Otc 2001

Offshore
Technology
Conference
2001
Authored Papers
ABS & Affiliated Companies
Authored Papers
ABS & Affiliated Companies
Integrated Risk Based Design of FPSO Topsides,
Structural and Marine Systems
Risk Based ‘Optimum’ Inspection for FPSO Hulls
Reliability Methods for Deepwater Position-Mooring

Design and Analysis
FPSO Standards and Recommended Practices
A Comparitive Risk Analysis of FPSO’s with Other Deepwater

Production Systems in the Gulf of Mexico
OTC 12948
Integrated Risk Based Design of FPSO Topsides, Structural and Marine Systems
Andrew J. Wolford, James C. Lin, James K. Liming, Andrew Lidstone, & Robert E. Sheppard, EQE International, Inc.
Copyright 2001, Offshore Technology Conference

paper. The model represents systems failures more
This paper was prepared for presentation at the 2001 Offshore Technology Conference held in comprehensively than any other offshore risk assessment
Houston, Texas, 30 April–3 May 2001.
known to the authors. This level of detail was sought in order
This paper was selected for presentation by the OTC Program Committee following review of to achieve specific goals: 1) Develop explicit risk measures of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Offshore Technology Conference and are subject to class rules to facilitate prioritization and optimization, thus
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Offshore Technology Conference or its officers. Electronic reproduction, allowing one to focus resources on the greatest risk
distribution, or storage of any part of this paper for commercial purposes without the written contributors, 2) Develop a consistent means for performing
consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print
is restricted to an abstract of not more than 300 words; illustrations may not be copied. The risk tradeoffs (i.e. demonstration of equivalent level of safety),
abstract must contain conspicuous acknowledgment of where and by whom the paper was
presented. and 3) Provide a vehicle for expansion of Class or Group
services to risk significant systems, components, structures, or
human actions not currently included in Class scope.
Abstract
FPSOs and other floating offshore facilities typically follow
Model Development
“prescriptive based” classification rules for design of the hull,
The overall model is comprised of a collection of initiating
mooring and marine systems. In some cases the process
events, facility response model and consequence calculations.
facilities are also classified. An alternative approach is to use
These are categorized into discrete damage states (see Figure
a more open framework, “risk based” design approach that
1). Also illustrated in Figure 1, the facility response model is
allows variation from prescriptive rules provided system risks
comprised of support event trees, frontline event trees and end
are maintained at acceptable levels. The various classification
states. The support event trees represent the failed/operational
societies currently allow such risk-based alternatives [2, 3].
state of support systems (e.g. utilities, instrumentation and
Although numerous “component” risk studies for FPSOs have
control, emergency) required for successful operation of the
been conducted and published, this is one of the first that
frontline (main system) event trees. The end states are simply
accounts for the integration and linking of risks and risk
a discrete categorization of the various failed configurations of
tradeoffs among the hull, mooring system, marine systems,
the facility. Frontline systems were segregated into three
topside process plant and the utility, power and control
categories for convenience: process, marine and structural
systems that support them. The model and basis are first
systems. Each of these model partitions are described in
described, followed by application to a prototype deepwater,
summary in Table 1, and the development activities are
turret moored FPSO with gas handling. Example cases are
described in the following sections.
shown to demonstrate use of the model to make design
decisions to the various FPSO components.
Process Systems
The process model follows a conventional offshore QRA
Introduction
approach [1,4]:
• Development of isolatable sections,
Classification rules are established based on engineering
principles, experience, testing & expert judgement. They are • Summarize the loss of containment frequency by using a
intended to ensure probabilities of accidents are low, but this parts count approach
is not explicit. Changes to developing and implementing ABS • Identifying spatial interactions that could lead to
rules are being explored through risk based approaches. An escalation
alternative, risk-based approach to classification of Floating These steps were completed for 54 isolatable sections, using
Production, Storage and Offloading systems is being the event tree structure of Figure 2. Three hole sizes were
investigated by ABS as part of a major internal technology selected to represent the hole size distribution of various
development project. The project is comprised of model, process equipment. Multi-phase releases (oil / gas / water)
database and methodology development and training at were treated where applicable. Leak frequencies were derived
multiple levels throughout the organization. The prototype primarily from generic databases (e.g. E&P Forum, OREDA,
model was completed early in 2001 and is the focus of this Offshore Hydrocarbon Release Statistics). The model
2 WOLFORD ET AL. OTC 12948
explicitly accounts for emergency detection as well as process • Explosion overpressure impacts to humans and equipment
control response to a loss of containment event, so that risk • Simple evacuation of personnel on board.
importance measures can be associated with rule and standard-
required I&C. Example Applications
Two applications are used to highlight the utility of the model
Marine Systems described in this paper:
The marine modeling followed a broadly similar approach to • Comparison of loss of containment frequency for a
the process modeling, except the scope of marine events was production separator among material failure mechanisms
broader than loss of containment of hydrocarbons. The vs. transient induced leaks
majority of the marine event trees addressed fires (fueled, • Comparison of the design of mooring line redundancy
electrical, other). Marine event scenarios were represented requirements
with 89 unique initiating events, 12 frontline system event
trees, one support tree and 141 marine fault trees of which 89 Leak Initiators
developed specific initiating events and 52 modeled system Various codes and rules [2,3] specify the use of process
response functions (see Table 1). Over 2 billion unique event instrumentation to protect against accidental overpressure of
sequences were evaluated. Fire initiating event frequencies hydrocarbon containing equipment. These prescriptive rules
were developed for 70 individual hazard zones combined with are effective in minimizing overpressure transients that can
an assessment of initiator density. An explicit parts count potentially lead to significant loss of containment. As such,
approach was not utilized for marine fire initiation based on these measures provide non-explicit risk mitigation of offshore
unavailability of component-specific fire initiator frequency production facilities. A comparative exercise was performed
data. to determine the extent to which individual instrumentation
and control component should be represented in the risk
Structures model. The exercise formulated a fault tree model of a
The modeling of structural failures also followed a broadly prototypical production separator, and quantified the transient
similar approach, but deviated from developing a component- induced leak frequency, for major leaks, and compared to the
wise model and relied substantially upon subject matter historical, “inherent” leak frequency. The fault tree is shown
experts to identify, prioritize and structure the event in Figure 6. The results, using point estimates only, show:
sequences. The approach is illustrated in Figure 3. As shown
in that figure, structure subsystems were broken down into Inherent Leak Frequency All Leaks: 0.42/yr
mooring, turret, topside structures and hull. Team meetings Major Leaks 0.04/yr
were held to construct event sequence diagrams – which Transient Induced Leak Frequency Major Leaks: ~1E-8/yr
illustrate the sequential (hardware and human) event sequence
of events resulting in a system failure. The event sequence The result of the above exercise shows that the transient
diagram has proven to be a superior tool to portray and elicit induced leak frequency is substantially lower than the inherent
necessary expert input. An example is shown for mooring in leak frequency, hence the need to treat the individual
Figure 4. From this intermediate step, system event trees were equipment failures explicitly is unwarranted, unless a process
developed analogous to those for the process and marine system departs radically from existing codes and standards.
frontline event trees. The mooring event tree is shown in
Figure 5. Structural event scenarios were represented with 46 Mooring Line Redundancy
unique initiating events, 13 frontline system event trees and An 8-leg, external turret mooring system was evaluated with
one support tree. three hypothetical design criteria
• No redundancy
Consequence Modeling
• Single line failure (per ABS Rules)
Risk metrics were defined to measure health, safety,
environmental and financial impacts from the FPSO. The • 2 Line failure
endpoint metrics selected were: fatalities, oil spill, capital loss Parameters were modified in the FPSO risk model to
and business interruption (the latter two only if associated with incorporate modifications in failure rates and consequences.
a scenario with potential for fatality or spill, i.e. the entire The mooring event tree is shown in Figure 5. Frequencies
scope of financial impacts ere not included in this phase of the were tabulated for the four risk metrics. Figure 7 shows that
model, only a subset). substantial risk reduction can be achieved using a dual line
The physical phenomena represented in this model failure criterion.
drew upon de facto standards used in offshore QRA [1].
Summary
• Release modeling, multi-phase, near field flow regime,
A comprehensive risk model has been constructed to assist in
internal pressure-time history
FPSO design risk tradeoffs. This model is envisioned to be
• Thermal radiation effects to humans and equipment from
useful in supporting risk based classification activities.
jet fires and pool fires
OTC 12948 INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS 3
References
1. “A Guide to Risk Assessment for Offshore Installations

Part I”, John Sponge and Edward Smith, DNV Technica,
Revision 1, An MTD Multi-Sponsored Project, January
1995.
2. Guide for Building and Classing Facilities on Offshore
Installations, American Bureau of Shipping, June 2000.
3. Guide for Building and Classing Floating Production
Installations, American Bureau of Shipping, June 2000.
4. Guidance Notes on Risk Assessment Application for the
Marine and Offshore Oil and Gas Industries, American
Bureau of Shipping, June 2000.
5. “Hydrocarbon Leak and Ignition Database”, Report
11.4/180, DNV Technica, prepared for E&P Forum, June
1992.
6. “Hydrocarbon Release Statistics Review”, David
Mansfield, AEA Technology, A report produced for
UKOOA, January 1998.
7. Offshore Reliability Data, OREDA Participants,
distributed by DNV Technica, 1993.
Table 1 Description of Model Partitions
System Initiating Events Event Trees

Category
• 171 Loss of • Comprised of 54 process section
Process Containment • Each of the 2 phases in 3 separators is
• Fault Tree modeled as a separate initiating event
modeling each of • 3 Hole sizes (small, medium, large)
the 171 initiating • Model Functions
events (Parts Ø Ignition/Explosion
Count) Ø Isolation
• 57 Escalation Ø Blowdown
Events Ø Fire Suppression
• 89 (70 fires by • Failure of cargo management
Marine zone) • Ballast control failure
• Flooding from seawater system
• Flooding from cargo oil system
• Rupture of marine pressure vessels
• Energetic Release – turbine breakup
• Marine fire
• Crude oil spill
• Diesel fuel oil spill
• Pump/engine room explosion
• Inadvertent discharge of oily waste due to
bilge system failure
• Inadvertent discharge of oily waste due to
surface runoff
• 46 Structural • Single mooring line failure
Structura Damage • Corrosion holes or fatigue crack in turret
l shell
• Vessel impact with turret
• Hull damage following vessel impact or
helicopter crash
• Reduced weather vaning
• Turret superstructure or foundation damage
• Turret superstructure underdeck damage
• Process support damage
• Process support underdeck damage
• Transverse bulkhead damage
• Longitudinal bulkhead damage
• Ship hull damage during extreme weather
• Turret events following fire and explosion
Total 363 Initiating 17 Billion individual event sequences
Events
Table 2 Top Events for Mooring Trees Table 3 Top Events for Process Tree
TOP
TOP EVENT DESCRIPTION
EVENT DESCRIPTION IE Initiating event
IE Initiating Event IGNE Immediate ignition
MOORL Single mooring line fails PSL Failure of PSL for isolation
OFST Vessel drift exceeds design limits LSL Failure of LSL for isolation
MOORS Mooring system fails (Loss of 2 or more lines) PISO Failure of isolation by Process PSL/LSL
STATN Loss of station GDET Gas detection system
RSSTR Production risers experience stresses greater IGNL Late ignition
than ultimate limits FDET Fire detection
RSLOC Riser cracking occurs breaching containment ESDO Manual actuation of ESD and blowdown
TUG Tug not available for recovery operator ISOL Failure of isolation
GRND Vessel runs aground BLDN Blowdown valves
VSSL Vessel impacts passing vessel FWSP Water spray deluge suppression
FIXD Vessel impacts fixed installation FOAM Foam/water fire suppression
Figure 1. Overall FPSO Risk Model Structure
Initiating Consequence
Facility Response Model
Event Model
Damage
I.E.1 Support Frontline End X1

Event Tree Event Tree States
I.E.2 X2
E1
• Consequence •
•
E2 •
S1 S2 ••• • Analysis
• F1 F2 • • • • •
• • •
•
• • •
I.E.n Em Xl
S1
System
Component Generic Into

Population Data
Facility-Specific Data
Failure Mode ø
Figure 2 Process Loss of Containment Event Tree
Figure 3 Structural Failure Modeling Approach
System Subject
Familiariza- Matter
tion Expert
Subject
System
Matter
Breakdown
Expert
Turret Topside Hull

Mooring
Structural System
System System
System
Subject
Matter
Expert
Moderated
Meetings Subject
Matter
Expert
Build Event
Sequence
Diagrams
Build Event
Tree
Diagrams
Fre-
quency
Input
Quantification
End State
Assign-
ment
Figure 4a Mooring Event Sequence Diagram
Single Mooring No adverse

No
Line Failure effect
Yes
Design Offset Shut-in Inititaite

No No Repairs
Exceeded Production
Yes
Initiate
Yes Repairs
Over-
Mooring stress
System Failure No
Risers
Yes
Loss of
Station
Over-
stress Shut-in Inititaite
No No
Risers Production Repairs
Yes
Initiate
Yes
Repairs
Release well Shut-in Inititaite

No No
fluids Production Repairs
Yes
Initiate
Yes
Repairs
Shut-in Inititaite
No
Production Repairs
Yes
Initiate
Repairs
Figure 4b Mooring Event Sequence Diagram
Over-
Loss of
stress
Station No
Risers
Yes
Over-
stress
Risers Severed No
Risers
Release well
fluids
Shut-in
Production
Controlled Vessel Initiate

Vessel Drift
(not available for extreme weater events) Recovery Repairs
Uncontrolled
Grounding
Loss of Asset
Polution
Boat Impact
Loss of Asset
Polution
Impact with
other offshore
installation
Loss of Asset
Polution
Vessel Initiate
Recovery Repairs
Figure 5 Mooring Event Tree
IE MOORS STATN RSSTR RSLOC TUG GRND VSSL FIXD B# S#
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
Figure 6a Transient Induced Leak Frequency Fault Tree

Figure 6b Transient Induced Leak Frequency Fault Tree

Figure 7 Mooring Redundancy Risk Results
Fatalities = None
Minimum Consequences
Oil Release = 0 barrels
1E+00
Repair Cost = $10-100 Thousand
1E-01 Downtime = None
1E-02
1E-03
1E-04
Frequency
1E-05
1E-06
1E-07
1E-08
1E-09
1E-10
None
Single Line (ABS
Rules) Two Lines (BP
Proposal)
Mooring Line Redundancy
Fatalities = 1
Maximum Consequences
Oil Release = 30 - 40,000 barrels
1E+00 Repair Cost = $3-10 Million

Downtime = 1 - 3 months
1E-01
1E-02
1E-03 Low Value Facility

(well protecter)
Frequency
1E-04
1E-05
Typical Offshore
1E-06 Facility (fixed platform)
1E-07
1E-08
1E-09 High Value Facility
1E-10
None
Single Line
(ABS Rules) Two Lines (BP
Proposal)
Mooring Line Redundancy
ABS BIOGRAPHIES
ANDY WOLFORD JAMES K. LIMING
Dr. Wolford has worked in industrial James K. Liming is a Senior
risk assessment for 16 years. He has Reliability and Risk Management
directed risk applications on a Consultant and Corporate Associate
diverse range of engineered systems, with over 17 years of experience in
including offshore and onshore oil managing and performing complex
and gas installations, mobile offshore engineered facility reliability
drilling units, and marine trans- engineering, risk analysis, and
portation systems in the U.S., operations and maintenance support.
Central and South America, the He has a diverse, wellbalanced
North Sea, and offshore Malaysia background including handson
and Australia. With a focus on risk analysis and reliability power plant operating and maintenance experience as well
engineering, Dr. Wolford has worked with numerous as extensive analytical expertise. He has served as project
organizations and companies to develop quantitative risk manager or project engineer on several major industry and
assessments, which could be utilized to make more informed government risk management projects worldwide, and is
business decisions. Dr. Wolford earned his Sc.D. from the noted as a leading practitioner of probabilistic risk
Massachusetts Institute of Technology. assessment (PRA) and risk-informed asset management. In
addition to providing direct analytical support for clients, he
has also developed and presented many training workshops
JAMES C. LIN and technical papers on risk and reliability analysis
James Lin is a Senior Consultant at applications tools and techniques. He has authored or co-
EQE International with over 17 years authored over 70 publications on risk and reliability analysis
of experience in system engineering, methods and applications. He is a former fully qualified U.
reliability and availability analysis, S. Navy nuclear submarine officer, and he is currently a
and probabilistic safety assessment. Captain (O-6) and Naval Sea Systems Command unit
Mr. Lin has a wide range of Commanding Officer in the U. S. Naval Reserve. He holds a
experience that includes acting as a B. S. degree from the U. S. Naval Academy and an S. M.
project manager, performing human degree in nuclear engineering from the Massachusetts
reliability analysis, and analyzing Institute of Technology (MIT).
seismic risk. He has lectured PSA
courses and has responded to NRC review comments on
IPEs/PRA. Mr. Lin earned his B.S. in Nuclear Engineering ANDY LIDSTONE
from National Tsing Hua University, Taiwan and his M.S. in (picture unavailable)
Nuclear Engineering from University of California, Los Andy Lidstone is a Senior Engineer with EQE International
Angeles. He is a Registered Nuclear Engineer, California with over fifteen years of industry experience for offshore
and a Certified Reliability Engineer, American Society of and onshore oil and gas facilities and chemical plants. Mr.
Quality Control. Lidstone’s areas of expertise include the preparation of
hazard assessment reports, HSE Cases and probabilistic
safety assessments and reliability analysis. Mr. Lidstone is
ROBERT E. SHEPPARD equally comfortable as a teacher / lecturer of Safety Case
(picture unavailable) materials or as a project manager for the development of
Robert Sheppard, a Principal Engineer with EQE Failure Modes and Effects Analyses (FMEAs) for a full
International, has over twelve years of experience in range of drilling structures. Mr. Lidstone earned a B.Sc.
structural engineering and reliability analysis. Mr. Sheppard (Hons) in Physics from the University of Salford, England
has been involved in projects across many industries and is a Chartered Physicist and member of the Institute of
including offshore oil, onshore petrochemical, nuclear Physics.
power generation and other commercial and industrial
facilities. He has specialized in the assessment of natural and
man-made hazards including wind, hurricane, blast and
earthquake risks, and their effects on structures and systems.
Mr. Sheppard earned a B.S. in Civil Engineering from Rice
University and an M.S. in Structural Engineering from the
University of California Berkeley and he is a registered Civil
Engineer in California.
OTC 12949
Risk Based ‘Optimum’ Inspection for FPSO Hulls

T. Xu, Tao Xu & Associates, Yong Bai, American Bureau of Shipping, Mark Wang, Aker Engineering, R. G. Bea,
University of California at Berkeley

• Determination of condition of structural elements and
This paper was prepared for presentation at the 2001 Offshore Technology Conference held structural system,
•
in Houston, Texas, April 30–May 3, 2001
Disclosure of defects (design, construction, operation,
This paper was selected for presentation by the OTC Program Committee following review and maintenance),
of information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Offshore Technology Conference and are subject • Assurance of conformance with plans, specifications,
to correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Offshore Technology Conference or its officers. Electronic reproduction, guidelines, rules, and quality requirements,
distribution, or storage of any part of this paper for commercial purposes without the written
consent of the Offshore Technology Conference is prohibited. Permission to reproduce in
• Disclosure of damage, and
print is restricted to an abstract of not more than 300 words; illustrations may not be copied. • Development of information to improve design,
The abstract must contain conspicuous acknowledgment of where and by whom the paper
was presented. construction, operation, and maintenance procedures.
FPSO inspections have several levels of intensity:
ABSTRACT • General (global conditions),

• Specific (basic aspects of defects and damage), and
The increase in deepwater exploration activity has generated • Detailed (precise descriptions of flaws and other items
increased use of the Floating Production/Storage Offloading of operation and maintenance concern).
Systems (FPSO′s). Converting existing tankers is, in many FPSO inspections should be life-cycle oriented and include
cases, more economically feasible and faster than building quality assurance and control measures in:
new FPSO′s for the same purpose. For conversion of tankers
• Design,
to FPSO′s, inspection and subsequent fitness for purpose
assessment are crucially important. • Construction,
• Operation,
The objective of this paper is to present the principles and • Maintenance, and
strategies of in-service inspection programs for FPSO′s. The • Accidents / casualties.
paper summarizes the technical basis for three levels of
inspection strategies: 1) probability-based inspection FPSO inspections should be full-scope and include quality
method, 2) risk-based inspection method, and 3) ‘optimum’ assurance and control measures of the structure, equipment,
inspection method. facilities, and personnel.
Research in inspection has been conducted in regulatory
INTRODUCTION organizations, universities, and other leading organizations
[1, 2, 3, 4]. Some important guidelines have been developed
FPSO′s has many attractive features including relative low for offshore inspection practices.
cost, large working area, large water surface area, and good
The objective of this paper is to present the principles of
stability and floatability.
development of in-service inspection strategies for FPSO′s.
FPSO′s in many cases, are converted from existing tankers.
It is therefore important to have a rational and reliable PROBABILITY BASED INSPECTION
inspection method to provide information and knowledge
concerning the proposed, present, and future integrity of A number of limitations of in-service inspections have been
FPSO′s. identified [3, 4, 5], especially the significant uncertainties in
design, fabrication and damage detection, as well as the
FPSO inspections should be focused on:
adequacy of examining only a limited amount of the
structural elements. The usefulness of probabilistic models to
2 T. XU, Y. BAI, M. WANG, R. BEA OTC 12949
deal with uncertainties, as well as Bayesian models has been POD Updating
recognized. The probability-based inspection method is thus
developed to include Bayesian analysis, and probability- The POD updating is based on Bayesian approach to update
based inspection planning. of the multi-variant probability distributions, see Fig. 1. The
updated multi-variant probability distribution is used to
Bayesian Analysis recalculate the failure probability. Assuming that the defect
size distribution is FA(a) and that the probability of detection
The bulk of research on Bayesian analysis in engineering of a is PD(a). By means of Bayes' theorem (Bayes form):
application was first conducted in early 1970’s. This
P[p | q r H]• P[q r | H]
approach, defined as the probability of detection (POD) P[q r | pH] = n (5)
∑ P[q
updating method, was widely applied in aeronautical
engineering systems, such as airframes, gas turbine engines i | H]• P[p | q i H]
[6]. In the late 1980s, an alternative Bayesian approach was i= 1
developed as event updating using First Order Reliability

