OTC 2001
Content
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 Reliabili bility Methods for De Deepwa pwater Posi Positiontion-M Mooring ooring 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 This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, 30 April–3 May 2001. This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
Abstract FPSOs and other floating offshore facilities typically follow “prescriptive based” classification rules for design of the hull, mooring and marine systems. In some cases the process facilities are also classified. An alternative approach is to use a more open framework, “risk based” design approach that allows variation from prescriptive rules provided system risks are maintained at acceptable levels. The various classification societies currently allow such risk-based alternatives [2, 3]. Although numerous “component” risk studies for FPSOs have been conducted and published, this is one of the first that accounts for the integration and linking of risks and risk tradeoffs among the hull, mooring system, marine systems, topside process plant and the utility, power and control systems that support them. The model and basis are first described, followed by application to a prototype deepwater, turret moored FPSO with gas handling. Example cases are shown to demonstrate use of the model to make design decisions to the various FPSO components. Introduction Classification rules are established based on engineering principles, experience, testing & expert judgement. They are intended to ensure probabilities of accidents are low, but this is not explicit. Changes to developing and implementing ABS rules are being explored through risk based approaches. An alternative, risk-based approach to classification of Floating Production, Storage and Offloading systems is being investigated by ABS as part of a major internal technology development project. The project is comprised of model, database and methodology development and training at multiple levels throughout the organization. The prototype model was completed early in 2001 and is the focus of this
paper. The model represents systems failures more comprehensively than any other offshore risk assessment known to the authors. This level of detail was sought in order to achieve specific goals: 1) Develop explicit risk measures of class rules to facilitate prioritization and optimization, thus allowing one to focus resources on the greatest risk contributors, 2) Develop a consistent means for performing risk tradeoffs (i.e. demonstration of equivalent level of safety), 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.
Model Development The overall model is comprised of a collection of initiating events, facility response model and consequence calculations. These are categorized into discrete damage states (see Figure 1). Also illustrated in Figure 1, the facility response model is comprised of support event trees, frontline event trees and end states. The support event trees represent the failed/operational state of support systems (e.g. utilities, instrumentation and control, emergency) required for successful operation of the frontline (main system) event trees. The end states are simply a discrete categorization of the various failed configurations of the facility. Frontline systems were segregated into three categories for convenience: process, marine and structural systems. Each of these model partitions are described in summary in Table 1, and the development activities are described in the following sections. Process Systems The process model follows a conventional offshore QRA approach [1,4]: • Development of isolatable sections, • Summarize the loss of containment frequency by using a parts count approach • Identifying spatial interactions that could lead to escalation These steps were completed for 54 isolatable sections, using the event tree structure of Figure 2. Three hole sizes were selected to represent the hole size distribution of various process equipment. Multi-phase releases (oil / gas / water) were treated where applicable. Leak frequencies were derived primarily from generic databases (e.g. E&P Forum, OREDA, Offshore Hydrocarbon Release Statistics). The model
2
WOLFORD ET AL.
explicitly accounts for emergency detection as well as process control response to a loss of containment event, so that risk importance measures can be associated with rule and standardrequired I&C. Marine Systems The marine modeling followed a broadly similar approach to the process modeling, except the scope of marine events was broader than loss of containment of hydrocarbons. The majority of the marine event trees addressed fires (fueled, electrical, other). Marine event scenarios were represented with 89 unique initiating events, 12 frontline system event trees, one support tree and 141 marine fault trees of which 89 developed specific initiating events and 52 modeled system response functions (see Table 1). Over 2 billion unique event sequences were evaluated. Fire initiating event frequencies were developed for 70 individual hazard zones combined with an assessment of initiator density. An explicit parts count approach was not utilized for marine fire initiation based on unavailability of component-specific fire initiator frequency data. Structures The modeling of structural failures also followed a broadly similar approach, but deviated from developing a componentwise model and relied substantially upon subject matter experts to identify, prioritize and structure the event sequences. The approach is illustrated in Figure 3. As shown in that figure, structure subsystems were broken down into mooring, turret, topside structures and hull. Team meetings were held to construct event sequence diagrams – which illustrate the sequential (hardware and human) event sequence of events resulting in a system failure. The event sequence diagram has proven to be a superior tool to portray and elicit necessary expert input. An example is shown for mooring in Figure 4. From this intermediate step, system event trees were developed analogous to those for the process and marine frontline event trees. The mooring event tree is shown in Figure 5. Structural event scenarios were represented with 46 unique initiating events, 13 frontline system event trees and one support tree. Consequence Modeling Risk metrics were defined to measure health, safety, environmental and financial impacts from the FPSO. The endpoint metrics selected were: fatalities, oil spill, capital loss and business interruption (the latter two only if associated with a scenario with potential for fatality or spill, i.e. the entire scope of financial impacts ere not included in this phase of the model, only a subset). The physical phenomena represented in this model drew upon de facto standards used in offshore QRA [1]. • Release modeling, multi-phase, near field flow regime, internal pressure-time history • Thermal radiation effects to humans and equipment from jet fires and pool fires
• •
OTC 12948
Explosion overpressure impacts to humans and equipment Simple evacuation of personnel on board.
Example Applications Two applications are used to highlight the utility of the model described in this paper: • Comparison of loss of containment frequency for a production separator among material failure mechanisms vs. transient induced leaks • Comparison of the design of mooring line redundancy requirements Leak Initiators Various codes and rules [2,3] specify the use of process instrumentation to protect against accidental overpressure of hydrocarbon containing equipment. These prescriptive rules are effective in minimizing overpressure transients that can potentially lead to significant loss of containment. As such, these measures provide non-explicit risk mitigation of offshore production facilities. A comparative exercise was performed to determine the extent to which individual instrumentation and control component should be represented in the risk model. The exercise formulated a fault tree model of a prototypical production separator, and quantified the transient induced leak frequency, for major leaks, and compared to the historical, “inherent” leak frequency. The fault tree is shown in Figure 6. The results, using point estimates only, show:
Inherent Leak Frequency Transient Induced Leak Frequency
All Leaks: 0.42/yr Major Leaks 0.04/yr Major Leaks: ~1E-8/yr
The result of the above exercise shows that the transient induced leak frequency is substantially lower than the inherent leak frequency, hence the need to treat the individual equipment failures explicitly is unwarranted, unless a process system departs radically from existing codes and standards. Mooring Line Redundancy An 8-leg, external turret mooring system was evaluated with three hypothetical design criteria • No redundancy • Single line failure (per ABS Rules) • 2 Line failure Parameters were modified in the FPSO risk model to incorporate modifications in failure rates and consequences. The mooring event tree is shown in Figure 5. Frequencies were tabulated for the four risk metrics. Figure 7 shows that substantial risk reduction can be achieved using a dual line failure criterion.
Summary A comprehensive risk model has been constructed to assist in FPSO design risk tradeoffs. This model is envisioned to be useful in supporting risk based classification activities.
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
References 1.
2. 3. 4.
5.
6.
7.
“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. Guide for Building and Classing Facilities on Offshore Installations, American Bureau of Shipping, June 2000. Guide for Building and Classing Floating Production Installations, American Bureau of Shipping, June 2000. Guidance Notes on Risk Assessment Application for the Marine and Offshore Oil and Gas Industries, American Bureau of Shipping, June 2000. “Hydrocarbon Leak and Ignition Database”, Report 11.4/180, DNV Technica, prepared for E&P Forum, June 1992. “Hydrocarbon Release Statistics Review”, David Mansfield, AEA Technology, A report produced for UKOOA, January 1998. Offshore Reliability Data, OREDA Participants, distributed by DNV Technica, 1993.
3
4
WOLFORD ET AL.
Table 1 Description of Model Partitions System Category
Initiating Events
Event Trees
•
• •
Process
•
• • Marine
• Structura l
Total
171 Loss of Containment Fault Tree modeling each of the 171 initiating events (Parts Count) 57 Escalation Events 89 (70 fires by zone)
46 Structural Damage
363 Initiating Events
Comprised of 54 process section Each of the 2 phases in 3 separators is modeled as a separate initiating event • 3 Hole sizes (small, medium, large) • Model Functions Ø Ignition/Explosion Ø Isolation Ø Blowdown Ø Fire Suppression • Failure of cargo management • 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 • Single mooring line failure • Corrosion holes or fatigue crack in turret 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 17 Billion individual event sequences
OTC 12948
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
Table 2 Top Events for Mooring Trees
TOP EVENT
IE MOORL OFST MOORS STATN RSSTR RSLOC TUG GRND VSSL FIXD
5
Table 3 Top Events for Process Tree TOP EVENT IE IGNE PSL LSL PISO GDET IGNL FDET ESDO ISOL BLDN FWSP FOAM
DESCRIPTION
Initiating Event Single mooring line fails Vessel drift exceeds design limits Mooring system fails (Loss of 2 or more lines) Loss of station Production risers experience stresses greater than ultimate limits Riser cracking occurs breaching containment Tug not available for recovery operator Vessel runs aground Vessel impacts passing vessel Vessel impacts fixed installation
DESCRIPTION Initiating event Immediate ignition Failure of PSL for isolation Failure of LSL for isolation Failure of isolation by Process PSL/LSL Gas detection system Late ignition Fire detection Manual actuation of ESD and blowdown Failure of isolation Blowdown valves Water spray deluge suppression Foam/water fire suppression
Figure 1. Overall FPSO Risk Model Structure
Initiating Event I.E.1 I.E.2
Support Event Tree
Frontline Event Tree
End States E1 E2
• •
Consequence Model
Facility Response Model
S1 S2
• • •
F1 F2
•
•
• • •
X1 X2 Consequence Analysis
• • • • • •
• •
•
• •
•
•
• •
•
I.E.n
Xl
Em
S1
System
Component
Failure Mode
Generic Into Population Data Facility-Specific Data
Damage
ø
Figure 2 Process Loss of Containment Event Tree
Figure 3 Structural Failure Modeling Approach
Mooring System
System Familiarization
Subject Matter Expert
System Breakdown
Subject Matter Expert
Turret System
Topside Structural System
Moderated Meetings
Hull System
Subject Matter Expert Subject Matter Expert
Build Event Sequence Diagrams
Build Event Tree Diagrams Frequency Input Quantification End State Assignment
8
WOLFORD ET AL.
OTC 12948
Figure 4a Mooring Event Sequence Diagram
Single Mooring Line Failure
No
No adverse effect
No
Shut-in Production
Yes
Design Offset Exceeded
No
Yes Initiate Repairs
Yes
Mooring System Failure
No
Overstress Risers
Yes Loss of Station
Overstress Risers
No
Shut-in Production
No
Inititaite Repairs
No
Inititaite Repairs
Yes Initiate Repairs
Yes
Release well fluids
No
Shut-in Production
Yes Initiate Repairs
Yes
Shut-in Production
Yes Initiate Repairs
No
Inititaite Repairs
Inititaite Repairs
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
Figure 4b Mooring Event Sequence Diagram
Loss of Station
No
Overstress Risers
Yes
Risers Severed
No
Overstress Risers
Release well fluids
Shut-in Production
Vessel Drift
Controlled (not available for extreme weater events)
Uncontrolled Grounding
Loss of Asset
Polution
Boat Impact
Loss of Asset
Polution
Impact with other offshore installation
Loss of Asset
Polution
Vessel Recovery
Initiate Repairs
Vessel Recovery
Initiate Repairs
9
10
WOLFORD ET AL.
OTC 12948
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
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
Figure 6a Transient Induced Leak Frequency Fault Tree
11
12
WOLFORD ET AL.
Figure 6b Transient Induced Leak Frequency Fault Tree
OTC 12948
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
13
Figure 7 Mooring Redundancy Risk Results
Minimum Consequences 1E+00
Fatalities = None Oil Release = 0 barrels Repair Cost = $10-100 Thousand Downtime = None
1E-01 1E-02 1E-03 y c n e u q e r F
1E-04 1E-05 1E-06 1E-07 1E-08 1E-09 1E-10 None Single Line (ABS Rules)
Two Lines (BP Proposal)
Mooring Line Redundancy
Maximum Consequences
Fatalities = 1 Oil Release = 30 - 40,000 barrels Repair Cost = $3-10 Million
1E+00
Downtime = 1 - 3 months
1E-01 1E-02 1E-03
Low Value Facility (well protecter)
y 1E-04 c n e u 1E-05 q e r F 1E-06
Typical Offshore Facility (fixed platform)
1E-07 1E-08
High Value Facility
1E-09 1E-10 None Single Line (ABS Rules)
Two Lines (BP Proposal)
Mooring Line Redundancy
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 transportation 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.
J AMES C . LIN James Lin is a Senior Consultant at EQE International with over 17 years of experience in system engineering, reliability and availability analysis, and probabilistic safety assessment. Mr. Lin has a wide range of experience that includes acting as a project manager, performing human reliability analysis, and analyzing 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 from National Tsing Hua University, Taiwan and his M.S. in Nuclear Engineering from University of California, Los Angeles. He is a Registered Nuclear Engineer, California and a Certified Reliability Engineer, American Society of Quality Control.
ROBERT E. SHEPPARD (picture unavailable) Robert Sheppard, a Principal Engineer with EQE International, has over twelve years of experience in structural engineering and reliability analysis. Mr. Sheppard has been involved in projects across many industries including offshore oil, onshore petrochemical, nuclear 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.
J AMES K. LIMING James K. Liming is a Senior Reliability and Risk Management Consultant and Corporate Associate with over 17 years of experience in managing and performing complex engineered facility reliability engineering, risk analysis, and operations and maintenance support. He has a diverse, wellbalanced background including handson power plant operating and maintenance experience as well as extensive analytical expertise. He has served as project manager or project engineer on several major industry and government risk management projects worldwide, and is noted as a leading practitioner of probabilistic risk 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 and technical papers on risk and reliability analysis applications tools and techniques. He has authored or coauthored over 70 publications on risk and reliability analysis methods and applications. He is a former fully qualified U. S. Navy nuclear submarine officer, and he is currently a Captain (O-6) and Naval Sea Systems Command unit Commanding Officer in the U. S. Naval Reserve. He holds a B. S. degree from the U. S. Naval Academy and an S. M. degree in nuclear engineering from the Massachusetts Institute of Technology (MIT).
ANDY LIDSTONE (picture unavailable) Andy Lidstone is a Senior Engineer with EQE International with over fifteen years of industry experience for offshore and onshore oil and gas facilities and chemical plants. Mr. Lidstone’s areas of expertise include the preparation of hazard assessment reports, HSE Cases and probabilistic safety assessments and reliability analysis. Mr. Lidstone is equally comfortable as a teacher / lecturer of Safety Case materials or as a project manager for the development of Failure Modes and Effects Analyses (FMEAs) for a full range of drilling structures. Mr. Lidstone earned a B.Sc. (Hons) in Physics from the University of Salford, England and is a Chartered Physicist and member of the Institute of Physics.
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 Copyright 2001, Offshore Technology Conference This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, April 30–May 3, 2001 This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
• • • • •
Determination of condition of structural elements and structural system, Disclosure of defects (design, construction, operation, and maintenance), Assurance of conformance with plans, specifications, guidelines, rules, and quality requirements, Disclosure of damage, and Development of information to improve design, construction, operation, and maintenance procedures.
FPSO inspections have several levels of intensity:
ABSTRACT The increase in deepwater exploration activity has generated increased use of the Floating Production /Storage Offloading Systems (FPSO′s). Converting existing tankers is, in many cases, more economically feasible and faster than building new FPSO′s for the same purpose. For conversion of tankers to FPSO′s, inspection and subsequent fitness for purpose assessment are crucially important. The objective of this paper is to present the principles and strategies of in-service inspection programs for FPSO′s. The paper summarizes the technical basis for three levels of inspection strategies: 1) probability-based inspection method, 2) risk-based inspection method, and 3) ‘optimum’ inspection method.
INTRODUCTION FPSO′s has many attractive features including relative low cost, large working area, large water surface area, and good stability and floatability. FPSO′s in many cases, are converted from existing tankers. It is therefore important to have a rational and reliable inspection method to provide information and knowledge concerning the proposed, present, and future integrity of FPSO′s. FPSO inspections should be focused on:
• • •
General (global conditions), Specific (basic aspects of defects and damage), and Detailed (precise descriptions of flaws and other items of operation and maintenance concern).
FPSO inspections should be life-cycle oriented and include quality assurance and control measures in:
• • • • •
Design, Construction, Operation, Maintenance, and Accidents / casualties.
FPSO inspections should be full-scope and include quality assurance and control measures of the structure, equipment, facilities, and personnel. Research in inspection has been conducted in regulatory organizations, universities, and other leading organizations [1, 2, 3, 4 ]. Some important guidelines have been developed for offshore inspection practices. The objective of this paper is to present the principles of development of in-service inspection strategies for FPSO ′s.
PROBABILITY BASED INSPECTION A number of limitations of in-service inspections have been identified [3, 4, 5], especially the significant uncertainties in design, fabrication and damage detection, as well as the 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
deal with uncertainties, as well as Bayesian models has been recognized. The probability-based inspection method is thus developed to include Bayesian analysis, and probabilitybased inspection planning.
Bayesian Analysis The bulk of research on Bayesian analysis in engineering application was first conducted in early 1970’s. This approach, defined as the probability of detection (POD) updating method, was widely applied in aeronautical engineering systems, such as airframes, gas turbine engines [6]. In the late 1980s, an alternative Bayesian approach was developed as event updating using First Order Reliability Method (FOSM) in the offshore industry [7]. Itagaki et al [8] developed Bayesian estimation approach in ship structures based on Bayesian point estimation method.
Event Updating
POD Updating The POD updating is based on Bayesian approach to update of the multi-variant probability distributions, see Fig. 1. The updated multi-variant probability distribution is used to recalculate the failure probability. Assuming that the defect size distribution is F A(a) and that the probability of detection of a is P D(a). By means of Bayes' theorem (Bayes form):
P[p|q r H] • P[q r | H]
P[q r | pH] =
n
∑
P[q i | H] • P[p| q i H]
(5)
i= 1
The updated probability density can be determined as follows:
q r | H : a ≤ A ≤ a + da ,
Let
i.e.
P[q r | H] = f A (a )da , and p | q r H : inspection resulting in no crack detection:
This approach is to update the probability of events, such as fatigue failure directly. Bayes theorem is applied here to update the failure probability conditioned on additional events, such as inspection event. It is expressed as
P (q r | pH ) =
OTC 12949
P ( qr | H ) P ( p | H )
f A,up (a) =
[1 − PD (a)] • f A (a)
∞
∫
(1 − PD (a)) • f A (a)
(7)
0
The original safety event q r|H can be formulated by using the critical size (e.g., critical crack size for fatigue or critical thickness reduction for corrosion) as the failure criteria
−a ≤ 0
(6)
and the updated probability density is:
(1)
where, qr|H - The original safety event , q r|pH - The fatigue safety event after inspections, p|H - inspection event
q r | H = a c
P[ p | q r H] = 1 − PD (a )
(2)
The same methodology can be applied in the crack detection inspection where the updated probability density is :
f A,up (a) =
∞
f A (a) • PD (a)
∫
f A (a) • PD (a)da
(8)
0
Bayesian Estimation
The critical size is selected, perhaps based on serviceability consideration. Other safety event such as brittle fracture or corrosion threshold can be formulated.
This approach applies Bayesian approach in statistical parameter (mean, standard deviation) estimation, see Fig. 2.
The inspection event p|H is formulated to describe the additional information from inspections. Two types of event margins are classified based on inspection results. p | H = a ( N i ) − A d ≤ 0 (3)
The statistical parameter to describe the distribution of a random variable X in the fatigue problem is defined as the random variable µ . The Bayesian estimation of µ is to use the extended Bayes theorem in probability density as:
and
p | H = a ( N i ) − A = 0
(4)
Equation (3) is the event that a defect size a(N i) is not detected by an in-service inspection with the crack delectability limit A d at the inspection time N i. Ad is described by the probability of detection (POD). Equation (4) is the event that a particular defect size A is detected and measured at the inspection time N i.
f " (µ) = cL(µ)f ' (µ)
(9)
'
Where f (µ) = the prior distribution, L(µ ) = the likelihood function of the observed data (objective information), c= a "
normalizing factor, and f (µ ) = the posterior distribution incorporating the objective data. For a sample of observations x i , i=1,2,..., of X from inspections, the likelihood of the sample is proportional to the product of the probability densities of X at x 1, ...., x n :
OTC 12949
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
•
n ' f (µ ) ∝ ∏ f x (x i | µ) f (µ) i=1 "
(10)
The updated probability distribution of X is thus expressed as:
∫
f x (x) = f x|µ (x | µ )f (µ)dµ "
(11)
The updated probability of failure is computed using the updated probability density f x ( x ) .
With the development of Bayesian analysis, research on probability-based inspection planning was developed. Itagaki, et. al [8] and Førli [9] discussed probability based inspection planning for ship and offshore structures. In probability-based inspection planning, the failure probabilities are generally expressed in terms of intersections of the events of inspection and failure. Information gathered by inspection is accounted for by updating the probabilities using Bayesian analysis. Three approaches have been developed to extend the Bayesian analysis procedure to inspection planning: (1) target safety margins [4,8], (2) optimum life cycle cost [4,12], and (3) combination of (1) and (2) [4,10].
RISK BASED INSPECTION Probability based inspection can be used to establish the component/element in-service inspection schedules on the basis of reliability requirement of the individual critical components (elements). However, FPSO ′s usually contain a large number of components/elements. A rational inspection program should be developed based on system considerations. The development of a system-level, risk-based inspection process includes the prioritization of systems, subsystems, and components/elements using risk measures, and definition of an inspection strategy (i.e., the frequency, method, and scope/sample size) for performing the actual inspections. The process also includes decisions about the maintenance and repair strategies. Finally, there is a strategy for updating the inspection program for a given system, subsystem, or component/element, using the results of the inspections that have been performed.
The use of a ‘living’ process that is flexible, strives for completeness, and can be easily implemented and updated. The use of qualitative and quantitative risk measures The use of effective and efficient qualitative and quantitative methods that provide results familiar to inspection personnel.
Figure 3 illustrates the overall risk-based inspection process. The process is composed of the following steps:
• •
Probability-based Inspection Planning
•
• •
Definition of the system for inspection Use of a qualitative risk assessment that utilizes expert judgment and experience in identifying failure modes, causes, and consequences for initial ranking of systems and components/elements in inspection. Application of quantitative risk analysis methods, primarily using an enhanced Failure Modes, Effects, and Critically Analysis (FEMCA) and treating uncertainties, as necessary, to focus the inspection efforts on systems and components/elements associated with the highest calculated safety, economic, or environmental risk. Development of the inspection program for the components, using decision analysis to include costbenefit considerations. The inspection strategy are being updated and implemented, based on the findings and experience from the previous inspections.
Several feedback loops are shown in Figure 3 to represent a living process for the definition of the system, the ranking of components/elements, and the inspection strategy for each component/element.
System Definition A key step in defining a system for inspection, as shown in the first box of Figure 3, is the assembly of information that is needed for the risk-based inspection approach. In particular, the interviewing of key personnel, who are knowledgeable for degradation mechanisms or errors that may not be documented, is vital to the process.
Inspection Prioritization
be
The qualitative risk assessment, as included in the second box of Figure 3, utilizes expert judgment and experience in prioritizing systems, and components/elements for inspection. A key element for this assessment is to identify potential failure modes and causes, including design, operational, and maintenance errors and potential degradation mechanisms.
The use of a multidisciplinary, top-down approach that starts at the system level before focusing the inspection at the component/element level.
