities of the caisson and the second pile,again failure occurring close to the sea bed.

The high capacity of the jacket structure comes from the X-bracing system. The triangularbase of the 3-pile Monotower with its Z-bracing also provides an efficient load sharingarrangement resulting in high capacity. The Vierendeel Tower appears to suffer from the K-bracing system with its pushover capacity being governed by the two compressive and twotensile braces in the vertical frames in line with the wave.

The system reliability of a structure under extreme storm condition is closely related to theLoad Factor on environmental loading and not the absolute capacity of the structure. As aresult, the system reliability indices for the structures are surprisingly close to each other,with all the four structures showing comfortably high reliability levels. The system failureprobability for the Vierendeel Structure is about one order of magnitude higher than that ofthe 4-pile Jacket.

The results given in Table 7.2 correspond to the failure of the primary sub-structure with thefoundation failure being suppressed by artificially strengthening the piles. When foundationfailure was allowed, it is seen that the collapse load factors are in the range of 2.5 to 3.2with failure occurring in the foundation for all the structures. Therefore, it should be notedthat the high reliability values given in Table 7.2 will not be achieved in practice for thesestructures.

7.4 Performance Under Fatigue Condition

The degradation in pushover reliability as a result of fatigue deterioration was studied bycarrying out system reliability analyses under combined fatigue and extreme stormconditions. For each structure, various combinations of one, two, three or four tubular jointsfailing in sequence due to fatigue was studied. A service life of 20 years was used and thestructures were assumed not to undergo any inspection and maintenance over the 20 yearperiod. The probability of each fatigue failure sequence and the conditional probability ofsystem collapse due to environmental overload given the initial damage due to fatigue wereevaluated as described in Section 3. The results are summarised in Table 7.3 for all the fourstructures. For comparison the results for pure extreme storm condition are given in the lastcolumn of the table.

Page 74: Prepared by WS Atkins Consultants Ltd for the Health and

61

Table 7.3: Comparison of performance under fatigue condition

Structure Any single jointfailure by

fatigue, ββ afcf

System reliabilityunder combined

loading

Systemreliability under

pushoverloading

4-legged Jacket 1.82 5.43 5.73

3-pile Monotower 2.57 3.14 5.91

Vierendeel Tower 2.70 4.10 5.58

Braced Caisson 2.85 6.23 6.23

From the above table, it can be seen that the 3-pile Monotower and the Vierendeel Towerstructures show a significant influence of fatigue. The 4-leg Jacket shows a moderatereduction in reliability when fatigue deterioration is considered. The Braced Caisson showsno influence of fatigue as its pushover capacity is largely derived from the piles and thecaisson.

The fatigue-sensitive joints of the 3-pile Monotower all form part of the primary members ofthe underwater truss structure. It was found that progressive fatigue failure of any twotubular joints in sequence was sufficient to significantly reduce the pushover capacity of thestructure.

The fatigue-critical joints in the Vierendeel Tower are in the four pyramidal raking bracessupporting the Vierendeel frame. Fatigue failure of joints in any two out of these four braceswill lead to the platform collapsing under its self-weight. The system reliability index is thusdetermined by the probability of these fatigue sequences occurring.

7.5 Performance Under Ship Collision Condition

Time-domain, non-linear, ship/structure interaction analyses were carried out to study theperformance of the selected structures against collision from a supply vessel. The analysesaimed to determine the maximum impact capacities of the structures ignoring the presence, ifany, of conductors/risers within the impacted members (leg/caisson). A limiting dent depthof 0.7 of the impacted member diameter was chosen, and the analysis was stopped whenthis dent depth was reached or the structure failed during the impact itself. Following theimpact, a post-impact pushover analysis was carried out to determine the reduction inpushover capacity as a result of ship impact damage.

Based on a review of HSE’s collision database and considering the typical vessel sizesoperating in the Southern and Central North Sea fields, credible limits on supply vessel sizeswere set as 500 to 3500 tonne mass and impact velocity of up to 2.5 m/sec. For eachstructure ship impact analyses were carried out for a number of vessel mass and velocitycombinations to determine the ship impact capacities of the structures as detailed in Section4. The key results are summarised in Table 7.4.