Method (FOSM) in the offshore industry[7]. Itagaki et al [8] The updated probability density can be determined as
developed Bayesian estimation approach in ship structures q r | H : a ≤ A ≤ a + da ,
follows: Let i.e.
based on Bayesian point estimation method.
P[q r | H] = f A (a )da , and p | q r H : inspection resulting in
Event Updating no crack detection:
This approach is to update the probability of events, such as P[ p | q r H] = 1 − PD (a ) (6)

fatigue failure directly. Bayes theorem is applied here to and the updated probability density is:
update the failure probability conditioned on additional
events, such as inspection event. It is expressed as [1 − PD (a)]• fA (a)
fA ,up (a) = ∞
P ( qr | H )
P(q r | pH ) = (7)
P( p | H )
(1) ∫ (1− P (a)) • f
D A (a)
0
The same methodology can be applied in the crack detection

where, qr|H - The original safety event , qr|pH - The fatigue inspection where the updated probability density is :
safety event after inspections, p|H - inspection event fA (a) • PD (a)
fA ,up (a) = ∞
(8)
The original safety event qr|H can be formulated by using the
critical size (e.g., critical crack size for fatigue or critical ∫ fA (a) • PD (a)da
thickness reduction for corrosion) as the failure criteria 0
qr | H = ac − a ≤ 0 (2)
Bayesian Estimation
The critical size is selected, perhaps based on serviceability
consideration. Other safety event such as brittle fracture or This approach applies Bayesian approach in statistical
corrosion threshold can be formulated. parameter (mean, standard deviation) estimation, see Fig. 2.
The inspection event p|H is formulated to describe the The statistical parameter to describe the distribution of a
additional information from inspections. Two types of event random variable X in the fatigue problem is defined as the
margins are classified based on inspection results. random variable µ . The Bayesian estimation of µ is to use
p | H = a(Ni ) − Ad ≤ 0 (3) the extended Bayes theorem in probability density as:
and f " (µ) = cL(µ )f ' (µ) (9)

p | H = a ( Ni ) − A = 0 (4) Where f (µ) = the prior distribution, L(µ ) = the likelihood
'
Equation (3) is the event that a defect size a(Ni) is not function of the observed data (objective information), c= a
normalizing factor, and f (µ ) = the posterior distribution
"
detected by an in-service inspection with the crack
delectability limit Ad at the inspection time Ni. Ad is incorporating the objective data.
described by the probability of detection (POD).
For a sample of observations xi , i=1,2,..., of X from
Equation (4) is the event that a particular defect size A is inspections, the likelihood of the sample is proportional to
detected and measured at the inspection time Ni. the product of the probability densities of X at x1, ...., xn :
OTC 12949 RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS 3
 n  • The use of a ‘living’ process that is flexible, strives for

f " (µ ) ∝ ∏ fx (x i | µ)  f ' (µ) (10) completeness, and can be easily implemented and
 i=1  updated.
The updated probability distribution of X is thus expressed • The use of qualitative and quantitative risk measures
as: • The use of effective and efficient qualitative and
quantitative methods that provide results familiar to
f x (x) = ∫ fx|µ (x | µ )f " (µ)dµ (11) inspection personnel.
The updated probability of failure is computed using the Figure 3 illustrates the overall risk-based inspection process.
The process is composed of the following steps:
updated probability density f x ( x ) .
• Definition of the system for inspection
Probability-based Inspection Planning • Use of a qualitative risk assessment that utilizes expert
judgment and experience in identifying failure modes,
With the development of Bayesian analysis, research on causes, and consequences for initial ranking of systems
probability-based inspection planning was developed. and components/elements in inspection.
Itagaki, et. al [8] and Førli [9] discussed probability based • Application of quantitative risk analysis methods,
inspection planning for ship and offshore structures. In primarily using an enhanced Failure Modes, Effects, and
probability-based inspection planning, the failure Critically Analysis (FEMCA) and treating uncertainties,
probabilities are generally expressed in terms of intersections as necessary, to focus the inspection efforts on systems
of the events of inspection and failure. Information gathered and components/elements associated with the highest
by inspection is accounted for by updating the probabilities calculated safety, economic, or environmental risk.
using Bayesian analysis. Three approaches have been • Development of the inspection program for the
developed to extend the Bayesian analysis procedure to components, using decision analysis to include cost-
inspection planning: (1) target safety margins [4,8], (2) benefit considerations.
optimum life cycle cost [4,12], and (3) combination of (1) • The inspection strategy are being updated and
and (2) [4,10]. implemented, based on the findings and experience from
the previous inspections.
RISK BASED INSPECTION Several feedback loops are shown in Figure 3 to represent a
living process for the definition of the system, the ranking of
Probability based inspection can be used to establish the components/elements, and the inspection strategy for each
component/element in-service inspection schedules on the component/element.
basis of reliability requirement of the individual critical
components (elements). However, FPSO′s usually contain a System Definition
large number of components/elements. A rational inspection
program should be developed based on system A key step in defining a system for inspection, as shown in
considerations. the first box of Figure 3, is the assembly of information that
The development of a system-level, risk-based inspection is needed for the risk-based inspection approach. In
process includes the prioritization of systems, subsystems, particular, the interviewing of key personnel, who are
and components/elements using risk measures, and definition knowledgeable for degradation mechanisms or errors that
of an inspection strategy (i.e., the frequency, method, and may not be documented, is vital to the process.
scope/sample size) for performing the actual inspections.
The process also includes decisions about the maintenance Inspection Prioritization
and repair strategies. Finally, there is a strategy for updating
the inspection program for a given system, subsystem, or The qualitative risk assessment, as included in the second
component/element, using the results of the inspections that box of Figure 3, utilizes expert judgment and experience in
have been performed. prioritizing systems, and components/elements for
inspection. A key element for this assessment is to identify
Methodology potential failure modes and causes, including design,
operational, and maintenance errors and potential
The methodology for risk-based inspection may be degradation mechanisms.
summarized as: Figure 4 shows an example result of a qualitative risk
• The use of a multidisciplinary, top-down approach that assessment in which each box is representative of a given
starts at the system level before focusing the inspection component, and a box is used to show the range of estimated
at the component/element level. consequence and failure probability. Once numbers are
placed on the axes, the risk assessment become quantitative, failure probability of the component/element is
with uncertainty being represented by the size of boxes. estimated.
• Choose appropriate action and update state of
The risk is defined as:
knowledge and information: Following the performance
Risk = (likelihood of failure)x(failure consequence) (12) of the inspection, a critical decision is faced. That is,
should the component/element be repaired or replaced if
Consequences of failure can be measured in a variety of
ways, such as injuries/deaths, economic loss, environmental significant findings occur, or should nothing be done
damages, dollars. In Figure 4, region A is high risk, region except to redefine the inspection program (going back to
B is intermediate risk, and region C is low risk. part 1 of the overall process shown in Figure 3)? Should
Components/elements are grouped according to the region in the existing inspection, maintenance, repair system be
which they fall. changed? This depends on the fitness for purpose
evaluation to determine the inspection findings and
The FMECA (Failure Modes, Effects, and Criticality potential actions on the failure probabilities. In any
Analysis) in the third box of Figure 3 is an element of the case, all of the results related to inspection should be
subjective probabilistic ranking. It provides an efficient used to update the FMECA information on a periodic
means of integrating the information required for a risk- basis to re-rank the components/elements on the basis of
based prioritization. Information on systems or risk and to redefine the inspection program.
components/elements is gathered from
• design information, RISK BASED ‘OPTIMUM’ INSPECTION
• operating experience (including prior inspection
results), Experience with in-service inspections of ship and offshore
• structural reliability and risk analysis (SRRA) structures has adequately demonstrated that there are two
results, and distinct categories of defects and damage that are found:
• expert opinion to define failure modes, failure
causes, and (perhaps) failure probability. • Those due to intrinsic causes - those that could have
been or were anticipated (natural, predictable), and
In this way, the key information is integrated to provide the • Those due to extrinsic causes - those that could not
safety, economic, or environmental risk associated with the have been anticipated (human caused, unpredictable).
systems, subsystems, and elements to develop the inspection
rankings for different systems/components. Experience with fatigue and corrosion damage found in ship
and offshore structures clearly indicates that a substantial
The probability-based inspection method developed based on amount (if not a majority) of damage falls in the second
probabilistic structural mechanics is perhaps the essential category - unpredictable and due to the ‘erroneous’ actions
part of the FMECA since it provides a rational framework to and inactions of people.
estimate failure probabilities for components/elements.
Quantitative inspection analyses (e.g. probability or risk
based inspection methods and programs) can help address
Inspection Program Development the first category of defects by providing insights into when,
where, and how to inspect and repair. However, such an
Once the FMECA is completed, and the analysis cannot be relied upon to provide information that
components/elements are ranked or categorized, the next addresses the second category of defects. Expert observation
step is to develop an inspection program for each group of and deduction (diagnostic) techniques must be used to
components. This is the bottom box in Figure 3 that is address the second category of defects.
schematically shown in Figure 5. It can also be used to Such recognizations lead to the development of the
establish an inspection program for an individual ‘optimum’ inspection method for FPSO′s. The overall
component/element or a system, as necessary. The objective of the ‘optimum’ inspection method is to develop
recommended process is divided into three basic steps: an effective and efficient safety and quality control system in
• Choose potential inspection strategies that define the the life cycle management of the FPSO′s.
frequency, method, scope, and sampling procedure for
inspection: The method of inspection includes the Inspection Performance
procedure, equipment, and level of personnel
qualification to perform the inspection. The inspection Inspection performance is influenced by the vessel, the
strategy can also take advantage of monitoring systems inspector, and the environment (Figure 6).
and maintenance test program. The vessel factor can be divided into two categories: design
• Choose an inspection strategy and perform inspection: factors and condition/maintenance factors. Design factors,
From the potential inspection strategies, defined in the including structural layout, size, and coating, are fixed at the
above step, the effect of each of these strategies on the initial design or through the redesign that may accompany
repair. Condition/maintenance factors reflect the change in a (experience) and analysis to define the elements that are
vessel as it ages, including the operation history and critical to maintaining the integrity of a FPSO. The
characteristics of individual damages/defects (crack, likelihood evaluation focuses on defining those elements that
corrosion, bucking), its size, and its location. have high Likelihood’s of being damaged and defective.
Experience and analyses are complementary means of
The person (inspector) carries out an inspection can greatly
identifying these elements.
influence the inspection performance. Performance varies
not only from inspector to inspector, but also from inspection
What?
to inspection with the same inspector based on mental and
physical condition. Factors associated with the inspector
include experience, training, fatigue and motivation. A substantial amount (if not the majority) of the damage is
unpredictable and due to the unanticipated ‘erroneous’
The environment in which the inspection is carried out has a actions and inactions of people [13].
major influence on performance. The environment factors
can be divided into two categories: external factors which Current experience also indicates that the majority of
cannot be modified by inspection procedures and procedure damage that is associated with accidents (collisions, dropped
factors that can be modified. External factors include objects) is discovered after the incident occurs [13]. About
weather and location of the vessel, that is, whether the 60% of fatigue and corrosion damage is detected during
inspection is performed while underway, while in port, or routine inspections. However, the balance of 40% is
while in dry-dock. Procedural factors reflect the condition discovered accidentally or during non-routine inspections.
during the inspection (lighting, cleanliness, temperature,
ventilation), the way in which the inspection is conducted How ?
(access method, inspection method, crew support, time
available) and the overall specification for inspection The methods to be used in FPSO inspections are basically
(inspection type). visual. In one form or another, these methods are primarily
focused on getting an inspector close enough to the surface
Inspection Strategies to be inspected so that he can visually determine if there are
significant defects or damages. However, ultrasonic gauging,
Inspections, data recording, data archiving (storage), and magnetic particle, radiographic and other nondestructive
data analysis should all be a part of a comprehensive and methods sometimes are necessary for FPSO.
optimum inspection system. Records and thorough
understanding of the information contained in these records When?
are a key aspect of inspection programs.
Inspection is one part of a ‘system’ that is intended to help There are no general answers to the timing of inspections.
disclose the presence of ‘anticipated’ and ‘unanticipated’ The timing of inspections is dependent on:
defects and damage. Development of inspection programs
• The initial and long-term durability characteristics of the
should address:
FPSO structure;
• Elements to be inspected (where and how many?), • The margins that the operator wants in place over
• Defects, degradation, and damaged to be detected minimums so that there is sufficient time to plan and
(what?), implement effective repairs;
• Methods to be used to inspect, record, archive, and • The quality of the inspections and repairs; and
report results (how?), • The basis for maintenance – ‘on demand’ (repair when it
• Timing and scheduling (when?), ‘breaks or leaks’ or ‘programmed’ (repair or replace on
• Organization, selection, training, verification, conflict standard time basis).
resolution, and responsibilities (who?), and
• Objectives (why?). Who?
• Where and How Many?
Experience has adequately demonstrated that the single most
Definition of the elements to be inspected is based on two important part of the inspection system is the inspector. The
principal aspects: skills, knowledge, motivation, integrity of the inspector are
• Consequences of defects and damage, and critically important. Equally important are the
• Likelihood of defects and damage. organizational influences exerted on the inspector, the
procedures and processes that he is required to follow, the
The consequence evaluation essentially focuses on defining environments in which he must work, and the support
those elements, and components that have a major influence hardware and systems that are provided for him to perform
on the quality and safety of a FPSO. Evaluation of the his work. Thus, the inspector is significantly influenced by
potential consequences should be based on historical data
1) organizations, 2) procedures, 3) hardware (facilities), and • To confirm what is thought: to address the intrinsic
4) environments. damages/defects that can be predicted based results from
technical analyses,
Much has been learned about how to improve the
• To disclose what is not known before inspection; to
effectiveness and efficiency of the inspector [12]. As one
address damage/defects that can not be predicted based
designs new inspection systems, it is important that the
on technical analyses, and
inspector be recognized as a part of this system [4]. • To control the predictable and unpredictable damages; to
develop high quality maintenance and repair program.
Why?
The ‘optimum’ inspection program should be started with
Inspection should have objectives of several levels: first, it the design of the structure (conception), proceed through the
provides the general information and knowledge for the life of the structure, and conclude with its scrapping (life-
quality of the in-service structures for fitness for purpose cycle). The optimum inspection program should include not
evaluation (general condition), second, it is to detect the only the hull structure, but as well, its equipment and its
damage/defects as many as possible so that effective and personnel (full scope). The optimum inspections should
efficient maintenance and repair program can be become the means to assess the general conditions of the
implemented to correct these damages/defects (quality whole structure. The optimum inspections are also the means
control and assurance), third, it is a safety control tool to to detect unpredictable flaws and damages of the structural
prevent the failure or loss of the in-service structures during elements, and permit appropriate measures to be taken to
the inspection interval (safety control and assurance). preserve the safety and integrity of the structure. The
optimum inspections are also the means to assure that all is
The inspection strategies (when, where, how, who) for going as expected, that the structural elements are
different level objectives should be different. The first level performing as expected, and that corrosion protection and
inspection should select typical elements/components to mitigation (e.g. patching pits, renewing locally excessively
provide general information about the in-service structures corroded plate) is maintained.
for fitness for purpose evaluation. It is more frequent less
detail inspection associated with long-term maintenance and The ‘optimum’ inspection method starts from the survey for
repair program. The second level (quality control) the intrinsic damage that is common for the class of
inspection should focus on the critical components/elements structures. Based on the experience, the inspection for the
to detect damage/defects as many as possible. It is intrinsic damage can be conducted in the rational way. The
associated with the short-term maintenance and repair existing risk-based inspection method discussed early this
program. The third level inspection (safety control) is to paper is the framework for the intrinsic damages/defects for
prevent the most critical damage/defects or errors to ensure the structural system. The probability-based inspection
the safe operation during the inspection interval. It is the method can be applied for the specific elements/components
most detailed and difficult inspection to identify the safety- based on the results of risk-based inspection. For the
related predictable or unpredictable damages/defects and extrinsic damage for each individual structure, the
errors. Every inspection practice for a specific fleet should knowledge-based diagnosis method should be developed.
be a combination of these three different inspection The step-by-step knowledge-based diagnosis process is the
strategies. potential means to identify the extrinsic damages.
The value of the inspection for objectives of different levels Knowledge systems now routinely do diagnosis reasoning
should also be different. The value of the first level using three methods: model-based diagnosis, heuristic
inspection is about the decision on whether or not the classification, and case-based reasoning. Our system uses a
existing structure can fulfill the purpose for extended service. combination of each of these methods: Model-Based
The value of the second level inspection is about the decision Diagnosis (MBD) to identify the details of a large class of
whether or not we should change the maintenance and repair possible problems, heuristic classification to identify the
program. The value of the third level inspection is about the presence a set of idiosyncratic problems, and Case-Based
decision whether or not we should take any intermediate Reasoning (CBR) to compare observation with previously
actions. Value analysis (value of information) can help identified cases.
make these decisions. An ‘optimum’ inspection method could include:
‘Optimum’ Inspection Method • Developing standard task checklists to ensure that
relevant data and tasks are not lost because of
The ‘optimum’ inspection method can be proactive (focused distractions or workload,
on prevention) or it can be reactive (focused on correction). • Performing global surveys to develop situation
It should have four functions: awareness for potential expected and unexpected
damage and defects,
• Assess the general conditions of the in-service offshore
structures,
• Inspecting high likelihood of damage or defect ‘parts’ Inspection Data System