Figure 4 shows an example result of a qualitative risk assessment in which each box is representative of a given component, and a box is used to show the range of estimated consequence and failure probability. Once numbers are
Methodology The methodology summarized as:
•
• •
3
for
risk-based
inspection
may
4
T. XU, Y. BAI, M. WANG, R. BEA
placed on the axes, the risk assessment become quantitative, with uncertainty being represented by the size of boxes. The risk is defined as: Risk = (likelihood of failure)x(failure consequence)
•
(12)
Consequences of failure can be measured in a variety of ways, such as injuries/deaths, economic loss, environmental damages, dollars. In Figure 4, region A is high risk, region B is intermediate risk, and region C is low risk. Components/elements are grouped according to the region in which they fall. The FMECA (Failure Modes, Effects, and Criticality Analysis) in the third box of Figure 3 is an element of the subjective probabilistic ranking. It provides an efficient means of integrating the information required for a riskbased prioritization. Information on systems or components/elements is gathered from • design information, • operating experience (including prior inspection results), • structural reliability and risk analysis (SRRA) results, and • expert opinion to define failure modes, failure causes, and (perhaps) failure probability. In this way, the key information is integrated to provide the safety, economic, or environmental risk associated with the systems, subsystems, and elements to develop the inspection rankings for different systems/components. The probability-based inspection method developed based on probabilistic structural mechanics is perhaps the essential part of the FMECA since it provides a rational framework to estimate failure probabilities for components/elements.
Inspection Program Development Once the FMECA is completed, and the components/elements are ranked or categorized, the next step is to develop an inspection program for each group of components. This is the bottom box in Figure 3 that is schematically shown in Figure 5. It can also be used to establish an inspection program for an individual component/element or a system, as necessary. The recommended process is divided into three basic steps:
•
•
Choose potential inspection strategies that define the frequency, method, scope, and sampling procedure for inspection: The method of inspection includes the procedure, equipment, and level of personnel qualification to perform the inspection. The inspection strategy can also take advantage of monitoring systems and maintenance test program. Choose an inspection strategy and perform inspection: From the potential inspection strategies, defined in the above step, the effect of each of these strategies on the
OTC 12949
failure probability of the component/element is estimated. Choose appropriate action and update state of knowledge and information: Following the performance of the inspection, a critical decision is faced. That is, should the component/element be repaired or replaced if significant findings occur, or should nothing be done except to redefine the inspection program (going back to part 1 of the overall process shown in Figure 3)? Should the existing inspection, maintenance, repair system be changed? This depends on the fitness for purpose evaluation to determine the inspection findings and potential actions on the failure probabilities. In any case, all of the results related to inspection should be used to update the FMECA information on a periodic basis to re-rank the components/elements on the basis of risk and to redefine the inspection program.
RISK BASED ‘OPTIMUM’ INSPECTION Experience with in-service inspections of ship and offshore structures has adequately demonstrated that there are two distinct categories of defects and damage that are found:
• •
Those due to intrinsic causes - those that could have been or were anticipated (natural, predictable), and Those due to extrinsic causes - those that could not have been anticipated (human caused, unpredictable).
Experience with fatigue and corrosion damage found in ship and offshore structures clearly indicates that a substantial amount (if not a majority) of damage falls in the second category - unpredictable and due to the ‘erroneous’ actions and inactions of people. Quantitative inspection analyses (e.g. probability or risk based inspection methods and programs) can help address the first category of defects by providing insights into when, where, and how to inspect and repair. However, such an analysis cannot be relied upon to provide information that addresses the second category of defects. Expert observation and deduction (diagnostic) techniques must be used to address the second category of defects. Such recognizations lead to the development of the ‘optimum’ inspection method for FPSO ′s. The overall objective of the ‘optimum’ inspection method is to develop an effective and efficient safety and quality control system in the life cycle management of the FPSO ′s.
Inspection Performance Inspection performance is influenced by the vessel, the inspector, and the environment (Figure 6). The vessel factor can be divided into two categories: design factors and condition/maintenance factors. Design factors, including structural layout, size, and coating, are fixed at the initial design or through the redesign that may accompany
OTC 12949
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
repair. Condition/maintenance factors reflect the change in a vessel as it ages, including the operation history and characteristics of individual damages/defects (crack, corrosion, bucking), its size, and its location. The person (inspector) carries out an inspection can greatly influence the inspection performance. Performance varies not only from inspector to inspector, but also from inspection to inspection with the same inspector based on mental and physical condition. Factors associated with the inspector include experience, training, fatigue and motivation. The environment in which the inspection is carried out has a major influence on performance. The environment factors can be divided into two categories: external factors which cannot be modified by inspection procedures and procedure factors that can be modified. External factors include weather and location of the vessel, that is, whether the inspection is performed while underway, while in port, or while in dry-dock. Procedural factors reflect the condition during the inspection (lighting, cleanliness, temperature, ventilation), the way in which the inspection is conducted (access method, inspection method, crew support, time available) and the overall specification for inspection (inspection type).
Inspection Strategies Inspections, data recording, data archiving (storage), and data analysis should all be a part of a comprehensive and optimum inspection system. Records and thorough understanding of the information contained in these records are a key aspect of inspection programs. Inspection is one part of a ‘system’ that is intended to help disclose the presence of ‘anticipated’ and ‘unanticipated’ defects and damage. Development of inspection programs should address:
• • • • • • •
Elements to be inspected (where and how many?), Defects, degradation, and damaged to be detected (what?), Methods to be used to inspect, record, archive, and report results (how?), Timing and scheduling (when?), Organization, selection, training, verification, conflict resolution, and responsibilities (who?), and Objectives (why?). Where and How Many?
Definition of the elements to be inspected is based on two principal aspects:
• •
Consequences of defects and damage, and Likelihood of defects and damage.
The consequence evaluation essentially focuses on defining those elements, and components that have a major influence on the quality and safety of a FPSO. Evaluation of the potential consequences should be based on historical data
5
(experience) and analysis to define the elements that are critical to maintaining the integrity of a FPSO. The likelihood evaluation focuses on defining those elements that have high Likelihood’s of being damaged and defective. Experience and analyses are complementary means of identifying these elements.
What? A substantial amount (if not the majority) of the damage is unpredictable and due to the unanticipated ‘erroneous’ actions and inactions of people [13]. Current experience also indicates that the majority of damage that is associated with accidents (collisions, dropped objects) is discovered after the incident occurs [13]. About 60% of fatigue and corrosion damage is detected during routine inspections. However, the balance of 40% is discovered accidentally or during non-routine inspections.
How ? The methods to be used in FPSO inspections are basically visual. In one form or another, these methods are primarily focused on getting an inspector close enough to the surface to be inspected so that he can visually determine if there are significant defects or damages. However, ultrasonic gauging, magnetic particle, radiographic and other nondestructive methods sometimes are necessary for FPSO.
When? There are no general answers to the timing of inspections. The timing of inspections is dependent on:
• • • •
The initial and long-term durability characteristics of the FPSO structure; The margins that the operator wants in place over minimums so that there is sufficient time to plan and implement effective repairs; The quality of the inspections and repairs; and The basis for maintenance – ‘on demand’ (repair when it ‘breaks or leaks’ or ‘programmed’ (repair or replace on standard time basis).
Who? Experience has adequately demonstrated that the single most important part of the inspection inspection system is the inspector. inspector. The skills, knowledge, motivation, integrity of the inspector are critically important. Equally important are the organizational influences exerted on the inspector, the procedures and processes that he is required to follow, the environments in which he must work, and the support hardware and systems that are provided for him to perform his work. Thus, the inspector inspector is significantly significantly influenced influenced by
6
T. XU, Y. BAI, M. W AN G, R. BEA
1) organizations, 2) procedures, 3) hardware (facilities), and 4) environments.
•
Much has been learned about how to improve the effectiveness and efficiency of the inspector [12]. As one designs new inspection systems, it is important that the inspector be recognized as a part of this system [4].
• •
Why? Inspection should have objectives of several levels: first, it provides the general information and knowledge for the quality of the in-service structures for fitness for purpose evaluation (general condition), second, it is to detect the damage/defects as many as possible so that effective and efficient maintenance and repair program can be implemented to correct these damages/defects (quality control and assurance), third, it is a safety control tool to prevent the failure or loss of the in-service structures during the inspection interval (safety control and assurance). The inspection strategies (when, where, how, who) for different level objectives should be different. The first level inspection should select typical elements/components to provide general information about the in-service structures for fitness for purpose evaluation. It is more frequent less detail inspection associated with long-term maintenance and repair program. The second level (quality control) inspection should focus on the critical components/elements to detect damage/defects as many as possible. It is associated with the short-term maintenance and repair program. The third level inspection (safety control) is to prevent the most critical damage/defects or errors to ensure the safe operation during the inspection interval. It is the most detailed and difficult inspection to identify the safetyrelated predictable or unpredictable damages/defects and errors. Every inspection practice for a specific fleet should be a combination of these three different inspection strategies.
OTC 12949
To confirm what is thought: to address the intrinsic damages/defects that can be predicted based results from technical analyses, To disclose what is not known before inspection; to address damage/defects that can not be predicted based on technical analyses, and To control the predictable and unpredictable damages; to develop high quality maintenance and repair program.
The ‘optimum’ inspection program should be started with the design of the structure (conception), proceed through the life of the structure, and conclude with its scrapping (lifecycle). The optimum inspection program should include not only the hull structure, but as well, its equipment and its personnel (full scope). The optimum inspections should become the means to assess the general conditions of the whole structure. The optimum inspections are also the means to detect unpredictable flaws and damages of the structural elements, and permit appropriate measures to be taken to preserve the safety and integrity of the structure. The optimum inspections are also the means to assure that all is going as expected, that the structural elements are performing as expected, and that corrosion protection and mitigation (e.g. patching pits, renewing locally excessively corroded plate) is maintained. The ‘optimum’ inspection method starts from the survey for the intrinsic damage that is common for the class of structures. Based on the experience, the inspection for the intrinsic damage can be conducted in the rational way. The existing risk-based inspection method discussed early this paper is the framework for the intrinsic damages/defects for the structural system. The probability-based inspection method can be applied for the specific elements/components based on the results of risk-based inspection. For the extrinsic damage for each individual structure, the knowledge-based diagnosis method should be developed. The step-by-step knowledge-based diagnosis process is the potential means to identify the extrinsic damages.
The value of the inspection for objectives of different levels should also be different. different. The value of the first level inspection is about the decision on whether or not the existing structure can fulfill the purpose for extended service. The value of the second level inspection is about the decision whether or not we should change the maintenance and repair program. The value of the third level inspection is about the decision whether or not we should take any intermediate actions. Value analysis (value of information) can help make these decisions.
Knowledge systems now routinely do diagnosis reasoning using three methods: model-based diagnosis, heuristic classification, classifica tion, and case-based reasoning. Our system uses a combination of each of these methods: Model-Based Diagnosis (MBD) to identify the details of a large class of possible problems, heuristic classification to identify the presence a set of idiosyncratic problems, and Case-Based Reasoning (CBR) to compare observation with previously identified cases.
‘Optimum’ Inspection Method
•
The ‘optimum’ inspection method can be proactive (focused on prevention) or it can be reactive (focused on correction). It should have four functions:
•
•
Assess the general conditions of the in-service offshore structures ,
An ‘optimum’ inspection method could include: Developing standard task checklists to ensure that relevant data and tasks are not lost because of distractions or workload, Performing global surveys to develop situation awareness for potential expected and unexpected damage and defects,
OTC 12949
•
• • • • • •
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
Inspecting high likelihood of damage or defect ‘parts’ and high consequence parts; if something ‘suspicious’ is found, the inspection is intensified by model-based diagnosis, heuristic classification, and case-based reasoning until root causes (not symptoms) are determined, Inspecting periodically decreasing the time between inspections as the rate of degradation or likelihood of defects and damage increase, Inspecting after accidents or ‘early warning’ signals are sensed, Implement the long-term and short-term maintenance and repair strategies based on the inspection results, Update the IMMR (Inspection, Maintenance, Monitoring and Repair) plan based on the survey results and the results from maintenance and repair, Performing inspections that are independent from the circumstances that cause potential defects and damage, and Using qualified and experienced inspectors that have sufficient resources and incentives to perform quality inspections.
For each FPSO, standard checklist and procedures should be established from the FPSO Structural Life-Cycle Information Management System, in order to carry out an effective evaluation of the general condition, prior to the commencement of any general survey and include:
• • • • • • • • •
Structural drawing, Operating history and conditions, Previous damage/defects inspection results, Condition and extent of protective coatings, Classification status, including any outstanding conditions of class, Previous repair and maintenance work, Previous information on unpredictable damage or defects, Expert’s judgment and comments, Relevant information from its sister structures.
With this information and previous inspection guidelines regarding critical elements/subsystems in the FPSO structural systems considered to be sites of potential damage/defects based on historical data, analyses results, and expert’s judgment, it is possible to target the appropriate inspection strategies for the potential areas within the structure for general survey and the initial scope of the inspection. After the initial inspection to determine the general condition of the system, the inspector can develop situation awareness to identify some potential unpredictable critical damage/defect sites. Further knowledge-based diagnosis should be conducted for these suspicious areas. The knowledge-based diagnosis is conducted together with detailed or specific inspections.
7
Inspection Data System The single weakest component that has been found in present inspection systems for FPSO ′s regards the data and information that is developed during and from inspections. Little thought has been given to the efficient gathering of data and information, even less thought to what is done with this data and information when it is obtained, and far less thought given to the archiving, analysis, and reporting of the data. The interfaces in the data gathering, archiving, analysis, and reporting activities also have received a little systematic thought. Current work has not been able to identify a single coherent and optimum inspection data system for FPSO. Advances in information technology have resulted in better ways to use information for the management of safe and efficient ship and offshore structures. The integration of stand-alone systems combined with improved information recording, organization and communication offers substantial benefits for the life-cycle management of ship and offshore offshore structures. A life cycle Structural Information Management System (SMIS) is intended to facilitate the lifecycle management of FPSO. This includes areas from design and construction as well as operations including Inspection, Maintenance, Monitoring and Repair (IMMR). The inspection data system is a component of the IMMR module in SMIS. The general objectives of an inspection data system development development are:
• • • • • • •
Collect inspection data, Store the data, Provide means for logic inspection data management, Allow for the organization of the inspection data in a form suitable for fitness or purpose analyses, and failure analyses, Analyze the data, Show trends of the information such as damage/defects associated with structural integrity, Communicate and report the data.
Once a FPSO is ready for service, a series of inspections are scheduled according to inspection program. The objective objective and scope of of the internal tank inspections are defined. The access methods and data recording methods are chosen, and then the inspections are performed. The inspection results including corrosion gauging, cracking, status of coating and corrosion protection systems and other structure/equipment defects are updated into the corresponding database. Using the inspectional data, maintenance and repair strategies can be developed and the repair are finally carried out.
8
T. XU, Y. BAI, M. WANG, R. BEA
OTC 12949
REENGINEERING THE IMMR PROCESS
ACKNOWLEDGEMENT
FPSO in-service inspections and repairs are components in an Inspection, Maintenance, Monitoring and Repair (IMR) system that is intended to help disclose the presence of ‘anticipated’ and ‘unanticipated’ defects and damage to critical structural details [4,12].
The technical views expressed in this paper are those of the authors and are not necessarily those of the institutions they are affiliated with.
An IMMR system is a critical part of the maintenance of inservice quality of a FPSO. The IMMR process should be in place, working, and being further developed during the entire lifetime of the structure. The IMMR process is responsible for maintaining the quality of the structure during the useful lifetime of the structure. A fundamental and essential part of the IMMR process is knowledge. The IMMR process can be no more effective or efficient than the knowledge, data, and experience that form the basis for the process.
REFERENCES 1.
2.
3.
4.
Xu et al [4] indicates that organization of the FPSO inspection should be developed to:
• • • • •
Define inspection processes in which only that information that is absolutely necessary to assure acceptable quality in the FPSO is gathered, Determine how to minimize the people and man-hours required in the entire IMMR process, Define how to minimize the steps, interfaces, multiple processing, and paper required in the IMMR process, Determine how to minimize the checks, controls, reporting, and reconciliation required in the IMMR process, and Define how take full advantage of present computing, communications, and information technologies (CCIT).
5.
6.
7.
8.
CONCLUSIONS This paper addresses the development of ‘optimum’ inspection strategies for FPSO ′s. It details with the technical basis for 1) probability-based inspection method, 2) riskbased inspection method, and 3) ‘optimum ’ inspection method. The probability based inspection methods are recognized to be able to address only part of the potentials for damage in FPSO ′s. The risk-based inspection addresses the component/element’s damages/defects from the system risk point of view. The ‘optimum’ inspection method is a full-scope, life-cycle development of inspection system for the risk management. The ‘optimum’ inspection method not only includes use of expert observation and analysis (deductive) methods but also the use of structural information system. It is demonstrated that application of this comprehensive and integrated approach will result in better allocation of the resources used to develop effective inspection strategies.
9.
10.
11. 12.
13.
14.
15.
Tanker Structural Co-operative Forum (1986) Guidance Manuals for the Inspection and Condition Assessment of Tanker Structures. Tanker Structure Co-operate Forum, (1990) Inspection, Assessment and Experience of Old Tankers, Witherby Marine Publication. Marine Technology Directorate, (1989) "Underwater Inspection of Steel Offshore Installations: Implementation of a New Approach", Report 89/104. Xu, T., Bea, R. G., (1996) " Inspection of Marine Structures", Report for Joint Industry Research, Marine Technology and Management Group, Dept. of Civil Engineering, University of California at Berkeley, Berkeley, CA 94720. Demsetz, L.A., Cario, R., and Schulte-Strathaus, R., (1995) Inspection of Marine Structures, Report for Maritime Administration under Project Number DTMA91-93-G00040. Yang, J.N. and Chen, S., (1985) "Fatigue Reliability of Structural Components Under Scheduled Inspection and Repair Maintenance," Probabilistic Methods in Mechanics of Solids and Structures, edited by S. Eggwertz and N.C.Lind, Springer-Verlag, Berlin, pp. 103-110. Moan, T., (1993) "Reliability and Risk Analysis for Design and Operations Planning of Offshore Structures", Proc of the 6st Intl Conf on Struct Safety and Reliability, ICOSSAR'93. Itagaki, H., Akita, Y. and Nitta, A., (1983) "Application of Subjective Reliability Analysis to the Evaluation of Inspection Procedures on Ship Structures", Proceedings of the International Symposium on the Role of Design, Inspection and Redundancy on Marine Structural Reliability, National Academy Press, Nov. 13-16. Førli, O., (1990) "The Reliability and Cost-Effectiveness of Offshore Inspections", Proc. Intl Conf on Monitoring, Surveillance and Predictive Maintenance of Plants and Structures, Sicily, Italy. Shinozuka, M., (1990) "Relation of Inspection Findings to Fatigue Reliability", Ship Structural Committee Report, SSC-355 Marine Technology Directorate, (1994) "Review of Repairs to Offshore Structures and Pipelines, Report 94/102. Bea, R.G, (1993) Ship Structural Maintenance: Recent Research Results and Experience, Transactions, The Institution of Marine Engineers, London Jones, R. B. (1995). "Risk-Based Management - A Reliability-Centered Approach", Gulf Publishing Co., Houston, Texas. Platten, J. L., (1984) “Periodic (in-service) Inspection of Nuclear Station Piping Welds, for the Minimum overall th Radiation Risk”, Proceedings of the 5 International Meeting on Thermal Nuclear Reactor Safety, Vol. 1. 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
OTC 12949
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
9
Committee on Human Performance, organizational Systems, and Maritime Safety, National Academy of Engineering, National Research Council, Washington, D. C.
f A(a) Probability of Failure ac
f a(a)
f a(a) Experiment/Inspection
Experiment/Inspection Posterior
I d n i i t s i t a r i l b c u r t i a o c k n
Prior
Original distribution Mean Crack Size
N Inspection
Updated Number of Cycles distribution
Figure 1 Illustration of POD Updating
a
a
Figure 2 Illustration of Bayesian Estimation
10
T. XU, Y. BAI, M. WANG, R. BEA
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
OTC 12949
OTC 12949
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
11
CAUSE OF DAMAGE Operating error Design error C
o
r
r
o
Installation fault
A
Fabrication fault Dropped Object
B
C
o F
C
l a
0
0
l t
0
0
i i
0
.
1
.
FREQUENCY 0F PLATFORM DAMAGE (% / year) Consequences
Figure 5 Causes of Damage to North Sea Structures
Figure 4 Definition of Risk Index
1. Select Potential Inspection Strategies * Define Potential for Damage States * Define Potential for Inspection Damage * Define Reliability of Inspection Methods
Update
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 Appropriate IMR System
4. Later
(Sensitive Studies)
Now
(2) Implement IMR System
Update State Knowledge and Information
Figure 6 Development of In-Service Inspection Program
1
.
12
T. XU, Y. BAI, M. WANG, R. BEA
VESSEL
OTC 12949
ENVIRONMENT
Design
Inspection Performance
External - Weather - Location of Vessel
- Structural Layout - Size
Procedure
-Coatings
- Lighting - Cleanliness
Condition/Maintenance Cargo Defects
INSPECTOR Experience Fatigue Training Motivation
- Temperatures/Humidity - Ventilation - Access Method - Inspection Method - 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 participating 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 Copyright 2001, Offshore Technology Conference This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, 30 April–3 May 2001. This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
Abstract This paper describes the reliability methods for position mooring design and analysis with various degrees of complexity and with a particular emphasis on deepwater applications. The limit state design and safety factor formats for position mooring systems of various floating platforms are first addressed. This paper addresses technical issues of both passive and active mooring systems including winch or thruster assisted and dynamic positioning systems. The modeling uncertainties of numerical analysis, mooring components, environmental conditions and reliability methods are addressed. This paper presents a logical approach to reliability analysis for the calibration of partial safety factors for both passive and active mooring systems. Introduction As the development of offshore oil and gas is moving into deeper waters, floating platforms are used for drilling and/or production operations. For these operations, position mooring systems are required in order to keep the floating platforms on station under the design environmental criteria of wind, current and waves. Various types of floating platforms and mooring systems can be considered for offshore applications. For mooring strength limit states, partial safety factor formats have been proposed recently in the offshore industry. It is important, however, to fully understand how to use the reliability methods to calibrate the partial safety factors selected. The subject matter is significant to the offshore industry for enhancing the reliability of position mooring systems with cost-effective designs. This is especially important for deepwater applications because the consequences of position mooring system failure would be much more costly than in relatively shallow waters.