Page 75: Prepared by WS Atkins Consultants Ltd for the Health and

62

Table 7.4: Comparison of performance under ship collision condition

Item 4-PileJacket

3-PileMonotower

VierendeelTower

BracedCaisson

Impact location Legat -2 m

Caissonat + 4m

Columnat -2m

Caissonat + 2 m

Impacted member diameter, D (m) 1.42 2.40 0.84 2.13

Dent depth due to impact, d (m) 1.00 0.55 0.59 1.00

Relative dent depth (d/D) 0.70 0.23 0.70 0.47

Max. vessel mass (tonnes) 3500 2500 3500 3500

Limiting velocity (m/sec) 2.5 2.0 2.5 2.5

Post-impact pushover Load Factor 3.05 3.71 2.76 3.76

Load Factor with no impact damage 3.05 3.71 2.76 3.76

From these results it can be seen that, except for the 3-pile Monotower, all the otherstructures showed adequate capacities to resist collision from vessels up to the maximumcredible values of 3500 tonne mass and 2.5 m/sec impact velocity. For the 3-pileMonotower, collapse occurred during the impact itself at velocities higher than that shown inthe table. Furthermore, although significant denting occurred at the impact location, it didnot reduce the pushover capacity of any of the structures as the impacted members did notparticipate in the collapse mechanism under pushover condition.

The 3-pile Monotower structure failed during the ship impact for certain combinations ofmass and velocity. Using these values, a reliability analysis was carried out for failure duringship impact. Taking also into account the probability of a ship colliding with the structure, asystem reliability index of 4.39 was obtained which is lower than the reliability index of 5.13under pushover condition without ship impact damage.

It should be noted that the above results ignore any damage to conductors and riser withinthe impacted member (caisson/leg) which in practice will limit the size of the vessel andimpact velocities to much lower values. For the 3-pile Monotower, Vierendeel Tower andthe Braced Caisson, the values given in Table 7.4 should therefore be considered as upperbounds to the impact capacities of these structures, while those given in Table 2.2, which arebased on component failure, should be taken as lower bound values. For the 4-pile jacket,in which the conductors and the riser are located centrally within the jacket, the ship impactcapacity can well correspond to the values given in Table 7.4.

Page 76: Prepared by WS Atkins Consultants Ltd for the Health and

63

7.6 Performance Under Human and Organisational Errors

Five human and organisational error (HOE) scenarios, covering design, fabrication,installation, and operation phases, were identified based on judgement and historical data onincidents of human error. For each of these scenarios, the probability of error occurring wasevaluated using information on generic error rates for the basic tasks involved as explained inSection 6. These error rates take account of the influences of organisations, procedures,hardware, and environments. The error probabilities were modified to take account ofcorrelation in the different error sources and also the effect of QA/QC procedures inreducing the likelihood of error.

The likely levels of defects/damage suffered by the structures as a result of human errorwere evaluated or simply estimated in some cases. The system reliability of the structureswith these defects/damage were carried out considering either pure pushover loading orcombined fatigue and pushover loading depending on whichever is the governing criterionfor the considered HOE scenario.

The key results from the above analyses are presented in Table 7.5. The “intrinsic” or“error-free” probabilities of system failure, Pf

I, under pure pushover and combined fatigueand pushover loading are given in the first row of the table. The probabilities of systemfailure as a result of “extrinsic” or “human error” causes, Pf

E, are given for each HOEscenario, and these were obtained by multiplying the human error probability (given incolumn 2) with the conditional probability of system failure given the HOE. The totalprobability of system failure is then obtained as Pf

T = PfI + Pf

E.

From the results in Table 7.5 it can be seen that, the first two HOE scenarios: (1) Omissionof pile driving stresses during design and not making adequate provisions for alignment ofpiles during driving, and (2) Fit-up and welding flaws introduced during fabrication, both ofwhich affect fatigue strength, have the most significant influence in degrading the systemreliability of 3-pile Monotower and Vierendeel Tower structures. These two structures aretherefore less robust under these HOE scenarios. The 4-pile Jacket shows only a marginalinfluence due to HOE scenario (2), while the Braced Caisson shows practically no influencefrom these HOE scenarios.