and high consequence parts; if something ‘suspicious’ is
found, the inspection is intensified by model-based The single weakest component that has been found in present
diagnosis, heuristic classification, and case-based inspection systems for FPSO′s regards the data and
reasoning until root causes (not symptoms) are information that is developed during and from inspections.
determined, Little thought has been given to the efficient gathering of
• Inspecting periodically decreasing the time between data and information, even less thought to what is done with
inspections as the rate of degradation or likelihood of this data and information when it is obtained, and far less
defects and damage increase, thought given to the archiving, analysis, and reporting of the
• Inspecting after accidents or ‘early warning’ signals are data. The interfaces in the data gathering, archiving,
sensed, analysis, and reporting activities also have received a little
• Implement the long-term and short-term maintenance systematic thought. Current work has not been able to
and repair strategies based on the inspection results, identify a single coherent and optimum inspection data
• Update the IMMR (Inspection, Maintenance, system for FPSO.
Monitoring and Repair) plan based on the survey results Advances in information technology have resulted in better
and the results from maintenance and repair, ways to use information for the management of safe and
• Performing inspections that are independent from the efficient ship and offshore structures. The integration of
circumstances that cause potential defects and damage, stand-alone systems combined with improved information
and recording, organization and communication offers
• Using qualified and experienced inspectors that have substantial benefits for the life-cycle management of ship
sufficient resources and incentives to perform quality and offshore structures. A life cycle Structural Information
inspections. Management System (SMIS) is intended to facilitate the life-
cycle management of FPSO. This includes areas from design
For each FPSO, standard checklist and procedures should be
and construction as well as operations including Inspection,
established from the FPSO Structural Life-Cycle Information
Maintenance, Monitoring and Repair (IMMR). The
Management System, in order to carry out an effective
inspection data system is a component of the IMMR module
evaluation of the general condition, prior to the
in SMIS.
commencement of any general survey and include:
The general objectives of an inspection data system
• Structural drawing,
development are:
• Operating history and conditions,
• Previous damage/defects inspection results, • Collect inspection data,
• Condition and extent of protective coatings, • Store the data,
• Classification status, including any outstanding • Provide means for logic inspection data management,
conditions of class, • Allow for the organization of the inspection data in a
• Previous repair and maintenance work, form suitable for fitness or purpose analyses, and failure
• Previous information on unpredictable damage or analyses,
defects, • Analyze the data,
• Expert’s judgment and comments, • Show trends of the information such as damage/defects
• Relevant information from its sister structures. associated with structural integrity,
• Communicate and report the data.
With this information and previous inspection guidelines
regarding critical elements/subsystems in the FPSO Once a FPSO is ready for service, a series of inspections are
structural systems considered to be sites of potential scheduled according to inspection program. The objective
damage/defects based on historical data, analyses results, and and scope of the internal tank inspections are defined. The
expert’s judgment, it is possible to target the appropriate access methods and data recording methods are chosen, and
inspection strategies for the potential areas within the then the inspections are performed. The inspection results
structure for general survey and the initial scope of the including corrosion gauging, cracking, status of coating and
inspection. After the initial inspection to determine the corrosion protection systems and other structure/equipment
general condition of the system, the inspector can develop defects are updated into the corresponding database. Using
situation awareness to identify some potential unpredictable the inspectional data, maintenance and repair strategies can
critical damage/defect sites. Further knowledge-based be developed and the repair are finally carried out.
diagnosis should be conducted for these suspicious areas.
The knowledge-based diagnosis is conducted together with
detailed or specific inspections.
REENGINEERING THE IMMR PROCESS ACKNOWLEDGEMENT
FPSO in-service inspections and repairs are components in The technical views expressed in this paper are those of the
an Inspection, Maintenance, Monitoring and Repair (IMR) authors and are not necessarily those of the institutions they
system that is intended to help disclose the presence of are affiliated with.
‘anticipated’ and ‘unanticipated’ defects and damage to
critical structural details [4,12]. REFERENCES
An IMMR system is a critical part of the maintenance of in- 1. Tanker Structural Co-operative Forum (1986) Guidance
service quality of a FPSO. The IMMR process should be in Manuals for the Inspection and Condition Assessment of
place, working, and being further developed during the entire Tanker Structures.
lifetime of the structure. The IMMR process is responsible 2. Tanker Structure Co-operate Forum, (1990) Inspection,
for maintaining the quality of the structure during the useful Assessment and Experience of Old Tankers, Witherby
Marine Publication.
lifetime of the structure. A fundamental and essential part of 3. Marine Technology Directorate, (1989) "Underwater
the IMMR process is knowledge. The IMMR process can be Inspection of Steel Offshore Installations: Implementation
no more effective or efficient than the knowledge, data, and of a New Approach", Report 89/104.
experience that form the basis for the process. 4. Xu, T., Bea, R. G., (1996) " Inspection of Marine
Structures", Report for Joint Industry Research, Marine
Xu et al [4] indicates that organization of the FPSO Technology and Management Group, Dept. of Civil
inspection should be developed to: Engineering, University of California at Berkeley, Berkeley,
CA 94720.
• Define inspection processes in which only that 5. Demsetz, L.A., Cario, R., and Schulte-Strathaus, R., (1995)
information that is absolutely necessary to assure Inspection of Marine Structures, Report for Maritime
acceptable quality in the FPSO is gathered, Administration under Project Number DTMA91-93-G-
• Determine how to minimize the people and man-hours 00040.
required in the entire IMMR process, 6. Yang, J.N. and Chen, S., (1985) "Fatigue Reliability of
• Define how to minimize the steps, interfaces, multiple Structural Components Under Scheduled Inspection and
processing, and paper required in the IMMR process, Repair Maintenance," Probabilistic Methods in Mechanics
of Solids and Structures, edited by S. Eggwertz and
• Determine how to minimize the checks, controls, N.C.Lind, Springer-Verlag, Berlin, pp. 103-110.
reporting, and reconciliation required in the IMMR 7. Moan, T., (1993) "Reliability and Risk Analysis for Design
process, and and Operations Planning of Offshore Structures", Proc of
• Define how take full advantage of present computing, the 6st Intl Conf on Struct Safety and Reliability,
communications, and information technologies (CCIT). ICOSSAR'93.
8. Itagaki, H., Akita, Y. and Nitta, A., (1983) "Application of
Subjective Reliability Analysis to the Evaluation of
Inspection Procedures on Ship Structures", Proceedings of
CONCLUSIONS the International Symposium on the Role of Design,
Inspection and Redundancy on Marine Structural
This paper addresses the development of ‘optimum’ Reliability, National Academy Press, Nov. 13-16.
inspection strategies for FPSO′s. It details with the technical 9. Førli, O., (1990) "The Reliability and Cost-Effectiveness of
basis for 1) probability-based inspection method, 2) risk- Offshore Inspections", Proc. Intl Conf on Monitoring,
based inspection method, and 3) ‘optimum’ inspection Surveillance and Predictive Maintenance of Plants and
method. The probability based inspection methods are Structures, Sicily, Italy.
10. Shinozuka, M., (1990) "Relation of Inspection Findings to
recognized to be able to address only part of the potentials Fatigue Reliability", Ship Structural Committee Report,
for damage in FPSO′s. The risk-based inspection addresses SSC-355
the component/element’s damages/defects from the system 11. Marine Technology Directorate, (1994) "Review of Repairs
risk point of view. The ‘optimum’ inspection method is a to Offshore Structures and Pipelines, Report 94/102.
full-scope, life-cycle development of inspection system for 12. Bea, R.G, (1993) Ship Structural Maintenance: Recent
the risk management. The ‘optimum’ inspection method not Research Results and Experience, Transactions, The
only includes use of expert observation and analysis Institution of Marine Engineers, London
(deductive) methods but also the use of structural 13. Jones, R. B. (1995). "Risk-Based Management - A
Reliability-Centered Approach", Gulf Publishing Co.,
information system. It is demonstrated that application of Houston, Texas.
this comprehensive and integrated approach will result in 14. Platten, J. L., (1984) “Periodic (in-service) Inspection of
better allocation of the resources used to develop effective Nuclear Station Piping Welds, for the Minimum overall
inspection strategies. Radiation Risk”, Proceedings of the 5th International
Meeting on Thermal Nuclear Reactor Safety, Vol. 1.
15. Drury, C. G., Lock, M. W. B. (1996). “Ergonomics in Civil
Aircraft Inspection,” Human Error in Aircraft Inspection
and Maintenance, Paper Presented to Marine Board
Committee on Human Performance, organizational Systems,

and Maritime Safety, National Academy of Engineering,
National Research Council, Washington, D. C.
fA(a)
Probability of Failure
fa(a) fa(a)
ac Experiment/Inspection
Experiment/Inspection
Posterior
Initial crack
distribution
Original Prior
distribution Mean Crack
Size
N
a a
Updated Number of Cycles
distribution
Inspection
Figure 2 Illustration of Bayesian Estimation
Figure 1 Illustration of POD Updating
System Definition
* Defines System, System Boundary,
and fitness for purpose criteria
* Collect Information
Risk Analysis
* Define Failure Modes
* Define Failure Criteria
* Identify Consequence
* Rank Subsystem e.g. stiffened panel
* Rank Components/Elements
(1) Failure Modes, Effects, and Critically Analysis

* Redefine Failure Modes
* Redefine Failure Causes
* Redefine Failure Consequence
* Assess Failure Probabilities for the Fitness for Purpose
* Assess Conseuquences
* Risk Evaluation
* Risk Based Ranking
Risk
Analysis
(2) Development of Risk Based Inspection Program

* Choose Potential Inspection Strategies
(Frequence, Methods, Sampling Procedures)
* Define Potential for Damage States
* Define Potential Damage for Inspection Damage
* Estimate Effect of Inspection on Failure Probabilities
* Choose Inspection Stratey and Perform Inspection
* Perform Sensitive Studies
* Choose Appropriate Inspection, Maintenance, Repair (IMR) System
Figure 3 Risk-based Inspection Process

CAUSE OF DAMAGE
Operating error
Design error
C o r r o s
Installation fault
A Fabrication fault
B Dropped Object
C o l l i
F a t i g
C
0 0 0 0 0 . 1 . 1 . .
FREQUENCY 0F PLATFORM DAMAGE (% / year)
Consequences
Figure 4 Definition of Risk Index Figure 5 Causes of Damage to North Sea Structures
1. Select Potential Inspection Strategies

* Define Potential for Damage States Update
* Define Potential for Inspection Damage
* Define Reliability of Inspection Methods
2. Choose an Inspection Strategy and Perform Inspection

* Estimate Effect of Inspection on Failure Probabilities
* Estimate Effect on Potential Degradation Mechanism
* Estimate Effect of Potential Loading Condition
* Fitness for Purpose and Sensitivity Studies
(1) Obtain More

Information
3. Select (Sensitive Studies)
Appropriate
IMR System
(2) Implement
Now IMR System
4.
Later
Update State Knowledge
and Information
Figure 6 Development of In-Service Inspection Program

VESSEL ENVIRONMENT
Design External
Inspection Performance
- Weather
- Structural Layout - Location of Vessel
- Size Procedure
-Coatings
- Lighting
- Cleanliness
Condition/Maintenance - Temperatures/Humidity
INSPECTOR - Ventilation
-
Cargo - Access Method
- Experience Fatigue
- Inspection Method
Defects Training Motivation - Inspection Strategy
type age - Area to be Inspected
size location - Crew Support
Number - Time Available

- Inspection Type
Figure 7 Factors that affect Inspection Performance

ABS BIOGRAPHIES
YONG BAI
Yong Bai is Manager of Offshore

Technology in the ABS Technology
Group. He is leading and partici-
pating in the preparation and
updating of offshore classification
guides on pipelines and risers,
floating production installations and
guidance notes on ultimate strength,
fatigue/fracture and structural
analysis of jack-ups. He is also actively conducting research
on offshore structural reliability and FPSO’s.
Yong has a MSc. degree and a Ph.D. degree in naval

architecture. When he was a professor of offshore structures,
he wrote books on “Pipelines and Risers” and “Marine
Structural Design.” He has been actively involved with
structural design and analysis of pipelines, risers and
offshore platforms.
OTC 13269
Reliability Methods for Deepwater Position-Mooring Design and Analysis