Limit States of Mooring Design Limit state design methods are first described with particular reference to the design of position mooring systems for floating installations. The limit state design philosophy may be used to provide a rational framework for the design of safe and serviceable structures or structural components, by accounting for uncertainties and variabilities in the basic 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 state formulation, including the deterministic and reliability analysis methods used in evaluating the probability that the limit state will be exceeded, to ensure that existing accepted safety levels are achieved. The performance of a structural system, or a component of the structure, may be described by a set of limit states (limiting conditions) beyond which the structure, or component, is no longer deemed to satisfy the design requirements. Limit states can be regarded as a discrete representation of a more general continuous loss function. In general, the failure of a structural system or component to satisfy the design requirements may be represented by an inequality of the form: (1) g X 1 , K , X n , C ≤ 0
(
)
Where, g is the limit state function, X 1 ,…, X n are the basic variables associated with the limit state, and C are constraints describing the limits of acceptable accelerations, offsets, clearances, etc. The acceptable performance of position mooring systems depends on system responses to environmental actions and the system’s strength, in intact and damaged conditions. Additionally, moored installations may depend on the availability of active systems, such as thrusters and winches, to control or reduce critical system responses. In general, the evaluation of mooring system limit states is more complicated than the evaluation of component limit states. In the analysis of position mooring systems, computer programs, databases, model tests, statistical models, empirical rules and assumptions are used in evaluating the complex nonlinear coupled behavior of the vessel, mooring lines, and risers. All of this is symbolized by the limit state function g in eqn. (1). For example, linear potential theory may be used to synthesize approximate first and second order vessel forces
2
KEN HUANG AND YONG BAI
and motions in irregular seas. Wind and current force coefficients may be derived from databases for generic vessel types or from wind tunnel tests on scale models. Characteristic and extreme values of response parameters may be calculated from statistical models and assumed distributions and calculated output parameters from one analysis program are often used to define input variables or coefficients in a second program. The basic variables that determine the behavior and responses of moored systems fall into one of the following categories. (1) Action or load variables - associated with intensity and spatial variations of actions or loads, such as; wind and wave intensities, spectra, and directions, current profile, velocity, and direction, mooring line pretensions, vessel draft/ballast, etc. (2) Component and material variables - associated with component and material properties, such as; mooring component break strength, fatigue endurance, density, elasticity, soil strength, etc. (3) Geometric variables - associated with the structural system or component geometry, such as; vessel shapes/dimensions, mooring line diameter, line segment length, line angles, water depth, etc. (4) State variables - associated with combined properties of the system or component and the environment, such as; wind, wave drift, and current force and moment coefficients, response amplitude operators (RAOs), wave frequency natural periods and damping, mooring line and vessel drag and added mass coefficients, low frequency damping, etc. (5) Active system variables - associated with the performance and mean time to failure and repair of the mechanical and electrical, distribution, monitoring, and control systems and the experience and training (human factors) of the personnel operating the active systems. (6) Modeling uncertainty variables - associated with the physical (hydrodynamic and structural theories) and statistical models and the reliability methods, which are used in calculating system responses, component and system strengths, endurance, etc. and the notional reliability levels of components and systems. For position mooring systems, strength limit states are usually associated with the maximum installation, survival, operational, disconnect, and reconnect criteria in their most extreme design environments. Depending upon the inspection and maintenance program, strength may also depend on damage accumulation; that is on the fatigue, corrosion, and wear endurance limit states, generally in the long-term environment. While, operational performance is usually related to offset and motion limit states in the maximum operational environment. These limit states are normally subdivided further. For example, in the maximum design environment individual limit states will address extreme line loads, total turret loads, anchor loads and line angles, vessel offsets, motions, and clearances etc. for intact, damaged, and transient cases. Design requirements, criteria, and the safety factors
OTC 13269
recommended in present codes developed and evolved to serve the needs of mobile offshore drilling units (MODUs). In particular, the methods of analysis and design criteria contained in API codes were originally developed based mainly on the operational experience of chain and wire catenary spread moored semisubmersible drilling units, [3 – 7]. Over the past twenty years these methods of designing mooring systems for MODUs have been used worldwide and have proved to provide adequate safety levels for conventionally moored semisubmersibles, [8]. More recently, spread moored semisubmersibles, spars, and turret moored ships, with or without thruster assistance, using taut or catenary mooring systems, and synthetic mooring line components as well as chain and wire, have been used as permanent production installations. API and ISO have developed codes that also address the design requirements of these types of installations, [9 – 12]. However, in these codes the combined effects of variability in environments and system responses and strengths, for different types of vessel, mooring systems, and environments, have not been fully accounted for. Consequently, safety levels inherent in today’s codes are largely arbitrary and cannot ensure either an optimum level of safety or economy in construction. Due to the historical experience on which code development has been based, the analysis methods and safety factors recommended by the present API and ISO codes, [10, 11], are most applicable to conventionally moored semisubmersibles. Safety Factor Formats When limit states are evaluated using deterministic methods, the basic variables ( X 1 ,…, X n) are replaced by their characteristic design values ( y1 ,…, yn), and a set of safety factors (γ 1 ,…, γ m) or safety elements are introduced which are chosen to ensure a minimum level of safety, [2]. This allows the limit state checking equation used in deterministic design methods to be represented by: (2) g ( y 1 , K , y n , C , γ 1 , K , γ m ) ≤ 0
For example in evaluating the mooring system’s strength limit state, the characteristic values of wind, wave, and current are usually chosen as the most probable maximum values associated with the design return period, and conservative ranges (safety elements) of peak periods and relative wind, wave, and current directions are considered. While the mooring line break strength is represented by the catalog break strength of the component, including a reduction in diameter (safety element) to allow for corrosion losses. In this traditional design method some safety elements are preapplied (spectral peak period and wind, wave, and current directions) while intact, damaged, and transient line tension safety factors are post-applied . API RP 2SK and the ISO station keeping standard, [10, 11], both use a total safety factor that is applied to the mooring system responses, post-applied , and contain some references to pre-applied safety elements. Generally, line strength postapplied safety factor formats for intact, damaged, and transient cases may be expressed as:
OTC 13269 BS
γ m
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
≥ γ µ T µ ( y1 , K , y n ) + γ LF T LF (y1 ,K
, yn
) + γ
T
WF WF
(y ,K , y ) 1
n
(3)
or
(
(
BS ≥ γ m γ µT µ y1 , K , yn
) + γ
T
LF LF
(y ,K , y ) + γ 1
n
(
T WF y1 ,K , yn
WF
))
Where, BS = catalog break strength of mooring component T µ = mean tension, nonlinearly dependent on ( y1 ,…, yn) T LF = low frequency tension, nonlinearly dependent on ( y1 ,…, yn) T WF = wave frequency tension, nonlinearly dependent on ( y1 ,…, yn) ( y1 ,…, yn) = characteristic design values of the basic variables on which line tensions depend γ m = mooring component partial safety factor = mean tension partial safety factor γ µ = low frequency tension partial safety factor γ LF = wave frequency tension partial safety factor γ WF Design methods that are based on a safety factor format of this type, eqn. (3), are known as load and resistance factored design (LRFD) methods. In the case of the API and ISO mooring codes only a single or total safety factor is specified. Where there is uncertainty or variability in the basic (input) variables, post-applied safety factor formats of the LRFD type will produce inconsistent safety levels for mooring systems that have different nonlinear response characteristics. Additionally, inconsistent levels of safety may also result where the variability in environmental conditions differs from site to site. For example, the coefficient of variation associated with 100-year return period wind speeds for Gulf of Mexico hurricanes and North Sea winter storms are approximately 13% and 5% respectively, and wind loads vary as the square of the wind speed. Consequently, mooring systems designed to the same post-applied (LRFD) safety factors in the Gulf of Mexico and the North Sea will be associated with different notional reliability or safety levels, this is true even for systems which display identical, linear or nonlinear, response characteristics. It may be possible to achieve more consistent designs, in terms of safety levels, by the use of more complicated safety factor formats. One possibility is to use a post-applied safety factor format, eqn. (3), with the values of the partial safety factors defined as functions of vessel/positioning system type and environment. Alternatively, a pre-applied partial safety factor format, eqn. (2), could be developed, where the values of the partial factors are related to the uncertainty (coefficient of variation) in the basic variables. For example, the characteristic design value of environmental variables could be defined as the value associated with the mean plus nstandard deviations, instead of the most probable (modal) value that is currently used. At present the industry has no experience with mooring codes that use pre-applied safety factor formats, and such design codes would represent a major departure from the design methods that have evolved with the offshore industry. Additionally, it has yet to be demonstrated that a reasonably
3
simple and practical pre-applied safety factor format that will reduce the variability in safety levels across the range of types of positioning systems and environments can be devised. Limit State Design Limit states define the limiting conditions beyond which the structural system or component is no longer deemed to satisfy the design requirements, consequently, there is a correspondence between limit states and traditional design criteria. By calibrating individual limit states, the limit state design method may be used to achieve more consistent position mooring designs than is presently possible by the use of the traditional (un-calibrated) design methods specified by today’s codes. In some cases it is not practical, or possible, to express design criteria in terms of a limit state function that allows numerical evaluation. For example, protection against corrosion and wear of mooring line components is normally provided for by increasing the component diameter, jacketing the line, using blocking compounds, providing galvanic protection etc. and specifying inspection and maintenance programs. Not all operational limit states depend upon the position mooring system, for example processing operations may be limited by wave frequency vessel motions and accelerations which in many cases are unaffected by the mooring system. In general, position mooring systems are redundant seriessystems, in which system responses, including line-loads, are nonlinearly dependent on the environmental (wind, wave, and current) actions. Redundancy is provided for in the number of mooring lines (load paths), and recognized in mooring codes by specifying intact and broken line factors of safety. Single anchor leg moorings (SALMs), and some portions of the load paths in single point moorings (SPMs) and catenary anchor leg moorings (CALMs) may not provide redundancy. These systems may be susceptible to single point failure and details of their design require special consideration. For example, in the design of SALMs API RP 2SK recommends that the level of structural redundancy be recognized in selecting an appropriate safety factor. Mooring codes, [10, 11], recognize redundancy in position mooring systems by specifying design requirements and safety factors for intact, damaged, and transient cases. However, these codes do not allow for differences in nonlinear system response characteristics, which depend upon the type of vessel, type of mooring system, and water depth. Or for differences in mooring line (series-system) strength, which depends on the type of mooring component (chain, wire, or synthetic rope) and the length of the line segment. Nor do these codes allow for differences in the variability of the environmental actions, which depend upon geographic region and season. In addition, position mooring systems may be entirely passive systems, or they may depend on active intervention. Active station keeping systems make use of thruster assistance or adjustment of mooring line lengths to reduce critical system responses. While some turret designs require power generation and functioning hydraulic equipment
4
KEN HUANG AND YONG BAI
to release brakes that hold the turret in its normally locked condition. Thrusters may be used to assist the mooring system by controlling the vessel’s heading, reducing the mean environmental forces, damping low frequency motions, or a combination of these functions. In high current environments adjustment of mooring lines may be used to position the vessel up current. Mooring lines may be adjusted in preparation for storm conditions. If the position mooring system depends upon the use of thrusters or line length adjustments to satisfy the design requirements, then the capability of the thruster system or the ability to perform winching operations, in the associated environmental conditions, must be assessed. In evaluating the capability of active systems, it is necessary to consider the equipment that support and control the thrusters or winches, their modes of failure, repair times, and the training and experience of the personnel operating the systems. Even for the most advanced mechanical and electrical systems there is a risk, albeit small, of a total loss of all electrical power (blackout). Consequently, wherever possible active systems should be designed to “fail-safe” and the consequences of a power blackout should be known for all high-risk operations. For passive mooring systems, structural reliability methods use models furnished by physics and probability theory to calculate the notional probability that a limit state will be exceeded. Evaluation of notional probabilities of failure for active position mooring limit states requires the evaluation of the probability, availability, that the active systems (mechanical and electrical, distribution, monitoring, control, operators, etc.) perform on demand in the relevant environmental conditions, in addition to the evaluation of the mooring system’s passive structural reliability. That is, the evaluation of notional safety levels for active position mooring systems requires both availability and reliability analyses of the active and passive systems on which station keeping depends. If limit state design methods are to be used to arrive at optimum designs for position mooring systems, through reliability based calibration, then the calibration of the limit states must allow for differences that exist in: • System redundancy, which depends on type of mooring system • System response characteristics, which depend on the type of vessel, type of mooring system and components, and water depth • Mooring line segment strength, which depends on mooring component material and construction and length of the line segment • Environmental variability, which depends on the geographic region and season • Passive or active system reliability, which depends on structural reliability, mechanical and electrical system availability, operator training, and human factors
OTC 13269
Reliability Analysis of Position-Mooring Systems Guidance and suggestions are discussed next to assist in the application of reliability methods to the design of position mooring systems. Methods of probabilistic structural analysis, reliability analysis, provide a rational means of dealing with the uncertainty and unpredictability in the basic variables on which the design of mooring systems depend, by means of the application of probability theory, statistics, and deterministic structural theory. Reliability analysis methods are concerned with quantifying the ‘measured chance’ that a structure will support the loads to which it is subjected, [1]. In particular, reliability methods provide approximate means for evaluating the notional probability of failure of limit states under defined actions (loads). In principle, the notional probability that the structure as whole will satisfy the design requirements over its design life may be synthesized from individual limit state notional probabilities of failure. Some examples of position mooring systems and environmental distributions are illustrated on Figure 1 and a general discussion encompassing other vessel types and station keeping systems is contained in API RP 2SK, [10]. The vessel and mooring system are subject to wind, wave, and current actions and mooring system responses are complex, coupled, and nonlinear. Mean, low, and wave frequency vessel offsets, motions, and line tensions are system responses that are nonlinearly dependent on the environmental actions and interactions (coupling) among the vessel, mooring system and environments. The degree and type of system nonlinearity is a function of vessel type, mooring system type and water depth, and the wind, wave, and current intensities and directions. These system responses enter into the limit state functions that define acceptable performance of the position mooring system, thus, the design of position mooring systems is concerned with the global analysis of nonlinear systems. Additionally, the variability in environmental actions depends upon the geographic region and season, which determine the type of storm that the system is exposed to. Absolute values of component or system probabilities of failure can only be calculated if the deterministic mooring analysis method, and the data and statistical models used to represent the vessel, mooring system, and the environment are completely known and the reliability analysis method is perfect. However, reliability methods provide a powerful tool for comparing the safety implications of different designs. Target safety levels for new designs, may be established by performing reliability calculations for old designs that have a history of satisfactory performance, in this way absolute values of failure probabilities are not required. For these reasons, probabilities of failure calculated using reliability methods are referred to as notional probabilities of failure, and target notional failure probabilities for limit states of new designs are established by calibrating to old designs for which satisfactory past experience exists. The results of reliability computations will depend upon: (1) The Deterministic Mooring Analysis – The hydrodynamic and engineering/analytical models and
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
their ability to predict the nonlinear coupled behavior of different system types. (2) The Basic Variables – The variables treated as random variables in the reliability analysis and their statistical descriptions, including modeling uncertainties, mooring component and line strength models, and the modeling of environmental conditions. (3) The Reliability Method – The methodology and assumptions used to simplify the reliability analysis and synthesize system notional probabilities of failure. If the resulting notional probabilities of failure, and comparisons between notional probabilities of failure across different types of moored systems, are to be meaningful then these aspects of the reliability analysis methodology and their influence on the final results must be understood and accounted for. Deterministic Analysis of Position Mooring Systems The results of reliability computations depend directly upon the deterministic analysis used to model the physical behavior of the position mooring system. For each position mooring system, a large number of deterministic mooring analyses (typically of the order of 103 to 105) are required to calculate intact and damaged notional failure probabilities using Level 2 or Monte Carlo reliability methods. As a consequence reliability analyses are usually based on frequency domain methods of analyzing the mooring system. Frequency domain analysis methods decompose the total response of the mooring system into mean, low frequency, and wave frequency responses. Statistical values of the low and wave frequency responses are then calculated and combined with the mean response to yield the maximum response in a specified storm duration. There are a number of ways in which system responses may be decomposed into mean, low frequency, and wave frequency responses and by which the statistical values of low and wave frequency responses may be calculated. For example, the mean equilibrium position of a spar may include the effects of pitch and roll on the wind, wave, and current force coefficients and allow for vessel setdown in calculating mean line tensions. The low frequency modal decomposition may use surge, sway, and yaw modes for all position mooring systems, or may use different modal decompositions for turret moored ships, spars, and spread moored semisubmersibles. The definition of characteristic low and wave frequency tension responses may be calculated based on a single vessel position, or may make use of different vessel positions for each mooring line. In this section details of the analysis methodology that have a bearing on the notional probabilities of failure calculated in the reliability analysis of position mooring systems are discussed. M ean response
The vessel’s mean offset position is calculated by requiring that the mean environmental loads and the net mooring system restoring forces and moments are in equilibrium. The mean environmental load is composed of the
5
mean wave drift and drag forces and moments, and the mean wind and current forces and moments. Most mooring analysis programs consider force and moment equilibrium for the surge, sway, and yaw degrees of freedom. In general, mean wave, wind, and current force and moment coefficients depend on the vessel’s mean pitch, roll, and yaw angles with respect to the environmental actions. For turret moored ships, the mean offset position depends on the mean yaw equilibrium heading and its stability, which must be established before the mean surge and sway forces are calculated. While for spars, mean environmental loads are also sensitive to the mean pitch and roll angles, requiring an iterative solution. Due to the small water plane area of spars, vessel set-down, heave, may also significantly effect mean system responses. For all vessel types, the mean current velocity and direction affect wave drift and drag forces and moments, and the relative importance of the wave forces and moments depend upon the type of vessel and environment, through the environmental force coefficients and relative magnitude of the environmental actions. Low and wave frequency system responses are dependent on the mean equilibrium position, consequently any error in the mean vessel position will also result in errors in the prediction of low and wave frequency responses. The capability of the mooring analysis method embodied in the computer program(s) used to calculate mean loads and system responses must be assessed for the types of vessel, mooring systems, water depths, and environments for which reliability analyses are performed. L ow f r equency modal r esponses
Ideally, low frequency responses of the vessel and mooring system should be decomposed into the system’s normal modes (eigen-modes), or an approximation of the normal modes. Generally, the normal modes will depend on the type of vessel and mooring system. For spread moored semisubmersibles, the surge, sway, and yaw modes provide a good approximation to the normal modes. Whereas for turret moored ships, sway and yaw responses are coupled, while for spread moored spars, the surge and pitch responses and sway and roll responses are coupled. The methods of decomposing low frequency responses recommended by API RP 2SK and the ISO code, [10,11], are most applicable to spread moored semisubmersibles. Low frequency motions about the vessel’s mean offset position are excited by the slowly varying components of the wind, wave reflection, wave drag, and current forces and moments. The calculation of the energy dissipated by the vessel, mooring lines, and risers, during the low frequency oscillations of the vessel (the low frequency damping) is one of the most difficult problems in the analysis of moored systems. In general, low frequency system damping consists of the following components: • Still water damping • Damping due to wind • Damping due to current • Damping due to wave reflection • Damping due to wave drag
6
KEN HUANG AND YONG BAI
•
Damping due to mooring line and riser motions
Low frequency damping is nonlinear and increases with the severity of the environment and the amplitude of the low and wave frequency motions of the vessel, mooring lines, and risers. Consequently, low frequency damping can only be calculated as part of a coupled analysis in which low frequency damping is solved for iteratively in time domain. From the system properties at the mean offset position, the low frequency force and moment spectra, and the low frequency modal damping, the statistical values of low frequency modal responses are calculated. API RP 2SK and the ISO code present formulae for a linear frequency domain method of calculating significant and maximum low frequency modal responses. However, linearization of the combined mean environmental and mooring system forces and moments with respect to offset, can introduce large errors in the prediction of significant and extreme low frequency offsets and line tensions. For a simple case, differences in the low frequency surge response predicted by linear and nonlinear methods are illustrated on Figure 2. The upper diagram on Figure 2 shows the 3x3 grouped mooring pattern with the vessel at the mean equilibrium position and defines the x-y coordinate system. The net mooring system restoring force, F x, increases more rapidly (stiffer) for negative offsets than for positive offsets, this is shown on the graph on Figure 2. Both equally spaced and grouped mooring patterns display non-linear force versus offset characteristics. Figure 2 also shows the mooring system potential energy, calculated by integrating the nonlinear mooring system restoring force, F x, with respect to offset, x, and the quadratic approximation to the potential energy calculated from the linearized system stiffness, K x. For both linear and nonlinear methods, the mean energy and the associated standard deviation, σ x, of the surge motion (rms response) is calculated from the low frequency surge force spectrum and the linearized system properties and damping at the mean offset position. For a given storm duration the maximum energy in the surge mode, E max, is related to the mean energy, E mean, by:
( )
E max = E mean2ln N x
where 1 2 mean energy of the low frequency surge res E mean = K xσ x = 2 K linear mooring system surge stiffness at the mean off x =
σ x = standard deviation (rms) of the low frequency surge r T , T is the storm duration, andT is the surge period N x = T N N
(4) The minimum and maximum low frequency offsets that are calculated from the (nonlinear) system potential energy and the (linear) quadratic approximation to the potential energy are also shown on Figure 2. The nonlinear method yields different results for minimum and maximum offsets in the stiff and soft offset directions, which are as expected.