Under HOE scenario (4), only the 3-pile Monotower structure shows a significant reductionin pushover reliability as a result of ship impact damage, while the other three structuresshow high level of robustness.

The HOE scenarios (3) and (5) involving damage to one of the braces do not show asignificant impact on the system reliability of any of the four structures.

Page 77: Prepared by WS Atkins Consultants Ltd for the Health and

64

Table 7.5: Comparison of the performance of structures under various error scenarios

Scenario DescriptionError Prob.

(No QA/QC)

4-PileJacket

3-PileMonotower

VierendeelTower

BracedCaisson

“Error-free” PfI -

Pushover: 5.0E-9

Fatigue : 3.0E-8

Pushover: 2.0E-9

Fatigue : 8.0E-4

Pushover: 1.0E-8

Fatigue : 2.0E-5

Pushover: 2.5E-10

Fatigue : 2.5E-10

1. Design: Omission of piledriving stresses → Fatigue 7.0E-03

PfE = not credible

PfT= 3.0E-8

PfE = 2.5E-3

PfT= 3.3E-3

PfE = 6.7E-4

PfT= 6.9E-4

PfE = negligible

PfT= 2.5E-10

2. Fabrication: Welding and fit-up flaws → Fatigue 1.4E-03

PfE = 2.4E-7

PfT= 2.7E-7

PfE = 5.2E-4

PfT= 1.3E-3

PfE = 1.3E-4

PfT= 1.5E-4

PfE = negligible

PfT= 2.5E-10

3. Installation: Pile insertiondamage → Pushover Strength 2.8E-03

PfE = not credible

PfT= 5.0E-9

PfE = negligible

PfT= 2.0E-9

PfE = negligible

PfT= 1.0E-8

PfE = negligible

PfT= 2.5E-10

4. Production: Ship impact →Pushover Strength 6.0E-03

PfE = negligible

PfT= 5.0E-9

PfE = 3.8E-6

PfT= 8.0E-4

PfE = negligible

PfT= 2.0E-5

PfE = negligible

PfT= 2.5E-10

5. Production: Dropped objectdamage → Pushover Strength 3.5E-03

PfE = negligible

PfT= 5.0E-9

PfE = negligible

PfT= 8.0E-4

PfE = negligible

PfT= 2.0E-5

PfE = negligible

PfT= 2.5E-10

Notes: PfI = Intrinsic probability of failure of “error-free” structure

PfE = Extrinsic probability of failure of a structure damaged by human error

PfT = Total probability of system failure (Pf

I + PfE)

Page 78: Prepared by WS Atkins Consultants Ltd for the Health and

65

7.7 Overall Comparison

All the four structures show comfortably high levels of system reliability under extreme stormcondition when foundation failure is suppressed. In practice, the pushover reliability of allthe four structures will be limited by foundation capacity.

Considering fatigue deterioration over the service life of 20 years and not accounting for anyinspection and repair, the system reliability of the 3-pile Monotower and the VierendeelTower structures are reduced considerably. The reliability of the 4-leg Jacket reducesmarginally while the Braced Caisson shows no influence of fatigue.

All the structures, except for the 3-pile Monotower show adequate capacities to resistcollisions from vessels over the range of mass and impact velocities considered. Althoughsignificant denting was caused at the impact location, this did not reduce the pushovercapacity of any of the structures. The 3-pile Monotower failed during the impact event forcertain combinations of vessel mass and velocity. Although this reduced the reliability of thisstructure, the resulting value is still comfortably high.

Of the five human error scenarios considered, two scenarios which affect fatigue strengthshow considerable influence on the 3-pile Monotower and the Vierendeel Tower structures.Hence these two structures can be regarded as less robust against human and organisationalerrors.

The Braced Caisson, despite having far fewer members than the other three structures,shows a remarkably high system reliability and high robustness against HOE. This isbecause its ultimate strength is derived primarily by the central caisson and the two pileswhich are all large diameter stocky members.