Ken Huang and Yong Bai, American Bureau of Shipping, Houston
This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Limit States of Mooring Design
Limit state design methods are first described with particular
This paper was selected for presentation by the OTC Program Committee following review of reference to the design of position mooring systems for
presented, have not been reviewed by the Offshore Technology Conference and are subject to floating installations. The limit state design philosophy may
position of the Offshore Technology Conference or its officers. Electronic reproduction, be used to provide a rational framework for the design of safe
distribution, or storage of any part of this paper for commercial purposes without the written and serviceable structures or structural components, by
is restricted to an abstract of not more than 300 words; illustrations may not be copied. The accounting for uncertainties and variabilities in the basic
presented. variables affecting the design, [1, 2]. This is achieved by
describing these uncertainties and variabilities statistically,
using data from offshore practice, and calibrating the limit
Abstract
state formulation, including the deterministic and reliability
This paper describes the reliability methods for position
analysis methods used in evaluating the probability that the
mooring design and analysis with various degrees of
limit state will be exceeded, to ensure that existing accepted
complexity and with a particular emphasis on deepwater
safety levels are achieved.
applications. The limit state design and safety factor formats
The performance of a structural system, or a component of
for position mooring systems of various floating platforms are
the structure, may be described by a set of limit states
first addressed. This paper addresses technical issues of both
(limiting conditions) beyond which the structure, or
passive and active mooring systems including winch or
component, is no longer deemed to satisfy the design
thruster assisted and dynamic positioning systems. The
requirements. Limit states can be regarded as a discrete
modeling uncertainties of numerical analysis, mooring
representation of a more general continuous loss function. In
components, environmental conditions and reliability methods
general, the failure of a structural system or component to
are addressed. This paper presents a logical approach to
satisfy the design requirements may be represented by an
reliability analysis for the calibration of partial safety factors
inequality of the form:
for both passive and active mooring systems.
g ( X 1 ,K , X n , C ) ≤ 0 (1)
Introduction Where, g is the limit state function, X1,…, Xn are the basic
As the development of offshore oil and gas is moving into variables associated with the limit state, and C are constraints
deeper waters, floating platforms are used for drilling and/or describing the limits of acceptable accelerations, offsets,
production operations. For these operations, position mooring clearances, etc. The acceptable performance of position
systems are required in order to keep the floating platforms on mooring systems depends on system responses to
station under the design environmental criteria of wind, environmental actions and the system’s strength, in intact and
current and waves. Various types of floating platforms and damaged conditions. Additionally, moored installations may
mooring systems can be considered for offshore applications. depend on the availability of active systems, such as thrusters
For mooring strength limit states, partial safety factor formats and winches, to control or reduce critical system responses.
have been proposed recently in the offshore industry. It is In general, the evaluation of mooring system limit states is
important, however, to fully understand how to use the more complicated than the evaluation of component limit
reliability methods to calibrate the partial safety factors states. In the analysis of position mooring systems, computer
selected. The subject matter is significant to the offshore programs, databases, model tests, statistical models, empirical
industry for enhancing the reliability of position mooring rules and assumptions are used in evaluating the complex non-
systems with cost-effective designs. This is especially linear coupled behavior of the vessel, mooring lines, and
important for deepwater applications because the risers. All of this is symbolized by the limit state function g in
consequences of position mooring system failure would be eqn. (1). For example, linear potential theory may be used to
much more costly than in relatively shallow waters. synthesize approximate first and second order vessel forces
2 KEN HUANG AND YONG BAI OTC 13269
and motions in irregular seas. Wind and current force recommended in present codes developed and evolved to
coefficients may be derived from databases for generic vessel serve the needs of mobile offshore drilling units (MODUs). In
types or from wind tunnel tests on scale models. particular, the methods of analysis and design criteria
Characteristic and extreme values of response parameters may contained in API codes were originally developed based
be calculated from statistical models and assumed mainly on the operational experience of chain and wire
distributions and calculated output parameters from one catenary spread moored semisubmersible drilling units, [3 –
analysis program are often used to define input variables or 7]. Over the past twenty years these methods of designing
coefficients in a second program. mooring systems for MODUs have been used worldwide and
The basic variables that determine the behavior and have proved to provide adequate safety levels for
responses of moored systems fall into one of the following conventionally moored semisubmersibles, [8]. More recently,
categories. spread moored semisubmersibles, spars, and turret moored
(1) Action or load variables - associated with intensity and ships, with or without thruster assistance, using taut or
spatial variations of actions or loads, such as; wind and wave catenary mooring systems, and synthetic mooring line
intensities, spectra, and directions, current profile, velocity, components as well as chain and wire, have been used as
and direction, mooring line pretensions, vessel draft/ballast, permanent production installations. API and ISO have
etc. developed codes that also address the design requirements of
(2) Component and material variables - associated with these types of installations, [9 – 12]. However, in these codes
component and material properties, such as; mooring the combined effects of variability in environments and
component break strength, fatigue endurance, density, system responses and strengths, for different types of vessel,
elasticity, soil strength, etc. mooring systems, and environments, have not been fully
(3) Geometric variables - associated with the structural system accounted for. Consequently, safety levels inherent in today’s
or component geometry, such as; vessel shapes/dimensions, codes are largely arbitrary and cannot ensure either an
mooring line diameter, line segment length, line angles, water optimum level of safety or economy in construction. Due to
depth, etc. the historical experience on which code development has been
(4) State variables - associated with combined properties of based, the analysis methods and safety factors recommended
the system or component and the environment, such as; wind, by the present API and ISO codes, [10, 11], are most
wave drift, and current force and moment coefficients, applicable to conventionally moored semisubmersibles.
response amplitude operators (RAOs), wave frequency natural
periods and damping, mooring line and vessel drag and added Safety Factor Formats
mass coefficients, low frequency damping, etc. When limit states are evaluated using deterministic
(5) Active system variables - associated with the performance methods, the basic variables (X1,…, Xn) are replaced by their
and mean time to failure and repair of the mechanical and characteristic design values (y1,…, yn), and a set of safety
electrical, distribution, monitoring, and control systems and factors (γ1,…, γm) or safety elements are introduced which are
the experience and training (human factors) of the personnel chosen to ensure a minimum level of safety, [2]. This allows
operating the active systems. the limit state checking equation used in deterministic design
(6) Modeling uncertainty variables - associated with the methods to be represented by:
physical (hydrodynamic and structural theories) and statistical g ( y 1 ,K , y n , C , γ 1 , K , γ m ) ≤ 0 (2)
models and the reliability methods, which are used in For example in evaluating the mooring system’s strength
calculating system responses, component and system limit state, the characteristic values of wind, wave, and current
strengths, endurance, etc. and the notional reliability levels of are usually chosen as the most probable maximum values
components and systems. associated with the design return period, and conservative
For position mooring systems, strength limit states are ranges (safety elements) of peak periods and relative wind,
usually associated with the maximum installation, survival, wave, and current directions are considered. While the
operational, disconnect, and reconnect criteria in their most mooring line break strength is represented by the catalog break
extreme design environments. Depending upon the inspection strength of the component, including a reduction in diameter
and maintenance program, strength may also depend on (safety element) to allow for corrosion losses. In this
damage accumulation; that is on the fatigue, corrosion, and traditional design method some safety elements are pre-
wear endurance limit states, generally in the long-term applied (spectral peak period and wind, wave, and current
environment. While, operational performance is usually directions) while intact, damaged, and transient line tension
related to offset and motion limit states in the maximum safety factors are post-applied.
operational environment. These limit states are normally API RP 2SK and the ISO station keeping standard, [10,
subdivided further. For example, in the maximum design 11], both use a total safety factor that is applied to the mooring
environment individual limit states will address extreme line system responses, post-applied, and contain some references
loads, total turret loads, anchor loads and line angles, vessel to pre-applied safety elements. Generally, line strength post-
offsets, motions, and clearances etc. for intact, damaged, and applied safety factor formats for intact, damaged, and transient
transient cases. cases may be expressed as:
Design requirements, criteria, and the safety factors
OTC 13269 RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS 3
BS
≥ γ µ Tµ ( y1 , K , y n ) + γ TLF (y1 ,K , yn ) + γ WF TWF (y1 , K , yn ) simple and practical pre-applied safety factor format that will
γm LF
reduce the variability in safety levels across the range of types
(3)
or of positioning systems and environments can be devised.
(
BS ≥ γ m γ µTµ (y1 , K , yn ) + γ LF )
TLF ( y1 , K , y n ) + γ WF TWF (y1 , K , yn )
Limit State Design
Where,
Limit states define the limiting conditions beyond which
BS = catalog break strength of mooring component
the structural system or component is no longer deemed to
Tµ = mean tension, nonlinearly dependent on (y1,…, yn)
satisfy the design requirements, consequently, there is a
TLF = low frequency tension, nonlinearly dependent on
correspondence between limit states and traditional design
(y1,…, yn)
criteria. By calibrating individual limit states, the limit state
TWF = wave frequency tension, nonlinearly dependent on
design method may be used to achieve more consistent
(y1,…, yn)
position mooring designs than is presently possible by the use
(y1,…, yn) = characteristic design values of the basic variables
of the traditional (un-calibrated) design methods specified by
on which line tensions depend
today’s codes.
γm = mooring component partial safety factor In some cases it is not practical, or possible, to express
γµ = mean tension partial safety factor design criteria in terms of a limit state function that allows
γLF = low frequency tension partial safety factor numerical evaluation. For example, protection against
γWF = wave frequency tension partial safety factor corrosion and wear of mooring line components is normally
provided for by increasing the component diameter, jacketing
Design methods that are based on a safety factor format of the line, using blocking compounds, providing galvanic
this type, eqn. (3), are known as load and resistance factored protection etc. and specifying inspection and maintenance
design (LRFD) methods. In the case of the API and ISO programs. Not all operational limit states depend upon the
mooring codes only a single or total safety factor is specified. position mooring system, for example processing operations
Where there is uncertainty or variability in the basic may be limited by wave frequency vessel motions and
(input) variables, post-applied safety factor formats of the accelerations which in many cases are unaffected by the
LRFD type will produce inconsistent safety levels for mooring mooring system.
systems that have different nonlinear response characteristics. In general, position mooring systems are redundant series-
Additionally, inconsistent levels of safety may also result systems, in which system responses, including line-loads, are
where the variability in environmental conditions differs from nonlinearly dependent on the environmental (wind, wave, and
site to site. For example, the coefficient of variation current) actions. Redundancy is provided for in the number of
associated with 100-year return period wind speeds for Gulf of mooring lines (load paths), and recognized in mooring codes
Mexico hurricanes and North Sea winter storms are by specifying intact and broken line factors of safety. Single
approximately 13% and 5% respectively, and wind loads vary anchor leg moorings (SALMs), and some portions of the load
as the square of the wind speed. Consequently, mooring paths in single point moorings (SPMs) and catenary anchor leg
systems designed to the same post-applied (LRFD) safety moorings (CALMs) may not provide redundancy. These
factors in the Gulf of Mexico and the North Sea will be systems may be susceptible to single point failure and details
associated with different notional reliability or safety levels, of their design require special consideration. For example, in
this is true even for systems which display identical, linear or the design of SALMs API RP 2SK recommends that the level
nonlinear, response characteristics. of structural redundancy be recognized in selecting an
It may be possible to achieve more consistent designs, in appropriate safety factor.
terms of safety levels, by the use of more complicated safety Mooring codes, [10, 11], recognize redundancy in position
factor formats. One possibility is to use a post-applied safety mooring systems by specifying design requirements and safety
factor format, eqn. (3), with the values of the partial safety factors for intact, damaged, and transient cases. However,
factors defined as functions of vessel/positioning system type these codes do not allow for differences in nonlinear system
and environment. Alternatively, a pre-applied partial safety response characteristics, which depend upon the type of
factor format, eqn. (2), could be developed, where the values vessel, type of mooring system, and water depth. Or for
of the partial factors are related to the uncertainty (coefficient differences in mooring line (series-system) strength, which
of variation) in the basic variables. For example, the depends on the type of mooring component (chain, wire, or
characteristic design value of environmental variables could synthetic rope) and the length of the line segment. Nor do
be defined as the value associated with the mean plus n- these codes allow for differences in the variability of the
standard deviations, instead of the most probable (modal) environmental actions, which depend upon geographic region
value that is currently used. and season. In addition, position mooring systems may be
At present the industry has no experience with mooring entirely passive systems, or they may depend on active
codes that use pre-applied safety factor formats, and such intervention. Active station keeping systems make use of
design codes would represent a major departure from the thruster assistance or adjustment of mooring line lengths to
design methods that have evolved with the offshore industry. reduce critical system responses. While some turret designs
Additionally, it has yet to be demonstrated that a reasonably require power generation and functioning hydraulic equipment
to release brakes that hold the turret in its normally locked Reliability Analysis of Position-Mooring Systems
condition. Guidance and suggestions are discussed next to assist in the
Thrusters may be used to assist the mooring system by application of reliability methods to the design of position
controlling the vessel’s heading, reducing the mean mooring systems.
environmental forces, damping low frequency motions, or a Methods of probabilistic structural analysis, reliability
combination of these functions. In high current environments analysis, provide a rational means of dealing with the
adjustment of mooring lines may be used to position the vessel uncertainty and unpredictability in the basic variables on
up current. Mooring lines may be adjusted in preparation for which the design of mooring systems depend, by means of the
storm conditions. If the position mooring system depends application of probability theory, statistics, and deterministic
upon the use of thrusters or line length adjustments to satisfy structural theory. Reliability analysis methods are concerned
the design requirements, then the capability of the thruster with quantifying the ‘measured chance’ that a structure will
system or the ability to perform winching operations, in the support the loads to which it is subjected, [1]. In particular,
associated environmental conditions, must be assessed. In reliability methods provide approximate means for evaluating
evaluating the capability of active systems, it is necessary to the notional probability of failure of limit states under defined
consider the equipment that support and control the thrusters actions (loads). In principle, the notional probability that the
or winches, their modes of failure, repair times, and the structure as whole will satisfy the design requirements over its
training and experience of the personnel operating the design life may be synthesized from individual limit state
systems. Even for the most advanced mechanical and notional probabilities of failure.
electrical systems there is a risk, albeit small, of a total loss of Some examples of position mooring systems and
all electrical power (blackout). Consequently, wherever environmental distributions are illustrated on Figure 1 and a
possible active systems should be designed to “fail-safe” and general discussion encompassing other vessel types and
the consequences of a power blackout should be known for all station keeping systems is contained in API RP 2SK, [10].
high-risk operations. The vessel and mooring system are subject to wind, wave, and
For passive mooring systems, structural reliability methods current actions and mooring system responses are complex,
use models furnished by physics and probability theory to coupled, and nonlinear. Mean, low, and wave frequency
calculate the notional probability that a limit state will be vessel offsets, motions, and line tensions are system responses
exceeded. Evaluation of notional probabilities of failure for that are nonlinearly dependent on the environmental actions
active position mooring limit states requires the evaluation of and interactions (coupling) among the vessel, mooring system
the probability, availability, that the active systems and environments. The degree and type of system
(mechanical and electrical, distribution, monitoring, control, nonlinearity is a function of vessel type, mooring system type
operators, etc.) perform on demand in the relevant and water depth, and the wind, wave, and current intensities
environmental conditions, in addition to the evaluation of the and directions. These system responses enter into the limit
mooring system’s passive structural reliability. That is, the state functions that define acceptable performance of the
evaluation of notional safety levels for active position mooring position mooring system, thus, the design of position mooring
systems requires both availability and reliability analyses of systems is concerned with the global analysis of nonlinear
the active and passive systems on which station keeping systems. Additionally, the variability in environmental
depends. actions depends upon the geographic region and season, which
If limit state design methods are to be used to arrive at determine the type of storm that the system is exposed to.
optimum designs for position mooring systems, through Absolute values of component or system probabilities of
reliability based calibration, then the calibration of the limit failure can only be calculated if the deterministic mooring
states must allow for differences that exist in: analysis method, and the data and statistical models used to
• System redundancy, which depends on type of mooring represent the vessel, mooring system, and the environment are
system completely known and the reliability analysis method is
• System response characteristics, which depend on the type perfect. However, reliability methods provide a powerful tool
of vessel, type of mooring system and components, and for comparing the safety implications of different designs.
water depth Target safety levels for new designs, may be established by
• Mooring line segment strength, which depends on mooring performing reliability calculations for old designs that have a
component material and construction and length of the line history of satisfactory performance, in this way absolute
segment values of failure probabilities are not required. For these
• Environmental variability, which depends on the reasons, probabilities of failure calculated using reliability
geographic region and season methods are referred to as notional probabilities of failure, and
• Passive or active system reliability, which depends on target notional failure probabilities for limit states of new
structural reliability, mechanical and electrical system designs are established by calibrating to old designs for which
availability, operator training, and human factors satisfactory past experience exists.
The results of reliability computations will depend upon:
(1) The Deterministic Mooring Analysis –
The hydrodynamic and engineering/analytical models and
their ability to predict the nonlinear coupled behavior of mean wave drift and drag forces and moments, and the mean
different system types. wind and current forces and moments. Most mooring analysis
(2) The Basic Variables – programs consider force and moment equilibrium for the
The variables treated as random variables in the reliability surge, sway, and yaw degrees of freedom. In general, mean
analysis and their statistical descriptions, including modeling wave, wind, and current force and moment coefficients
uncertainties, mooring component and line strength models, depend on the vessel’s mean pitch, roll, and yaw angles with
and the modeling of environmental conditions. respect to the environmental actions. For turret moored ships,
(3) The Reliability Method – the mean offset position depends on the mean yaw equilibrium
The methodology and assumptions used to simplify the heading and its stability, which must be established before the
reliability analysis and synthesize system notional mean surge and sway forces are calculated. While for spars,
probabilities of failure. mean environmental loads are also sensitive to the mean pitch
If the resulting notional probabilities of failure, and and roll angles, requiring an iterative solution. Due to the
comparisons between notional probabilities of failure across small water plane area of spars, vessel set-down, heave, may
different types of moored systems, are to be meaningful then also significantly effect mean system responses. For all vessel
these aspects of the reliability analysis methodology and their types, the mean current velocity and direction affect wave drift
influence on the final results must be understood and and drag forces and moments, and the relative importance of
accounted for. the wave forces and moments depend upon the type of vessel
and environment, through the environmental force coefficients
Deterministic Analysis of Position Mooring Systems and relative magnitude of the environmental actions. Low
The results of reliability computations depend directly and wave frequency system responses are dependent on the
upon the deterministic analysis used to model the physical mean equilibrium position, consequently any error in the mean
behavior of the position mooring system. For each position vessel position will also result in errors in the prediction of
mooring system, a large number of deterministic mooring low and wave frequency responses. The capability of the
analyses (typically of the order of 103 to 105) are required to mooring analysis method embodied in the computer
calculate intact and damaged notional failure probabilities program(s) used to calculate mean loads and system responses
using Level 2 or Monte Carlo reliability methods. As a must be assessed for the types of vessel, mooring systems,
consequence reliability analyses are usually based on water depths, and environments for which reliability analyses
frequency domain methods of analyzing the mooring system. are performed.
Frequency domain analysis methods decompose the total
response of the mooring system into mean, low frequency, and Low frequency modal responses
wave frequency responses. Statistical values of the low and Ideally, low frequency responses of the vessel and mooring
wave frequency responses are then calculated and combined system should be decomposed into the system’s normal modes
with the mean response to yield the maximum response in a (eigen-modes), or an approximation of the normal modes.
specified storm duration. There are a number of ways in Generally, the normal modes will depend on the type of vessel
which system responses may be decomposed into mean, low and mooring system. For spread moored semisubmersibles,
frequency, and wave frequency responses and by which the the surge, sway, and yaw modes provide a good
statistical values of low and wave frequency responses may be approximation to the normal modes. Whereas for turret
calculated. For example, the mean equilibrium position of a moored ships, sway and yaw responses are coupled, while for
spar may include the effects of pitch and roll on the wind, spread moored spars, the surge and pitch responses and sway
wave, and current force coefficients and allow for vessel set- and roll responses are coupled. The methods of decomposing
down in calculating mean line tensions. The low frequency low frequency responses recommended by API RP 2SK and
modal decomposition may use surge, sway, and yaw modes the ISO code, [10,11], are most applicable to spread moored
for all position mooring systems, or may use different modal semisubmersibles.
decompositions for turret moored ships, spars, and spread Low frequency motions about the vessel’s mean offset
moored semisubmersibles. The definition of characteristic position are excited by the slowly varying components of the
low and wave frequency tension responses may be calculated wind, wave reflection, wave drag, and current forces and
based on a single vessel position, or may make use of different moments. The calculation of the energy dissipated by the
vessel positions for each mooring line. In this section details vessel, mooring lines, and risers, during the low frequency
of the analysis methodology that have a bearing on the oscillations of the vessel (the low frequency damping) is one
notional probabilities of failure calculated in the reliability of the most difficult problems in the analysis of moored
analysis of position mooring systems are discussed. systems. In general, low frequency system damping consists
of the following components:
Mean response • Still water damping
The vessel’s mean offset position is calculated by • Damping due to wind
requiring that the mean environmental loads and the net • Damping due to current
mooring system restoring forces and moments are in • Damping due to wave reflection
equilibrium. The mean environmental load is composed of the • Damping due to wave drag
• Damping due to mooring line and riser motions While, the linear method incorrectly predicts the same
minimum and maximum surge offsets for both stiff and soft
Low frequency damping is nonlinear and increases with offset directions.
the severity of the environment and the amplitude of the low The example above illustrates the effect of nonlinear
and wave frequency motions of the vessel, mooring lines, and mooring system force response on minimum and maximum
risers. Consequently, low frequency damping can only be low frequency surge offsets. For turret moored vessels the
calculated as part of a coupled analysis in which low mean environmental moment contributes to the rotational
frequency damping is solved for iteratively in time domain. stiffness of the long period sway-yaw mode, and the
From the system properties at the mean offset position, the nonlinearity of the mean environmental moment with respect
low frequency force and moment spectra, and the low to the vessel’s heading will generally result in large
frequency modal damping, the statistical values of low differences between the linearly and nonlinearly calculated
frequency modal responses are calculated. API RP 2SK and minimum and maximum modal responses. Moored systems
the ISO code present formulae for a linear frequency domain display different degrees of nonlinearity, where reliability
method of calculating significant and maximum low frequency analyses are used to compare safety levels for different vessel
modal responses. However, linearization of the combined and mooring system types, the ability of the analysis
mean environmental and mooring system forces and moments method(s) to predict low frequency responses should be
with respect to offset, can introduce large errors in the investigated and where possible calibrated against model test
prediction of significant and extreme low frequency offsets results.
and line tensions.
For a simple case, differences in the low frequency surge Combining low frequency modal responses
response predicted by linear and nonlinear methods are API RP 2SK and the ISO code recommend that wave
illustrated on Figure 2. The upper diagram on Figure 2 shows frequency responses are calculated at the mean plus significant
the 3x3 grouped mooring pattern with the vessel at the mean low frequency offset position and at the mean plus maximum
equilibrium position and defines the x-y coordinate system. low frequency offset position. However, neither code
The net mooring system restoring force, Fx, increases more specifies how the combined significant and maximum low
rapidly (stiffer) for negative offsets than for positive offsets, frequency offsets are to be calculated from the low frequency
this is shown on the graph on Figure 2. Both equally spaced modal responses, nor do the codes discuss the need to define
and grouped mooring patterns display non-linear force versus these positions differently depending on the system response
offset characteristics. Figure 2 also shows the mooring system under investigation. The low frequency meanderings of the
potential energy, calculated by integrating the nonlinear vessel will involve, at least, surge, sway, and yaw motions,
mooring system restoring force, Fx, with respect to offset, x, which may be mapped in a three-dimensional space where the
and the quadratic approximation to the potential energy (x, y, z) axes represent surge, sway, and yaw offsets measured
calculated from the linearized system stiffness, Kx. For both from the mean equilibrium position. In this space the
linear and nonlinear methods, the mean energy and the significant and maximum low frequency offset positions
associated standard deviation, σx, of the surge motion (rms referred to by API RP 2SK and the ISO code are represented
response) is calculated from the low frequency surge force by an infinite combination (a surface) of surge, sway, and yaw
spectrum and the linearized system properties and damping at values. Mooring codes contain no guidance or discussion as
the mean offset position. For a given storm duration the to how the low frequency significant and maximum design
maximum energy in the surge mode, Emax, is related to the values should be selected from the significant and maximum
mean energy, Emean, by: low frequency surfaces. The “worst” position on the
significant and maximum low frequency surfaces will depend
Emax = Emean2ln(Nx ) upon the system response under investigation. For example,
where in the calculation of maximum line tensions the “worst”
position of the vessel associated with the mean plus maximum
1 2
Emean = Kxσx = mean energy of the low frequency surge res low frequency response will depend upon the line under
2 investigation. In the case of surge only motions illustrated on
Kx = linear mooring system surge stiffness at the mean offs Figure 2, the mean plus maximum low frequency tension for
σx = standard deviation (rms) of the low frequency surge re the bow lines will occur at the minimum extreme vessel offset
Nx = T T , T is the storm duration,
andTN is the surge period (turn-around) point. While the mean plus maximum low
N frequency tensions for lines in the other two groups will occur
(4) at the maximum extreme vessel offset point.
The minimum and maximum low frequency offsets that In reliability analyses of strength limit states, system
are calculated from the (nonlinear) system potential energy notional probabilities of one and two line failures are
and the (linear) quadratic approximation to the potential synthesized from individual line notional probabilities of
energy are also shown on Figure 2. The nonlinear method failure. If the individual mooring line notional probabilities of
yields different results for minimum and maximum offsets in failure are to be calculated consistently, then the mooring
the stiff and soft offset directions, which are as expected. analysis software must use the individual line’s “worst”
significant and maximum low frequency vessel position in its own will show considerable scatter for nominally identical
calculating mooring line wave frequency tensions. This is experiments. If reliability calculations are used comparatively
particularly important where reliability analysis of grouped then the modeling uncertainty must be carefully investigated
and equally spaced moored systems are being compared. and defined for the cases being compared.
For full scale mooring systems the definition of modeling
Wave frequency line tensions uncertainty requires a comprehensive instrumentation
Present codes, [10, 11], provide formulae for calculating package, monitoring environmental loading parameters and
significant and maximum wave frequency line tensions from system response parameters, and the occurrence of severe
the standard deviation of the line tension. These formulae weather to ensure that the responses of interest are captured.
assume a frequency domain method that is based on; Alternatively, wave basin tests on scale models may be used to
transforming the vessel motion RAOs to the fairleads, assist in defining the model uncertainty across a range of
synthesizing the fairlead motion spectra from the wave vessel and mooring system types, for different environmental
spectrum and the fairlead RAOs, and solving linearized criteria. For the mean, low frequency, and wave frequency
dynamic equations for mooring line tensions in the frequency components of mooring line tensions, their modeling
domain. The principle result of the frequency domain analysis uncertainties are post-applied, as factors, to the line tension
is the standard deviation of the mooring line tension, from components calculated by the deterministic mooring analysis,
which significant and maximum line tensions are calculated in which the distributions of the other basic variables are pre-
assuming a Rayleigh distribution. Alternatively, the applied. The modeling uncertainty is meant to include the
significant and maximum fairlead motions may be calculated effects of conservative or non-conservative assumptions, bias,
from the fairlead motion spectra, assuming a Rayleigh in the methods used to predict line tensions, as well as the
distribution, and significant and maximum line tensions may naturally occurring scatter in the line tensions.
then be calculated dynamically by imposing the significant
and maximum fairlead motions. These are both frequency Progressive failure and environmental return periods
domain line dynamic methods of calculating wave frequency The definition of the basic variables describing the
tensions, however due to differences in the way in which the environmental parameters determine, the reoccurrence period
linearization is performed the two methods will result in that the notional probability of failure is associated with.
different predictions of wave frequency line tensions. The When a mooring line fails it will be retrieved and replaced at
reliability analyst should be aware of the assumptions used in the earliest possible opportunity, therefore the use of long-
the deterministic analysis and their effects on the calculated term environmental distributions is not sensible when
wave frequency line tensions. evaluating damaged line limit states. Consider the case in
which the i’th and j’th lines are the first and second lines to
Basic Variables and Probability Distributions fail in a storm. The probability that this occurs is given by
In reliability analyses the uncertainty in the basic variables Pf(i)Pf(j | i), where Pf(i) is the intact notional probability of
that enter into the limit state function are represented by failure of the i’th line and Pf(j | i) is the conditional probability
probability distributions. Sensitivity studies may be used to of failure of the j’th line, given that the i’th line has already
assist in deciding which variables should be described as failed. Clearly, Pf(i) and Pf(j | i) cannot be calculated based on
random and which may take fixed values in the reliability the long-term environmental distribution, as both lines must
analysis. The results of reliability analyses are generally fail within a short time interval of each other. Thus, the
sensitive to the tails of a few key probability distributions and notional probability of failure of limit states associated with
in most cases there is insufficient data available to accurately damaged mooring systems cannot be evaluated using long-
define the distributions and parameters of the basic variables. term environmental distributions. However, notional
The basic variables and their distributions for the description probabilities of failure for damaged limit states may be
of uncertainty in the reliability analysis of position mooring synthesized from notional intact and damaged probabilities of
systems are further discussed below. failure calculated for individual storms, i.e. using the “short-
term” environmental distributions.
Modeling uncertainty If the reliability analysis is performed using the design
Regardless of the sophistication of the deterministic tool (100-year) return period environmental distributions, then the
used to analyze the system, it is necessary to define the calculated notional probability of intact and damaged failure is
modeling uncertainty, as the modeling uncertainty provides the conditional probability of failure given the occurrence of
the link between the results of theory with reality. Neglecting the design (100-year) return period storm. This has the
to include the model uncertainty in the reliability analysis is advantage that it preserves a direct connection between the
identical to defining it as having a (mean) value of 1.0 with a intact and damaged safety factors used in the deterministic
standard deviation of zero. This is equivalent to making the design analysis and the notional intact and damaged
two claims that; theory is perfect and cannot be improved probabilities of failure, as both are associated with the same
upon and the experimental data show absolutely no scatter. In (100-year) metocean conditions. However, for geographic
many cases of interest real responses are sensitive to boundary regions with very different extreme environmental
conditions or initial conditions and the experimental data on distributions, such as winter storm and tropical revolving
storm areas, comparisons of notional probabilities of failure Hence, if the parent distribution is a Weibull distribution
conditional on the occurrence of the design return period for maxima, the distribution for N-year return values is also a
storm may not be a true indicator of comparative safety levels. Weibull distribution for maxima with threshold (εN), scale
From the response-based point of view, the better approach to (αN), and shape (βN) parameters given by:
obtain annual notional failure probabilities is to perform εN = ε
reliability analyses over a range of return period weather bins,
α (8)
and summing the product of the conditional notional αN = β
probabilities of failure with the annual probability of N1
occurrence of the weather bins. βN = β
This self-locking property applies to all three of the
Environmental basic variables extreme value distributions, [19], and is extremely useful in
Gumbel (FT Type I) or Weibull (FT Type III) extreme deriving consistent distributions for series systems.
value distributions are generally used to describe the
intensities of environmental parameters. The 100-year return Line segment strength basic variables
period distributions of wind and current velocity and the Similarities in the maximum storm event (the maximum
significant wave height are defined by Weibull distributions value in a series of N years) and mooring line strength (the
for maxima. A method for deriving consistent distributions of minimum value in series of N test lengths), allow the same
environmental parameters for arbitrary return periods is statistical methods to be used for both problems. A mooring
outlined below. line is a typical series system and the prediction of mooring
Typical metocean design criteria for the Gulf of Mexico line strength is one of the classic problems in extreme value
and North Sea, based on ISO 13819-1, [18], are summarized theory. If F(x) is the distribution function for the break
in the Tables on Figure 3. Parameters for distributions of strength of test lengths, then [1 - F(x)] is the probability that
annual maxima, which are consistent with the design criteria, the strength of a test length exceeds x. For a mooring line
are readily obtained by fitting the distributions to the 10 and segment composed of N similar independent and identically
100-year (P = 0.9 and P = 0.99) return period design values. distributed lengths the probability that the strength of all N
Weibull distributions for Gulf of Mexico and North Sea lengths exceeds x is given by [1 - F(x)]N, and the probability
annual maxima are shown at the bottom of Figure 3. These that the strength of the line segment is less then x is
distributions are consistent with the 10 and 100-year design 1 - [1 - F(x)]N. That is, the strength distribution, FN(x), of a
criteria, note that the probability of exceedence (1 – P) is line segment composed of N test lengths in series is given by:
plotted on the y-axis.
F N ( x ) = 1 − [1 − F ( x ) ]
N
The Weibull distribution for maxima is given by: (9)
  β  
 exp  −  ε − x   ; x ≤ ε , α > 0, β > 0  The Weibull distribution for minima is given by:
(x) =    α     
FWmax   β 