OTC 13269
While, the linear method incorrectly predicts the same minimum and maximum surge offsets for both stiff and soft offset directions. The example above illustrates the effect of nonlinear mooring system force response on minimum and maximum low frequency surge offsets. For turret moored vessels the mean environmental moment contributes to the rotational stiffness of the long period sway-yaw mode, and the nonlinearity of the mean environmental moment with respect to the vessel’s heading will generally result in large differences between the linearly and nonlinearly calculated minimum and maximum modal responses. Moored systems display different degrees of nonlinearity, where reliability analyses are used to compare safety levels for different vessel and mooring system types, the ability of the analysis method(s) to predict low frequency responses should be investigated and where possible calibrated against model test results. Combin in g low fr equency modal responses
API RP 2SK and the ISO code recommend that wave frequency responses are calculated at the mean plus significant low frequency offset position and at the mean plus maximum low frequency offset position. However, neither code specifies how the combined significant and maximum low frequency offsets are to be calculated from the low frequency modal responses, nor do the codes discuss the need to define these positions differently depending on the system response under investigation. The low frequency meanderings of the vessel will involve, at least, surge, sway, and yaw motions, which may be mapped in a three-dimensional space where the ( x, y, z ) axes represent surge, sway, and yaw offsets measured from the mean equilibrium position. In this space the significant and maximum low frequency offset positions referred to by API RP 2SK and the ISO code are represented by an infinite combination (a surface) of surge, sway, and yaw values. Mooring codes contain no guidance or discussion as to how the low frequency significant and maximum design values should be selected from the significant and maximum low frequency surfaces. The “worst” position on the significant and maximum low frequency surfaces will depend upon the system response under investigation. For example, in the calculation of maximum line tensions the “worst” position of the vessel associated with the mean plus maximum low frequency response will depend upon the line under investigation. In the case of surge only motions illustrated on Figure 2, the mean plus maximum low frequency tension for the bow lines will occur at the minimum extreme vessel offset (turn-around) point. While the mean plus maximum low frequency tensions for lines in the other two groups will occur at the maximum extreme vessel offset point. In reliability analyses of strength limit states, system notional probabilities of one and two line failures are synthesized from individual line notional probabilities of failure. If the individual mooring line notional probabilities of failure are to be calculated consistently, then the mooring analysis software must use the individual line’s “worst”
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
significant and maximum low frequency vessel position in calculating mooring line wave frequency tensions. This is particularly important where reliability analysis of grouped and equally spaced moored systems are being compared. Wave fr equency li ne tensions
Present codes, [10, 11], provide formulae for calculating significant and maximum wave frequency line tensions from the standard deviation of the line tension. These formulae assume a frequency domain method that is based on; transforming the vessel motion RAOs to the fairleads, synthesizing the fairlead motion spectra from the wave spectrum and the fairlead RAOs, and solving linearized dynamic equations for mooring line tensions in the frequency domain. The principle result of the frequency domain analysis is the standard deviation of the mooring line tension, from which significant and maximum line tensions are calculated assuming a Rayleigh distribution. Alternatively, the significant and maximum fairlead motions may be calculated from the fairlead motion spectra, assuming a Rayleigh distribution, and significant and maximum line tensions may then be calculated dynamically by imposing the significant and maximum fairlead motions. These are both frequency domain line dynamic methods of calculating wave frequency tensions, however due to differences in the way in which the linearization is performed the two methods will result in different predictions of wave frequency line tensions. The reliability analyst should be aware of the assumptions used in the deterministic analysis and their effects on the calculated wave frequency line tensions. Basic Variables and Probability Distributions In reliability analyses the uncertainty in the basic variables that enter into the limit state function are represented by probability distributions. Sensitivity studies may be used to assist in deciding which variables should be described as random and which may take fixed values in the reliability analysis. The results of reliability analyses are generally sensitive to the tails of a few key probability distributions and in most cases there is insufficient data available to accurately define the distributions and parameters of the basic variables. The basic variables and their distributions for the description of uncertainty in the reliability analysis of position mooring systems are further discussed below. M odeli ng uncertainty
Regardless of the sophistication of the deterministic tool used to analyze the system, it is necessary to define the modeling uncertainty, as the modeling uncertainty provides the link between the results of theory with reality. Neglecting to include the model uncertainty in the reliability analysis is identical to defining it as having a (mean) value of 1.0 with a standard deviation of zero. This is equivalent to making the two claims that; theory is perfect and cannot be improved upon and the experimental data show absolutely no scatter. In many cases of interest real responses are sensitive to boundary conditions or initial conditions and the experimental data on
7
its own will show considerable scatter for nominally identical experiments. If reliability calculations are used comparatively then the modeling uncertainty must be carefully investigated and defined for the cases being compared. For full scale mooring systems the definition of modeling uncertainty requires a comprehensive instrumentation package, monitoring environmental loading parameters and system response parameters, and the occurrence of severe weather to ensure that the responses of interest are captured. Alternatively, wave basin tests on scale models may be used to assist in defining the model uncertainty across a range of vessel and mooring system types, for different environmental criteria. For the mean, low frequency, and wave frequency components of mooring line tensions, their modeling uncertainties are post-applied, as factors, to the line tension components calculated by the deterministic mooring analysis, in which the distributions of the other basic variables are preapplied. The modeling uncertainty is meant to include the effects of conservative or non-conservative assumptions, bias, in the methods used to predict line tensions, as well as the naturally occurring scatter in the line tensions. Progressive fail ur e and envir onmental return peri ods
The definition of the basic variables describing the environmental parameters determine, the reoccurrence period that the notional probability of failure is associated with. When a mooring line fails it will be retrieved and replaced at the earliest possible opportunity, therefore the use of longterm environmental distributions is not sensible when evaluating damaged line limit states. Consider the case in which the i’th and j’th lines are the first and second lines to fail in a storm. The probability that this occurs is given by Pf (i)Pf (j | i), where Pf (i) is the intact notional probability of failure of the i’th line and Pf (j | i) is the conditional probability of failure of the j’th line, given that the i’th line has already failed. Clearly, Pf (i) and Pf (j | i) cannot be calculated based on the long-term environmental distribution, as both lines must fail within a short time interval of each other. Thus, the notional probability of failure of limit states associated with damaged mooring systems cannot be evaluated using longterm environmental distributions. However, notional probabilities of failure for damaged limit states may be synthesized from notional intact and damaged probabilities of failure calculated for individual storms, i.e. using the “shortterm” environmental distributions. If the reliability analysis is performed using the design (100-year) return period environmental distributions, then the calculated notional probability of intact and damaged failure is the conditional probability of failure given the occurrence of the design (100-year) return period storm. This has the advantage that it preserves a direct connection between the intact and damaged safety factors used in the deterministic design analysis and the notional intact and damaged probabilities of failure, as both are associated with the same (100-year) metocean conditions. However, for geographic regions with very different extreme environmental distributions, such as winter storm and tropical revolving
8
KEN HUANG AND YONG BAI
storm areas, comparisons of notional probabilities of failure conditional on the occurrence of the design return period storm may not be a true indicator of comparative safety levels. From the response-based point of view, the better approach to obtain annual notional failure probabilities is to perform reliability analyses over a range of return period weather bins, and summing the product of the conditional notional probabilities of failure with the annual probability of occurrence of the weather bins. Envir onmental basic vari ables
Gumbel (FT Type I) or Weibull (FT Type III) extreme value distributions are generally used to describe the intensities of environmental parameters. The 100-year return period distributions of wind and current velocity and the significant wave height are defined by Weibull distributions for maxima. A method for deriving consistent distributions of environmental parameters for arbitrary return periods is outlined below. Typical metocean design criteria for the Gulf of Mexico and North Sea, based on ISO 13819-1, [18], are summarized in the Tables on Figure 3. Parameters for distributions of annual maxima, which are consistent with the design criteria, are readily obtained by fitting the distributions to the 10 and 100-year (P = 0.9 and P = 0.99) return period design values. Weibull distributions for Gulf of Mexico and North Sea annual maxima are shown at the bottom of Figure 3. These distributions are consistent with the 10 and 100-year design criteria, note that the probability of exceedence (1 – P) is plotted on the y-axis. The Weibull distribution for maxima is given by:
F Wmax ( x )
exp − ε − α = 1
x
β
; x
;
x > ε
≤
ε , α > 0, β > 0
(5) The three parameters ε, α, and β are the threshold, scale, and shape parameters respectively. If F(x) is the distribution function of annual maximum values, and these annual events are independent and identically distributed, then the probability, F N(x), that in N-years the random quantity X does not exceed x is given by: N
N
F ( x )
=
∏ i
N
prob ( X
≤
x)
=1
=
∏ i
F ( x )
= [ F ( x ) ]
N
=1
(6) If the distribution for annual maxima is a Weibull distribution, then the distribution of N-year return values, F N(x), is given by:
N
F ( x )
=
[ F
Wmax
( x )
]
N
e − = exp − a
N x = b
exp
β ε − x − α 1 β N
(7)
OTC 13269
Hence, if the parent distribution is a Weibull distribution for maxima, the distribution for N-year return values is also a Weibull distribution for maxima with threshold ( ε N), scale (α N), and shape (β N) parameters given by:
ε N
=ε
α N
=
β N
=
α
(8)
1β
N
β
This self-locking property applies to all three of the extreme value distributions, [19], and is extremely useful in deriving consistent distributions for series systems. L in e segment strength basic var iabl es
Similarities in the maximum storm event (the maximum value in a series of N years) and mooring line strength (the minimum value in series of N test lengths), allow the same statistical methods to be used for both problems. A mooring line is a typical series system and the prediction of mooring line strength is one of the classic problems in extreme value theory. If F(x) is the distribution function for the break strength of test lengths, then [1 - F(x)] is the probability that the strength of a test length exceeds x. For a mooring line segment composed of N similar independent and identically distributed lengths the probability that the strength of all N lengths exceeds x is given by [1 - F(x)] N, and the probability that the strength of the line segment is less then x is 1 - [1 - F(x)] N. That is, the strength distribution, F N(x), of a line segment composed of N test lengths in series is given by: F N ( x )
= 1 − [1 − F ( x ) ] N
(9)
The Weibull distribution for minima is given by: β 1 − exp − x − ε ; x ≥ ε , α > 0, β > 0 α ( )= F Wmin x 0 ; x < ε
(10) If the distribution of test length break strengths, F(x) in eqn. (9), is a Weibull distribution for minima, then the distribution of strength for a line segment composed of N test lengths, F N(x), is also a Weibull distribution for minima, with threshold (ε N), scale (α N), and shape (β N) parameters given by eqn. (8). The formulae above allow break strength to be defined consistently for line segments of different lengths (number of test lengths N). Mooring lines may be made of chain, steel rope or 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 statistical analysis of break test data. As mooring line segments are typically of the order of 10 to 1,000 times longer than the lengths used in break tests the strength of a line segment will depend upon the left tail of the test break strength distribution. The method used to determine the “best-
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
fit” parameters for the break test data should be chosen carefully. Plotting the “best-fit” distributions on the same graph as the break data, see Figure 4, will allow the fit to the all-important left tail of the strength distribution to be examined.
distributions of basic variables, rather than the computing requirements of the Monte Carlo method. Monte Carlo methods are often used to validate other methods, [20]. However, due to the speed of present day computers, there is no reason why Monte Carlo methods should not be used for all the reliability analyses of position moored systems, rather than simply for validation, [8]. Level 3 methods evaluate eqn. (11) by “exact” methods. For Level 3 methods to be feasible, the joint distribution of the basic variables X1,…, Xn must be known. The failure probability Pf can then be written as the n-dimensional integral
Sensiti vity studi es and other basic var iabl es
In addition to modeling uncertainty, environment, and strength, other variables whose uncertainties are judged to be important, either by experience or by sensitivity studies, should be represented as random variables in the reliability analysis. Deterministic sensitivity studies which encompass the range of vessel and mooring system types, water depths, and environmental conditions over which reliability analyses are to be performed can be used to assist in identifying those variables that should be represented as random variables. The sensitivity plots on Figure 5 show the tension in the most loaded line for variations of ±15% in the 1-hour average wind velocity, pretension in line 2 and heave natural period. The probability distribution used for each basic variable should be based, as far as possible, on the statistical analysis of supporting background data. In general, normal, lognormal, Gumbel, or Weibull distributions should be used. When no detailed information is available, the three-parameter lognormal and Weibull distributions should be used to represent variables that are known to have lower or upper limits, while the normal and Gumbel distributions should be used for distributions that are unlimited.
∫ ∫ f ( x)dx K
Classification of Reliability Methods Reliability theory provides methods for the approximate evaluation of the failure probability, (11) P f = Prob g X 1 , K , X n , C ≤ 0
((
)
)
In order that eqn. (11) can be evaluated the joint probability distribution of the basic variables has to be known, or alternatively independence of X1,…, Xn is assumed and knowledge of the marginal distributions suffices. The choice of the marginal distributions arises from statistical studies analyzing available data for the respective basic variables. The Law of Large Numbers, [19], allows the following frequency interpretation of the probability of failure for large l , P f
=
1 l
×
number of indices
i = 1, K , l with g ( X i , C )
≤
(12) Where Xi, i = 1, 2,…, l is an independent sequence of random vectors each having the distribution of X=(X 1,…,Xn)T. Monte Carlo methods exploit eqn. (12) to arrive at an approximate value of Pf , [20], using independent random samples from the joint distributions, if they are known, of the variables, X1,…, Xn. As the failure probability of the limit state function decreases the computing time required by Monte Carlo methods increases. However, in practice the accuracy with which small probabilities of failure can be calculated will usually be controlled by the lack of data available to define the tails, extreme values, of the
9
0
x
1
K dx n ,
{x ∈ℜ n : g( x ) ≤ 0}
(13) Where f X denotes the density of the distribution of X=(X 1,…, Xn). Numerical integration may then be used for the evaluation of the integral above. However, this is not practical for multiple degrees of integration. These methods are often not feasible due to a lack of knowledge of the "true" joint distribution of the variables, or because of the complexity of the integration required, [1, 20]. Level 2 methods are approximate techniques for calculating Pf by replacing the limit state function, eqn. (11), by an approximating function, which renders the problem amenable to analysis. In Level 2 methods the original basic variables, (X1,…, Xn), are transformed into independent standard normal variables, (Z1,…, Zn), which implies a transformation of the original g-function to a new limit state function f. Working with the n-dimensional standard normal distribution is computationally convenient due to the rotational symmetry of its density and because the probability content of a half-space can be easily represented as Φ(-β), where Φ is the standard normal distribution and β denotes the distance of the separating hyperplane from the origin. Level 2 methods involve an iterative search for the minimum distance of the failure surface from the origin, in the n-dimensional hyperspace, [8], as shown on Figure 6. First order reliability methods (FORM) approximate the limit state surface, at the closest point to the origin, by its tangent hyperplane, and the probability of failure is approximated by the probability content of the half-space lying near the failure region, [22, 23]. Second order reliability methods (SORM) also utilize the curvatures of the failure surface in arriving at the approximation of the failure probability, [22]. Level 1 methods are design methods that use a number of partial safety factors, which are related to pre-defined characteristic values of the basic variables, to provide the necessary level of structural reliability. That is, Level 1 methods are not methods of structural reliability, but methods of safety checking used in design, [2]. The main limitations of level 2 methods, illustrated on Figure 6, are those associated with any optimization process which are: (1) Convergence of the processes does not guarantee that the global minima has been found. (2) Only one limit state may be evaluated at time, i.e.
10
KEN HUANG AND YONG BAI
separate iterative searches for minima must be performed for each mooring line in intact and damaged conditions. The Monte Carlo method may be used to evaluate the reliability of position mooring systems. Monte Carlo methods do not involve an iterative search, consequently any number of limit states may be evaluated simultaneously. This is particularly useful for mooring systems in which a number of the mooring lines contribute significantly to the total notional intact and damaged probabilities of one and two line failures.
Conclusion The following can be concluded from this paper: (1) Where there is uncertainty or variability in the basic (input) variables, post-applied safety factor formats of the LRFD type will produce inconsistent safety levels for mooring systems that have different nonlinear response characteristics. (2) The use of pre-applied safety factor formats has yet to be demonstrated that a reasonably simple and practical format can be devised. (3) The ability of the analysis method(s) to predict low frequency responses should be investigated and where possible calibrated against model test results. (4) The reliability analyst should be aware of the assumptions used in the deterministic analysis and their effects on the calculated wave frequency line tensions. (5) Sensitivity studies can be used to assist in deciding which variables should be described as random and which may take fixed values in the reliability analysis. (6) Notional probabilities of failure for damaged limit states should be synthesized from notional intact and damaged probabilities of failure calculated for individual storms, i.e. using the “short-term” environmental distributions. (7) Finally, reliability methods can be a powerful tool for comparing the safety implications of different designs, and target safety levels for new designs, can be established by performing reliability calculations for old designs that have a history of satisfactory performance.
8.
9.
10.
11.
12.
13.
14.
15.
16. 17. 18.
19.
References 1. Rationalization of safety and serviceability factors in structural codes, CIRIA, Report 63, 1977. 2. General principles for the verification of the safety of structures, ISO 2394, International Standards Organization, 1986. 3. Recommended Practice for the Analysis of Spread Mooring Systems for Floating Drilling Units, API RP 2P, 2nd Ed, American Petroleum Institute, 1987. 4. Recommended Practice for In-Service Inspection of Mooring Hardware for Floating Drilling Units, API RP 2I, 1st Ed., American Petroleum Institute, 1987. 5. Specification for Mooring Chain, API Spec 2F, 6 th Ed., American Petroleum Institute. 6. Qualification Testing of Steel Anchor Designs for Floating Structures, API RP 2M, 2 nd Ed., 1996, American Petroleum Institute, 1997. 7. Recommended Practice for Design, Selection,
20.
21.
22.
23.
OTC 13269
Operation, and Maintenance of Marine Drilling Riser Systems, API RP 2Q, 2 nd Ed., American Petroleum Institute, 1993. Calibration of ABS, API, HSE, and NMD Mooring Design Codes for Drilling and Production Platforms, Mooring Code JIS, Noble, Denton & Associates, Inc. Rpt. No. H3687, 1995. Recommended Practice for Design, Analysis, and Maintenance of For Floating Production Systems, API RP 2FP1, 1st Ed., American Petroleum Institute, 1993. Recommended Practice for Design and Analysis of Stationkeeping System for Floating Structures, API RP 2SK, 1st Ed., American Petroleum Institute, 1997. Draft ISO Standard for Design and Analysis of Stationkeeping Systems for Floating Structures, Informative April 14, 1999, Normative Nov. 8, 1999, ISO/TC67/SC7/WG5/P5 Committee Correspondence. Recommended Practice for Design, Manufacture, Installation, and Maintenance of Synthetic Fiber Ropes for Offshore Moorings, API RP 2SM, Draft, American Petroleum Institute, 1999. ABS Guide for Building and Classing Floating Production Installations, American Bureau of Shipping, June, 2000. Recommended Practice for Design, Selection, Operation, and Maintenance of Marine Drilling Riser Systems, API RP 16Q, 2 nd Ed., American Petroleum Institute, 1993. Recommended Practice for Design and Operation of Subsea Production Systems, API RP 17A, 1 st Ed., American Petroleum Institute, 1987. Recommended Practice for Flexible Pipe, API RP 17B, 1st Ed., American Petroleum Institute, 1988. Comparison of Marine Drilling Riser Analyses, API Bull. 16J, 1st Ed., American Petroleum Institute, 1992. Draft ISO Standard for Fixed Steel Offshore Structures, General Annexes, Draft C ISO 13819-1, ISO/TC67/SC7/WG3 Committee Correspondence, 1997. Mann, N.R., Schafer, R.E., and Scigpunwalla, N.D., Methods for Statistical Analysis of Reliability and Life, Wiley, N.Y., 1974. Sweeting, T.J., and Finn A.F., A Monte Carlo Method Based on First- and Second-Order Reliability Approximations, Structural Safety, 11, pp. 203-212, 1992. Shinozuka, M., Basic Analysis of Structural Safety, Journal of Structural Engineering, ASCE, Vol. 109, No. 3, pp. 721-740, 1983. Madsen, H.O., First Order vs. Second Order Reliability Analysis Of Series Structures, Structural Safety, Vol. 2, pp. 207-214, 1985. Dolinski, K., First-order second-moment approximation in reliability of structural systems: critical review and alternative approach, Structural Safety, Vol.1, pp. 211231, 1983.
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
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
11
12
KEN HUANG AND YONG BAI
3x3 Grouped Mooring Pattern
Figure 2 Comparison of Linear and Nonlinear Frequency Domain Methods of Calculating Low Frequency Offsets
OTC 13269
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
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
13
14
KEN HUANG AND YONG BAI
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
OTC 13269
ABS BIOGRAPHIES KEN HUANG Ken Huang is a graduate of the Texas A&M University with a M.S. in Ocean Engineering and Hydrodynamics. 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 participating 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
Copyright 2001, Offshore Technology Conference This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, 30 April–3 May 2001. This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
Abstract The oil and gas production industry has expressed an interest in being able to utilize Floating, Production, Storage and Offloading (FPSO) units as a development option in the deepwater areas of the United States (U.S.) Gulf of Mexico (GOM) outer continental shelf (OCS). Operators will need regulatory approval from both the United States Coast Guard (USCG) and the Minerals Management Service (MMS) for a FPSO project. Neither the USCG nor MMS currently have regulations specifically for the design and operation of FPSOs on the OCS. A workgroup was formed under the Offshore Operators Committee’s Deepwater Subcommittee to assist MMS and USCG in reviewing the existing regulations and body of standards, specifications, recommended practices and classification society rules and guides concerning the design and operation of FPSOs on the OCS for the Gulf of Mexico. The effort also aimed to identify gaps in the regulations and industry standards. This paper provides a summary of the major findings of the workgroup. In addition to the items discussed in this paper, the workgroup report identified other areas where additional modifications to regulations or industry standards may be warranted. Introduction As operators have moved into the deeper waters of the GOM over the last several years, interest has been growing in potentially utilizing FPSOs as a development option to the floating production systems (tension leg platforms (TLP), spars, etc) and subsea tie backs to either floating production systems or fixed platforms that are currently being utilized. In discussions with MMS and USCG, it became apparent that several studies would need to be conducted to confirm the acceptability of these systems for the GOM. MMS indicated
that an Environmental Impact Statement (EIS) would be prepared for the first FPSO proposed to be utilized in the GOM. In working with industry, MMS agreed to do a programmatic EIS on the generic use of FPSOs in the GOM if industry would fund the study. The joint industry project DeepStar agreed to fund the EIS 1. The draft EIS on the Proposed Use of Floating Production, Storage and Offloading Systems on the Gulf of Mexico Outer Continental Shelf in the Western and Central Planning Areas 2,3 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 Decision to be published no earlier than 30 days from the publication of the final EIS. The second study effort was to do a comparative risk assessment to evaluate and compare oil spill and fatality risks for the FPSO with a spar, a TLP and a shallow-water jacket serving as a hub and host to deepwater production. The Offshore Technology Research Center completed that study for MMS4. The third step in the process was to identify any gaps in the existing regulations and to develop a regulatory model that could be used by MMS and the USCG in the review and approval of a FPSO project. On March 22, 2000, Mr. Chris Oynes, MMS GOM Regional Director, sponsored a meeting between MMS, USCG and industry to discuss the regulatory requirements for FPSOs in the GOM, should they be found to be an acceptable development option. Although it was recognized that MMS and the USCG would have to agree among themselves the appropriate regulations and regulatory split between the two agencies, they agreed that it would be beneficial and appropriate to have industry provide input on the model. It was decided that a workgroup would be formed under the direction of the OOC Deepwater Subcommittee and consists of industry representatives and classification society representatives along with personnel from MMS and the USCG. A report was prepared by the workgroup and submitted to MMS and the USCG in September 2000 for their consideration5.
Regulatory Model Workgroup Goal The overall goal of the workgroup was to review the existing regulations and industry standards covering the design, construction and operation of FPSOs in the GOM and identify
2
WANDA J. PARKER AND TODD W. GROVE
any gaps in either the regulations or standards that needed to be addressed prior to bringing FPSOs into the GOM. The workgroup was focused on the design and operational considerations for the FPSO. The work group did not address the design of shuttle tankers or operational considerations once they were disconnected from the FPSO. Participants The workgroup was formed under the direction of the OOC Deepwater Subcommittee and met five times following the initial meeting. It was agreed that a cooperative effort with open discussions between the regulatory agencies and industry was desired and would produce the best work product. Due to the broad scope of the discussions, it was necessary to include a large number of participants. MMS agreed to have Mr. James Regg, Section Chief, Technical Assessment and Operations Support serve in the workgroup along with support from other MMS personnel. LCDR Bill Daughdrill from the Eighth USCG District represented the USCG in the work group. Personnel from the MMS Headquarters and USCG Headquarters groups were kept informed of the workgroup’s activities through e-mail. Twenty-five persons representing 17 companies participated in one or more of the meetings. Since classification societies have historically played a large role in the approval of FPSOs worldwide, it was felt that it was important to have broad representation from the major classification societies who currently class FPSOs. Four classification societies; American Bureau of Shipping, Bureau Veritas, Det Norske Veritas and Lloyd’s Register of Shipping were represented in the workgroup. Tim Sampson represented the American Petroleum Institute (API). Wanda Parker agreed to chair the workgroup for OOC. A complete listing of all workgroup participants is in the workgroup report5. All of these individuals dedicated a considerable amount of time and expertise to this effort. FPSO Characteristics For consistency, the workgroup decided to use the FPSO characteristics used in the EIS and CRA studies as the FPSO model for this effort. The workgroup considered the regulations that would apply to a US flag FPSO or an undocumented FPSO that is designed to US flag requirements (similar to the existing GOM floating production systems). Limited discussions were held on the differences in permitting a US flag FPSO and a foreign flag FPSO. The FPSO was considered to be ship-shaped with limited discussions on the differences between a ship-shaped FPSO and a non shipshaped FPSO. Discussions were limited to a permanently moored FPSO for simplicity since a disconnectable FPSO introduces many complicating factors into the discussion. Finally, discussions were focused on systems that are unique to FPSOs with only limited discussion of systems that are common to fixed platforms or other floating production systems. This list of system includes: 1. In Hull Cargo Storage Systems 2. Cargo Piping and Transfer 3. Turret/Mooring/Stationkeeping/Swivel
4. 5. 6. 7. 8. 9. 10. 11. 12.
OTC 13170
Propulsion Stability Motions/Global Performance Risers Offloading Systems Layout Operational Considerations Discharges Manning
Regulations MOU It is recognized that both MMS and USCG have a large body of regulations that could be applied to FPSOs. In the Memorandum of Understanding (MOU) between MMS and USCG6 signed on December 16, 1998, the delineation of jurisdictions regarding floating production system components, operations and issues is addressed. The workgroup reviewed the MOU for completeness in coverage of systems on a FPSO and put together a table showing the applicable regulations, industry standards and classification society rules for each system and sub-system in the MOU. While this table was not exhaustive, it quickly pointed out areas where regulations and standards were well established and systems where either regulations or standards were lacking. The work group made some specific recommendations concerning the implementation of the MOU that will be discussed below. MMS MMS regulations that are applicable to FPSOs are primarily located in 30 CFR 250, Subparts H and I. In conjunction with their regulations for specific systems, MMS has said that they intend to utilize the Deepwater Operations Plan (DWOP) process in their review of a FPSO project7. The workgroup reviewed the MMS regulations for adequacy and made some specific recommendations that will be discussed below. USCG The USCG has said that FPSOs will be regulated as vessels and therefore will be required to meet specific vessel design and operational regulations8,9. USCG regulations that are applicable to FPSOs are primarily located in 46 CFR, Subchapters D and IA and 33 CFR Subchapter N. On Dec 7, 1999, the USCG published a Notice of Proposed Rulemaking for 33 CFR Subchapter N that includes proposed regulations applicable to FPSOs. The workgroup considered these proposed regulations in addition to the established regulations. The workgroup report was submitted to the record as a comment to the proposed regulations. The specific recommendations will be discussed below. Industry Standards Both USCG and MMS regulations incorporate by reference a large number of industry specifications and recommended practices, particularly API documents. Many of these are applicable to FPSO system design or operation. In addition,
OTC 13170
FPSO STANDARDS AND RECOMMENDED PRACTICES
there are a host of other international standards that may be applicable to FPSOs. Most of these standards cover individual systems or subsystems that may be used in conjunction with a wide variety of types of installations. It was outside of the scope of the workgroup to conduct a thorough review of the adequacy of the individual industry specifications and recommended practices. Rather, the workgroup focused on a few key standards and made recommendations of standards the agencies should consider incorporating by reference. Classification Society Rules and Guides Another body of standards applicable to FPSOs and recognized worldwide are those developed and applied by a number of the major classification societies. The rules and guides of the societies are utilized in assessing the fitness-for purpose of FPSOs and are focused on safety aspects. The classification requirements address design requirements as well as those for fabrication, installation, and commissioning. Unlike the snapshot nature of certification, classification is an ongoing process by which the societies survey a FPSO periodically during its operational life to ensure compliance with the rules. Given their unique independent role internationally, classification societies are also often delegated statutory responsibilities by flag and coastal state authorities to act on behalf of the administration. USCG has, to varying degrees, delegated to several classification societies approval responsibilities for existing GOM floating production units. Expectations are that similar delegations will be forthcoming as experience is gained in GOM FPSO applications. Several classification societies have also been active to varying degrees in the MMS Certified Verification Agent program for existing structures, which may also be extended to FPSOs. The relevant practical experience from classification and statutory responsibilities is utilized in maintaining the rules and guides in a current and relevant form. The four participating societies contributed citations to the regulatory matrix for the most prevalent rule and guide requirements applicable to the various aspects of the FPSO.