Considering also the fabrication and installation effort, the Braced Caisson may be the firstchoice in the Gulf-of-Mexico where the installation costs are relatively low, while the highinstallation cost and the high risk associated with the significant offshore work could makethis concept unattractive in the North Sea. The 3-pile Monotower and the VierendeelTower structures could be the preferred options for the North Sea, provided these areadequately designed for fatigue and ship collision, and effective QA/QC procedures are putin place to safeguard against human and organisational errors. Apart from a slightly higherinitial cost, the conventional 4-leg jacket can still be a very attractive concept for the NorthSea.

Page 79: Prepared by WS Atkins Consultants Ltd for the Health and

66

8. CONCLUSIONS AND RECOMMENDATIONS

8.1 Conclusions

The life-cycle reliability characteristics of the four structures have been compared in terms ofthe following factors:

• Fabrication and installation effort

• Performance under extreme storm condition

• Performance under fatigue condition

• Performance under ship collision condition

• Robustness against human and organisational errors

Considering each of the above factors in turn, the following observations and conclusionscan be made based on the results obtained within the JIP.

Fabrication and Installation Effort

• The total weight of steel (jacket + pile) for all the four structures is very similar with theweights for the three minimum structures being at the most 10% lower than the 4-pileJacket.

• All of the structural concepts have certain strengths and weaknesses with regard to theease of fabrication which are influenced by design features. The 4-pile Jacket andVierendeel Tower benefit from the use of relatively small size tubulars allowing point-to-point construction without the need for expensive PWHT but need many braces and alarge number of joints to make. The 3-pile Monotower and the Braced Caisson, on theother hand, have much fewer braces and joints but they use large diameter rolledmembers needing nodal construction and expensive PWHT.

• For installation, the piles for the Vierendeel Tower and the Monotower can be driven insingle sections thus reducing the installation time. The Jacket and the Braced Caissonmay need multiple sections, however, the piling is done above the water level which maybe an advantage in places where underwater hammers are not readily available orexpensive.

Because of the above, the total fabrication and installation costs in the North Sea for thethree minimum structures are not likely to be considerably lower than the 4-pile Jacket. TheBraced Caisson may not be preferred in the North Sea because of the considerablefabrication/assembly work required offshore which would increase the installation risks andpush up the costs. In the Gulf-of-Mexico, due to the generally calm weather conditions andthe availability of a greater range of installation vessels, the Braced Caisson may be thecheapest option in terms of initial costs. Provided the chosen yard has the necessary plantand experience, the 3-pile Monotower and Vierendeel Tower concepts may be cheaper tofabricate and install in the North Sea.

Page 80: Prepared by WS Atkins Consultants Ltd for the Health and

67

Performance under Extreme Storm Condition

• The ultimate capacities of the three minimum structures, measured in terms of thewave + current base shear at collapse, are about half of that for the 4-pile Jacket. However, the 4-pile jacket attracts 40% to 50% higher hydrodynamic loading thanthe minimum structures for the same wave height.

• Despite the high ultimate capacity of the 4-leg Jacket, the environmental load factorson the 100-year loading at which collapse occurs are relatively similar for all thestructures, ranging from 4.0 to 5.0, with the highest value for the Braced Caissonstructure.

• Similarly, the system reliability indices under extreme storm condition are broadlysimilar for all the structures with the β index ranging from 5.58 to 6.23. Again thehighest value is associated with the Braced Caisson.

• The above results correspond to the case when failure of the foundation wassuppressed by artificially strengthening the piles and increasing their penetration depth.When foundation failure was allowed, the collapse load factors reduce to between 2.5and 3.2.

Performance under Fatigue Condition

• The 3-pile Monotower and Vierendeel Tower structures show a significant influence offatigue on system reliability. The 4-leg Jacket shows a moderate reduction in reliabilitywhen fatigue deterioration is considered. The Braced Caisson shows no influence offatigue as its pushover capacity is largely derived from the piles and the caisson.

• The fatigue-sensitive joints of the 3-pile Monotower all form part of the primary membersof the underwater truss structure. It was found that progressive fatigue failure of any twotubular joints in sequence was sufficient to significantly reduce the pushover capacity ofthe structure.