1 ; x > ε

 1 − exp −  x − ε   ; x ≥ ε , α > 0, β > 0 
(5) FWmin ( x ) =    α   
 
The three parameters ε, α, and β are the threshold, scale, 0 ; x <ε 
and shape parameters respectively. If F(x) is the distribution (10)
function of annual maximum values, and these annual events If the distribution of test length break strengths, F(x) in
are independent and identically distributed, then the eqn. (9), is a Weibull distribution for minima, then the
probability, FN(x), that in N-years the random quantity X does distribution of strength for a line segment composed of N test
not exceed x is given by: lengths, FN(x), is also a Weibull distribution for minima, with
N N
∏ ∏ F ( x) = [ F (x )]
N
(x) = prob ( X ≤ x ) =
N
F threshold (εN), scale (αN), and shape (βN) parameters given by
i =1 i=1
(6) eqn. (8). The formulae above allow break strength to be

defined consistently for line segments of different lengths
If the distribution for annual maxima is a Weibull
(number of test lengths N).
distribution, then the distribution of N-year return values,
Mooring lines may be made of chain, steel rope or
FN(x), is given by:
synthetic rope, etc. of different sizes, constructions, and grades
resulting in different strength distributions for line segments.
The parameters of the parent distribution for the strength of
mooring line test lengths should be determined from a
   β 
    statistical analysis of break test data. As mooring line
   e − x  b  
N
 ε − x  
[ FWmax ( x )] =  exp  −     = exp  −  segments are typically of the order of 10 to 1,000 times longer
N
F N (x) = 
   a     α 
   1 β  
   N   than the lengths used in break tests the strength of a line
(7) segment will depend upon the left tail of the test break
strength distribution. The method used to determine the “best-
fit” parameters for the break test data should be chosen distributions of basic variables, rather than the computing
carefully. Plotting the “best-fit” distributions on the same requirements of the Monte Carlo method. Monte Carlo
graph as the break data, see Figure 4, will allow the fit to the methods are often used to validate other methods, [20].
all-important left tail of the strength distribution to be However, due to the speed of present day computers, there is
examined. no reason why Monte Carlo methods should not be used for all
the reliability analyses of position moored systems, rather than
Sensitivity studies and other basic variables simply for validation, [8].
In addition to modeling uncertainty, environment, and Level 3 methods evaluate eqn. (11) by “exact” methods.
strength, other variables whose uncertainties are judged to be For Level 3 methods to be feasible, the joint distribution of the
important, either by experience or by sensitivity studies, basic variables X1,…, Xn must be known. The failure
should be represented as random variables in the reliability probability Pf can then be written as the n-dimensional
analysis. Deterministic sensitivity studies which encompass integral
∫ K ∫ f (x)dx K dx
the range of vessel and mooring system types, water depths, , {x ∈ℜ n : g(x) ≤ 0}
x 1 n
and environmental conditions over which reliability analyses
are to be performed can be used to assist in identifying those (13)
variables that should be represented as random variables. The Where fX denotes the density of the distribution of X=(X1,…,
sensitivity plots on Figure 5 show the tension in the most Xn). Numerical integration may then be used for the
loaded line for variations of ±15% in the 1-hour average wind evaluation of the integral above. However, this is not practical
velocity, pretension in line 2 and heave natural period. for multiple degrees of integration. These methods are often
The probability distribution used for each basic variable not feasible due to a lack of knowledge of the "true" joint
should be based, as far as possible, on the statistical analysis distribution of the variables, or because of the complexity of
of supporting background data. In general, normal, the integration required, [1, 20].
lognormal, Gumbel, or Weibull distributions should be used. Level 2 methods are approximate techniques for
When no detailed information is available, the three-parameter calculating Pf by replacing the limit state function, eqn. (11),
lognormal and Weibull distributions should be used to by an approximating function, which renders the problem
represent variables that are known to have lower or upper amenable to analysis. In Level 2 methods the original basic
limits, while the normal and Gumbel distributions should be variables, (X1,…, Xn), are transformed into independent
used for distributions that are unlimited. standard normal variables, (Z1,…, Zn), which implies a
transformation of the original g-function to a new limit state
Classification of Reliability Methods function f. Working with the n-dimensional standard normal
Reliability theory provides methods for the approximate distribution is computationally convenient due to the rotational
evaluation of the failure probability, symmetry of its density and because the probability content of
P f = Prob (g ( X 1 , K , X n , C ) ≤ 0 ) (11) a half-space can be easily represented as Φ(-β), where Φ is the
standard normal distribution and β denotes the distance of the
In order that eqn. (11) can be evaluated the joint
separating hyperplane from the origin. Level 2 methods
probability distribution of the basic variables has to be known,
involve an iterative search for the minimum distance of the
or alternatively independence of X1,…, Xn is assumed and
failure surface from the origin, in the n-dimensional
knowledge of the marginal distributions suffices. The choice
hyperspace, [8], as shown on Figure 6. First order reliability
of the marginal distributions arises from statistical studies
methods (FORM) approximate the limit state surface, at the
analyzing available data for the respective basic variables.
closest point to the origin, by its tangent hyperplane, and the
The Law of Large Numbers, [19], allows the following
probability of failure is approximated by the probability
frequency interpretation of the probability of failure for large
content of the half-space lying near the failure region, [22, 23].
l,
1 Second order reliability methods (SORM) also utilize the
Pf = × number of indices i = 1, K , l with g ( X i , C ) ≤ 0 curvatures of the failure surface in arriving at the
l
(12) approximation of the failure probability, [22].
Where Xi, i = 1, 2,…, l is an independent sequence of random Level 1 methods are design methods that use a number of
vectors each having the distribution of X=(X1,…,Xn)T. partial safety factors, which are related to pre-defined
Monte Carlo methods exploit eqn. (12) to arrive at an characteristic values of the basic variables, to provide the
approximate value of Pf, [20], using independent random necessary level of structural reliability. That is, Level 1
samples from the joint distributions, if they are known, of the methods are not methods of structural reliability, but methods
variables, X1,…, Xn. As the failure probability of the limit of safety checking used in design, [2].
state function decreases the computing time required by The main limitations of level 2 methods, illustrated on
Monte Carlo methods increases. However, in practice the Figure 6, are those associated with any optimization process
accuracy with which small probabilities of failure can be which are:
calculated will usually be controlled by the lack of data (1) Convergence of the processes does not guarantee that the
available to define the tails, extreme values, of the global minima has been found.
(2) Only one limit state may be evaluated at time, i.e.
separate iterative searches for minima must be performed Operation, and Maintenance of Marine Drilling Riser
for each mooring line in intact and damaged conditions. Systems, API RP 2Q, 2nd Ed., American Petroleum
The Monte Carlo method may be used to evaluate the Institute, 1993.
reliability of position mooring systems. Monte Carlo methods 8. Calibration of ABS, API, HSE, and NMD Mooring
do not involve an iterative search, consequently any number of Design Codes for Drilling and Production Platforms,
limit states may be evaluated simultaneously. This is Mooring Code JIS, Noble, Denton & Associates, Inc.
particularly useful for mooring systems in which a number of Rpt. No. H3687, 1995.
the mooring lines contribute significantly to the total notional 9. Recommended Practice for Design, Analysis, and
intact and damaged probabilities of one and two line failures. Maintenance of For Floating Production Systems, API
RP 2FP1, 1st Ed., American Petroleum Institute, 1993.
Conclusion 10. Recommended Practice for Design and Analysis of
The following can be concluded from this paper: Stationkeeping System for Floating Structures, API RP
(1) Where there is uncertainty or variability in the basic 2SK, 1st Ed., American Petroleum Institute, 1997.
(input) variables, post-applied safety factor formats of the 11. Draft ISO Standard for Design and Analysis of
LRFD type will produce inconsistent safety levels for Stationkeeping Systems for Floating Structures,
mooring systems that have different nonlinear response Informative April 14, 1999, Normative Nov. 8, 1999,
characteristics. ISO/TC67/SC7/WG5/P5 Committee Correspondence.
(2) The use of pre-applied safety factor formats has yet to be 12. Recommended Practice for Design, Manufacture,
demonstrated that a reasonably simple and practical Installation, and Maintenance of Synthetic Fiber Ropes
format can be devised. for Offshore Moorings, API RP 2SM, Draft, American
(3) The ability of the analysis method(s) to predict low Petroleum Institute, 1999.
frequency responses should be investigated and where 13. ABS Guide for Building and Classing Floating
possible calibrated against model test results. Production Installations, American Bureau of Shipping,
(4) The reliability analyst should be aware of the assumptions June, 2000.
used in the deterministic analysis and their effects on the 14. Recommended Practice for Design, Selection,
calculated wave frequency line tensions. Operation, and Maintenance of Marine Drilling Riser
(5) Sensitivity studies can be used to assist in deciding which Systems, API RP 16Q, 2nd Ed., American Petroleum
variables should be described as random and which may Institute, 1993.
take fixed values in the reliability analysis. 15. Recommended Practice for Design and Operation of
(6) Notional probabilities of failure for damaged limit states Subsea Production Systems, API RP 17A, 1st Ed.,
should be synthesized from notional intact and damaged American Petroleum Institute, 1987.
probabilities of failure calculated for individual storms, 16. Recommended Practice for Flexible Pipe, API RP 17B,
i.e. using the “short-term” environmental distributions. 1st Ed., American Petroleum Institute, 1988.
(7) Finally, reliability methods can be a powerful tool for 17. Comparison of Marine Drilling Riser Analyses, API
comparing the safety implications of different designs, Bull. 16J, 1st Ed., American Petroleum Institute, 1992.
and target safety levels for new designs, can be 18. Draft ISO Standard for Fixed Steel Offshore Structures,
established by performing reliability calculations for old General Annexes, Draft C ISO 13819-1,
designs that have a history of satisfactory performance. ISO/TC67/SC7/WG3 Committee Correspondence, 1997.
19. Mann, N.R., Schafer, R.E., and Scigpunwalla, N.D.,
References Methods for Statistical Analysis of Reliability and Life,
1. Rationalization of safety and serviceability factors in Wiley, N.Y., 1974.
structural codes, CIRIA, Report 63, 1977. 20. Sweeting, T.J., and Finn A.F., A Monte Carlo Method
2. General principles for the verification of the safety of Based on First- and Second-Order Reliability
structures, ISO 2394, International Standards Approximations, Structural Safety, 11, pp. 203-212,
Organization, 1986. 1992.
3. Recommended Practice for the Analysis of Spread 21. Shinozuka, M., Basic Analysis of Structural Safety,
Mooring Systems for Floating Drilling Units, API RP Journal of Structural Engineering, ASCE, Vol. 109, No.
2P, 2nd Ed, American Petroleum Institute, 1987. 3, pp. 721-740, 1983.
4. Recommended Practice for In-Service Inspection of 22. Madsen, H.O., First Order vs. Second Order Reliability
Mooring Hardware for Floating Drilling Units, API RP Analysis Of Series Structures, Structural Safety, Vol. 2,
2I, 1st Ed., American Petroleum Institute, 1987. pp. 207-214, 1985.
5. Specification for Mooring Chain, API Spec 2F, 6th Ed., 23. Dolinski, K., First-order second-moment approximation
American Petroleum Institute. in reliability of structural systems: critical review and
6. Qualification Testing of Steel Anchor Designs for alternative approach, Structural Safety, Vol.1, pp. 211-
Floating Structures, API RP 2M, 2nd Ed., 1996, 231, 1983.
American Petroleum Institute, 1997.
7. Recommended Practice for Design, Selection,
Turret and spread moored ships
Spread moored semisubmersibles Mooring System Types

Spread and turret moored
Catenary and taut mooring lines
Equally spaced and grouped mooring patterns
Chain, wire, and polyester line segments
Number of mooring lines
Water depth
Spread moored spars

(a) Vessel and Mooring System Types
b) Environmental Conditions
Nseg1 = Segment Length / Test Length Nseg2 = Segment Length / Test Length
Pseg = 1-(1-Ptest length)Nseg
(c) Line Strength of Series System Depends on Extreme Values of Component Strength and Length
Figure 1 Examples of Vessel and Mooring System Types, Environmental Conditions, and Line Types of
Position Mooring Systems
3x3 Grouped Mooring Pattern
Figure 2 Comparison of Linear and Nonlinear Frequency Domain Methods of Calculating Low Frequency Offsets
Metocean Design Criteria, Based on ISO 13819-1, Fixed Steel Structures Standard, General Annex 1.
Units are meters, seconds, meters/second, and degrees; directions are w.r.t. lead parameter
Normalized Wind, Wave and Current, X/100-year Characteristic Design Value
Figure 3 Metocean Design Criteria and Weibull (Parent) Distributions of Annual Maxima Used in Reliability
Modeling of Environmental Basic Variables for Wind Events
Figure 4 Break Test Data and Fitted Distributions
Figure 5 Line Tension Sensitivity Plots, Semisubmersible, Gulf of Mexico
Figure 6 Level 2 Iterative procedure to Find the Most Probable Point z* in the Standard Normal Space
ABS BIOGRAPHIES
KEN HUANG
Ken Huang is a graduate of the Texas

A&M University with a M.S. in
Ocean Engineering and Hydro-
dynamics. He has over 25 years of
industrial experience related to
offshore and marine engineering.
Ken was employed by Brown &
Root and Noble Denton prior to
joining ABS in 1997.
Ken is currently Group Head of the Global Performance

Team at Offshore Engineering Department of ABS
Americas, responsible for design review, plan approval,
independent analysis, rule development and R&D activities
covering design criteria, global performance, motions and
loads, mooring and riser systems, and model tests for
floating installations under ABS classification and/or
certification.
YONG BAI
Yong Bai is Manager of Offshore

Technology in the ABS Technology
Group. He is leading and partici-
pating in the preparation and
updating of offshore classification
guides on pipelines and risers,
floating production installations and
guidance notes on ultimate strength,
fatigue/fracture and structural
analysis of jack-ups. He is also actively conducting research
on offshore structural reliability and FPSO’s.
Yong has a MSc. degree and a Ph.D. degree in naval

architecture. When he was a professor of offshore structures,
he wrote books on “Pipelines and Risers” and “Marine
Structural Design.” He has been actively involved with
structural design and analysis of pipelines, risers and
offshore platforms.
OTC 13170
FPSO Standards and Recommended Practices

Wanda J. Parker, WJP Enterprises, and Todd W. Grove, American Bureau of Shipping

that an Environmental Impact Statement (EIS) would be
This paper was prepared for presentation at the 2001 Offshore Technology Conference held in prepared for the first FPSO proposed to be utilized in the
GOM. In working with industry, MMS agreed to do a
This paper was selected for presentation by the OTC Program Committee following review of programmatic EIS on the generic use of FPSOs in the GOM if
presented, have not been reviewed by the Offshore Technology Conference and are subject to industry would fund the study. The joint industry project
position of the Offshore Technology Conference or its officers. Electronic reproduction, DeepStar agreed to fund the EIS1. The draft EIS on the
distribution, or storage of any part of this paper for commercial purposes without the written Proposed Use of Floating Production, Storage and Offloading
is restricted to an abstract of not more than 300 words; illustrations may not be copied. The Systems on the Gulf of Mexico Outer Continental Shelf in the
presented. Western and Central Planning Areas2,3 was published in
August 2000 for comment. The final EIS is expected to be
published in the first quarter of 2001 with a Record of
Abstract
Decision to be published no earlier than 30 days from the
The oil and gas production industry has expressed an interest
publication of the final EIS.
in being able to utilize Floating, Production, Storage and
The second study effort was to do a comparative risk
Offloading (FPSO) units as a development option in the
assessment to evaluate and compare oil spill and fatality risks
deepwater areas of the United States (U.S.) Gulf of Mexico
for the FPSO with a spar, a TLP and a shallow-water jacket
(GOM) outer continental shelf (OCS). Operators will need
serving as a hub and host to deepwater production. The
regulatory approval from both the United States Coast Guard
Offshore Technology Research Center completed that study
(USCG) and the Minerals Management Service (MMS) for a
for MMS4.
FPSO project. Neither the USCG nor MMS currently have
The third step in the process was to identify any gaps in
regulations specifically for the design and operation of FPSOs
the existing regulations and to develop a regulatory model that
on the OCS. A workgroup was formed under the Offshore
could be used by MMS and the USCG in the review and
Operators Committee’s Deepwater Subcommittee to assist
approval of a FPSO project. On March 22, 2000, Mr. Chris
MMS and USCG in reviewing the existing regulations and
Oynes, MMS GOM Regional Director, sponsored a meeting
body of standards, specifications, recommended practices and
between MMS, USCG and industry to discuss the regulatory
classification society rules and guides concerning the design
requirements for FPSOs in the GOM, should they be found to
and operation of FPSOs on the OCS for the Gulf of Mexico.
be an acceptable development option. Although it was
The effort also aimed to identify gaps in the regulations and
recognized that MMS and the USCG would have to agree
industry standards. This paper provides a summary of the
among themselves the appropriate regulations and regulatory
major findings of the workgroup. In addition to the items
split between the two agencies, they agreed that it would be
discussed in this paper, the workgroup report identified other
beneficial and appropriate to have industry provide input on
areas where additional modifications to regulations or industry
the model. It was decided that a workgroup would be formed
standards may be warranted.
under the direction of the OOC Deepwater Subcommittee and
consists of industry representatives and classification society
Introduction
representatives along with personnel from MMS and the
As operators have moved into the deeper waters of the GOM
USCG. A report was prepared by the workgroup and
over the last several years, interest has been growing in
submitted to MMS and the USCG in September 2000 for their
potentially utilizing FPSOs as a development option to the
consideration5.
floating production systems (tension leg platforms (TLP),
spars, etc) and subsea tie backs to either floating production
Regulatory Model Workgroup
systems or fixed platforms that are currently being utilized. In
Goal
discussions with MMS and USCG, it became apparent that
The overall goal of the workgroup was to review the existing
several studies would need to be conducted to confirm the
regulations and industry standards covering the design,
acceptability of these systems for the GOM. MMS indicated
construction and operation of FPSOs in the GOM and identify
2 WANDA J. PARKER AND TODD W. GROVE OTC 13170
any gaps in either the regulations or standards that needed to 4. Propulsion