Gaps and Recommendations During the review process, the workgroup identified a number of recommendations either for ways to close existing gaps in the regulations or for additional work that should be considered. These recommendations were broken out into areas for consideration by MMS and USCG and for industry. MMS In 30 CFR 250, Subpart I, MMS has established a platform verification program requiring a third party certification of the platform design. These regulations were established for fixed platforms, but MMS has been applying them to floating production facilities. It is recommended that MMS revise these regulations to update them for floating facilities, including FPSOs. The workgroup recommended adding the turret, risers and mooring systems to the verification program.
In lieu of the agency writing prescriptive regulations for
3
FPSOs, the workgroup recommended that MMS consider incorporating by reference additional industry standards and recommended practices as they become available. The following were identified as candidates for MMS review and consideration for incorporation in their entirety 10,11,12,13,14: 1. API RP 2FPS, Planning, Designing and Constructing Floating Production Systems 2. API RP 2SM, Design and Analysis of Synthetic Moorings 3. API RP 2SK, Design and analysis of Stationkeeping Systems for Floating Structures 4. API RP 2RD, Design of Risers for Floating Production Systems (FPSs) and Tension-Leg Platforms (TLPs) 5. API Spec 17J, Specification for Unbonded Flexible Pipe USCG On December 7, 1999, a Notice of Proposed Making was published revising 33 CFR, Subpart N. In that revision, both Mobile Offshore Drilling Unit (MODU) regulations located in 46 CFR, Subchapter I-A and tank vessel regulations located in 46 CFR, Subchapter D were referenced. However, the regulation was not clear in many cases which regulation should be followed if both regulations cover the same system or subsystem. Also, in many cases, modifications to these regulations may be needed to address the unique circumstances of FPSO or floating system operations versus a tank vessel or MODU. The workgroup recommends that specific regulations for floating facilities should be written in lieu of pointing to regulations for various types of vessels, which may not be completely applicable to floating production facilities. The workgroup also recommended that their report be included in the comment record for the proposed rulemaking. The workgroup recommended that marine crew manning and qualification regulations should be codified for FPSOs and for other floating production systems. The Eighth Coast Guard District has issued a policy letter 15 for marine crew manning for floating production systems other than those storing oil in bulk, which could serve as an appropriate starting base for non-self propelled FPSOs. It was recognized that self-propelled FPSOs would need additional consideration. MMS has adopted API RP 500/505 16,17 for area classification while the USCG has prescriptive regulations in 46 CFR, Subpart J. The workgroup recommended that the USCG adopt API RP 500/505 for floating production systems including FPSOs. Adopting common standards would minimize confusion and duplication of effort for industry since both agencies have jurisdiction for area classification. Although the USCG has been given sole jurisdiction over fire fighting systems for floating production systems, including FPSOs, the USCG has not yet proposed regulations for fire fighting systems in the production area. It is recommended that the USCG adopt fire-fighting regulations for the production area.
4
WANDA J. PARKER AND TODD W. GROVE
Both MMS and USCG have voluntary safety management system programs that could be applicable to FPSOs. MMS has recognized API RP 75 18 as an acceptable basis for a safety management system for fixed and floating production systems on the US OCS. Most oil and gas production companies have based their safety management programs for operations on the US OCS on API RP 75. However, the USCG has recognized the International Safety Management Code (ISM) 19, which is applicable to vessels that must comply with Chapter IX of SOLAS. While both programs have merit, and individual companies may want to base their programs on either standard, a combination of the standards or some other standard, the workgroup recommends that the USCG recognize API RP 75 as an acceptable basis for a safety management program in addition to ISM. In the event that foreign flagged FPSOs are acceptable for use in the GOM, the USCG will issue a Letter of Compliance (LOC)8. Since a full design review is not normally conducted for a LOC, the workgroup recommended that the USCG develop a LOC checklist that could be either used with existing foreign flagged FPSOs or new-built foreign flagged FPSOs proposed for operations in the GOM. Although the USCG has general lightering regulations and additional operational regulations that apply to the designated lightering areas in the GOM, there are no specific operational regulations that apply to FPSOS or to tandem offloading. It is recommended that a work group be formed to review existing international standards to determine their adequacy for GOM operations. Joint MMS and USCG In the MOU, both MMS and USCG have been given jurisdiction for reviewing and approving the design of the turret and mooring system. The workgroup agreed that technology is rapidly evolving for these systems and that it would be burdensome on the regulatory agencies to have personnel fully knowledgeable about these systems as they change. It is recommended that a verification agent acceptable to both agencies be selected to review and certify the design. In the MOU, both MMS and USCG have been given jurisdiction for reviewing and approving various portions of the integrated monitoring and safety systems. It is recommended that a work group consisting of representatives of Industry, MMS and the USCG be formed to address the integration of these systems. In the MOU, both MMS and USCG have been given jurisdiction over piping systems. It is recommended that for cargo tank piping that the specification break between MMS and USCG jurisdiction occur at the 1 st valve downstream of the last processing vessel (and its control valves and safety system) prior to the oil entering the cargo storage tanks. It is recommended that a work group be formed consisting of representatives of industry, MMS and USCG to review other similar systems and agree to where the specification breaks between the systems should occur. These breaks should be codified into the MMS and USCG regulations. Alternatively, MMS and USCG should consider adopting consistent industry
OTC 13170
standards for piping systems. Neither MMS nor USCG regulations address integral hull tanks used as process vessels (such as wet/dry oil tanks). The work group recommended that all integral hull tanks be under USCG jurisdiction for structural design. For tanks used as process vessels, the safety system, control valves, and piping to and from the process vessels should be under MMS jurisdiction. It is recommended that piping specification breaks should occur at the first flange outside the tank. Industry Both the Environmental Protection Agency (EPA) and the USCG have regulations concerning discharges to the ocean that could occur from FPSOs. It is recommended that a work group under the OOC Deepwater Subcommittee be formed to review the overboard discharge regulations of both agencies to ensure that all discharges are adequately addressed for FPSO operations. Although the USCG has regulations that provide for the inspection of foreign flagged tank vessels, no formal determination has been requested or received from US Customs that indicates if foreign flagged FPSOs will or will not be allowed to operated on the OCS (i.e. Will Customs interpretation and policy for foreign flagged FPSOs be similar to their current interpretation and policy for foreign flagged MODUs?) The workgroup recommended that industry develop a strategy for obtaining a formal, written determination from Customs. It was recommended that the various API standards and recommended practices for mooring systems and riser design be reviewed to determine if the inspection guidelines given in those documents are adequate for floating production systems and updating the documents as needed. It was recommended that API RP 14C be reviewed and revised if needed for floating production systems, including taking into account the effect environmentally induced motions may have on the safety and monitoring systems. In addition, it is recommended that the safety and monitoring systems for swivels, integral hull tank process vessels and other unique systems to a FPSO be covered in the recommended practice. It was recommended that API RP 14E 20 be reviewed for adequacy and updated as needed for the effects of motion and piping support. It was recommended that the various API documents on composite materials should be reviewed for adequacy for floating facilities including FPSOS. It was recommended that API RP 75 be reviewed and updated as needed to ensure the document is an adequate basis for a safety management system for FPSO systems and operations. Applicable portions of ISO 9000/14000 could be incorporated, if desired. The work group recognized that standards for cargo tank cleaning for FPSOs might be different from trading tankers since inspections may occur when the FPSO is on location and in operation and not in dry dock. It is recommended that appropriate standards for cargo tank cleaning be confirmed.
OTC 13170
FPSO STANDARDS AND RECOMMENDED PRACTICES
Conclusions While neither MMS nor USCG has regulations that specifically apply to the design or operation of FPSOs, there is a multitude of regulations that are applicable to FPSOs and with the modifications indicated in the work group’s recommendations, the existing framework is adequate. A large body of industry recommended practices and standards consisting of classification society rules and guides, API standards, specifications and recommended practices and international standards exist that cover FPSO system design and operation. In some cases these need to be reviewed and updated, where needed, to ensure they are applicable to GOM FPSO operations. Many of the recommendations identified for FPSOs are also applicable to other types of floating facilities. The majority of the discussions by the workgroup focused on systems that are unique to FPSOs. The FPSO model used for the discussions was a ship-shaped permanently moored FPSO. It is recognized that a FPSO is not dependent on shape and some variations occur as you move between ship-shaped and non-ship-shaped FPSOs that will need additional attention. Likewise, a disconnectable, self-propelled FPSO has many aspects that are different from a permanently moored FPSO that will need to be considered. The review of the existing regulations was conducted without representation from the USCG Marine Safety Center or Headquarters groups. These groups need to be fully engaged before final determinations can be made on the appropriate regulations for FPSO design and operations. The representatives from the Eighth Coast Guard District participated fully in the discussions, but it was recognized by the workgroup that they were not authorized to speak definitively on USCG policy or regulation. OOC appreciated the opportunity to take the lead role in this cooperative effort between MMS, USCG and Industry. By working together and pooling our thoughts and ideas, regulations that meet the needs of all concerned can be put in place for FPSOs in the GOM. A large number of industry representatives and classification society representatives dedicated a considerable amount of time and expertise to this task. The efforts of Mr. Jim Regg and LCDR Bill Daughdrill were recognized for their active participation in the work group. As policies and rulemaking for FPSOs are drafted, continuing the cooperative effort between the regulators, industry and the classification societies will be beneficial. Acknowledgments The authors would like to thank the Offshore Operators Committee for allowing us to publish the results of the work group. References 1. 2.
Verret, Allen J., Hays, Paul R., “Deepstar’s Program Related to FPSO’s”, Offshore Technology Conference, 10703, May 1999. “Proposed Use of Floating Production, Storage, and Offloading Systems on the Gulf of Mexico Outer Continental Shelf,
3.
4.
5.
6.
7.
8.
9. 10.
11.
12.
13.
14. 15.
16.
17.
18.
19.
20.
5
Western and Central Planning Areas, Draft Environmental Impact Statement”, Minerals Management Service, Gulf of Mexico OCS Region, August 2000. George, J.E., Parker, W.J., Cranswick, D.J., “FPSO Environmental Impact Statement: What is Happening?”, Offshore Technology Conference, 10705, May 1999. Gilbert, R.B., Ward, E.G., Wolford, A.J., “A Comparative Risk Analysis of FPSO’s with Other Deepwater Production Systems in the Gulf of Mexico”, Offshore Technology Conference, 13173, May 2001. Letter to Carolita Kallaur, Minerals Management Service, from Allen Verret, Executive Director, Offshore Operators Committee, Regarding Regulatory Framework – Floating Production Storage and Offloading Systems, September 2000. Memorandum of Understanding (MOU) Between the Minerals Management Service and the United States Coast Guard”, Federal Register, Vol. 64, No. 10, January 15, 1999. Regg, James B., “Floating Production, Storage and Offloading Systems in the Gulf of Mexico OCS: A Regulatory Perspective”, Offshore Technology Conference, 10701, May 1999. Daughdrill, W.H., Brown, M.J., “The Regulatory Scheme Applicable to Floating Production, Storage, and Offloading Systems”, Offshore Technology Conference, 10702, May 1999. Letter to Carolita Kallaur, Minerals Management Service, from RADM North, United States Coast Guard, November 16, 1998. “API Recommended Practice for Planning, Designing, and Constructing Floating Production Systems”. American Petroleum Institute, 2000 Draft. “Recommended Practice for Design, Manufacture, Installation, and Maintenance of Synthetic Fiber Ropes for Offshore Mooring”. American Petroleum Institute, November 1999 Draft. “API Recommended Practice for Design and Analysis of Stationkeeping Systems for Floating Structures”. American Petroleum Institute, December 1996. “API Recommended Practice for Design of Risers for Floating Production (FPSs) and Tension-Leg Platforms (TLPs)”. American Petroleum Institute, June 1998. “API Specification for Unbonded Flexible Pipe”. American Petroleum Institute, July 2000. The Eighth Coast Guard District policy letter for marine crew manning for floating production system other than those storing oil in bulk “API Recommend Practice for Classification of Locations for Electrical Installations at Petroleum Facilities Classified as Class I, Division 1 and Division 2”. American Petroleum Institute, November 1997. “API Recommend Practice for Classification of Locations for Electrical Installations at Petroleum Facilities Classified as Class I, Zone 0, Zone 1 and Zone 2”. American Petroleum Institute, November 1997. “API Recommend Practice for Development of a Safety and Environmental Management Program for Outer Continental Shelf Operations and Facilities”. American Petroleum Institute, July 1998. “International Management Code for the Safe Operation of Ships and for Pollution Prevention”, International Maritime Organization, Assembly Resolution A.741(18), 1993. “API Recommend Practice for Design and Installation of Offshore Production Platform Piping Systems”. American Petroleum Institute, October 1991.
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 Divisional 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. Copyright 2001, Offshore Technology Conference This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, 30 April–3 May 2001. This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
Abstract This paper describes a study to compare the risks of Floating Production Storage and Offloading Systems (FPSO’s), which have never been used in the Gulf of Mexico, with the risks for existing deepwater production systems in the Gulf of Mexico. The major conclusion is that the expected risks for fatalities and oil spills associated with FPSO’s are comparable to those for already accepted alternatives for deepwater production. In addition, the oil spill risks are dominated by spills that occur during transportation of oil from the production facility to the shore with either pipelines or shuttle tankers. The major recommendation is to periodically update these results so that they serve as a baseline for future analyses of risk in the Gulf of Mexico. This study was undertaken for the Minerals Management Service to provide information for their use in developing a policy for FPSO’s in the Gulf of Mexico. Introduction To date, deepwater (more than 3,000-foot water depth) reserves in the Gulf of Mexico have been developed primarily with the following types of production systems: Spars; Tension Leg Platforms (TLP’s); and Subsea Well Systems tied back to these floating systems or to shallow water jackets that may also serve as hubs for other deepwater production systems (Hub/Host Jacket). All three of these types of systems rely on pipelines to transport oil to shore. A potentially attractive alternative to these systems is a tanker based Floating Production Storage and Offloading (FPSO) system with oil transportation to shore via shuttle tankers. Floating Production Storage and Offloading systems have been used in many areas of the world, but not the Gulf of Mexico.
The Minerals Management Service (MMS) funded the Offshore Technology Research Center (a National Science Foundation engineering research center located at Texas A&M University and The University of Texas at Austin), with EQE International, Inc. as a subcontractor, to conduct a Comparative Risk Analysis (CRA). The purpose of this study was to assess and compare the system risks for FPSO’s with those for existing deepwater production systems, specifically 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 Mexico 1. Information from both the Comparative Risk Analysis and the EIS will be used by the MMS in developing policies concerning the use of FPSO’s in the Gulf of Mexico. The primary objectives of the Comparative Risk Analysis were the following: 1. Assess and compare the system risks for FPSO’s with those for existing deepwater production systems, specifically Spars, TLP’s, and Hub/Host Jackets; and 2. Understand the contributions to system risk by subsystems and phases of operation.
Approach The approach used to conduct the Comparative Risk Analysis was developed with the following goals in mind: 1. Provide the MMS with information that can be used for a consistent and objective comparison of the risks associated with the four production systems; 2. Provide the MMS with a level of detail necessary to compare and understand overall system risks for typical production systems; and 3. Incorporate industry data, experience and expertise to the greatest extent possible into evaluating the risks. The approach used to achieve these goals involved teams of experts and a series of workshops. Participation of Technical Experts. Historical data on actual failures, particularly those very infrequent failures with large consequences that tend to drive overall risks, are scarce, and the risks must be estimated by other means. For this study, we chose to directly involve the expertise and experiences of engineers involved in the design and operation of these
2
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
production systems. The Deepstar consortia facilitated and coordinated the participation of industry engineers in this project. There was active participation by experienced engineers representing all segments of the industry, including oil companies, consultants, manufacturers, contractors, classification societies, as well as the regulatory agencies. They brought a detailed understanding of the nature of these risks as well as design and operational options to manage these risks. Many of the industry engineers who were involved in this study had also participated in risk studies within their companies, which are often undertaken either in the design of a specific system or to compare several systems in selecting the most appropriate system for a given project. The practical experience and perspective that these engineers brought to the study was deemed critical to the success of this study. Separate teams were formed for each of the four production systems, the Spar, the TLP, the Hub/Host Jacket, and the FPSO. These teams were made up of invited participants from industry and representatives from the MMS and the U. S. Coast Guard (USCG), the government agencies responsible for regulating the deployment and operation of deepwater production systems. The teams were designed to include engineers with expertise and experience in the design, construction and operation of the overall systems as well as the subsystems and components that make up the systems. There was an average of about ten members per team. The companies that provided one or more participants to these system teams are listed in Table 1. It is worth noting that these companies represent a large measure of the offshore industry’s deepwater experience and expertise. They have been very active in the design, operation, and/or certification of deepwater production systems in the Gulf of Mexico and elsewhere. Of particular importance is their direct involvement and experience with the deepwater production systems used in this study: Spars, TLP’s, Hub/Host Jackets, and FPSO’s. The teams were balanced to include members with overall systems expertise as well as those with expertise in various sub-systems, components, and operations, including: • Platform and subsea well systems; • Drilling and well intervention operations for both platform and subsea wells; • Topsides (processing facilities, equipment); • Production operations; • Pipelines and flowlines; • Tanker and FPSO design and operations; • Structures (hulls, decks, mooring systems, riser systems); • Helicopter operations (personnel transport); • Supply boat operations (material & personnel transport); and • Diving operations. Thus the teams were able to focus on risks at the sub-system, component, and operational levels as well as to focus on
OTC
overall system risks. Additional contributions from industry included input on detailed hazard identifications through participation in formal specialty interview sessions, and various other interactions in which individuals provided data, input, or advice. Technical experts from the companies and organizations listed in Table 2 as well as from most of those listed previously in Table 1 contributed in these areas. In all, over 100 of the industry’s more experienced engineers directly participated in the study either through the system teams and workshops, or the specialty interview sessions. The names and affiliations of these participants are summarized in Gilbert et al. 2. The average experience level for these experts was approximately 20 years. The level of participation by the industry experts was substantial. Their direct involvement in the workshops (preparation, participation, and review) and the specialty interviews involved an estimated 5,000 man-hours. Further, these experts often sought additional input and review from colleagues in their companies, and gathered additional relevant information for the study. Workshop Process. A flowchart for the workshops is shown in Fig. 1. Individual, one-day workshops were conducted for each system during the first three phases (Workshops #1 to #3). The final workshop was held collectively over a two-day period. The activities conducted between workshops are also indicated on Fig. 1. The objective of Workshop #1 was to develop conceptual system descriptions for the four production systems. This work included establishing study boundaries in space and time and describing the physical and operational features for each system. Draft system descriptions were distributed to the workshop participants ahead of the workshop and then used as the starting point in the workshop. The objective of Workshop #2 was to perform hazard identifications for each system. A list of possible adverse events (or initiating events) that could contribute to risk was developed for each study system and organized by sub-system or activity. Detailed hazard identifications were developed through a series of specialty interviews with technical experts before Workshop #2. The participants and topics for these interview sessions are summarized in Gilbert et al. 2. These detailed lists were then reviewed during Workshop #2 and used to develop a framework for the quantitative risk assessment. The objective of Workshop #3 was to elicit quantitative information about frequencies and consequences for oil spills and fatalities. A preliminary quantitative risk assessment based entirely on raw data was distributed to the workshop participants before the workshop. This preliminary risk assessment was then refined during Workshop #3 and needs for additional information and studies were identified. The objective of Workshop #4 was to review and refine the risk assessment. Information from additional studies conducted between Workshops #3 and #4 was incorporated into this review.