• The fatigue-critical joints in the Vierendeel Tower are in the four pyramidal raking bracessupporting the Vierendeel frame. Fatigue failure of joints in any two out of these fourbraces will lead to the platform collapsing under its self-weight. The system reliabilityindex is thus determined by the probability of these fatigue sequences occurring.

Performance under Ship Collision Condition

• For the range of ship mass (500-3500 tonnes) and velocities (up to 2.5m/s) considered,only the 3-pile Monotower structure failed during the ship impact event (for certain Mand v combinations). This has the effect of reducing the baseline pushover reliabilityindex for this structure from 5.1 to 4.4.

• The ship impact generally introduced very deep dent at the impact location, in certaincases the dent depths of up to 0.7 of the member diameter were obtained without global

Page 81: Prepared by WS Atkins Consultants Ltd for the Health and

68

collapse. In practice, conductors/riser inside the impacted member (caisson/leg) may beseverely damaged before this dent depth is reached. This was ignored in the analyseswith a view to quantify the maximum capacities of the structures.

• Although the dent must have a weakening effect on the impacted member, the memberdid not participate in the collapse mechanism under pushover conditions. Therefore,provided the structure survived the ship impact itself, the reserve strength undersubsequent pushover conditions remained unaffected for all the four structures.

• Since damage to well/riser inside the impacted member has been ignored, the resultsfrom ship impact analyses should be treated as upper bound values for the three minimumstructures. The 4-pile jacket can withstand the full range of impact scenarios consideredas there is no risk of damaging the wells. The impact capacities calculated duringconceptual design, which give critical velocities in the range 1.0 - 1.8 m/sec for a 1000tonne vessel should be taken as lower bound values for the three minimum structures asthey are based on component checks.

Performance under Human and Organisational Error (HOE) Scenarios

• Five potential human error scenarios covering the design, fabrication, installation, andoperation phases of a structure were identified. The likelihood of error for thesescenarios range from 1.0E-03 to 9.0E-03 which can be reduced by about one order ofmagnitude using effective QA/QC measures.

• Two HOE scenarios (omission of pile driving stresses, and fit-up and welding flaws)which affect fatigue strength, have the most significant influence in degrading the systemreliability of 3-pile Monotower and Vierendeel Tower structures. These two structuresare therefore less robust under these HOE scenarios. The 4-pile Jacket shows only amarginal influence while the Braced Caisson shows practically no influence from theseHOE scenarios.

• The dropped object damage to one of the critical braces did not have a significant effecton the ship collision capacity of the structures and had only a marginal effect on pushovercapacity. The relative dent depths caused by the dropped object (0.5 MJ energy) were< 0.05 for the Vierendeel Tower and the Braced Caisson, while for the Monotower andthe Jacket structures, they were around 0.2.

• The pile insertion damage during installation to one of the critical braces had no significanteffect on pushover reliability.

In summary, provided the 3-pile Monotower and the Vierendeel Tower structuresare adequately designed for fatigue and ship collision, and effective QA/QCmeasures are put in place to safeguard against human and organisational errors, allthe four structures can achieve comfortably high reliability levels.

Page 82: Prepared by WS Atkins Consultants Ltd for the Health and

69

8.2 Limitations

The focus of the JIP has been to compare the inherent reliability of the sub-structure (i.e. thejacket), and for this reason, the following considerations were specifically excluded from thestudy:

• failure of the foundation,

• damage to conductors/risers,

• wave impingement on the deck,

• wave breaking,

• fire and blast effects.

The above factors are very important, and in practice may actually govern the systemreliability of the concepts studied. For this reason, the results and conclusions from thisstudy should be used with caution.

Initial studies showed that the reliability under extreme storm condition will be limited byfoundation failure for all structures, particularly severely for the 3-pile Monotower and theBraced Caisson. Except for the 4-pile Jacket, the reliability under ship impact will begoverned by the need to limit dent depths in order to prevent damage to conductors/riserinside the impacted leg/caisson member.

The various analyses carried out in the project have been based on the North Seaenvironmental and geotechnical conditions, and standard North Sea design, fabrication,installation, and operation procedures have been assumed. Although, some effort has beenmade to generalise the reliability results to other environmental conditions, care should beexercised in extrapolating the results to other geographical locations with wholly differentenvironmental and geotechnical conditions, and design, fabrication, installation, operationand maintenance practices.