be addressed prior to bringing FPSOs into the GOM. The 5. Stability
workgroup was focused on the design and operational 6. Motions/Global Performance
considerations for the FPSO. The work group did not address 7. Risers
the design of shuttle tankers or operational considerations 8. Offloading Systems
once they were disconnected from the FPSO. 9. Layout
10. Operational Considerations
Participants 11. Discharges
The workgroup was formed under the direction of the OOC 12. Manning
Deepwater Subcommittee and met five times following the
initial meeting. It was agreed that a cooperative effort with Regulations
open discussions between the regulatory agencies and industry MOU
was desired and would produce the best work product. Due to It is recognized that both MMS and USCG have a large body
the broad scope of the discussions, it was necessary to include of regulations that could be applied to FPSOs. In the
a large number of participants. MMS agreed to have Mr. Memorandum of Understanding (MOU) between MMS and
James Regg, Section Chief, Technical Assessment and USCG6 signed on December 16, 1998, the delineation of
Operations Support serve in the workgroup along with support jurisdictions regarding floating production system
from other MMS personnel. LCDR Bill Daughdrill from the components, operations and issues is addressed. The
Eighth USCG District represented the USCG in the work workgroup reviewed the MOU for completeness in coverage
group. Personnel from the MMS Headquarters and USCG of systems on a FPSO and put together a table showing the
Headquarters groups were kept informed of the workgroup’s applicable regulations, industry standards and classification
activities through e-mail. Twenty-five persons representing 17 society rules for each system and sub-system in the MOU.
companies participated in one or more of the meetings. Since While this table was not exhaustive, it quickly pointed out
classification societies have historically played a large role in areas where regulations and standards were well established
the approval of FPSOs worldwide, it was felt that it was and systems where either regulations or standards were
important to have broad representation from the major lacking. The work group made some specific
classification societies who currently class FPSOs. Four recommendations concerning the implementation of the MOU
classification societies; American Bureau of Shipping, Bureau that will be discussed below.
Veritas, Det Norske Veritas and Lloyd’s Register of Shipping
were represented in the workgroup. Tim Sampson represented MMS
the American Petroleum Institute (API). Wanda Parker agreed MMS regulations that are applicable to FPSOs are primarily
to chair the workgroup for OOC. A complete listing of all located in 30 CFR 250, Subparts H and I. In conjunction with
workgroup participants is in the workgroup report5. All of their regulations for specific systems, MMS has said that they
these individuals dedicated a considerable amount of time and intend to utilize the Deepwater Operations Plan (DWOP)
expertise to this effort. process in their review of a FPSO project7. The workgroup
reviewed the MMS regulations for adequacy and made some
FPSO Characteristics specific recommendations that will be discussed below.
For consistency, the workgroup decided to use the FPSO
characteristics used in the EIS and CRA studies as the FPSO USCG
model for this effort. The workgroup considered the The USCG has said that FPSOs will be regulated as vessels
regulations that would apply to a US flag FPSO or an and therefore will be required to meet specific vessel design
undocumented FPSO that is designed to US flag requirements and operational regulations8,9. USCG regulations that are
(similar to the existing GOM floating production systems). applicable to FPSOs are primarily located in 46 CFR,
Limited discussions were held on the differences in permitting Subchapters D and IA and 33 CFR Subchapter N. On Dec 7,
a US flag FPSO and a foreign flag FPSO. The FPSO was 1999, the USCG published a Notice of Proposed Rulemaking
considered to be ship-shaped with limited discussions on the for 33 CFR Subchapter N that includes proposed regulations
differences between a ship-shaped FPSO and a non ship- applicable to FPSOs. The workgroup considered these
shaped FPSO. Discussions were limited to a permanently proposed regulations in addition to the established regulations.
moored FPSO for simplicity since a disconnectable FPSO The workgroup report was submitted to the record as a
introduces many complicating factors into the discussion. comment to the proposed regulations. The specific
Finally, discussions were focused on systems that are unique recommendations will be discussed below.
to FPSOs with only limited discussion of systems that are
common to fixed platforms or other floating production Industry Standards
systems. This list of system includes: Both USCG and MMS regulations incorporate by reference a
1. In Hull Cargo Storage Systems large number of industry specifications and recommended
2. Cargo Piping and Transfer practices, particularly API documents. Many of these are
3. Turret/Mooring/Stationkeeping/Swivel applicable to FPSO system design or operation. In addition,
OTC 13170 FPSO STANDARDS AND RECOMMENDED PRACTICES 3
there are a host of other international standards that may be FPSOs, the workgroup recommended that MMS consider
applicable to FPSOs. Most of these standards cover individual incorporating by reference additional industry standards and
systems or subsystems that may be used in conjunction with a recommended practices as they become available. The
wide variety of types of installations. It was outside of the following were identified as candidates for MMS review and
scope of the workgroup to conduct a thorough review of the consideration for incorporation in their entirety10,11,12,13,14:
adequacy of the individual industry specifications and 1. API RP 2FPS, Planning, Designing and Constructing
recommended practices. Rather, the workgroup focused on a Floating Production Systems
few key standards and made recommendations of standards 2. API RP 2SM, Design and Analysis of Synthetic
the agencies should consider incorporating by reference. Moorings
3. API RP 2SK, Design and analysis of Stationkeeping
Classification Society Rules and Guides Systems for Floating Structures
Another body of standards applicable to FPSOs and 4. API RP 2RD, Design of Risers for Floating
recognized worldwide are those developed and applied by a Production Systems (FPSs) and Tension-Leg
number of the major classification societies. The rules and Platforms (TLPs)
guides of the societies are utilized in assessing the fitness-for- 5. API Spec 17J, Specification for Unbonded Flexible
purpose of FPSOs and are focused on safety aspects. The Pipe
classification requirements address design requirements as
well as those for fabrication, installation, and commissioning. USCG
Unlike the snapshot nature of certification, classification is an On December 7, 1999, a Notice of Proposed Making was
ongoing process by which the societies survey a FPSO published revising 33 CFR, Subpart N. In that revision, both
periodically during its operational life to ensure compliance Mobile Offshore Drilling Unit (MODU) regulations located in
with the rules. Given their unique independent role 46 CFR, Subchapter I-A and tank vessel regulations located in
internationally, classification societies are also often delegated 46 CFR, Subchapter D were referenced. However, the
statutory responsibilities by flag and coastal state authorities to regulation was not clear in many cases which regulation
act on behalf of the administration. USCG has, to varying should be followed if both regulations cover the same system
degrees, delegated to several classification societies approval or subsystem. Also, in many cases, modifications to these
responsibilities for existing GOM floating production units. regulations may be needed to address the unique
Expectations are that similar delegations will be forthcoming circumstances of FPSO or floating system operations versus a
as experience is gained in GOM FPSO applications. Several tank vessel or MODU. The workgroup recommends that
classification societies have also been active to varying specific regulations for floating facilities should be written in
degrees in the MMS Certified Verification Agent program for lieu of pointing to regulations for various types of vessels,
existing structures, which may also be extended to FPSOs. which may not be completely applicable to floating production
The relevant practical experience from classification and facilities. The workgroup also recommended that their report
statutory responsibilities is utilized in maintaining the rules be included in the comment record for the proposed
and guides in a current and relevant form. The four rulemaking.
participating societies contributed citations to the regulatory The workgroup recommended that marine crew manning
matrix for the most prevalent rule and guide requirements and qualification regulations should be codified for FPSOs
applicable to the various aspects of the FPSO. and for other floating production systems. The Eighth Coast
Guard District has issued a policy letter15 for marine crew
Gaps and Recommendations manning for floating production systems other than those
During the review process, the workgroup identified a number storing oil in bulk, which could serve as an appropriate
of recommendations either for ways to close existing gaps in starting base for non-self propelled FPSOs. It was recognized
the regulations or for additional work that should be that self-propelled FPSOs would need additional
considered. These recommendations were broken out into consideration.
areas for consideration by MMS and USCG and for industry. MMS has adopted API RP 500/50516,17 for area
classification while the USCG has prescriptive regulations in
MMS 46 CFR, Subpart J. The workgroup recommended that the
In 30 CFR 250, Subpart I, MMS has established a platform USCG adopt API RP 500/505 for floating production systems
verification program requiring a third party certification of the including FPSOs. Adopting common standards would
platform design. These regulations were established for fixed minimize confusion and duplication of effort for industry
platforms, but MMS has been applying them to floating since both agencies have jurisdiction for area classification.
production facilities. It is recommended that MMS revise Although the USCG has been given sole jurisdiction over
these regulations to update them for floating facilities, fire fighting systems for floating production systems,
including FPSOs. The workgroup recommended adding the including FPSOs, the USCG has not yet proposed regulations
turret, risers and mooring systems to the verification program. for fire fighting systems in the production area. It is
recommended that the USCG adopt fire-fighting regulations
In lieu of the agency writing prescriptive regulations for for the production area.
4 WANDA J. PARKER AND TODD W. GROVE OTC 13170
Both MMS and USCG have voluntary safety management standards for piping systems.
system programs that could be applicable to FPSOs. MMS has Neither MMS nor USCG regulations address integral hull
recognized API RP 7518 as an acceptable basis for a safety tanks used as process vessels (such as wet/dry oil tanks). The
management system for fixed and floating production systems work group recommended that all integral hull tanks be under
on the US OCS. Most oil and gas production companies have USCG jurisdiction for structural design. For tanks used as
based their safety management programs for operations on the process vessels, the safety system, control valves, and piping
US OCS on API RP 75. However, the USCG has recognized to and from the process vessels should be under MMS
the International Safety Management Code (ISM)19, which is jurisdiction. It is recommended that piping specification
applicable to vessels that must comply with Chapter IX of breaks should occur at the first flange outside the tank.
SOLAS. While both programs have merit, and individual
companies may want to base their programs on either Industry
standard, a combination of the standards or some other Both the Environmental Protection Agency (EPA) and the
standard, the workgroup recommends that the USCG USCG have regulations concerning discharges to the ocean
recognize API RP 75 as an acceptable basis for a safety that could occur from FPSOs. It is recommended that a work
management program in addition to ISM. group under the OOC Deepwater Subcommittee be formed to
In the event that foreign flagged FPSOs are acceptable for review the overboard discharge regulations of both agencies to
use in the GOM, the USCG will issue a Letter of Compliance ensure that all discharges are adequately addressed for FPSO
(LOC)8. Since a full design review is not normally conducted operations.
for a LOC, the workgroup recommended that the USCG Although the USCG has regulations that provide for the
develop a LOC checklist that could be either used with inspection of foreign flagged tank vessels, no formal
existing foreign flagged FPSOs or new-built foreign flagged determination has been requested or received from US
FPSOs proposed for operations in the GOM. Customs that indicates if foreign flagged FPSOs will or will
Although the USCG has general lightering regulations and not be allowed to operated on the OCS (i.e. Will Customs
additional operational regulations that apply to the designated interpretation and policy for foreign flagged FPSOs be similar
lightering areas in the GOM, there are no specific operational to their current interpretation and policy for foreign flagged
regulations that apply to FPSOS or to tandem offloading. It is MODUs?) The workgroup recommended that industry
recommended that a work group be formed to review existing develop a strategy for obtaining a formal, written
international standards to determine their adequacy for GOM determination from Customs.
operations. It was recommended that the various API standards and
recommended practices for mooring systems and riser design
Joint MMS and USCG be reviewed to determine if the inspection guidelines given in
In the MOU, both MMS and USCG have been given those documents are adequate for floating production systems
jurisdiction for reviewing and approving the design of the and updating the documents as needed.
turret and mooring system. The workgroup agreed that It was recommended that API RP 14C be reviewed and
technology is rapidly evolving for these systems and that it revised if needed for floating production systems, including
would be burdensome on the regulatory agencies to have taking into account the effect environmentally induced
personnel fully knowledgeable about these systems as they motions may have on the safety and monitoring systems. In
change. It is recommended that a verification agent acceptable addition, it is recommended that the safety and monitoring
to both agencies be selected to review and certify the design. systems for swivels, integral hull tank process vessels and
In the MOU, both MMS and USCG have been given other unique systems to a FPSO be covered in the
jurisdiction for reviewing and approving various portions of recommended practice.
the integrated monitoring and safety systems. It is It was recommended that API RP 14E20 be reviewed for
recommended that a work group consisting of representatives adequacy and updated as needed for the effects of motion and
of Industry, MMS and the USCG be formed to address the piping support.
integration of these systems. It was recommended that the various API documents on
In the MOU, both MMS and USCG have been given composite materials should be reviewed for adequacy for
jurisdiction over piping systems. It is recommended that for floating facilities including FPSOS.
cargo tank piping that the specification break between MMS It was recommended that API RP 75 be reviewed and
and USCG jurisdiction occur at the 1st valve downstream of updated as needed to ensure the document is an adequate basis
the last processing vessel (and its control valves and safety for a safety management system for FPSO systems and
system) prior to the oil entering the cargo storage tanks. It is operations. Applicable portions of ISO 9000/14000 could be
recommended that a work group be formed consisting of incorporated, if desired.
representatives of industry, MMS and USCG to review other The work group recognized that standards for cargo tank
similar systems and agree to where the specification breaks cleaning for FPSOs might be different from trading tankers
between the systems should occur. These breaks should be since inspections may occur when the FPSO is on location and
codified into the MMS and USCG regulations. Alternatively, in operation and not in dry dock. It is recommended that
MMS and USCG should consider adopting consistent industry appropriate standards for cargo tank cleaning be confirmed.
OTC 13170 FPSO STANDARDS AND RECOMMENDED PRACTICES 5
Western and Central Planning Areas, Draft Environmental

Conclusions Impact Statement”, Minerals Management Service, Gulf of
While neither MMS nor USCG has regulations that Mexico OCS Region, August 2000.
specifically apply to the design or operation of FPSOs, there is 3. George, J.E., Parker, W.J., Cranswick, D.J., “FPSO
Environmental Impact Statement: What is Happening?”,
a multitude of regulations that are applicable to FPSOs and Offshore Technology Conference, 10705, May 1999.
with the modifications indicated in the work group’s 4. Gilbert, R.B., Ward, E.G., Wolford, A.J., “A Comparative Risk
recommendations, the existing framework is adequate. A large Analysis of FPSO’s with Other Deepwater Production Systems
body of industry recommended practices and standards in the Gulf of Mexico”, Offshore Technology Conference,
consisting of classification society rules and guides, API 13173, May 2001.
standards, specifications and recommended practices and 5. Letter to Carolita Kallaur, Minerals Management Service, from
international standards exist that cover FPSO system design Allen Verret, Executive Director, Offshore Operators
and operation. In some cases these need to be reviewed and Committee, Regarding Regulatory Framework – Floating
updated, where needed, to ensure they are applicable to GOM Production Storage and Offloading Systems, September 2000.
6. Memorandum of Understanding (MOU) Between the Minerals
FPSO operations. Many of the recommendations identified
Management Service and the United States Coast Guard”,
for FPSOs are also applicable to other types of floating Federal Register, Vol. 64, No. 10, January 15, 1999.
facilities. 7. Regg, James B., “Floating Production, Storage and Offloading
The majority of the discussions by the workgroup focused Systems in the Gulf of Mexico OCS: A Regulatory Perspective”,
on systems that are unique to FPSOs. The FPSO model used Offshore Technology Conference, 10701, May 1999.
for the discussions was a ship-shaped permanently moored 8. Daughdrill, W.H., Brown, M.J., “The Regulatory Scheme
FPSO. It is recognized that a FPSO is not dependent on shape Applicable to Floating Production, Storage, and Offloading
and some variations occur as you move between ship-shaped Systems”, Offshore Technology Conference, 10702, May 1999.
and non-ship-shaped FPSOs that will need additional 9. Letter to Carolita Kallaur, Minerals Management Service, from
RADM North, United States Coast Guard, November 16, 1998.
attention. Likewise, a disconnectable, self-propelled FPSO has
10. “API Recommended Practice for Planning, Designing, and
many aspects that are different from a permanently moored Constructing Floating Production Systems”. American
FPSO that will need to be considered. Petroleum Institute, 2000 Draft.
The review of the existing regulations was conducted 11. “Recommended Practice for Design, Manufacture, Installation,
without representation from the USCG Marine Safety Center and Maintenance of Synthetic Fiber Ropes for Offshore
or Headquarters groups. These groups need to be fully Mooring”. American Petroleum Institute, November 1999 Draft.
engaged before final determinations can be made on the 12. “API Recommended Practice for Design and Analysis of
appropriate regulations for FPSO design and operations. The Stationkeeping Systems for Floating Structures”. American
representatives from the Eighth Coast Guard District Petroleum Institute, December 1996.
13. “API Recommended Practice for Design of Risers for Floating
participated fully in the discussions, but it was recognized by
Production (FPSs) and Tension-Leg Platforms (TLPs)”.
the workgroup that they were not authorized to speak American Petroleum Institute, June 1998.
definitively on USCG policy or regulation. 14. “API Specification for Unbonded Flexible Pipe”. American
OOC appreciated the opportunity to take the lead role in Petroleum Institute, July 2000.
this cooperative effort between MMS, USCG and Industry. By 15. The Eighth Coast Guard District policy letter for marine crew
working together and pooling our thoughts and ideas, manning for floating production system other than those storing
regulations that meet the needs of all concerned can be put in oil in bulk
place for FPSOs in the GOM. A large number of industry 16. “API Recommend Practice for Classification of Locations for
representatives and classification society representatives Electrical Installations at Petroleum Facilities Classified as Class
I, Division 1 and Division 2”. American Petroleum Institute,
dedicated a considerable amount of time and expertise to this
November 1997.
task. The efforts of Mr. Jim Regg and LCDR Bill Daughdrill 17. “API Recommend Practice for Classification of Locations for
were recognized for their active participation in the work Electrical Installations at Petroleum Facilities Classified as Class
group. As policies and rulemaking for FPSOs are drafted, I, Zone 0, Zone 1 and Zone 2”. American Petroleum Institute,
continuing the cooperative effort between the regulators, November 1997.
industry and the classification societies will be beneficial. 18. “API Recommend Practice for Development of a Safety and
Environmental Management Program for Outer Continental
Acknowledgments Shelf Operations and Facilities”. American Petroleum Institute,
The authors would like to thank the Offshore Operators July 1998.
19. “International Management Code for the Safe Operation of
Committee for allowing us to publish the results of the work
Ships and for Pollution Prevention”, International Maritime
group. Organization, Assembly Resolution A.741(18), 1993.
20. “API Recommend Practice for Design and Installation of
References Offshore Production Platform Piping Systems”. American
1. Verret, Allen J., Hays, Paul R., “Deepstar’s Program Related to Petroleum Institute, October 1991.
FPSO’s”, Offshore Technology Conference, 10703, May 1999.
2. “Proposed Use of Floating Production, Storage, and Offloading
Systems on the Gulf of Mexico Outer Continental Shelf,
ABS BIOGRAPHIES
TODD GROVE
Todd Grove is a graduate of the

University of Michigan with a degree
in Naval Architecture and Marine
Engineering. He has been with
ABS for 19 years serving in the
Corporate office in the New York
area, Pacific Divisional Headquarters
in Singapore and Americas Divi-
sional Headquarters in Houston. He
was the manager of the Offshore Engineering Department in
Houston where he was responsible for ABS classification
and certification design review of structure, stability and
safety issues for Mobile Offshore Drilling Units, Floating
Production Systems and other Site-Specific Installations.
Currently Todd is the Director of ABS’ Offshore Project
Development Team where he coordinates ABS’ global
offshore resources for bid and proposal development.
OTC 13173
A Comparative Risk Analysis of FPSO’s with Other Deepwater Production Systems in

the Gulf of Mexico
R.B. Gilbert, Offshore Technology Research Center, The University of Texas at Austin, E.G. Ward, Offshore Technology
Research Center, Texas A&M University, and A.J. Wolford, EQE International, Inc.