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
The general work processes used for the workshops was as follows. Preliminary information that had been distributed to the participants was reviewed and refined through an open forum process that included time for discussion and developing a consensus regarding the input on risks. The open forum approach encouraged an iterative and synergistic discussion of risks and information from different perspectives. The participation of both the industry and the regulatory agencies helped to provide balance and objectivity in the discussions and input. Consensus was generally readily achieved, but when significant disagreement occurred between participants, votes were taken to achieve a consensus and dissenters’ opinions were recorded. The phased and progressive nature of the workshops provided the opportunity to seek and incorporate additional expertise and information as the study progressed and additional needs became apparent. Interim reports summarizing information from each of the first three workshops were distributed to participants after each workshop. These reports provided participants with opportunities for ongoing review and a means to ensure consistency in assumptions and approaches among the different systems. Finally, evaluations were conducted at the completion of the first three sets of workshops to continually improve the process. Risk Measures. Risk measures for the study systems were developed using the following criteria: • The measures of risk should provide relevant and useful input to MMS in their decision making process; • The measures of risk should be tractable and quantifiable; and • The measures of risk should be measures that are currently tracked and recorded so that (i) available data can be used to support the results of this risk analysis and (ii) future data can be used to validate and calibrate the results of this risk analysis. From these criteria, the risk measures listed in Table 3 were adopted for this study. The total number of fatalities is intended to measure the human safety risk. The volume of oil spilled is intended to measure the environmental risk. The environmental effects of an oil spill are not considered directly in this study because (1) there is a correlation between the magnitude of environmental damage and the volume of oil spilled; (2) environmental effects are difficult to measure and quantify; and (3) the environmental impacts of oil spills from FPSO’s are included in the scope of the EIS 1. The total volume of oil spilled in the 20-year lifetime is intended to measure chronic environmental risks. The maximum volume of oil spilled in a single incident is intended to measure acute environmental risks. The risk measures in Table 3 are practical simplifications that are intended to approximately capture the multitude of risks present. These measures of risk were not discounted with time. In addition, each measure was treated separately in comparisons and no attempt was made to combine them into a single
3
measure, such as equivalent cost. Descriptions of Study Systems. The following criteria, in order of decreasing importance, were used to develop conceptual descriptions for each of the representative study systems: 1. The study systems for the Spar, TLP, and Hub/Host Jacket should be typical of existing systems and technologies that are currently being used in the Gulf of Mexico because these systems and technologies have been approved and therefore represent acceptable risks. 2. The study system for the FPSO should be comparable to that already developed for the base case in the EIS1 in order to capitalize on the substantial effort devoted to developing this study basis. 3. The study systems should be as comparable to one another as possible so that differences in risks among them represent realistic differences among these types of systems and are not an unintended artifact of the study system descriptions. As an example of how these criteria were applied, consider a Spar. In order for the study Spar to be as comparable as possible to the study FPSO (criterion 3), which has oil transport via shuttle tankers consistent with the EIS 1 (criterion 2), the Spar would also have oil storage and oil transport via shuttle tankers. However, while this type of a Spar is possible, it is not typical of existing Spars in the Gulf of Mexico (criterion 1). Therefore, the description for the study Spar did not include oil storage and has oil transport via pipeline. The first step in the system description process was to establish a time frame for the risk assessment. The intent was to assess risks covering all aspects of offshore production including oil and gas production and processing offshore; drilling and well intervention during production; export of the oil and gas to shore; and transport of personnel to and from shore. The “lifetime” for a study system was defined to start when oil flows through the first production riser and end when the last well is shut in. For this study, a 20-year lifetime was used. Other phases in the actual lifetime for a system, such as construction, system installation, commissioning, decommissioning and system removal, were not included in this risk analysis. The second step in the system description process was to establish physical boundaries for the risk assessment. The study boundaries included the production facility, the pipelines and shuttle tankers used to transport oil to a shore terminal, and the supply vessels and helicopters used to support the production operations. These physical boundaries are shown schematically on Fig. 2. The third step in the system description process was to define the physical and operational attributes for each system. Detailed descriptions for these attributes are contained in Gilbert et al. 2, and the major attributes are summarized in Table 4. For the Spar, TLP and Hub/Host Jacket, operating experience from the Gulf of Mexico was directly drawn upon in developing the system descriptions. For the FPSO,
4
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
experience with tanker operations in the Gulf of Mexico was used together with operating experience for FPSO’s in other parts of the world, such as the North Sea and the South China Sea. It is important to note that these study systems represent typical or generic systems. Therefore, the range of risks associated with possible variations in hardware and operating practices is not captured in the results of this project. Quantitative Risk Assessment. The objective of the quantitative risk assessment was to quantitatively assess the risk measures listed in Table 3. These risk measures were quantified by estimating representative or average values for each study system. As an example, consider the total volume of oil spilled during the operational lifetime. If a fleet of Spars similar to the one defined in this study were installed and operated for 20 years in the Gulf of Mexico, then the total oil spill risk associated with this type of system would be the average value for the total volume of oil spilled from each Spar (that is, the sum of all the oil spilled from each Spar in its 20 year lifetime divided by the total number of Spars). Likewise, the average values for the maximum volume of oil spilled in a single incident from each Spar and the total number of fatalities on each Spar would represent the other measures of risk. Since there is an extremely limited experience base in the Gulf of Mexico for the types of production systems being analyzed in this study, it is not possible to obtain average values directly for the total number of fatalities, the total volume of oil spilled, and the maximum volume of oil spilled in a single incident. The goal of this study was to predict what the average values would be (the expected value) if each study system were hypothetically installed and operated in the future in the Gulf of Mexico. As with any prediction, there is uncertainty that the actual average value for each risk measure (obtained many years in the future) will be equal to the predicted value in this study. The range of possible values for the actual average was represented in this study by two quantities: the expected value and the standard deviation. The expected value represents the predicted value for the average, while the standard deviation represents the magnitude of uncertainty in the prediction. The expected value and standard deviation can be used to calculate confidence intervals for the prediction. For example, the 90 percent confidence intervals indicate that there is a ninety percent probability that the actual average will be within this interval. This section describes how the quantitative risk assessments were conducted through a process of developing preliminary assessments and then refining those assessments using the input of the technical experts. Preliminary Risk Assessments. Preliminary (or preworkshop) quantitative risk assessments played a very important role in this project because they were used to elicit quantitative information from the technical experts during Workshops #3 and #4 (Fig. 1). These preliminary risk assessments were developed to be objective, consistent and complete in order to maximize the value of the information
OTC
obtained from the technical experts during the workshops. The philosophy adopted in developing the preliminary risk assessments was to extrapolate directly from historical experience in the Gulf of Mexico to predict future performance. The primary data sources were the MMS3 and the USCG4. The methodology used to develop the preliminary risk assessments had the following steps: 1. Summarize Data for Sub-Systems: The data sets were first divided into sub-systems based on the hazard identification work in Workshop #2. These subsystems are listed in Table 5. The data for fatalities were then summarized as the total number of fatalities in the data record for each sub-system. The data for oil spills were further sub-divided into categories by the size of the spill, and then the number of incidents in the data record that had occurred for each spill-size category was compiled. The data for oil spills were divided into categories to facilitate the assessment since the range of spill volumes per incident covered five to six orders of magnitude and the frequency distribution for spill sizes was highly skewed. 2. Select Exposure Factors for Sub-Systems: The exposure for a risk is an indicator of the factors that influence the risk. In this way, the data can be extrapolated to each study system based on the exposure to the risk for that study system. The factors used to express the exposure for each subsystem category were selected based on the hazard identification information work in Workshop #2. These factors are listed in Table 5. 3. Estimate Frequencies of Occurrence for SubSystems: Estimates for the frequencies of occurrence for incidents (from Step 1) relative to the exposure factors (from Step 2) were developed using statistical methods that are described in Gilbert et al. 2. Both the expected value and the standard deviation for these frequencies were calculated. For the oil spill frequencies, it was assumed that a spill could occur in the next largest spill-size category above the maximum spill size observed in the historical data set. 4. Determine Sub-System Exposures for Study Systems: The exposure for each sub-system was determined from the system descriptions. 5. Assess Sub-System Risks for Study Systems: The estimated frequencies from the historical data (from Step 3) were then combined with the exposures for the study systems (from Step 4) to assess the subsystem risks. Both an expected value and a standard deviation were calculated for each risk measure (see Gilbert et al. 2 for details). 6. Assess System Risks from Sub-System Risks: The final step was to combine the information for the subsystem risks (from Step 5) to assess the total system risk (see Gilbert et al.2 for details). F inal Risk Assessments. The preliminary risk assessments
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
were then refined through the workshop process to develop final risk assessments. This process involved the following steps: 1. Start with data-based estimates that are as complete as possible (the preliminary risk assessments). 2. Evaluate the data sources and refine raw data sets as necessary so that they are relevant for predicting future performance of the study systems. As an example, the data set for oil spills from tankers in the Gulf of Mexico was limited to years after 1990 to account for the positive effects that the Oil Pollution Act of 19905 (OPA ’90) has had on recent performance and is anticipated to have on future performance. 3. Extrapolate predictions of future performance from the data set, applying corrections to the data-based estimates if necessary. As an example, the frequencies for small spills from subsea well systems were increased from the data-based estimates to account for differences between subsea well systems and the platform well systems that dominate the data set. 4. Account for all sources of uncertainty in the estimates, including the following: • the limited quality and quantity of relevant data records, especially for rare events; • the sometimes limited information available on the exposures corresponding to the data sets; and • the extrapolation of future performance from historical performance. 5. Document the whole process clearly and thoroughly. The detailed quantitative risk assessments for fatalities and oil spills are contained in Gilbert et al. 2.
Results The results from the quantitative risk assessment for fatalities and oil spills are presented, analyzed and discussed in this section. Risks for Fatalities. Results for the average total number of fatalities are shown on Fig. 3 for each study system. The results indicate that the fatality risks are very similar among the four study systems (Fig. 3). The expected contributions to the total fatality risk are shown on Fig. 4. Production and drilling and well intervention activities dominate the total fatality risk for all of the study systems. This result occurs because these activities require the bulk of the man-hours over a 20-year lifetime. The estimated frequencies of fatalities per man-hour worked for production and for drilling and well intervention activities (Gilbert et al. 2) are comparable to those reported for common industrial activities (AIChE6) and for the oil and gas industry throughout the world (OGP7). The contribution of drilling and intervention activities to the total fatality risk for the FPSO is not as large as for the other systems because all of the wells on
5
the FPSO are subsea wells that are subjected to less frequent well intervention compared to platform wells. Risks for Oil Spills. The results for the oil spill risks are presented and discussed in this section. First, the frequencies for different spill sizes are addressed. Next, the average total volume of oil spilled and the maximum volume spilled in a single incident over the lifetime are addressed. Frequencies of Spills. The frequencies of spills from production and transportation are first presented and discussed, and then the frequencies of spills from all sources are addressed. Frequencies of Production Spills. The annual frequencies for spills from production (Table 5) are shown on Fig. 5 for each of the study systems. Note that the frequency of spills tends to decrease as the spill size increases. Also, note that the relative magnitude of uncertainty in the estimated frequency increases as the spill size increases. This relative increase in uncertainty occurs because large spills are rare events, so there are few occurrences available from which to estimate frequencies. The information on Fig. 5 highlights the similarities and differences among the systems regarding oil spills from production. First, the Spar and the TLP are indistinguishable. This result is reasonable in that the elements of the designs on both systems that affect the potential for oil spills from production are nearly identical on these two study systems. Second, the Hub/Host Jacket tends to have smaller spill frequencies from production than the Spar, TLP and FPSO for spill sizes less than 1,000 bbl (Fig. 5). This difference is due to the smaller indigenous production rate on the shallow-water Hub/Host Jacket versus the deepwater floating production systems (Table 4). Third, the frequency of very small spills (less than 10 bbl) on the FPSO is less than that on the Spar and TLP, even though the production rates are similar on all of these study systems. This difference is due to the large deck area and the solid decking that exist on an FPSO; the deck would contain most small spills. Fourth, the frequency of spills between 100 and 10,000 bbl is slightly larger for the FPSO versus the other systems. This relative difference is because the FPSO has more subsea wells than the other systems; subsea wells were considered to have a higher leak frequency than platform wells because of a greater potential for sand erosion and cutouts due to high flow rates and detection difficulties for sand. In addition, the FPSO has a greater number of flowlines and flowline risers, which both contribute to the frequency of spills between 100 and 10,000 bbl. Fifth, the Spar and the TLP have the potential for very large spills (greater than 10,000 bbl), although the frequency for these spills is very small (Fig. 5). The potential source for these very large spills on the Spar and TLP is the dry tree risers. This risk does not exist on the FPSO study system because the trees that control the reservoir pressure and flow are on the seafloor (wet trees) rather than at the surface (dry trees), and it is negligible for the Hub/Host Jacket study
6
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
system because of the lack of movement for this non-floating system. A comparison with published information for the frequency of large spills from production is shown on Fig. 6. Anderson and LaBelle8 report a frequency for spills greater than 1,000 bbl in size. Their frequency was estimated using spill data from offshore platforms operating in the United States between the years 1974 and 1992. They report their frequency on the basis of the volume produced. In order to develop Fig. 6, this frequency has been converted to an annual frequency for the study systems using the total volume of oil produced in the 20-year lifetime for each system. The estimated frequencies for the study systems are less than the values obtained from Anderson and LaBelle8 (Fig. 6). There are two reasons for this result. First, two different data sets have been used. In the CRA project, data before 1990 were discarded due to the implementation of new regulations in 1990 (API RP14C9), which improved operating procedures on platforms. The Anderson and LaBelle data set extends back to 1974. Second, the CRA study systems are not representative of conventional, shallow-water platforms in the United States, which dominate the population of platforms in the Anderson and LaBelle data set. Note that the agreement between the CRA and Anderson and LaBelle is best for the Hub/Host Jacket study system (Fig. 6), which is most similar to the platforms in the Anderson and LaBelle data set. Frequencies of Transportation Spills. The annual frequencies for spills from transportation (Table 5) are shown on Fig. 7 for each of the study systems. The results highlight the similarities and differences among the systems regarding oil spills from transportation. First, compare the systems with pipelines. The Spar and the TLP are indistinguishable. This result is reasonable in that the elements of the designs on both systems that affect the potential for oil spills from transportation are nearly identical on these two study systems. The Hub/Host Jacket has slightly smaller spill frequencies from its pipeline than the Spar and TLP (Fig. 7). This difference is because there is a shorter length of pipeline exposed for the Hub/Host Jacket due to the shorter distance to the shore (Fig. 2). In addition, there is relatively less uncertainty in the estimated spill frequencies for the Hub/Host Jacket for spills less than 1,000 bbl (Fig. 7). The greater uncertainty for the Spar and TLP is mainly due to the potential for spills from the more flexible steel catenary export pipeline risers versus the more rigid risers on fixed jackets. The uncertainty for the Spar and TLP reflects that there are limited data concerning the performance of these risers in deepwater applications. Second, compare the two different types of transportation systems. There are notable differences between the pipelines for the Spar, TLP and Hub/Host Jacket and the in-field storage and shuttle tanker system for the FPSO. For very small spill sizes (less than 10 bbl), the frequency of spills for the FPSO is greater than from pipelines due to the potential for spills during offloading from hoses and valves. For spill sizes between 1,000 and 100,000 bbl, the annual frequencies of
OTC
spills for the shuttle tanker are smaller than the annual frequencies for pipelines (Fig. 7). One reason for this difference is that the potential for spills from the pipeline remains a constant as long as there is oil in the pipeline, regardless of the production rate. However, the potential for spills from the shuttle tanker will go down as the production rate decreases since fewer offloading events are required. Lastly, very large spill sizes (greater than 100,000 bbl) are not considered possible for pipelines due to operational and physical constraints (Gilbert et al. 2), while they are possible although infrequent for the FPSO. A spill between 100,000 and 500,000 bbl represents a major loss from the shuttle tanker due to a collision or explosion. A spill greater than 500,000 bbl represents a major loss from the FPSO due to a collision or explosion. A comparison with published information for the frequency of large spills from transportation is shown on Fig. 8. Anderson and LaBelle8 report frequencies for spills greater than 1,000 bbl in size from pipelines and tankers. Their frequencies were estimated using spill data from offshore operations in the United States between the years 1974 and 1992. They report their frequency on the basis of the volume produced. In order to develop Fig. 8, this frequency has been converted to an annual frequency for the study systems using the total volume of oil produced in the 20-year lifetime for each system. The estimated frequency for the Hub/Host Jacket is comparable to that from Anderson and LaBelle8 (Fig. 8). This result is reasonable since the pipeline from the Hub/Host Jacket is representative of the conventional, shallow-water pipelines that are in the Anderson and LaBelle data set. However, the estimated frequencies for the Spar and TLP study systems are less than those obtained from Anderson and LaBelle (Fig. 8). The primary reason for this difference is that the Anderson and LaBelle frequency for pipeline spills is proportional to the volume produced. However, the potential for spills from pipelines in the CRA study was related to the length of the pipeline and the time of exposure, not the volume of throughput. Therefore, the higher production rates for deepwater systems do not necessarily lead to proportionally higher spill frequencies for pipelines. The estimated frequency for the FPSO is lower than that from Anderson and LaBelle8 (Fig. 8). This result is due to the different data sets used to estimate the frequency. The Anderson and LaBelle data set extends back to 1974, and includes data from all U. S. coastal and offshore waters. In the CRA project, data before 1992 were discarded due to the implementation of the OPA ’903, which improved operating procedures on tankers and probably reduced the frequency of spills. Data for crude oil tankers in the Gulf of Mexico are summarized in Table 6 to support the hypothesis that data prior to the passage of OPA ’90 are not representative of existing conditions. In addition, data from outside of the Gulf of Mexico were not applied directly in the CRA project to estimate the shuttle tanker risk in the Gulf of Mexico. An analysis of tanker spills from 1992 to 1999 indicates that frequencies of spills between 50 and 5,000 bbl and of spills
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
greater than 5,000 bbl in the Gulf of Mexico are approximately 40 percent of those for the rest of the world (Gilbert et al. 2). Tanker spills are considered to be less likely on average in the Gulf of Mexico than in the rest of the world for the following reasons, in order of importance: 1. The regulatory environment in the Gulf of Mexico is more restrictive; 2. The environmental conditions in the Gulf of Mexico are less severe; 3. The consequences of grounding are significantly less due to the lack of rocky coasts in the Gulf of Mexico; 4. Shuttle tankers used in the Gulf of Mexico have a smaller parcel size on average; 5. The Gulf of Mexico has less congested waterways on average; and 6. Newer vessels are used in the Gulf of Mexico due to recent federal regulations. Frequencies of Spills from All Sources. The annual frequencies for spills from all sources, including production and transportation, are shown on Fig. 9 for each of the study systems. The frequencies for spills are generally dominated by production-related spills for spill sizes up to 1,000 bbl and by transportation-related spills for spill sizes greater than 1,000 bbl. Therefore, the similarities and differences among the study systems are related to those for production for spills less than 1,000 bbl (Fig. 5) and to those for transportation for spills greater than 1,000 bbl (Fig. 7). Note that the Spar and TLP are indistinguishable for all spill sizes. Total Volu me of Oil Spill ed over L if etime. Results for the average total volume are shown on Fig. 10 for each study system. These results indicate that the systems provide very comparable risks. The risk for the Hub/Host Jacket is slightly smaller than the risks for the other systems because it has a smaller production rate and a shorter transportation distance to the shore. The risks for all of the deepwater systems (Spar, TLP and FPSO) are nearly identical even though the frequencies for different spill sizes are not identical (Fig. 9). This result occurs because the risk is a measure of both frequency and consequence (spill size). While very large spills (greater than 100,000 bbl) are more likely with the FPSO than with the Spar or TLP, the annual frequencies are still small. Furthermore, the frequencies for spills less than 100,000 bbl for the FPSO are generally smaller than those for the TLP or Spar (Fig. 9). Therefore, the risks for the Spar, TLP and FPSO are comparable. In order to facilitate interpretation of the results on Fig. 10, the relative contribution of each spill-category to the total volume spilled is shown on Fig. 11. Note that the chronic environmental risk is dominated by large spills (greater than 1,000 bbl), which are low frequency but high consequence events. Therefore, most of the systems in a fleet of study systems will have small volumes of oil spilled. Occasionally, one of the systems may have a large spill and this large spill will dominate the average for the fleet. To emphasize this point, Table 7 summarizes the expected time between spills of different sizes for each of the study systems. Note that most of the risk for the Spar and TLP study systems comes from
7
spills between 10,000 and 100,000 bbl, which are only expected approximately once every 600 years of operation. Furthermore, most of the risk for the FPSO study system comes from spills between 100,000 and 500,000 bbl, which are only expected once every 4,500 years of operation. Table 7 and Fig. 11 show how the contributions to the risks for the Spar and TLP versus those for the FPSO are different even though the resulting risks are comparable (Fig. 10). One effect of the spill risk being dominated by rare, high consequence events is that the confidence intervals in the predicted average oil spill volumes range over nearly an order of magnitude (Fig. 10). This uncertainty reflects the typically limited quantity and quality of historical data available to estimate frequencies for rare events. Note that the confidence interval for the FPSO is wider than those for the other systems (Fig. 10) because there are relatively fewer data available for FPSO’s in the Gulf of Mexico and because the FPSO risk is dominated by very rare events with expected return periods of approximately 4,500 years. The contributions to the total oil spill risk from different sub-systems are shown on Fig. 12. Production, which dominates the smaller spill sizes (Fig. 5), does not contribute substantially to the total risk (Fig. 12). The main contributor to oil spills from production are related to the processing facilities (topsides on Fig. 12). Transportation, which dominates the larger spill sizes (Fig. 7), is the main contributor to the total oil spill risk (Fig. 12). M aximum Single Oil Spill in Li fetime. Results for the average single maximum spill are shown on Fig. 13 for each study system. The results indicate that the risks for the different study systems are comparable. Furthermore, these results emphasize that the maximum spill volume from a single incident dominates the average total spill volume. More than 70 percent of the total is expected to come from a single incident. The wide confidence intervals on Fig. 13 reflect the uncertainty inherent in estimating frequencies for rare events.
Conclusions and Recommendations A quantitative risk analysis was performed to assess and compare oil spill and fatality risks for four representative deepwater production systems in the Gulf of Mexico. Three of the study system types have already been operated successfully in the Gulf of Mexico: two floating production systems in deepwater with oil pipelines, a Spar and a Tension Leg Platform (TLP); and a shallow-water jacket serving as a hub and host to deepwater production. One of the study system types has not been used in the Gulf of Mexico: a tanker-based Floating Production Storage and Offloading (FPSO) system with oil transportation to shore via shuttle tankers. The objective of this analysis was to understand and compare the risks of the FPSO with those for currently acceptable alternatives for deepwater production. Conceptual system descriptions that are representative of existing and typical technology in the Gulf of Mexico were developed for the four systems. The scope of these descriptions included the entire production systems and
8
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
operations from the wells through the transport of product to the shore. Three risk measures were assessed and analyzed for each system: the total number of fatalities in a 20-year production life as a measure of the human safety risk, the total volume of oil spilled in a 20-year production life as a measure of the chronic environmental risk, and the maximum volume spilled in a single incident in a 20-year production life as a measure of the acute environmental risk. The process of developing the conceptual descriptions for the systems and then evaluating the risks has drawn on expertise from all facets of oil and gas production, including operators, contractors, manufacturers, class societies and regulators. Conclusions. The following major conclusions have been drawn from the results of this analysis: 1. There are no significant differences in the fatality risks among the four study systems. 2. There are no significant differences in the oil-spill risks among the four study systems. 3. The average total volume of oil spilled during the facility lifetime will be dominated by rare, large spills rather than frequent, small spills. 4. The major contribution to the oil spill risks for all systems is the transportation of oil from the production facility to the shore terminal by either pipelines or shuttle tankers. Spill risks for pipelines and shuttle tankers are comparable, although the frequencies and sizes of possible spills are different for pipelines versus shuttle tankers. The spill risks for pipelines are dominated by the possibility of spills between 10,000 and 100,000 bbl in size that are expected to occur once every 600 years on average. The spill risks for shuttle tankers are dominated by the possibility of spills between 100,000 and 500,000 bbl in size that are expected to occur on average once every 4,500 years. 5. The confidence intervals in predicted oil spill volumes range over about an order of magnitude, reflecting the limited quantity and quality of historical data available to estimate frequencies for rare events. Therefore, the expected risks associated with the FPSO are comparable to those for already accepted alternatives for deepwater production, including a Spar, a TLP and a shallowwater jacket serving as a hub and a host to deepwater production. Recommendations. The following recommendations have been developed from this work: 1. The results from this study should be periodically updated because they provide a valuable baseline for future analyses of risk in the Gulf of Mexico. The three measures of risk used in this study can all be readily measured and tracked for new and existing deepwater production facilities in the Gulf of Mexico.
OTC
2.
The quality of existing data sets for the Gulf of Mexico should be improved so that they are of greater value in future risk analyses. First, the type and quality of data that are currently collected should be evaluated, and any changes recommended from this evaluation should be implemented in a timely manner. Second, single agencies should be responsible for tracking and compiling similar types of data. Third, all data records should be reviewed annually by the industry and regulators to improve the clarity, quality and usefulness of the information in these records. Finally, the data should be published annually in a clear and an easily accessible format. 3. Additional information about the populations of offshore facilities and operations in the Gulf of Mexico should be collected on an annual basis. Specifically, the following information from federal and state waters in the Gulf of Mexico would be valuable: the length of active pipelines operating per year, the number of tanker on-loading and offloading events in ports and lightering zones per year, and the number of man-hours in production-related activities, supply vessel operations and tanker operations per year. 4. Uncertainty in the predicted performance for these four study systems should be considered carefully in drawing conclusions from and applying the results from this study. 5. The process used on this project to assess risks has been effective in obtaining valuable technical information from industry and regulators, and should be considered in supporting other analyses of new technology in the Gulf of Mexico.
Acknowledgments The authors wish to acknowledge the Minerals Management Service for funding this project. Specific individuals at MMS who have been instrumental in this study are Paul Martin, Charles Smith, James Regg and Cheryl Anderson. In addition, the following individuals have played key roles in the project: Professor Larry Lake and Research Assistant Jihad Jaber, The University of Texas at Austin; Joe Gebara, EQE International; Lieutenant Commander Bill Daughdrill and Joe Myers, United States Coast Guard; and Allen Verret, Private Consultant. Finally, the Deepstar Consortium has been instrumental in coordinating and facilitating the involvement of industry experts on workshop teams. References 1. MMS (2000), “Proposed Use of Floating Production, Storage, and Offloading Systems On the Gulf of Mexico Outer Continental Shelf, Western and Central Planning Areas, Draft Environmental Impact Statement,” Prepared Under MMS Contract 1435-01-99-CT-30962, Minerals Management Service, Gulf of Mexico OCS Region. 2. Gilbert, R. B., Ward, E. G. and Wolford, A. J. (2001),
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
3.
4.
5. 6.