Page 83: Prepared by WS Atkins Consultants Ltd for the Health and

70

8.3 Recommendations

All the four structures, despite being designed using the conventional deterministiccomponent-based design approach, have shown high levels of system reliability. Based onthe results from this study, the following generic recommendations are made for improvingthe life-cycle reliability characteristics of the selected concepts.

• Under extreme storm condition, the reliability of all the four structures, as designed, isseen to be governed by foundation failure. For the 3-pile Monotower and the BracedCaisson the failure is initiated in the foundation before the failure of any members in thejacket. For the Vierendeel Tower and the 4-pile Jacket, axial pile failures occur afterinitial failures in the jacket but would limit the final capacity and the reliability level. Aproper design of the foundation is necessary to fully mobilise jacket capacity and toachieve the high levels of reliability reported here.

• For the 4-pile Jacket, the final failure under extreme storm condition occurred due to thefailure of the four legs in the lowermost bay close to the seabed. The jacket ultimatecapacity could be further increased by extending the pile element inside the legs up to thesecond horizontal framing level. This feature has considerably enhanced the capacity ofthe 3-pile Monotower and the Vierendeel Tower structures.

• In order to achieve acceptable reliability levels under fatigue as observed in this study, thejoints should be designed to nominal lives of at least 5 times the service life of theplatform. For the select few joints of members which are critical to the integrity of thestructure under the extreme storm condition, designing to even longer fatigue lives wouldbe beneficial for the 3-pile Monotower and the Vierendeel Tower.

• Except for the 4-pile Jacket, in which the conductors and the riser are located centrally,the ship impact resistance of all the other structures will be governed by considerationsfor preventing damage to conductors/risers. Careful attention should therefore be givento the packaging of wells/risers within the caisson/leg and the guides should beadequately strengthened. Where possible higher gap should be provided between thewell/riser and the caisson.

• For the 3-pile Monotower, the caisson failed during impact at less than the maximumspecified vessel sizes and impact velocities. This could be prevented by increasing thebending capacity of the central column by using cans of high strength steel at the twonodes where it is connected to the base structure.

• In view of their greater sensitivity to ship impact, minimum structures should be explicitlydesigned for ship collision based on a dynamic ship/structure interaction analysis. Thecost/benefits of designing to a higher impact capacity initially to reduce the risk ofexpensive offshore repair and shut-down costs later during service should beinvestigated.

• The 3-pile Monotower and the Vierendeel Tower structures are seen to be particularlyvulnerable to potential human errors which affect the fatigue strength of critical welds.

Page 84: Prepared by WS Atkins Consultants Ltd for the Health and

71

Effective QA/QC measures should be put in place to safeguard these structures againstthese HOE scenarios. In addition, designing the critical welds to longer fatigue lives andthorough inspection of welds at the fabrication yard and after installation arerecommended to minimise the risk of gross errors.

This JIP has clearly demonstrated that a life-cycle system reliability assessment offersconsiderable benefits by providing a better insight into the performance of minimumstructures. It is therefore recommended that a system reliability assessment is performedinitially at the feasibility stage of a project to select the best concept for the particular fieldand operations requirements, and subsequently during the detailed design stage to furtherenhance the life-cycle reliability characteristics of the selected concept.

In summary, minimum structures can be made as reliable as conventional jackets by betterengineering:

ü by designing for ship impact to mitigate the risk of damage to wells/risers consideringthe dynamic interaction between the vessel and the structure;

ü by designing critical welds for fatigue lives >10 times the service life;

ü by using a life-cycle system reliability-based approach during design.

Page 85: Prepared by WS Atkins Consultants Ltd for the Health and

72

9. REFERENCES

1. RAMBØLL, “Conceptual Design - Summary Report”, Job No. 978503, doc. No.340_005, rev. 1, 1999-01-05, JIP on Comparative Evaluation of Minimum Structuresand Jackets.

2. RAMBØLL, “Design Report, Conceptual Design, 3-Pile Monotower”, Job No.978503, doc. No. 340_001, rev. 1, 1998-05-11, JIP on Comparative Evaluation ofMinimum Structures and Jackets.