The Minerals Management Service (MMS) funded the
This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Offshore Technology Research Center (a National Science
Foundation engineering research center located at Texas A&M
This paper was selected for presentation by the OTC Program Committee following review of University and The University of Texas at Austin), with EQE
presented, have not been reviewed by the Offshore Technology Conference and are subject to International, Inc. as a subcontractor, to conduct a
position of the Offshore Technology Conference or its officers. Electronic reproduction, Comparative Risk Analysis (CRA). The purpose of this study
distribution, or storage of any part of this paper for commercial purposes without the written was to assess and compare the system risks for FPSO’s with
is restricted to an abstract of not more than 300 words; illustrations may not be copied. The those for existing deepwater production systems, specifically
presented. TLP’s, Spars and Hub/Host Jackets. This study was
conducted concurrently with an Environmental Impact
Statement (EIS) study for FPSO’s in the Gulf of Mexico1.
Abstract
Information from both the Comparative Risk Analysis and the
This paper describes a study to compare the risks of Floating
EIS will be used by the MMS in developing policies
Production Storage and Offloading Systems (FPSO’s), which
concerning the use of FPSO’s in the Gulf of Mexico.
have never been used in the Gulf of Mexico, with the risks for
The primary objectives of the Comparative Risk Analysis
existing deepwater production systems in the Gulf of Mexico.
were the following:
The major conclusion is that the expected risks for fatalities
1. Assess and compare the system risks for FPSO’s with
and oil spills associated with FPSO’s are comparable to those
those for existing deepwater production systems,
for already accepted alternatives for deepwater production. In
specifically Spars, TLP’s, and Hub/Host Jackets; and
addition, the oil spill risks are dominated by spills that occur
2. Understand the contributions to system risk by sub-
during transportation of oil from the production facility to the
systems and phases of operation.
shore with either pipelines or shuttle tankers. The major
recommendation is to periodically update these results so that
Approach
they serve as a baseline for future analyses of risk in the Gulf
The approach used to conduct the Comparative Risk Analysis
of Mexico. This study was undertaken for the Minerals
was developed with the following goals in mind:
Management Service to provide information for their use in
1. Provide the MMS with information that can be used
developing a policy for FPSO’s in the Gulf of Mexico.
for a consistent and objective comparison of the risks
associated with the four production systems;
Introduction
2. Provide the MMS with a level of detail necessary to
To date, deepwater (more than 3,000-foot water depth)
compare and understand overall system risks for
reserves in the Gulf of Mexico have been developed primarily
typical production systems; and
with the following types of production systems: Spars;
3. Incorporate industry data, experience and expertise to
Tension Leg Platforms (TLP’s); and Subsea Well Systems tied
the greatest extent possible into evaluating the risks.
back to these floating systems or to shallow water jackets that
The approach used to achieve these goals involved teams of
may also serve as hubs for other deepwater production
experts and a series of workshops.
systems (Hub/Host Jacket). All three of these types of
systems rely on pipelines to transport oil to shore. A
Participation of Technical Experts. Historical data on actual
potentially attractive alternative to these systems is a tanker-
failures, particularly those very infrequent failures with large
based Floating Production Storage and Offloading (FPSO)
consequences that tend to drive overall risks, are scarce, and
system with oil transportation to shore via shuttle tankers.
the risks must be estimated by other means. For this study, we
Floating Production Storage and Offloading systems have
chose to directly involve the expertise and experiences of
been used in many areas of the world, but not the Gulf of
engineers involved in the design and operation of these
Mexico.
2 R.B. GILBERT, E.G. WARD AND A.J. WOLFORD OTC
production systems. The Deepstar consortia facilitated and overall system risks.
coordinated the participation of industry engineers in this Additional contributions from industry included input on
project. detailed hazard identifications through participation in formal
There was active participation by experienced engineers specialty interview sessions, and various other interactions in
representing all segments of the industry, including oil which individuals provided data, input, or advice. Technical
companies, consultants, manufacturers, contractors, experts from the companies and organizations listed in Table 2
classification societies, as well as the regulatory agencies. as well as from most of those listed previously in Table 1
They brought a detailed understanding of the nature of these contributed in these areas.
risks as well as design and operational options to manage In all, over 100 of the industry’s more experienced
these risks. Many of the industry engineers who were engineers directly participated in the study either through the
involved in this study had also participated in risk studies system teams and workshops, or the specialty interview
within their companies, which are often undertaken either in sessions. The names and affiliations of these participants are
the design of a specific system or to compare several systems summarized in Gilbert et al.2. The average experience level
in selecting the most appropriate system for a given project. for these experts was approximately 20 years.
The practical experience and perspective that these engineers The level of participation by the industry experts was
brought to the study was deemed critical to the success of this substantial. Their direct involvement in the workshops
study. (preparation, participation, and review) and the specialty
Separate teams were formed for each of the four interviews involved an estimated 5,000 man-hours. Further,
production systems, the Spar, the TLP, the Hub/Host Jacket, these experts often sought additional input and review from
and the FPSO. These teams were made up of invited colleagues in their companies, and gathered additional
participants from industry and representatives from the MMS relevant information for the study.
and the U. S. Coast Guard (USCG), the government agencies
responsible for regulating the deployment and operation of Workshop Process. A flowchart for the workshops is shown
deepwater production systems. The teams were designed to in Fig. 1. Individual, one-day workshops were conducted for
include engineers with expertise and experience in the design, each system during the first three phases (Workshops #1 to
construction and operation of the overall systems as well as #3). The final workshop was held collectively over a two-day
the subsystems and components that make up the systems. period. The activities conducted between workshops are also
There was an average of about ten members per team. The indicated on Fig. 1.
companies that provided one or more participants to these The objective of Workshop #1 was to develop conceptual
system teams are listed in Table 1. system descriptions for the four production systems. This
It is worth noting that these companies represent a large work included establishing study boundaries in space and time
measure of the offshore industry’s deepwater experience and and describing the physical and operational features for each
expertise. They have been very active in the design, system. Draft system descriptions were distributed to the
operation, and/or certification of deepwater production workshop participants ahead of the workshop and then used as
systems in the Gulf of Mexico and elsewhere. Of particular the starting point in the workshop.
importance is their direct involvement and experience with the The objective of Workshop #2 was to perform hazard
deepwater production systems used in this study: Spars, identifications for each system. A list of possible adverse
TLP’s, Hub/Host Jackets, and FPSO’s. events (or initiating events) that could contribute to risk was
The teams were balanced to include members with overall developed for each study system and organized by sub-system
systems expertise as well as those with expertise in various or activity. Detailed hazard identifications were developed
sub-systems, components, and operations, including: through a series of specialty interviews with technical experts
• Platform and subsea well systems; before Workshop #2. The participants and topics for these
• Drilling and well intervention operations for both interview sessions are summarized in Gilbert et al.2. These
platform and subsea wells; detailed lists were then reviewed during Workshop #2 and
• Topsides (processing facilities, equipment); used to develop a framework for the quantitative risk
• Production operations; assessment.
• Pipelines and flowlines; The objective of Workshop #3 was to elicit quantitative
• Tanker and FPSO design and operations; information about frequencies and consequences for oil spills
and fatalities. A preliminary quantitative risk assessment
• Structures (hulls, decks, mooring systems, riser
based entirely on raw data was distributed to the workshop
systems);
participants before the workshop. This preliminary risk
• Helicopter operations (personnel transport);
assessment was then refined during Workshop #3 and needs
• Supply boat operations (material & personnel for additional information and studies were identified.
transport); and The objective of Workshop #4 was to review and refine
• Diving operations. the risk assessment. Information from additional studies
Thus the teams were able to focus on risks at the sub-system, conducted between Workshops #3 and #4 was incorporated
component, and operational levels as well as to focus on into this review.
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO 3
The general work processes used for the workshops was as measure, such as equivalent cost.
follows. Preliminary information that had been distributed to
the participants was reviewed and refined through an open Descriptions of Study Systems. The following criteria, in
forum process that included time for discussion and order of decreasing importance, were used to develop
developing a consensus regarding the input on risks. The conceptual descriptions for each of the representative study
open forum approach encouraged an iterative and synergistic systems:
discussion of risks and information from different 1. The study systems for the Spar, TLP, and Hub/Host
perspectives. The participation of both the industry and the Jacket should be typical of existing systems and
regulatory agencies helped to provide balance and objectivity technologies that are currently being used in the Gulf
in the discussions and input. Consensus was generally readily of Mexico because these systems and technologies
achieved, but when significant disagreement occurred between have been approved and therefore represent
participants, votes were taken to achieve a consensus and acceptable risks.
dissenters’ opinions were recorded. The phased and 2. The study system for the FPSO should be comparable
progressive nature of the workshops provided the opportunity to that already developed for the base case in the EIS1
to seek and incorporate additional expertise and information as in order to capitalize on the substantial effort devoted
the study progressed and additional needs became apparent. to developing this study basis.
Interim reports summarizing information from each of the first 3. The study systems should be as comparable to one
three workshops were distributed to participants after each another as possible so that differences in risks among
workshop. These reports provided participants with them represent realistic differences among these
opportunities for ongoing review and a means to ensure types of systems and are not an unintended artifact of
consistency in assumptions and approaches among the the study system descriptions.
different systems. Finally, evaluations were conducted at the As an example of how these criteria were applied, consider
completion of the first three sets of workshops to continually a Spar. In order for the study Spar to be as comparable as
improve the process. possible to the study FPSO (criterion 3), which has oil
transport via shuttle tankers consistent with the EIS1 (criterion
Risk Measures. Risk measures for the study systems were 2), the Spar would also have oil storage and oil transport via
developed using the following criteria: shuttle tankers. However, while this type of a Spar is possible,
• The measures of risk should provide relevant and it is not typical of existing Spars in the Gulf of Mexico
useful input to MMS in their decision making (criterion 1). Therefore, the description for the study Spar did
process; not include oil storage and has oil transport via pipeline.
• The measures of risk should be tractable and The first step in the system description process was to
quantifiable; and establish a time frame for the risk assessment. The intent was
• The measures of risk should be measures that are to assess risks covering all aspects of offshore production
currently tracked and recorded so that (i) available including oil and gas production and processing offshore;
data can be used to support the results of this risk drilling and well intervention during production; export of the
analysis and (ii) future data can be used to validate oil and gas to shore; and transport of personnel to and from
and calibrate the results of this risk analysis. shore. The “lifetime” for a study system was defined to start
From these criteria, the risk measures listed in Table 3 when oil flows through the first production riser and end when
were adopted for this study. The total number of fatalities is the last well is shut in. For this study, a 20-year lifetime was
intended to measure the human safety risk. The volume of oil used. Other phases in the actual lifetime for a system, such as
spilled is intended to measure the environmental risk. The construction, system installation, commissioning,
environmental effects of an oil spill are not considered directly decommissioning and system removal, were not included in
in this study because (1) there is a correlation between the this risk analysis.
magnitude of environmental damage and the volume of oil The second step in the system description process was to
spilled; (2) environmental effects are difficult to measure and establish physical boundaries for the risk assessment. The
quantify; and (3) the environmental impacts of oil spills from study boundaries included the production facility, the
FPSO’s are included in the scope of the EIS1. The total pipelines and shuttle tankers used to transport oil to a shore
volume of oil spilled in the 20-year lifetime is intended to terminal, and the supply vessels and helicopters used to
measure chronic environmental risks. The maximum volume support the production operations. These physical boundaries
of oil spilled in a single incident is intended to measure acute are shown schematically on Fig. 2.
environmental risks. The risk measures in Table 3 are The third step in the system description process was to
practical simplifications that are intended to approximately define the physical and operational attributes for each system.
capture the multitude of risks present. Detailed descriptions for these attributes are contained in
These measures of risk were not discounted with time. In Gilbert et al.2, and the major attributes are summarized in
addition, each measure was treated separately in comparisons Table 4. For the Spar, TLP and Hub/Host Jacket, operating
and no attempt was made to combine them into a single experience from the Gulf of Mexico was directly drawn upon
in developing the system descriptions. For the FPSO,
experience with tanker operations in the Gulf of Mexico was obtained from the technical experts during the workshops.
used together with operating experience for FPSO’s in other The philosophy adopted in developing the preliminary risk
parts of the world, such as the North Sea and the South China assessments was to extrapolate directly from historical
Sea. It is important to note that these study systems represent experience in the Gulf of Mexico to predict future
typical or generic systems. Therefore, the range of risks performance. The primary data sources were the MMS3 and
associated with possible variations in hardware and operating the USCG4. The methodology used to develop the
practices is not captured in the results of this project. preliminary risk assessments had the following steps:
1. Summarize Data for Sub-Systems: The data sets were
Quantitative Risk Assessment. The objective of the first divided into sub-systems based on the hazard
quantitative risk assessment was to quantitatively assess the identification work in Workshop #2. These sub-
risk measures listed in Table 3. These risk measures were systems are listed in Table 5. The data for fatalities
quantified by estimating representative or average values for were then summarized as the total number of
each study system. As an example, consider the total volume fatalities in the data record for each sub-system. The
of oil spilled during the operational lifetime. If a fleet of Spars data for oil spills were further sub-divided into
similar to the one defined in this study were installed and categories by the size of the spill, and then the
operated for 20 years in the Gulf of Mexico, then the total oil number of incidents in the data record that had
spill risk associated with this type of system would be the occurred for each spill-size category was compiled.
average value for the total volume of oil spilled from each The data for oil spills were divided into categories to
Spar (that is, the sum of all the oil spilled from each Spar in its facilitate the assessment since the range of spill
20 year lifetime divided by the total number of Spars). volumes per incident covered five to six orders of
Likewise, the average values for the maximum volume of oil magnitude and the frequency distribution for spill
spilled in a single incident from each Spar and the total sizes was highly skewed.
number of fatalities on each Spar would represent the other 2. Select Exposure Factors for Sub-Systems: The
measures of risk. exposure for a risk is an indicator of the factors that
Since there is an extremely limited experience base in the influence the risk. In this way, the data can be
Gulf of Mexico for the types of production systems being extrapolated to each study system based on the
analyzed in this study, it is not possible to obtain average exposure to the risk for that study system. The
values directly for the total number of fatalities, the total factors used to express the exposure for each sub-
volume of oil spilled, and the maximum volume of oil spilled system category were selected based on the hazard
in a single incident. The goal of this study was to predict what identification information work in Workshop #2.
the average values would be (the expected value) if each study These factors are listed in Table 5.
system were hypothetically installed and operated in the future 3. Estimate Frequencies of Occurrence for Sub-
in the Gulf of Mexico. Systems: Estimates for the frequencies of occurrence
As with any prediction, there is uncertainty that the actual for incidents (from Step 1) relative to the exposure
average value for each risk measure (obtained many years in factors (from Step 2) were developed using statistical
the future) will be equal to the predicted value in this study. methods that are described in Gilbert et al.2. Both the
The range of possible values for the actual average was expected value and the standard deviation for these
represented in this study by two quantities: the expected value frequencies were calculated. For the oil spill
and the standard deviation. The expected value represents the frequencies, it was assumed that a spill could occur in
predicted value for the average, while the standard deviation the next largest spill-size category above the
represents the magnitude of uncertainty in the prediction. The maximum spill size observed in the historical data
expected value and standard deviation can be used to calculate set.
confidence intervals for the prediction. For example, the 90- 4. Determine Sub-System Exposures for Study Systems:
percent confidence intervals indicate that there is a ninety- The exposure for each sub-system was determined
percent probability that the actual average will be within this from the system descriptions.
interval. 5. Assess Sub-System Risks for Study Systems: The
This section describes how the quantitative risk estimated frequencies from the historical data (from
assessments were conducted through a process of developing Step 3) were then combined with the exposures for
preliminary assessments and then refining those assessments the study systems (from Step 4) to assess the sub-
using the input of the technical experts. system risks. Both an expected value and a standard
Preliminary Risk Assessments. Preliminary (or pre- deviation were calculated for each risk measure (see
workshop) quantitative risk assessments played a very Gilbert et al.2 for details).
important role in this project because they were used to elicit 6. Assess System Risks from Sub-System Risks: The
quantitative information from the technical experts during final step was to combine the information for the sub-
Workshops #3 and #4 (Fig. 1). These preliminary risk system risks (from Step 5) to assess the total system
assessments were developed to be objective, consistent and risk (see Gilbert et al.2 for details).
complete in order to maximize the value of the information Final Risk Assessments. The preliminary risk assessments
were then refined through the workshop process to develop the FPSO are subsea wells that are subjected to less frequent
final risk assessments. This process involved the following well intervention compared to platform wells.
steps:
1. Start with data-based estimates that are as complete Risks for Oil Spills. The results for the oil spill risks are
as possible (the preliminary risk assessments). presented and discussed in this section. First, the frequencies
2. Evaluate the data sources and refine raw data sets as for different spill sizes are addressed. Next, the average total
necessary so that they are relevant for predicting volume of oil spilled and the maximum volume spilled in a
future performance of the study systems. As an single incident over the lifetime are addressed.
example, the data set for oil spills from tankers in the Frequencies of Spills. The frequencies of spills from
Gulf of Mexico was limited to years after 1990 to production and transportation are first presented and
account for the positive effects that the Oil Pollution discussed, and then the frequencies of spills from all sources
Act of 19905 (OPA ’90) has had on recent are addressed.
performance and is anticipated to have on future Frequencies of Production Spills. The annual frequencies
performance. for spills from production (Table 5) are shown on Fig. 5 for
3. Extrapolate predictions of future performance from each of the study systems. Note that the frequency of spills
the data set, applying corrections to the data-based tends to decrease as the spill size increases. Also, note that the
estimates if necessary. As an example, the relative magnitude of uncertainty in the estimated frequency
frequencies for small spills from subsea well systems increases as the spill size increases. This relative increase in
were increased from the data-based estimates to uncertainty occurs because large spills are rare events, so there
account for differences between subsea well systems are few occurrences available from which to estimate
and the platform well systems that dominate the data frequencies.
set. The information on Fig. 5 highlights the similarities and
4. Account for all sources of uncertainty in the differences among the systems regarding oil spills from
estimates, including the following: production. First, the Spar and the TLP are indistinguishable.
• the limited quality and quantity of relevant This result is reasonable in that the elements of the designs on
data records, especially for rare events; both systems that affect the potential for oil spills from
• the sometimes limited information available production are nearly identical on these two study systems.
on the exposures corresponding to the data Second, the Hub/Host Jacket tends to have smaller spill
sets; and frequencies from production than the Spar, TLP and FPSO for
• the extrapolation of future performance from spill sizes less than 1,000 bbl (Fig. 5). This difference is due
historical performance. to the smaller indigenous production rate on the shallow-water
5. Document the whole process clearly and thoroughly. Hub/Host Jacket versus the deepwater floating production
The detailed quantitative risk assessments for systems (Table 4).
fatalities and oil spills are contained in Gilbert et al.2. Third, the frequency of very small spills (less than 10 bbl)
on the FPSO is less than that on the Spar and TLP, even
Results though the production rates are similar on all of these study
The results from the quantitative risk assessment for fatalities systems. This difference is due to the large deck area and the
and oil spills are presented, analyzed and discussed in this solid decking that exist on an FPSO; the deck would contain
section. most small spills.
Fourth, the frequency of spills between 100 and 10,000 bbl
Risks for Fatalities. Results for the average total number of is slightly larger for the FPSO versus the other systems. This
fatalities are shown on Fig. 3 for each study system. The relative difference is because the FPSO has more subsea wells
results indicate that the fatality risks are very similar among than the other systems; subsea wells were considered to have a
the four study systems (Fig. 3). The expected contributions to higher leak frequency than platform wells because of a greater
the total fatality risk are shown on Fig. 4. potential for sand erosion and cutouts due to high flow rates
Production and drilling and well intervention activities and detection difficulties for sand. In addition, the FPSO has a
dominate the total fatality risk for all of the study systems. greater number of flowlines and flowline risers, which both
This result occurs because these activities require the bulk of contribute to the frequency of spills between 100 and 10,000
the man-hours over a 20-year lifetime. The estimated bbl.
frequencies of fatalities per man-hour worked for production Fifth, the Spar and the TLP have the potential for very
and for drilling and well intervention activities (Gilbert et al.2) large spills (greater than 10,000 bbl), although the frequency
are comparable to those reported for common industrial for these spills is very small (Fig. 5). The potential source for
activities (AIChE6) and for the oil and gas industry throughout these very large spills on the Spar and TLP is the dry tree
the world (OGP7). The contribution of drilling and risers. This risk does not exist on the FPSO study system
intervention activities to the total fatality risk for the FPSO is because the trees that control the reservoir pressure and flow
not as large as for the other systems because all of the wells on are on the seafloor (wet trees) rather than at the surface (dry
trees), and it is negligible for the Hub/Host Jacket study
system because of the lack of movement for this non-floating spills for the shuttle tanker are smaller than the annual
system. frequencies for pipelines (Fig. 7). One reason for this
A comparison with published information for the difference is that the potential for spills from the pipeline
frequency of large spills from production is shown on Fig. 6. remains a constant as long as there is oil in the pipeline,
Anderson and LaBelle8 report a frequency for spills greater regardless of the production rate. However, the potential for
than 1,000 bbl in size. Their frequency was estimated using spills from the shuttle tanker will go down as the production
spill data from offshore platforms operating in the United rate decreases since fewer offloading events are required.
States between the years 1974 and 1992. They report their Lastly, very large spill sizes (greater than 100,000 bbl) are not
frequency on the basis of the volume produced. In order to considered possible for pipelines due to operational and
develop Fig. 6, this frequency has been converted to an annual physical constraints (Gilbert et al.2), while they are possible
frequency for the study systems using the total volume of oil although infrequent for the FPSO. A spill between 100,000
produced in the 20-year lifetime for each system. and 500,000 bbl represents a major loss from the shuttle tanker
The estimated frequencies for the study systems are less due to a collision or explosion. A spill greater than 500,000
than the values obtained from Anderson and LaBelle8 (Fig. 6). bbl represents a major loss from the FPSO due to a collision or
There are two reasons for this result. First, two different data explosion.
sets have been used. In the CRA project, data before 1990 A comparison with published information for the
were discarded due to the implementation of new regulations frequency of large spills from transportation is shown on Fig.
in 1990 (API RP14C9), which improved operating procedures 8. Anderson and LaBelle8 report frequencies for spills greater
on platforms. The Anderson and LaBelle data set extends than 1,000 bbl in size from pipelines and tankers. Their
back to 1974. Second, the CRA study systems are not frequencies were estimated using spill data from offshore
representative of conventional, shallow-water platforms in the operations in the United States between the years 1974 and
United States, which dominate the population of platforms in 1992. They report their frequency on the basis of the volume
the Anderson and LaBelle data set. Note that the agreement produced. In order to develop Fig. 8, this frequency has been
between the CRA and Anderson and LaBelle is best for the converted to an annual frequency for the study systems using
Hub/Host Jacket study system (Fig. 6), which is most similar the total volume of oil produced in the 20-year lifetime for
to the platforms in the Anderson and LaBelle data set. each system.
Frequencies of Transportation Spills. The annual The estimated frequency for the Hub/Host Jacket is
frequencies for spills from transportation (Table 5) are shown comparable to that from Anderson and LaBelle8 (Fig. 8). This
on Fig. 7 for each of the study systems. The results highlight result is reasonable since the pipeline from the Hub/Host
the similarities and differences among the systems regarding Jacket is representative of the conventional, shallow-water
oil spills from transportation. pipelines that are in the Anderson and LaBelle data set.
First, compare the systems with pipelines. The Spar and However, the estimated frequencies for the Spar and TLP
the TLP are indistinguishable. This result is reasonable in that study systems are less than those obtained from Anderson and
the elements of the designs on both systems that affect the LaBelle (Fig. 8). The primary reason for this difference is that
potential for oil spills from transportation are nearly identical the Anderson and LaBelle frequency for pipeline spills is
on these two study systems. proportional to the volume produced. However, the potential
The Hub/Host Jacket has slightly smaller spill frequencies for spills from pipelines in the CRA study was related to the
from its pipeline than the Spar and TLP (Fig. 7). This length of the pipeline and the time of exposure, not the volume
difference is because there is a shorter length of pipeline of throughput. Therefore, the higher production rates for
exposed for the Hub/Host Jacket due to the shorter distance to deepwater systems do not necessarily lead to proportionally
the shore (Fig. 2). In addition, there is relatively less higher spill frequencies for pipelines.
uncertainty in the estimated spill frequencies for the Hub/Host The estimated frequency for the FPSO is lower than that
Jacket for spills less than 1,000 bbl (Fig. 7). The greater from Anderson and LaBelle8 (Fig. 8). This result is due to the
uncertainty for the Spar and TLP is mainly due to the potential different data sets used to estimate the frequency. The
for spills from the more flexible steel catenary export pipeline Anderson and LaBelle data set extends back to 1974, and
risers versus the more rigid risers on fixed jackets. The includes data from all U. S. coastal and offshore waters. In the
uncertainty for the Spar and TLP reflects that there are limited CRA project, data before 1992 were discarded due to the
data concerning the performance of these risers in deepwater implementation of the OPA ’903, which improved operating
applications. procedures on tankers and probably reduced the frequency of
Second, compare the two different types of transportation spills. Data for crude oil tankers in the Gulf of Mexico are
systems. There are notable differences between the pipelines summarized in Table 6 to support the hypothesis that data
for the Spar, TLP and Hub/Host Jacket and the in-field storage prior to the passage of OPA ’90 are not representative of
and shuttle tanker system for the FPSO. For very small spill existing conditions. In addition, data from outside of the Gulf
sizes (less than 10 bbl), the frequency of spills for the FPSO is of Mexico were not applied directly in the CRA project to
greater than from pipelines due to the potential for spills estimate the shuttle tanker risk in the Gulf of Mexico. An
during offloading from hoses and valves. For spill sizes analysis of tanker spills from 1992 to 1999 indicates that
between 1,000 and 100,000 bbl, the annual frequencies of frequencies of spills between 50 and 5,000 bbl and of spills
greater than 5,000 bbl in the Gulf of Mexico are spills between 10,000 and 100,000 bbl, which are only
approximately 40 percent of those for the rest of the world expected approximately once every 600 years of operation.
(Gilbert et al.2). Tanker spills are considered to be less likely Furthermore, most of the risk for the FPSO study system
on average in the Gulf of Mexico than in the rest of the world comes from spills between 100,000 and 500,000 bbl, which
for the following reasons, in order of importance: are only expected once every 4,500 years of operation. Table
1. The regulatory environment in the Gulf of Mexico is 7 and Fig. 11 show how the contributions to the risks for the
more restrictive; Spar and TLP versus those for the FPSO are different even
2. The environmental conditions in the Gulf of Mexico though the resulting risks are comparable (Fig. 10).
are less severe; One effect of the spill risk being dominated by rare, high
3. The consequences of grounding are significantly less consequence events is that the confidence intervals in the
due to the lack of rocky coasts in the Gulf of Mexico; predicted average oil spill volumes range over nearly an order
4. Shuttle tankers used in the Gulf of Mexico have a of magnitude (Fig. 10). This uncertainty reflects the typically
smaller parcel size on average; limited quantity and quality of historical data available to
5. The Gulf of Mexico has less congested waterways on estimate frequencies for rare events. Note that the confidence
average; and interval for the FPSO is wider than those for the other systems
6. Newer vessels are used in the Gulf of Mexico due to (Fig. 10) because there are relatively fewer data available for
recent federal regulations. FPSO’s in the Gulf of Mexico and because the FPSO risk is
Frequencies of Spills from All Sources. The annual dominated by very rare events with expected return periods of
frequencies for spills from all sources, including production approximately 4,500 years.
and transportation, are shown on Fig. 9 for each of the study The contributions to the total oil spill risk from different
systems. The frequencies for spills are generally dominated sub-systems are shown on Fig. 12. Production, which
by production-related spills for spill sizes up to 1,000 bbl and dominates the smaller spill sizes (Fig. 5), does not contribute
by transportation-related spills for spill sizes greater than substantially to the total risk (Fig. 12). The main contributor
1,000 bbl. Therefore, the similarities and differences among to oil spills from production are related to the processing
the study systems are related to those for production for spills facilities (topsides on Fig. 12). Transportation, which
less than 1,000 bbl (Fig. 5) and to those for transportation for dominates the larger spill sizes (Fig. 7), is the main contributor
spills greater than 1,000 bbl (Fig. 7). Note that the Spar and to the total oil spill risk (Fig. 12).
TLP are indistinguishable for all spill sizes. Maximum Single Oil Spill in Lifetime. Results for the
Total Volume of Oil Spilled over Lifetime. Results for the average single maximum spill are shown on Fig. 13 for each
average total volume are shown on Fig. 10 for each study study system. The results indicate that the risks for the
system. These results indicate that the systems provide very different study systems are comparable. Furthermore, these
comparable risks. The risk for the Hub/Host Jacket is slightly results emphasize that the maximum spill volume from a
smaller than the risks for the other systems because it has a single incident dominates the average total spill volume.
smaller production rate and a shorter transportation distance to More than 70 percent of the total is expected to come from a
the shore. The risks for all of the deepwater systems (Spar, single incident. The wide confidence intervals on Fig. 13
TLP and FPSO) are nearly identical even though the reflect the uncertainty inherent in estimating frequencies for
frequencies for different spill sizes are not identical (Fig. 9). rare events.
This result occurs because the risk is a measure of both
frequency and consequence (spill size). While very large Conclusions and Recommendations
spills (greater than 100,000 bbl) are more likely with the A quantitative risk analysis was performed to assess and
FPSO than with the Spar or TLP, the annual frequencies are compare oil spill and fatality risks for four representative
still small. Furthermore, the frequencies for spills less than deepwater production systems in the Gulf of Mexico. Three
100,000 bbl for the FPSO are generally smaller than those for of the study system types have already been operated
the TLP or Spar (Fig. 9). Therefore, the risks for the Spar, successfully in the Gulf of Mexico: two floating production
TLP and FPSO are comparable. systems in deepwater with oil pipelines, a Spar and a Tension
In order to facilitate interpretation of the results on Fig. 10, Leg Platform (TLP); and a shallow-water jacket serving as a
the relative contribution of each spill-category to the total hub and host to deepwater production. One of the study
volume spilled is shown on Fig. 11. Note that the chronic system types has not been used in the Gulf of Mexico: a
environmental risk is dominated by large spills (greater than tanker-based Floating Production Storage and Offloading
1,000 bbl), which are low frequency but high consequence (FPSO) system with oil transportation to shore via shuttle
events. Therefore, most of the systems in a fleet of study tankers. The objective of this analysis was to understand and
systems will have small volumes of oil spilled. Occasionally, compare the risks of the FPSO with those for currently
one of the systems may have a large spill and this large spill acceptable alternatives for deepwater production.
will dominate the average for the fleet. To emphasize this Conceptual system descriptions that are representative of
point, Table 7 summarizes the expected time between spills of existing and typical technology in the Gulf of Mexico were
different sizes for each of the study systems. Note that most developed for the four systems. The scope of these
of the risk for the Spar and TLP study systems comes from descriptions included the entire production systems and
operations from the wells through the transport of product to 2. The quality of existing data sets for the Gulf of
the shore. Mexico should be improved so that they are of
Three risk measures were assessed and analyzed for each greater value in future risk analyses. First, the type
system: the total number of fatalities in a 20-year production and quality of data that are currently collected should
life as a measure of the human safety risk, the total volume of be evaluated, and any changes recommended from
oil spilled in a 20-year production life as a measure of the this evaluation should be implemented in a timely
chronic environmental risk, and the maximum volume spilled manner. Second, single agencies should be
in a single incident in a 20-year production life as a measure responsible for tracking and compiling similar types
of the acute environmental risk. The process of developing of data. Third, all data records should be reviewed
the conceptual descriptions for the systems and then annually by the industry and regulators to improve
evaluating the risks has drawn on expertise from all facets of the clarity, quality and usefulness of the information
oil and gas production, including operators, contractors, in these records. Finally, the data should be
manufacturers, class societies and regulators. published annually in a clear and an easily accessible
format.
Conclusions. The following major conclusions have been 3. Additional information about the populations of
drawn from the results of this analysis: offshore facilities and operations in the Gulf of
1. There are no significant differences in the fatality Mexico should be collected on an annual basis.
risks among the four study systems. Specifically, the following information from federal
2. There are no significant differences in the oil-spill and state waters in the Gulf of Mexico would be
risks among the four study systems. valuable: the length of active pipelines operating per
3. The average total volume of oil spilled during the year, the number of tanker on-loading and off-
facility lifetime will be dominated by rare, large spills loading events in ports and lightering zones per year,
rather than frequent, small spills. and the number of man-hours in production-related
4. The major contribution to the oil spill risks for all activities, supply vessel operations and tanker
systems is the transportation of oil from the operations per year.
production facility to the shore terminal by either 4. Uncertainty in the predicted performance for these
pipelines or shuttle tankers. Spill risks for pipelines four study systems should be considered carefully in
and shuttle tankers are comparable, although the drawing conclusions from and applying the results
frequencies and sizes of possible spills are different from this study.
for pipelines versus shuttle tankers. The spill risks 5. The process used on this project to assess risks has
for pipelines are dominated by the possibility of spills been effective in obtaining valuable technical
between 10,000 and 100,000 bbl in size that are information from industry and regulators, and should
expected to occur once every 600 years on average. be considered in supporting other analyses of new
The spill risks for shuttle tankers are dominated by technology in the Gulf of Mexico.
the possibility of spills between 100,000 and 500,000
bbl in size that are expected to occur on average once Acknowledgments
every 4,500 years. The authors wish to acknowledge the Minerals Management
5. The confidence intervals in predicted oil spill Service for funding this project. Specific individuals at MMS
volumes range over about an order of magnitude, who have been instrumental in this study are Paul Martin,
reflecting the limited quantity and quality of Charles Smith, James Regg and Cheryl Anderson. In addition,
historical data available to estimate frequencies for the following individuals have played key roles in the project:
rare events. Professor Larry Lake and Research Assistant Jihad Jaber, The
Therefore, the expected risks associated with the FPSO are University of Texas at Austin; Joe Gebara, EQE International;
comparable to those for already accepted alternatives for Lieutenant Commander Bill Daughdrill and Joe Myers, United
deepwater production, including a Spar, a TLP and a shallow- States Coast Guard; and Allen Verret, Private Consultant.
water jacket serving as a hub and a host to deepwater Finally, the Deepstar Consortium has been instrumental in
production. coordinating and facilitating the involvement of industry
experts on workshop teams.
Recommendations. The following recommendations have
been developed from this work: References
1. The results from this study should be periodically 1. MMS (2000), “Proposed Use of Floating Production, Storage,
updated because they provide a valuable baseline for and Offloading Systems On the Gulf of Mexico Outer
future analyses of risk in the Gulf of Mexico. The Continental Shelf, Western and Central Planning Areas, Draft
three measures of risk used in this study can all be Environmental Impact Statement,” Prepared Under MMS
Contract 1435-01-99-CT-30962, Minerals Management Service,
readily measured and tracked for new and existing Gulf of Mexico OCS Region.
deepwater production facilities in the Gulf of 2. Gilbert, R. B., Ward, E. G. and Wolford, A. J. (2001),
Mexico.
“Comparative Risk Analysis for Deepwater Production Chemical Engineers, New York, New York.
Systems,” Final Project Report, Prepared for Minerals 7. OGP (1999b), “Safety Performance of the Global E&P Industry,
Management Service, Washington, D. C. 1998,” Report No. 6.80/295, International Association of Oil
3. MMS (2000), “MMS OCS Spill Database,” Minerals and Gas Producers, London, England.
Management Services, Available: cheryl.anderson@mms.gov, 8. Anderson, C. M. and LaBelle, R. P. (1994), “Comparative
Accessed: January, 2000. Occurrence Rates for Offshore Oil Spills” Spill Science and
4. USCG (1999), “Marine Casualty and Pollution Database,” CD- Technology Bulletin, Vol. 1, No. 2, 131-141.
ROM, Subscription Order No. 5441INC, Available: National 9. API RP14C (1998), “Analysis, Design, Installation and Testing
Technical Information Services, Springfield, Virginia. of Basic Surface Safety Systems on Offshore Production
5. Oil Pollution Act of 1990 (OPA ’90), 33 USCA Sec. 2701-2761. Platforms,” Sixth Edition, American Petroleum Institute,
6. AIChE (1989), Chemical Process Quantitative Risk Analysis, Washington, D.C.
Center for Chemical Process Safety, American Institute of
Table 1 - Industry sources for workshop participants.