“Comparative Risk Analysis for Deepwater Production Systems,” Final Project Report, Prepared for Minerals Management Service, Washington, D. C. MMS (2000), “MMS OCS Spill Database,” Minerals Management Services, Available: [email protected], Accessed: January, 2000. USCG (1999), “Marine Casualty and Pollution Database,” CD ROM, Subscription Order No. 5441INC, Available: National Technical Information Services, Springfield, Virginia. Oil Pollution Act of 1990 (OPA ’90), 33 USCA Sec. 2701-2761. AIChE (1989), Chemical Process Quantitative Risk Analysis, Center for Chemical Process Safety, American Institute of
Chemical Engineers, New York, New York. 7. OGP (1999b), “Safety Performance of the Global E&P Industry, 1998,” Report No. 6.80/295, International Association of Oil and Gas Producers, London, England. 8. Anderson, C. M. and LaBelle, R. P. (1994), “Comparative Occurrence Rates for Offshore Oil Spills” Spill Science and Technology Bulletin, Vol. 1, No. 2, 131-141. 9. API RP14C (1998), “Analysis, Design, Installation and Testing of Basic Surface Safety Systems on Offshore Production Platforms,” Sixth Edition, American Petroleum Institute, Washington, D.C.
Table 1 - Industry sources for workshop participants. Oil Companies BP Amoco Chevron Conoco Elf ExxonMobil Marathon Oxy Shell Statoil Texaco
Consultants & Contractors EQE ABB Atlantia FMC Paragon McDermott Navion
Classification Societies ABS Lloyd’s Register DNV
Table 2 - Additional industry sources for technical expertise. Skaugen Petrotrans SBM IMODCO Global Maritime Aker R&B Falcon Transocean SedcoForex Edison Chouest Tidewater Marine HSAC Air Logistics PHI ERA Aviation
Association of Diving Contractors Oceaneering Cameron Mentor Bay Ltd. Spirit Energy Horizon Engineering Kerr McGee Mathews Daniels LOOP State of Louisiana
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. Water Depth (ft) Peak Production Oil (bopd) Gas (scfpd) Export Oil (bopd) Gas (scfpd) Wells Platform Subsea (MODU) Manning Production Marine Drilling – Platform Drilling MODU
9
Spar 4,000
TLP 4,000
Hub/Host Jacket 600
FPSO 5,000
150,000 200,000
150,000 200,000
50,000 50,000
150,000 200,000
150,000 200,000
150,000 200,000
250,000 550,000
150,000 200,000
6 3
6 3
6 3
0 9
30-45 6 65 65
30-45 6 65 65
30-45 0 50 65
30-45 10 0 65
10
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
OTC
Table 5 - Sub-system categories used in risk assessment. Risk Measure
Exposure Factor
Sub-System Category Production Drilling Supply Vessels Helicopter Transport Tanker Operations Major Accident Well Systems – Platform (or Surface) Well Systems – Subsea Dry Tree (or Production) Risers Flowlines Import Flowline Risers Topsides Supply Vessels Drilling and Intervention Pipelines Export Pipeline Risers Shuttle Tanker (Offloading in Field and at Port) FPSO Cargo Tank
Fatalities
Production System Oil Spills
Transportation System
man-hours man-hours docking calls passengers docking calls platform-years bbl produced bbl produced riser-years mile-years riser-years bbl processed docking calls man-hours mile-years riser-years docking calls platform-years
Table 6 - Summary of Oil Spills from Crude Tankers in Gulf of Mexico 1
Number of Spills Year
1-10 bbl
10-100 bbl
100-1,000 bbl
1,000-10,000 bbl
10,000-100,000 bbl
1991
0 2 4 5 3 7 5
1 1 0 1 1 2 0
0 0 0 0 2 1 0
0 0 0 0 0 0 0
0 0 0 1 0 0 0
30 28 5 15,401 1,146 266 17
Sub-Total
26
6
3
0
1
16,893
1992
0 2 2 0 0 1 1 1
0 1 0 0 0 0 1 0
0 1 0 1 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 191 8 179 0 2 22 9
7
2
2
0
0
411
1985 1986 1987 1988 1989 1990
1993 1994 1995 1996 1997 1998 1999 Sub-Total 1
Volume Spilled (bbl)
4
Note: Data from USCG .
Table 7 - Expected return periods for spills. System
1 – 10 bbl
10 – 100 bbl
Spar TLP Hub/Host Jacket FPSO
0.8 0.8 3 3
3 3 8 3
Expected Return Period between Spills (years) 100 - 1,000 1,000 - 10,000 10,000 - 100,000 100,000 - 500,000 bbl bbl bbl bbl 15 60 580 Not Credible 15 60 580 Not Credible 35 91 920 Not Credible 12 110 2,500 4,700
500,000 - 1,000,000 bbl Not Credible Not Credible Not Credible 300,000
1
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
Draft System Descriptions
Workshop #1 System Definition Develop System Descriptions
Develop Preliminary Event/Outcome Tables Gas Flange r (Port Terminal)
Workshop #2 Hazard Identification Elicit Event/Outcome Information Conduct Preliminary QRA
r
l Hub/Host
Workshop #3 Quantitative Risk Analysis Elicit Frequency/Consequence Input
Workshop #4 Review Review and Refine QRA
Gas Pipeline
Oil Pipeline
Refine QRA & Perform Additional Studies
Fig. 2 - Physical layout for study systems (plan view).
3.0
0.5
Shuttle Tanker
l l TLP FPSO Spar
Prepare Final Report
Fig. 1 - Flowchart for workshop process .
s 2.5 e i t i l a t a F 2.0 f o r e e m b i t m e u f i 1.5 N L l n a i t o T 1.0 e g a r e v A
Oil Flange (Port Terminal)
90% confidence intervals
0.0 Spar
TLP
Hub/Host Jacket
FPSO
System
Fig. 3 - Average total number of fatalities in lifetime .
11
12
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
OTC
Spar/TLP
Production Drilling and Intervention Supply Vessels Helicopter Transport Tanker Operations
Drilling and Intervention (52%)
Major Accident
FPSO
Hub/Host Jacket
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. ) r 1.0E+01 a e y r e p 1.0E+00 ( s l l i p S 1.0E-01 n o i t c 1.0E-02 u d o r P r 1.0E-03 o f y c n 1.0E-04 e u q e r F 1.0E-05 l a u n n A 1.0E-06
Spar TLP Hub/Host Jacket FPSO
Expected value
90% confidence interval
0 1 1
0 0 1 0 1
0 0 0 , 1 0 0 1
0 0 0 , 0 1 0 0 0 , 1
0 0 0 , 0 0 1 0 0 0 , 0 1
0 0 0 , 0 0 5 0 0 0 , 0 0 1
0 0 0 , 0 0 0 , 1 0 0 0 , 0 0 5
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.
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
0 0 0 , 1 n a h t r e t a e r G s l l i p ) S r a n e y o i t r c e u p d ( l o r b P b f o y c n e u q e r F l a u n n A
13
1.0E+00
1.0E-01 Anderson and LaBelle (1994)
1.0E-02
1.0E-03
CRA (90% confidence intervals)
1.0E-04
1.0E-05 Spar
TLP
Hub/Host Jacket
FPSO
System
Fig. 6 - Comparison of production spill (>1,000 bbl) frequencies with published data. CRA denotes the results from this study. ) r a 1.0E+01 e y r e p ( 1.0E+00 s l l i p S n 1.0E-01 o i t a t r 1.0E-02 o p s n a r T 1.0E-03 r o f y c 1.0E-04 n e u q e r 1.0E-05 F l a u n 1.0E-06 n A
Spar TLP Hub/Host Jacket FPSO
Expected value 90% confidence interval
0 1 1
0 0 1 0 1
0 0 0 , 1 0 0 1
0 0 0 , 0 1 0 0 0 , 1
0 0 0 , 0 0 1 0 0 0 , 0 1
0 0 0 , 0 0 5 0 0 0 , 0 0 1
0 0 0 , 0 0 0 , 1 0 0 0 , 0 0 5
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.
14
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
n a 1.0E+00 h t r e t a e r G 1.0E-01 s l l i p S ) n r o a e i t y a r 1.0E-02 t r e o ( p p l s b n b a r T 0 0 1.0E-03 f 0 , o 1 y c n e u q 1.0E-04 e r F l a u n n A 1.0E-05
OTC
Anderson and LaBelle (1994)
CRA (90% confidence intervals)
Spar
TLP
Hub/Host Jacket
FPSO
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 TLP Hub/Host Jacket
1.0E+00 ) r a e y r e p ( y c n e u q e r F l a u n n A
FPSO
1.0E-01 1.0E-02 Expected value 1.0E-03 90% confidence interval
1.0E-04 1.0E-05 1.0E-06 0 1 1
0 0 1 0 1
0 0 0 , 1 0 0 1
0 0 0 , 0 1 0 0 0 , 1
0 0 0 , 0 0 1 0 0 0 , 0 1
0 0 0 , 0 0 5 0 0 0 , 0 0 1
0 0 0 , 0 0 0 , 1 0 0 0 , 0 0 5
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.
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
10000 90% confidence intervals 9000 8000 d ) e l l l i b p b ( S e l i m O i f t e o f i e L m r u e v l o o V
Average Total
7000 6000 5000 4000 3000 2000 1000 0 Spar
TLP
Hub/Host Jacket
FPSO
System Fig. 10 - Average total volume of oil spilled over lifetime – all sources.
n i d 100% e l l i p S e 80% m u l o V l a t 60% o e T i m e t e g f i a r L 40% e v A o t n 20% o i t u b i r t n 0% o C
Spar
0 - 1 1
0 0 - 1 1 0
0 1 0
TLP
Hub/Host Jacket
FPSO
0 0 0 0 0 0 0 0 0 , 0 , 0 , 0 , 0 0 0 0 0 0 5 0 0 - 1 , - 1 0 1 0 0 0 0 0 0 0 0 0 1 , , , 0 0 0 1 0 0 , 1 0 0 5
0 0 0 1 , -
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.
15
16
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
Well Systems - Platform Well Systems - Subsea Dry Tree Risers Flowlines Import Flowline Risers Topsides Export Pipeline Risers
Spar/TLP
Pipelines Shuttle Tanker FPSO Cargo Supply Vessels Drilling and Intervention
Pipelines (72%)
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 9000
90% confidence intervals
8000 d ) e l l l i b p b ( S e l i m O i f t e o f i e L m r u e v l o o V
Average Total
7000
Average Single Maximum
6000 5000 4000 3000 2000 1000 0 Spar
TLP
Hub/Host Jacket
FPSO
System
Fig. 13 - Average maximum volume spilled from a single incident in the lifetime .
OTC
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 transportation 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.
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 Reliabili bility Methods for De Deepwa pwater Posi Positiontion-M Mooring ooring 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 This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, 30 April–3 May 2001. This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
Abstract FPSOs and other floating offshore facilities typically follow “prescriptive based” classification rules for design of the hull, mooring and marine systems. In some cases the process facilities are also classified. An alternative approach is to use a more open framework, “risk based” design approach that allows variation from prescriptive rules provided system risks are maintained at acceptable levels. The various classification societies currently allow such risk-based alternatives [2, 3]. Although numerous “component” risk studies for FPSOs have been conducted and published, this is one of the first that accounts for the integration and linking of risks and risk tradeoffs among the hull, mooring system, marine systems, topside process plant and the utility, power and control systems that support them. The model and basis are first described, followed by application to a prototype deepwater, turret moored FPSO with gas handling. Example cases are shown to demonstrate use of the model to make design decisions to the various FPSO components. Introduction Classification rules are established based on engineering principles, experience, testing & expert judgement. They are intended to ensure probabilities of accidents are low, but this is not explicit. Changes to developing and implementing ABS rules are being explored through risk based approaches. An alternative, risk-based approach to classification of Floating Production, Storage and Offloading systems is being investigated by ABS as part of a major internal technology development project. The project is comprised of model, database and methodology development and training at multiple levels throughout the organization. The prototype model was completed early in 2001 and is the focus of this
paper. The model represents systems failures more comprehensively than any other offshore risk assessment known to the authors. This level of detail was sought in order to achieve specific goals: 1) Develop explicit risk measures of class rules to facilitate prioritization and optimization, thus allowing one to focus resources on the greatest risk contributors, 2) Develop a consistent means for performing risk tradeoffs (i.e. demonstration of equivalent level of safety), 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.
Model Development The overall model is comprised of a collection of initiating events, facility response model and consequence calculations. These are categorized into discrete damage states (see Figure 1). Also illustrated in Figure 1, the facility response model is comprised of support event trees, frontline event trees and end states. The support event trees represent the failed/operational state of support systems (e.g. utilities, instrumentation and control, emergency) required for successful operation of the frontline (main system) event trees. The end states are simply a discrete categorization of the various failed configurations of the facility. Frontline systems were segregated into three categories for convenience: process, marine and structural systems. Each of these model partitions are described in summary in Table 1, and the development activities are described in the following sections. Process Systems The process model follows a conventional offshore QRA approach [1,4]: • Development of isolatable sections, • Summarize the loss of containment frequency by using a parts count approach • Identifying spatial interactions that could lead to escalation These steps were completed for 54 isolatable sections, using the event tree structure of Figure 2. Three hole sizes were selected to represent the hole size distribution of various process equipment. Multi-phase releases (oil / gas / water) were treated where applicable. Leak frequencies were derived primarily from generic databases (e.g. E&P Forum, OREDA, Offshore Hydrocarbon Release Statistics). The model
2
WOLFORD ET AL.
explicitly accounts for emergency detection as well as process control response to a loss of containment event, so that risk importance measures can be associated with rule and standardrequired I&C. Marine Systems The marine modeling followed a broadly similar approach to the process modeling, except the scope of marine events was broader than loss of containment of hydrocarbons. The majority of the marine event trees addressed fires (fueled, electrical, other). Marine event scenarios were represented with 89 unique initiating events, 12 frontline system event trees, one support tree and 141 marine fault trees of which 89 developed specific initiating events and 52 modeled system response functions (see Table 1). Over 2 billion unique event sequences were evaluated. Fire initiating event frequencies were developed for 70 individual hazard zones combined with an assessment of initiator density. An explicit parts count approach was not utilized for marine fire initiation based on unavailability of component-specific fire initiator frequency data. Structures The modeling of structural failures also followed a broadly similar approach, but deviated from developing a componentwise model and relied substantially upon subject matter experts to identify, prioritize and structure the event sequences. The approach is illustrated in Figure 3. As shown in that figure, structure subsystems were broken down into mooring, turret, topside structures and hull. Team meetings were held to construct event sequence diagrams – which illustrate the sequential (hardware and human) event sequence of events resulting in a system failure. The event sequence diagram has proven to be a superior tool to portray and elicit necessary expert input. An example is shown for mooring in Figure 4. From this intermediate step, system event trees were developed analogous to those for the process and marine frontline event trees. The mooring event tree is shown in Figure 5. Structural event scenarios were represented with 46 unique initiating events, 13 frontline system event trees and one support tree. Consequence Modeling Risk metrics were defined to measure health, safety, environmental and financial impacts from the FPSO. The endpoint metrics selected were: fatalities, oil spill, capital loss and business interruption (the latter two only if associated with a scenario with potential for fatality or spill, i.e. the entire scope of financial impacts ere not included in this phase of the model, only a subset). The physical phenomena represented in this model drew upon de facto standards used in offshore QRA [1]. • Release modeling, multi-phase, near field flow regime, internal pressure-time history • Thermal radiation effects to humans and equipment from jet fires and pool fires
• •
OTC 12948
Explosion overpressure impacts to humans and equipment Simple evacuation of personnel on board.
Example Applications Two applications are used to highlight the utility of the model described in this paper: • Comparison of loss of containment frequency for a production separator among material failure mechanisms vs. transient induced leaks • Comparison of the design of mooring line redundancy requirements Leak Initiators Various codes and rules [2,3] specify the use of process instrumentation to protect against accidental overpressure of hydrocarbon containing equipment. These prescriptive rules are effective in minimizing overpressure transients that can potentially lead to significant loss of containment. As such, these measures provide non-explicit risk mitigation of offshore production facilities. A comparative exercise was performed to determine the extent to which individual instrumentation and control component should be represented in the risk model. The exercise formulated a fault tree model of a prototypical production separator, and quantified the transient induced leak frequency, for major leaks, and compared to the historical, “inherent” leak frequency. The fault tree is shown in Figure 6. The results, using point estimates only, show:
Inherent Leak Frequency Transient Induced Leak Frequency
All Leaks: 0.42/yr Major Leaks 0.04/yr Major Leaks: ~1E-8/yr
The result of the above exercise shows that the transient induced leak frequency is substantially lower than the inherent leak frequency, hence the need to treat the individual equipment failures explicitly is unwarranted, unless a process system departs radically from existing codes and standards. Mooring Line Redundancy An 8-leg, external turret mooring system was evaluated with three hypothetical design criteria • No redundancy • Single line failure (per ABS Rules) • 2 Line failure Parameters were modified in the FPSO risk model to incorporate modifications in failure rates and consequences. The mooring event tree is shown in Figure 5. Frequencies were tabulated for the four risk metrics. Figure 7 shows that substantial risk reduction can be achieved using a dual line failure criterion.
Summary A comprehensive risk model has been constructed to assist in FPSO design risk tradeoffs. This model is envisioned to be useful in supporting risk based classification activities.
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
References 1.
2. 3. 4.
5.
6.
7.
“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. Guide for Building and Classing Facilities on Offshore Installations, American Bureau of Shipping, June 2000. Guide for Building and Classing Floating Production Installations, American Bureau of Shipping, June 2000. Guidance Notes on Risk Assessment Application for the Marine and Offshore Oil and Gas Industries, American Bureau of Shipping, June 2000. “Hydrocarbon Leak and Ignition Database”, Report 11.4/180, DNV Technica, prepared for E&P Forum, June 1992. “Hydrocarbon Release Statistics Review”, David Mansfield, AEA Technology, A report produced for UKOOA, January 1998. Offshore Reliability Data, OREDA Participants, distributed by DNV Technica, 1993.
3
4
WOLFORD ET AL.
Table 1 Description of Model Partitions System Category
Initiating Events
Event Trees
•
• •
Process
•
• • Marine
• Structura l
Total
171 Loss of Containment Fault Tree modeling each of the 171 initiating events (Parts Count) 57 Escalation Events 89 (70 fires by zone)
46 Structural Damage
363 Initiating Events
Comprised of 54 process section Each of the 2 phases in 3 separators is modeled as a separate initiating event • 3 Hole sizes (small, medium, large) • Model Functions Ø Ignition/Explosion Ø Isolation Ø Blowdown Ø Fire Suppression • Failure of cargo management • 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 • Single mooring line failure • Corrosion holes or fatigue crack in turret 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 17 Billion individual event sequences
OTC 12948
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
Table 2 Top Events for Mooring Trees
TOP EVENT
IE MOORL OFST MOORS STATN RSSTR RSLOC TUG GRND VSSL FIXD
5
Table 3 Top Events for Process Tree TOP EVENT IE IGNE PSL LSL PISO GDET IGNL FDET ESDO ISOL BLDN FWSP FOAM
DESCRIPTION
Initiating Event Single mooring line fails Vessel drift exceeds design limits Mooring system fails (Loss of 2 or more lines) Loss of station Production risers experience stresses greater than ultimate limits Riser cracking occurs breaching containment Tug not available for recovery operator Vessel runs aground Vessel impacts passing vessel Vessel impacts fixed installation
DESCRIPTION Initiating event Immediate ignition Failure of PSL for isolation Failure of LSL for isolation Failure of isolation by Process PSL/LSL Gas detection system Late ignition Fire detection Manual actuation of ESD and blowdown Failure of isolation Blowdown valves Water spray deluge suppression Foam/water fire suppression
Figure 1. Overall FPSO Risk Model Structure
Initiating Event I.E.1 I.E.2
Support Event Tree
Frontline Event Tree
End States E1 E2
• •
Consequence Model
Facility Response Model
S1 S2
• • •
F1 F2
•
•
• • •
X1 X2 Consequence Analysis
• • • • • •
• •
•
• •
•
•
• •
•
I.E.n
Xl
Em
S1
System
Component
Failure Mode
Generic Into Population Data Facility-Specific Data
Damage
ø
Figure 2 Process Loss of Containment Event Tree
Figure 3 Structural Failure Modeling Approach
Mooring System
System Familiarization
Subject Matter Expert
System Breakdown
Subject Matter Expert
Turret System
Topside Structural System
Moderated Meetings
Hull System
Subject Matter Expert Subject Matter Expert
Build Event Sequence Diagrams
Build Event Tree Diagrams Frequency Input Quantification End State Assignment
8
WOLFORD ET AL.
OTC 12948
Figure 4a Mooring Event Sequence Diagram
Single Mooring Line Failure
No
No adverse effect
No
Shut-in Production
Yes
Design Offset Exceeded
No
Yes Initiate Repairs
Yes
Mooring System Failure
No
Overstress Risers
Yes Loss of Station
Overstress Risers
No
Shut-in Production
No
Inititaite Repairs
No
Inititaite Repairs
Yes Initiate Repairs
Yes
Release well fluids
No
Shut-in Production
Yes Initiate Repairs
Yes
Shut-in Production
Yes Initiate Repairs
No
Inititaite Repairs
Inititaite Repairs
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
Figure 4b Mooring Event Sequence Diagram
Loss of Station
No
Overstress Risers
Yes
Risers Severed
No
Overstress Risers
Release well fluids
Shut-in Production
Vessel Drift
Controlled (not available for extreme weater events)
Uncontrolled Grounding
Loss of Asset
Polution
Boat Impact
Loss of Asset
Polution
Impact with other offshore installation
Loss of Asset
Polution
Vessel Recovery
Initiate Repairs
Vessel Recovery
Initiate Repairs
9
10
WOLFORD ET AL.
OTC 12948
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
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
Figure 6a Transient Induced Leak Frequency Fault Tree
11
12
WOLFORD ET AL.
Figure 6b Transient Induced Leak Frequency Fault Tree
OTC 12948
OTC 12948
INTEGRATED RISK BASED DESIGN OF FPSO TOPSIDES, STRUCTURAL AND MARINE SYSTEMS
13
Figure 7 Mooring Redundancy Risk Results
Minimum Consequences 1E+00
Fatalities = None Oil Release = 0 barrels Repair Cost = $10-100 Thousand Downtime = None
1E-01 1E-02 1E-03 y c n e u q e r F
1E-04 1E-05 1E-06 1E-07 1E-08 1E-09 1E-10 None Single Line (ABS Rules)
Two Lines (BP Proposal)
Mooring Line Redundancy
Maximum Consequences
Fatalities = 1 Oil Release = 30 - 40,000 barrels Repair Cost = $3-10 Million
1E+00
Downtime = 1 - 3 months
1E-01 1E-02 1E-03
Low Value Facility (well protecter)
y 1E-04 c n e u 1E-05 q e r F 1E-06
Typical Offshore Facility (fixed platform)
1E-07 1E-08
High Value Facility
1E-09 1E-10 None Single Line (ABS Rules)
Two Lines (BP Proposal)
Mooring Line Redundancy
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 transportation 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.
J AMES C . LIN James Lin is a Senior Consultant at EQE International with over 17 years of experience in system engineering, reliability and availability analysis, and probabilistic safety assessment. Mr. Lin has a wide range of experience that includes acting as a project manager, performing human reliability analysis, and analyzing 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 from National Tsing Hua University, Taiwan and his M.S. in Nuclear Engineering from University of California, Los Angeles. He is a Registered Nuclear Engineer, California and a Certified Reliability Engineer, American Society of Quality Control.
ROBERT E. SHEPPARD (picture unavailable) Robert Sheppard, a Principal Engineer with EQE International, has over twelve years of experience in structural engineering and reliability analysis. Mr. Sheppard has been involved in projects across many industries including offshore oil, onshore petrochemical, nuclear 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.
J AMES K. LIMING James K. Liming is a Senior Reliability and Risk Management Consultant and Corporate Associate with over 17 years of experience in managing and performing complex engineered facility reliability engineering, risk analysis, and operations and maintenance support. He has a diverse, wellbalanced background including handson power plant operating and maintenance experience as well as extensive analytical expertise. He has served as project manager or project engineer on several major industry and government risk management projects worldwide, and is noted as a leading practitioner of probabilistic risk 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 and technical papers on risk and reliability analysis applications tools and techniques. He has authored or coauthored over 70 publications on risk and reliability analysis methods and applications. He is a former fully qualified U. S. Navy nuclear submarine officer, and he is currently a Captain (O-6) and Naval Sea Systems Command unit Commanding Officer in the U. S. Naval Reserve. He holds a B. S. degree from the U. S. Naval Academy and an S. M. degree in nuclear engineering from the Massachusetts Institute of Technology (MIT).