3. RAMBØLL, “Design Report, Conceptual Design, 4-Legged Jacket”, Job No. 978503,doc. No. 340_002, rev. 1, 1998-09-17, JIP on Comparative Evaluation of MinimumStructures and Jackets.

4. RAMBØLL, “Conceptual Design Report, Vierendeel Tower Jacket”, Job No. 978503,doc. No. 340_003, rev. 1, 1998-09-24, JIP on Comparative Evaluation of MinimumStructures and Jackets.

5. RAMBØLL, “Conceptual Design Report, Braced Caisson”, Job No. 978503, doc. No.340_004, rev. 0, 1998-10-30, JIP on Comparative Evaluation of Minimum Structuresand Jackets.

6. American Petroleum Institute, “Recommended Practice for Planning, Designing andConstruction Fixed Offshore Structures - Working Stress Design”, RP 2A - WSD,Twentieth Edition, July 1, 1993.

7. UK Health and Safety Executive, “Guidance on Design, Construction and Certification.Offshore Installations” Third Amendment to Fourth Edition, 1995.

8. Det Norske Veritas, “Environmental Conditions and Environmental Loads”,Classification Notes, Note No. 30.5, March 1991.

9. Gierlinski, JT and Rozmarynowski, B. “Task I.2 and II.3: System Reliability of Intact andDamaged Structures Under Extreme Environment and Fatigue Conditions”, Report No.WSA/AM3681/TaskI.2/Dec.99. JIP on Comparative Evaluation of Minimum Structuresand Jackets.

10. "RASOS User's Manual - Version 4", WS Atkins, Doc. Ref. WSA/AM1703, July1996.

11. Holnicki-Szulc, J. and Gierliski, J.T.: “Structural Analysis, Design and Control By theVirtual Distortion Method”, John Wiley and Sons, 1995.

12. Shetty, N.K.: “Selective enumeration method for identification of dominant failure pathsth International OMAE Conference, 1994, Houston, Texas.

13. MSL Engineering Ltd. “Effect of vessel impact on intact and damaged structures”, Doc.Ref. C209R007-Rev 1, July 1999, JIP on Comparative Evaluation of MinimumStructures and Jackets.

Page 86: Prepared by WS Atkins Consultants Ltd for the Health and

73

14. USFOS: Ultimate Strength for Offshore Structures, Computer Software, SINTEF,Norway, 1998.

15. Pettersen, E. and Johnsen, K.R. “New Non-Linear Methods for Estimation of CollisionResistance of Mobile Offshore Units.” OTC Paper 4135, Offshore TechnologyConference, May 1981.

16. Det Norske Veritas. “Load Impacts from Boats.” DNV Technical Note TN A202,1981.

17. Bea, R.G. and Lawson, R.B. “Stage-II: Analysis of Human and Organisational Factors,Report by Univ. of California, December 1997, JIP on Comparative Evaluation ofMinimum Structures and Jackets.

18. Lawson, R.B. and Bea, R.G. “SYRAS System Risk Assessment Software, Ver. 1.0”,Report by Univ. of California, December 1997, JIP on Comparative Evaluation ofMinimum Structures and Jackets.

19. Paté-Cornell, M. E. "Organizational Aspects of Engineering System Safety: The Caseof Offshore Platforms." Science, Volume 250, Nov., 1990.

20. Bekkevold, E., Fagerjord, O., Berge, M., and Funnemark, E. "Offshore Accidents, DoWe Ever Learn? A 20 years report from VERITEC's World Wide Offshore AccidentDatabank (WOAD)", Proc. Offshore Safety Conference, Brazil,1990.

21. Marine Technology Directorate. “Review of Repairs to Offshore Structures andPipelines”, Report 94/102, London, 1992.

Page 87: Prepared by WS Atkins Consultants Ltd for the Health and

Printed and published by the Health and Safety ExecutiveC0.50 4/01

Page 88: Prepared by WS Atkins Consultants Ltd for the Health and

OTO 2001/062

£15.00 9 780717 623532

ISBN 0-7176-2353-X