Oil Companies Consultants & Contractors Classification Societies
BP Amoco EQE ABS
Chevron ABB Lloyd’s Register
Conoco Atlantia DNV
Elf FMC
ExxonMobil Paragon
Marathon McDermott
Oxy Navion
Shell
Statoil
Texaco
Table 2 - Additional industry sources for technical expertise.

Skaugen Petrotrans Association of Diving Contractors
SBM IMODCO Oceaneering
Global Maritime Cameron
Aker Mentor
R&B Falcon Bay Ltd.
Transocean SedcoForex Spirit Energy
Edison Chouest Horizon Engineering
Tidewater Marine Kerr McGee
HSAC Mathews Daniels
Air Logistics LOOP
PHI State of Louisiana
ERA Aviation
Table 3 - Summary of risk measures.

Risk Measure of Risk Unit
Human Safety Total Fatalities over Production Lifetime Number of Fatalities
Environmental – Chronic Total Volume of Oil Spilled over Production Lifetime bbl of Oil
Environmental – Acute Maximum Single Spill Volume in Production Lifetime bbl of Oil
Table 4 - Summary of attributes for study systems.

Spar TLP Hub/Host Jacket FPSO
Water Depth (ft) 4,000 4,000 600 5,000
Peak Production
Oil (bopd) 150,000 150,000 50,000 150,000
Gas (scfpd) 200,000 200,000 50,000 200,000
Export
Oil (bopd) 150,000 150,000 250,000 150,000
Gas (scfpd) 200,000 200,000 550,000 200,000
Wells
Platform 6 6 6 0
Subsea (MODU) 3 3 3 9
Manning
Production 30-45 30-45 30-45 30-45
Marine 6 6 0 10
Drilling – Platform 65 65 50 0
Drilling MODU 65 65 65 65
Table 5 - Sub-system categories used in risk assessment.

Risk Exposure
Sub-System Category
Measure Factor
Production man-hours
Drilling man-hours
Fatalities Supply Vessels docking calls
Helicopter Transport passengers
Tanker Operations docking calls
Major Accident platform-years
Well Systems – Platform (or Surface) bbl produced
Well Systems – Subsea bbl produced
Dry Tree (or Production) Risers riser-years
Production Flowlines mile-years
System Import Flowline Risers riser-years
Topsides bbl processed
Oil Spills
Supply Vessels docking calls
Drilling and Intervention man-hours
Pipelines mile-years
Transportation Export Pipeline Risers riser-years
System Shuttle Tanker (Offloading in Field and at Port) docking calls
FPSO Cargo Tank platform-years
Table 6 - Summary of Oil Spills from Crude Tankers in Gulf of Mexico

1
Number of Spills 1
Volume Spilled
Year 1-10 10-100 100-1,000 1,000-10,000 10,000-100,000 (bbl)
bbl bbl bbl bbl bbl
1985 0 1 0 0 0 30
1986 2 1 0 0 0 28
1987 4 0 0 0 0 5
1988 5 1 0 0 1 15,401
1989 3 1 2 0 0 1,146
1990 7 2 1 0 0 266
1991 5 0 0 0 0 17
Sub-Total 26 6 3 0 1 16,893
1992 0 0 0 0 0 0
1993 2 1 1 0 0 191
1994 2 0 0 0 0 8
1995 0 0 1 0 0 179
1996 0 0 0 0 0 0
1997 1 0 0 0 0 2
1998 1 1 0 0 0 22
1999 1 0 0 0 0 9
Sub-Total 7 2 2 0 0 411
1 4
Note: Data from USCG .
Table 7 - Expected return periods for spills.

Expected Return Period between Spills (years)
100 - 1,000 1,000 - 10,000 10,000 - 100,000 100,000 - 500,000 500,000 - 1,000,000
System 1 – 10 bbl 10 – 100 bbl
bbl bbl bbl bbl bbl
Spar 0.8 3 15 60 580 Not Credible Not Credible
TLP 0.8 3 15 60 580 Not Credible Not Credible
Hub/Host Jacket 3 8 35 91 920 Not Credible Not Credible
FPSO 3 3 12 110 2,500 4,700 300,000
Draft System Workshop #1 System Definition

Descriptions Develop System Descriptions
Develop Preliminary Event/Outcome Tables
Gas Flange r Oil Flange

Workshop #2 Hazard Identification (Port Terminal) r
Elicit Event/Outcome Information (Port Terminal)
Conduct Preliminary QRA

l Hub/Host
Shuttle
Workshop #3 Quantitative Risk Analysis
Gas Tanker
Elicit Frequency/Consequence Input
Oil Pipeline
Refine QRA & Perform Additional Studies Pipeline
Workshop #4 Review Prepare Final TLP

ll
Review and Refine QRA Report FPSO
Spar
Fig. 1 - Flowchart for workshop process. Fig. 2 - Physical layout for study systems (plan view).
3.0
Average Total Number of Fatalities
2.5
2.0
in Lifetime
1.5
1.0
0.5 90% confidence intervals
0.0
Hub/Host
Spar TLP FPSO
Jacket
System
Fig. 3 - Average total number of fatalities in lifetime.

Spar/TLP
Production
Drilling and Intervention
Supply Vessels
Helicopter Transport
Tanker Operations
Drilling and
Intervention Major Accident
(52%)
Hub/Host Jacket FPSO
Production
(32%)
Drilling and
Intervention
(52%)
Fig. 4 - Expected contributions to average total fatalities versus activity. The categories are shown clockwise in the order in the legend.
Annual Frequency for Production Spills (per year)
1.0E+01
Spar
1.0E+00 TLP
Hub/Host Jacket
1.0E-01 FPSO
1.0E-02
1.0E-03
Expected value
1.0E-04
90% confidence
1.0E-05 interval
1.0E-06
1-10
100,000-500,000
10-100
100-1,000
1,000-10,000
10,000-100,000
500,000-1,000,000
Spill Size (bbl)
Fig. 5 - Annual frequency for production spills versus spill size. The intervals within each spill size are shown in the order in the legend.
Annual Frequency of Production Spills Greater than 1,000

1.0E+00
1.0E-01
Anderson and LaBelle (1994)
bbl (per year)

1.0E-02
1.0E-03
CRA (90%
1.0E-04 confidence intervals)
1.0E-05
Hub/Host
Spar TLP FPSO
Jacket
System
Fig. 6 - Comparison of production spill (>1,000 bbl) frequencies with published data. CRA denotes the results from this study.
Annual Frequency for Transportation Spills (per year)
1.0E+01
Spar
1.0E+00 TLP
Hub/Host Jacket
1.0E-01 FPSO
1.0E-02
1.0E-03
1.0E-04 Expected value
1.0E-05 90% confidence

interval
1.0E-06
100-1,000
100,000-500,000
1-10
10-100
1,000-10,000
10,000-100,000
500,000-1,000,000
Spill Size (bbl)
Fig. 7 - Annual frequency for transportation spills versus spill size. The intervals within each spill size are shown in the order in the legend.
Annual Frequency of Transportation Spills Greater than

1.0E+00
Anderson and LaBelle (1994)

1.0E-01
1,000 bbl (per year)

1.0E-02
1.0E-03
CRA (90%
confidence intervals)
1.0E-04
1.0E-05
Hub/Host
Spar TLP FPSO
Jacket
System
Fig. 8 - Comparison of transportation spill (>1,000 bbl) frequencies with published data. CRA denotes the results from this study.
1.0E+01
Spar
1.0E+00 TLP
Hub/Host Jacket
Annual Frequency (per year)
FPSO
1.0E-01
1.0E-02
Expected value
1.0E-03
90% confidence
1.0E-04 interval
1.0E-05
1.0E-06
1-10
1,000-10,000
100,000-500,000
500,000-1,000,000
10-100
100-1,000
10,000-100,000
Spill Size (bbl)
Fig. 9 - Annual frequency for spills from all sources versus spill size. The intervals within each spill size are shown in the order in the legend.
10000
90% confidence intervals
9000
8000
Average Total
Volume of Oil Spilled

7000
over Lifetime (bbl)

6000
5000
4000
3000
2000
1000
0
Hub/Host
Spar TLP FPSO
Jacket
System
Fig. 10 - Average total volume of oil spilled over lifetime – all sources.
Spar TLP Hub/Host Jacket FPSO

Contribution to Average Total Volume Spilled in
100%
80%
60%
Lifetime
40%
20%
0%
00
0
00
00
00
0
00
00
-1
,0
,0
,0
-1
,0
0,
00
00
-1
1
00
-1
10
-5
,0
-1
0
-1
10
00
0
00
00
0
1,
,0
00
0,
10
10
0,
50
Spill Size (bbl)
Fig. 11 - Contribution to average total spill volume versus spill size. The bars within each spill size are shown in the order in the legend.
Spar/TLP Well Systems - Platform

Well Systems - Subsea
Dry Tree Risers
Flowlines
Import Flowline Risers
Topsides
Export Pipeline Risers
Pipelines
Pipelines (72%)
Shuttle Tanker
FPSO Cargo
Supply Vessels
Drilling and Intervention
Hub/Host Jacket FPSO
Pipelines (85%) Shuttle Tanker

(63%)
Fig. 12 - Contribution to average total spill volume versus spill source. The categories are shown clockwise in the order in the legend.
10000
90% confidence intervals
9000
8000
Average Total
7000
Volume of Oil Spilled
over Lifetime (bbl)
Average Single
6000 Maximum
5000
4000
3000
2000
1000
0
Hub/Host
Spar TLP FPSO
Jacket
System
Fig. 13 - Average maximum volume spilled from a single incident in the lifetime.
ABS BIOGRAPHIES
ANDY WOLFORD
Dr. Wolford has worked in industrial

risk assessment for 16 years. He has
directed risk applications on a
diverse range of engineered systems,
including offshore and onshore oil
and gas installations, mobile offshore
drilling units, and marine trans-
portation systems in the U.S.,
Central and South America, the
North Sea, and offshore Malaysia and Australia. With a
focus on risk analysis and reliability engineering, Dr.
Wolford has worked with numerous organizations and
companies to develop quantitative risk assessments, which
could be utilized to make more informed business decisions.
Dr. Wolford earned his Sc.D. from the Massachusetts
Institute of Technology.

Otc 2001

Uploaded by

Copyright:

Available Formats

Otc 2001

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Otc 2001

Uploaded by

Copyright:

Available Formats

Offshore

Risk Based ‘Optimum’ Inspection for FPSO Hulls

Reliability Methods for Deepwater Position-Mooring

FPSO Standards and Recommended Practices

A Comparitive Risk Analysis of FPSO’s with Other Deepwater

Copyright 2001, Offshore Technology Conference

1. “A Guide to Risk Assessment for Offshore Installations

Table 1 Description of Model Partitions

System Initiating Events Event Trees

Figure 1. Overall FPSO Risk Model Structure

I.E.1 Support Frontline End X1

Component Generic Into

Turret Topside Hull

Figure 4a Mooring Event Sequence Diagram

Single Mooring No adverse

Design Offset Shut-in Inititaite

Release well Shut-in Inititaite

Figure 4b Mooring Event Sequence Diagram

Controlled Vessel Initiate

Figure 5 Mooring Event Tree

IE MOORS STATN RSSTR RSLOC TUG GRND VSSL FIXD B# S#

Figure 6a Transient Induced Leak Frequency Fault Tree

Figure 6b Transient Induced Leak Frequency Fault Tree

Figure 7 Mooring Redundancy Risk Results

1E+00 Repair Cost = $3-10 Million

1E-03 Low Value Facility

1E-09 High Value Facility

Risk Based ‘Optimum’ Inspection for FPSO Hulls

Copyright 2001, Offshore Technology Conference

ABSTRACT • General (global conditions),

developed as event updating using First Order Reliability

This approach is to update the probability of events, such as P[ p | q r H] = 1 − PD (a ) (6)

The same methodology can be applied in the crack detection

and f " (µ) = cL(µ )f ' (µ) (9)

 n  • The use of a ‘living’ process that is flexible, strives for

• Inspecting high likelihood of damage or defect ‘parts’ Inspection Data System

REENGINEERING THE IMMR PROCESS ACKNOWLEDGEMENT

Committee on Human Performance, organizational Systems,

(1) Failure Modes, Effects, and Critically Analysis

(2) Development of Risk Based Inspection Program

Figure 3 Risk-based Inspection Process

1. Select Potential Inspection Strategies

2. Choose an Inspection Strategy and Perform Inspection

(1) Obtain More

Figure 6 Development of In-Service Inspection Program

Number - Time Available

Figure 7 Factors that affect Inspection Performance

Yong Bai is Manager of Offshore

Yong has a MSc. degree and a Ph.D. degree in naval

Reliability Methods for Deepwater Position-Mooring Design and Analysis

Copyright 2001, Offshore Technology Conference

(6) eqn. (8). The formulae above allow break strength to be

Turret and spread moored ships

Spread moored semisubmersibles Mooring System Types

Spread moored spars

Pseg = 1-(1-Ptest length)Nseg

3x3 Grouped Mooring Pattern

Normalized Wind, Wave and Current, X/100-year Characteristic Design Value

Figure 4 Break Test Data and Fitted Distributions