ANDY LIDSTONE (picture unavailable) Andy Lidstone is a Senior Engineer with EQE International with over fifteen years of industry experience for offshore and onshore oil and gas facilities and chemical plants. Mr. Lidstone’s areas of expertise include the preparation of hazard assessment reports, HSE Cases and probabilistic safety assessments and reliability analysis. Mr. Lidstone is equally comfortable as a teacher / lecturer of Safety Case materials or as a project manager for the development of Failure Modes and Effects Analyses (FMEAs) for a full range of drilling structures. Mr. Lidstone earned a B.Sc. (Hons) in Physics from the University of Salford, England and is a Chartered Physicist and member of the Institute of Physics.
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 Copyright 2001, Offshore Technology Conference This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, April 30–May 3, 2001 This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
• • • • •
Determination of condition of structural elements and structural system, Disclosure of defects (design, construction, operation, and maintenance), Assurance of conformance with plans, specifications, guidelines, rules, and quality requirements, Disclosure of damage, and Development of information to improve design, construction, operation, and maintenance procedures.
FPSO inspections have several levels of intensity:
ABSTRACT The increase in deepwater exploration activity has generated increased use of the Floating Production /Storage Offloading Systems (FPSO′s). Converting existing tankers is, in many cases, more economically feasible and faster than building new FPSO′s for the same purpose. For conversion of tankers to FPSO′s, inspection and subsequent fitness for purpose assessment are crucially important. The objective of this paper is to present the principles and strategies of in-service inspection programs for FPSO′s. The paper summarizes the technical basis for three levels of inspection strategies: 1) probability-based inspection method, 2) risk-based inspection method, and 3) ‘optimum’ inspection method.
INTRODUCTION FPSO′s has many attractive features including relative low cost, large working area, large water surface area, and good stability and floatability. FPSO′s in many cases, are converted from existing tankers. It is therefore important to have a rational and reliable inspection method to provide information and knowledge concerning the proposed, present, and future integrity of FPSO′s. FPSO inspections should be focused on:
• • •
General (global conditions), Specific (basic aspects of defects and damage), and Detailed (precise descriptions of flaws and other items of operation and maintenance concern).
FPSO inspections should be life-cycle oriented and include quality assurance and control measures in:
• • • • •
Design, Construction, Operation, Maintenance, and Accidents / casualties.
FPSO inspections should be full-scope and include quality assurance and control measures of the structure, equipment, facilities, and personnel. Research in inspection has been conducted in regulatory organizations, universities, and other leading organizations [1, 2, 3, 4 ]. Some important guidelines have been developed for offshore inspection practices. The objective of this paper is to present the principles of development of in-service inspection strategies for FPSO ′s.
PROBABILITY BASED INSPECTION A number of limitations of in-service inspections have been identified [3, 4, 5], especially the significant uncertainties in design, fabrication and damage detection, as well as the 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
deal with uncertainties, as well as Bayesian models has been recognized. The probability-based inspection method is thus developed to include Bayesian analysis, and probabilitybased inspection planning.
Bayesian Analysis The bulk of research on Bayesian analysis in engineering application was first conducted in early 1970’s. This approach, defined as the probability of detection (POD) updating method, was widely applied in aeronautical engineering systems, such as airframes, gas turbine engines [6]. In the late 1980s, an alternative Bayesian approach was developed as event updating using First Order Reliability Method (FOSM) in the offshore industry [7]. Itagaki et al [8] developed Bayesian estimation approach in ship structures based on Bayesian point estimation method.
Event Updating
POD Updating The POD updating is based on Bayesian approach to update of the multi-variant probability distributions, see Fig. 1. The updated multi-variant probability distribution is used to recalculate the failure probability. Assuming that the defect size distribution is F A(a) and that the probability of detection of a is P D(a). By means of Bayes' theorem (Bayes form):
P[p|q r H] • P[q r | H]
P[q r | pH] =
n
∑
P[q i | H] • P[p| q i H]
(5)
i= 1
The updated probability density can be determined as follows:
q r | H : a ≤ A ≤ a + da ,
Let
i.e.
P[q r | H] = f A (a )da , and p | q r H : inspection resulting in no crack detection:
This approach is to update the probability of events, such as fatigue failure directly. Bayes theorem is applied here to update the failure probability conditioned on additional events, such as inspection event. It is expressed as
P (q r | pH ) =
OTC 12949
P ( qr | H ) P ( p | H )
f A,up (a) =
[1 − PD (a)] • f A (a)
∞
∫
(1 − PD (a)) • f A (a)
(7)
0
The original safety event q r|H can be formulated by using the critical size (e.g., critical crack size for fatigue or critical thickness reduction for corrosion) as the failure criteria
−a ≤ 0
(6)
and the updated probability density is:
(1)
where, qr|H - The original safety event , q r|pH - The fatigue safety event after inspections, p|H - inspection event
q r | H = a c
P[ p | q r H] = 1 − PD (a )
(2)
The same methodology can be applied in the crack detection inspection where the updated probability density is :
f A,up (a) =
∞
f A (a) • PD (a)
∫
f A (a) • PD (a)da
(8)
0
Bayesian Estimation
The critical size is selected, perhaps based on serviceability consideration. Other safety event such as brittle fracture or corrosion threshold can be formulated.
This approach applies Bayesian approach in statistical parameter (mean, standard deviation) estimation, see Fig. 2.
The inspection event p|H is formulated to describe the additional information from inspections. Two types of event margins are classified based on inspection results. p | H = a ( N i ) − A d ≤ 0 (3)
The statistical parameter to describe the distribution of a random variable X in the fatigue problem is defined as the random variable µ . The Bayesian estimation of µ is to use the extended Bayes theorem in probability density as:
and
p | H = a ( N i ) − A = 0
(4)
Equation (3) is the event that a defect size a(N i) is not detected by an in-service inspection with the crack delectability limit A d at the inspection time N i. Ad is described by the probability of detection (POD). Equation (4) is the event that a particular defect size A is detected and measured at the inspection time N i.
f " (µ) = cL(µ)f ' (µ)
(9)
'
Where f (µ) = the prior distribution, L(µ ) = the likelihood function of the observed data (objective information), c= a "
normalizing factor, and f (µ ) = the posterior distribution incorporating the objective data. For a sample of observations x i , i=1,2,..., of X from inspections, the likelihood of the sample is proportional to the product of the probability densities of X at x 1, ...., x n :
OTC 12949
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
•
n ' f (µ ) ∝ ∏ f x (x i | µ) f (µ) i=1 "
(10)
The updated probability distribution of X is thus expressed as:
∫
f x (x) = f x|µ (x | µ )f (µ)dµ "
(11)
The updated probability of failure is computed using the updated probability density f x ( x ) .
With the development of Bayesian analysis, research on probability-based inspection planning was developed. Itagaki, et. al [8] and Førli [9] discussed probability based inspection planning for ship and offshore structures. In probability-based inspection planning, the failure probabilities are generally expressed in terms of intersections of the events of inspection and failure. Information gathered by inspection is accounted for by updating the probabilities using Bayesian analysis. Three approaches have been developed to extend the Bayesian analysis procedure to inspection planning: (1) target safety margins [4,8], (2) optimum life cycle cost [4,12], and (3) combination of (1) and (2) [4,10].
RISK BASED INSPECTION Probability based inspection can be used to establish the component/element in-service inspection schedules on the basis of reliability requirement of the individual critical components (elements). However, FPSO ′s usually contain a large number of components/elements. A rational inspection program should be developed based on system considerations. The development of a system-level, risk-based inspection process includes the prioritization of systems, subsystems, and components/elements using risk measures, and definition of an inspection strategy (i.e., the frequency, method, and scope/sample size) for performing the actual inspections. The process also includes decisions about the maintenance and repair strategies. Finally, there is a strategy for updating the inspection program for a given system, subsystem, or component/element, using the results of the inspections that have been performed.
The use of a ‘living’ process that is flexible, strives for completeness, and can be easily implemented and updated. The use of qualitative and quantitative risk measures The use of effective and efficient qualitative and quantitative methods that provide results familiar to inspection personnel.
Figure 3 illustrates the overall risk-based inspection process. The process is composed of the following steps:
• •
Probability-based Inspection Planning
•
• •
Definition of the system for inspection Use of a qualitative risk assessment that utilizes expert judgment and experience in identifying failure modes, causes, and consequences for initial ranking of systems and components/elements in inspection. Application of quantitative risk analysis methods, primarily using an enhanced Failure Modes, Effects, and Critically Analysis (FEMCA) and treating uncertainties, as necessary, to focus the inspection efforts on systems and components/elements associated with the highest calculated safety, economic, or environmental risk. Development of the inspection program for the components, using decision analysis to include costbenefit considerations. The inspection strategy are being updated and implemented, based on the findings and experience from the previous inspections.
Several feedback loops are shown in Figure 3 to represent a living process for the definition of the system, the ranking of components/elements, and the inspection strategy for each component/element.
System Definition A key step in defining a system for inspection, as shown in the first box of Figure 3, is the assembly of information that is needed for the risk-based inspection approach. In particular, the interviewing of key personnel, who are knowledgeable for degradation mechanisms or errors that may not be documented, is vital to the process.
Inspection Prioritization
be
The qualitative risk assessment, as included in the second box of Figure 3, utilizes expert judgment and experience in prioritizing systems, and components/elements for inspection. A key element for this assessment is to identify potential failure modes and causes, including design, operational, and maintenance errors and potential degradation mechanisms.
The use of a multidisciplinary, top-down approach that starts at the system level before focusing the inspection at the component/element level.
Figure 4 shows an example result of a qualitative risk assessment in which each box is representative of a given component, and a box is used to show the range of estimated consequence and failure probability. Once numbers are
Methodology The methodology summarized as:
•
• •
3
for
risk-based
inspection
may
4
T. XU, Y. BAI, M. WANG, R. BEA
placed on the axes, the risk assessment become quantitative, with uncertainty being represented by the size of boxes. The risk is defined as: Risk = (likelihood of failure)x(failure consequence)
•
(12)
Consequences of failure can be measured in a variety of ways, such as injuries/deaths, economic loss, environmental damages, dollars. In Figure 4, region A is high risk, region B is intermediate risk, and region C is low risk. Components/elements are grouped according to the region in which they fall. The FMECA (Failure Modes, Effects, and Criticality Analysis) in the third box of Figure 3 is an element of the subjective probabilistic ranking. It provides an efficient means of integrating the information required for a riskbased prioritization. Information on systems or components/elements is gathered from • design information, • operating experience (including prior inspection results), • structural reliability and risk analysis (SRRA) results, and • expert opinion to define failure modes, failure causes, and (perhaps) failure probability. In this way, the key information is integrated to provide the safety, economic, or environmental risk associated with the systems, subsystems, and elements to develop the inspection rankings for different systems/components. The probability-based inspection method developed based on probabilistic structural mechanics is perhaps the essential part of the FMECA since it provides a rational framework to estimate failure probabilities for components/elements.
Inspection Program Development Once the FMECA is completed, and the components/elements are ranked or categorized, the next step is to develop an inspection program for each group of components. This is the bottom box in Figure 3 that is schematically shown in Figure 5. It can also be used to establish an inspection program for an individual component/element or a system, as necessary. The recommended process is divided into three basic steps:
•
•
Choose potential inspection strategies that define the frequency, method, scope, and sampling procedure for inspection: The method of inspection includes the procedure, equipment, and level of personnel qualification to perform the inspection. The inspection strategy can also take advantage of monitoring systems and maintenance test program. Choose an inspection strategy and perform inspection: From the potential inspection strategies, defined in the above step, the effect of each of these strategies on the
OTC 12949
failure probability of the component/element is estimated. Choose appropriate action and update state of knowledge and information: Following the performance of the inspection, a critical decision is faced. That is, should the component/element be repaired or replaced if significant findings occur, or should nothing be done except to redefine the inspection program (going back to part 1 of the overall process shown in Figure 3)? Should the existing inspection, maintenance, repair system be changed? This depends on the fitness for purpose evaluation to determine the inspection findings and potential actions on the failure probabilities. In any case, all of the results related to inspection should be used to update the FMECA information on a periodic basis to re-rank the components/elements on the basis of risk and to redefine the inspection program.
RISK BASED ‘OPTIMUM’ INSPECTION Experience with in-service inspections of ship and offshore structures has adequately demonstrated that there are two distinct categories of defects and damage that are found:
• •
Those due to intrinsic causes - those that could have been or were anticipated (natural, predictable), and Those due to extrinsic causes - those that could not have been anticipated (human caused, unpredictable).
Experience with fatigue and corrosion damage found in ship and offshore structures clearly indicates that a substantial amount (if not a majority) of damage falls in the second category - unpredictable and due to the ‘erroneous’ actions and inactions of people. Quantitative inspection analyses (e.g. probability or risk based inspection methods and programs) can help address the first category of defects by providing insights into when, where, and how to inspect and repair. However, such an analysis cannot be relied upon to provide information that addresses the second category of defects. Expert observation and deduction (diagnostic) techniques must be used to address the second category of defects. Such recognizations lead to the development of the ‘optimum’ inspection method for FPSO ′s. The overall objective of the ‘optimum’ inspection method is to develop an effective and efficient safety and quality control system in the life cycle management of the FPSO ′s.
Inspection Performance Inspection performance is influenced by the vessel, the inspector, and the environment (Figure 6). The vessel factor can be divided into two categories: design factors and condition/maintenance factors. Design factors, including structural layout, size, and coating, are fixed at the initial design or through the redesign that may accompany
OTC 12949
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
repair. Condition/maintenance factors reflect the change in a vessel as it ages, including the operation history and characteristics of individual damages/defects (crack, corrosion, bucking), its size, and its location. The person (inspector) carries out an inspection can greatly influence the inspection performance. Performance varies not only from inspector to inspector, but also from inspection to inspection with the same inspector based on mental and physical condition. Factors associated with the inspector include experience, training, fatigue and motivation. The environment in which the inspection is carried out has a major influence on performance. The environment factors can be divided into two categories: external factors which cannot be modified by inspection procedures and procedure factors that can be modified. External factors include weather and location of the vessel, that is, whether the inspection is performed while underway, while in port, or while in dry-dock. Procedural factors reflect the condition during the inspection (lighting, cleanliness, temperature, ventilation), the way in which the inspection is conducted (access method, inspection method, crew support, time available) and the overall specification for inspection (inspection type).
Inspection Strategies Inspections, data recording, data archiving (storage), and data analysis should all be a part of a comprehensive and optimum inspection system. Records and thorough understanding of the information contained in these records are a key aspect of inspection programs. Inspection is one part of a ‘system’ that is intended to help disclose the presence of ‘anticipated’ and ‘unanticipated’ defects and damage. Development of inspection programs should address:
• • • • • • •
Elements to be inspected (where and how many?), Defects, degradation, and damaged to be detected (what?), Methods to be used to inspect, record, archive, and report results (how?), Timing and scheduling (when?), Organization, selection, training, verification, conflict resolution, and responsibilities (who?), and Objectives (why?). Where and How Many?
Definition of the elements to be inspected is based on two principal aspects:
• •
Consequences of defects and damage, and Likelihood of defects and damage.
The consequence evaluation essentially focuses on defining those elements, and components that have a major influence on the quality and safety of a FPSO. Evaluation of the potential consequences should be based on historical data
5
(experience) and analysis to define the elements that are critical to maintaining the integrity of a FPSO. The likelihood evaluation focuses on defining those elements that have high Likelihood’s of being damaged and defective. Experience and analyses are complementary means of identifying these elements.
What? A substantial amount (if not the majority) of the damage is unpredictable and due to the unanticipated ‘erroneous’ actions and inactions of people [13]. Current experience also indicates that the majority of damage that is associated with accidents (collisions, dropped objects) is discovered after the incident occurs [13]. About 60% of fatigue and corrosion damage is detected during routine inspections. However, the balance of 40% is discovered accidentally or during non-routine inspections.
How ? The methods to be used in FPSO inspections are basically visual. In one form or another, these methods are primarily focused on getting an inspector close enough to the surface to be inspected so that he can visually determine if there are significant defects or damages. However, ultrasonic gauging, magnetic particle, radiographic and other nondestructive methods sometimes are necessary for FPSO.
When? There are no general answers to the timing of inspections. The timing of inspections is dependent on:
• • • •
The initial and long-term durability characteristics of the FPSO structure; The margins that the operator wants in place over minimums so that there is sufficient time to plan and implement effective repairs; The quality of the inspections and repairs; and The basis for maintenance – ‘on demand’ (repair when it ‘breaks or leaks’ or ‘programmed’ (repair or replace on standard time basis).
Who? Experience has adequately demonstrated that the single most important part of the inspection inspection system is the inspector. inspector. The skills, knowledge, motivation, integrity of the inspector are critically important. Equally important are the organizational influences exerted on the inspector, the procedures and processes that he is required to follow, the environments in which he must work, and the support hardware and systems that are provided for him to perform his work. Thus, the inspector inspector is significantly significantly influenced influenced by
6
T. XU, Y. BAI, M. W AN G, R. BEA
1) organizations, 2) procedures, 3) hardware (facilities), and 4) environments.
•
Much has been learned about how to improve the effectiveness and efficiency of the inspector [12]. As one designs new inspection systems, it is important that the inspector be recognized as a part of this system [4].
• •
Why? Inspection should have objectives of several levels: first, it provides the general information and knowledge for the quality of the in-service structures for fitness for purpose evaluation (general condition), second, it is to detect the damage/defects as many as possible so that effective and efficient maintenance and repair program can be implemented to correct these damages/defects (quality control and assurance), third, it is a safety control tool to prevent the failure or loss of the in-service structures during the inspection interval (safety control and assurance). The inspection strategies (when, where, how, who) for different level objectives should be different. The first level inspection should select typical elements/components to provide general information about the in-service structures for fitness for purpose evaluation. It is more frequent less detail inspection associated with long-term maintenance and repair program. The second level (quality control) inspection should focus on the critical components/elements to detect damage/defects as many as possible. It is associated with the short-term maintenance and repair program. The third level inspection (safety control) is to prevent the most critical damage/defects or errors to ensure the safe operation during the inspection interval. It is the most detailed and difficult inspection to identify the safetyrelated predictable or unpredictable damages/defects and errors. Every inspection practice for a specific fleet should be a combination of these three different inspection strategies.
OTC 12949
To confirm what is thought: to address the intrinsic damages/defects that can be predicted based results from technical analyses, To disclose what is not known before inspection; to address damage/defects that can not be predicted based on technical analyses, and To control the predictable and unpredictable damages; to develop high quality maintenance and repair program.
The ‘optimum’ inspection program should be started with the design of the structure (conception), proceed through the life of the structure, and conclude with its scrapping (lifecycle). The optimum inspection program should include not only the hull structure, but as well, its equipment and its personnel (full scope). The optimum inspections should become the means to assess the general conditions of the whole structure. The optimum inspections are also the means to detect unpredictable flaws and damages of the structural elements, and permit appropriate measures to be taken to preserve the safety and integrity of the structure. The optimum inspections are also the means to assure that all is going as expected, that the structural elements are performing as expected, and that corrosion protection and mitigation (e.g. patching pits, renewing locally excessively corroded plate) is maintained. The ‘optimum’ inspection method starts from the survey for the intrinsic damage that is common for the class of structures. Based on the experience, the inspection for the intrinsic damage can be conducted in the rational way. The existing risk-based inspection method discussed early this paper is the framework for the intrinsic damages/defects for the structural system. The probability-based inspection method can be applied for the specific elements/components based on the results of risk-based inspection. For the extrinsic damage for each individual structure, the knowledge-based diagnosis method should be developed. The step-by-step knowledge-based diagnosis process is the potential means to identify the extrinsic damages.
The value of the inspection for objectives of different levels should also be different. different. The value of the first level inspection is about the decision on whether or not the existing structure can fulfill the purpose for extended service. The value of the second level inspection is about the decision whether or not we should change the maintenance and repair program. The value of the third level inspection is about the decision whether or not we should take any intermediate actions. Value analysis (value of information) can help make these decisions.
Knowledge systems now routinely do diagnosis reasoning using three methods: model-based diagnosis, heuristic classification, classifica tion, and case-based reasoning. Our system uses a combination of each of these methods: Model-Based Diagnosis (MBD) to identify the details of a large class of possible problems, heuristic classification to identify the presence a set of idiosyncratic problems, and Case-Based Reasoning (CBR) to compare observation with previously identified cases.
‘Optimum’ Inspection Method
•
The ‘optimum’ inspection method can be proactive (focused on prevention) or it can be reactive (focused on correction). It should have four functions:
•
•
Assess the general conditions of the in-service offshore structures ,
An ‘optimum’ inspection method could include: Developing standard task checklists to ensure that relevant data and tasks are not lost because of distractions or workload, Performing global surveys to develop situation awareness for potential expected and unexpected damage and defects,
OTC 12949
•
• • • • • •
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
Inspecting high likelihood of damage or defect ‘parts’ and high consequence parts; if something ‘suspicious’ is found, the inspection is intensified by model-based diagnosis, heuristic classification, and case-based reasoning until root causes (not symptoms) are determined, Inspecting periodically decreasing the time between inspections as the rate of degradation or likelihood of defects and damage increase, Inspecting after accidents or ‘early warning’ signals are sensed, Implement the long-term and short-term maintenance and repair strategies based on the inspection results, Update the IMMR (Inspection, Maintenance, Monitoring and Repair) plan based on the survey results and the results from maintenance and repair, Performing inspections that are independent from the circumstances that cause potential defects and damage, and Using qualified and experienced inspectors that have sufficient resources and incentives to perform quality inspections.
For each FPSO, standard checklist and procedures should be established from the FPSO Structural Life-Cycle Information Management System, in order to carry out an effective evaluation of the general condition, prior to the commencement of any general survey and include:
• • • • • • • • •
Structural drawing, Operating history and conditions, Previous damage/defects inspection results, Condition and extent of protective coatings, Classification status, including any outstanding conditions of class, Previous repair and maintenance work, Previous information on unpredictable damage or defects, Expert’s judgment and comments, Relevant information from its sister structures.
With this information and previous inspection guidelines regarding critical elements/subsystems in the FPSO structural systems considered to be sites of potential damage/defects based on historical data, analyses results, and expert’s judgment, it is possible to target the appropriate inspection strategies for the potential areas within the structure for general survey and the initial scope of the inspection. After the initial inspection to determine the general condition of the system, the inspector can develop situation awareness to identify some potential unpredictable critical damage/defect sites. Further knowledge-based diagnosis should be conducted for these suspicious areas. The knowledge-based diagnosis is conducted together with detailed or specific inspections.
7
Inspection Data System The single weakest component that has been found in present inspection systems for FPSO ′s regards the data and information that is developed during and from inspections. Little thought has been given to the efficient gathering of data and information, even less thought to what is done with this data and information when it is obtained, and far less thought given to the archiving, analysis, and reporting of the data. The interfaces in the data gathering, archiving, analysis, and reporting activities also have received a little systematic thought. Current work has not been able to identify a single coherent and optimum inspection data system for FPSO. Advances in information technology have resulted in better ways to use information for the management of safe and efficient ship and offshore structures. The integration of stand-alone systems combined with improved information recording, organization and communication offers substantial benefits for the life-cycle management of ship and offshore offshore structures. A life cycle Structural Information Management System (SMIS) is intended to facilitate the lifecycle management of FPSO. This includes areas from design and construction as well as operations including Inspection, Maintenance, Monitoring and Repair (IMMR). The inspection data system is a component of the IMMR module in SMIS. The general objectives of an inspection data system development development are:
• • • • • • •
Collect inspection data, Store the data, Provide means for logic inspection data management, Allow for the organization of the inspection data in a form suitable for fitness or purpose analyses, and failure analyses, Analyze the data, Show trends of the information such as damage/defects associated with structural integrity, Communicate and report the data.
Once a FPSO is ready for service, a series of inspections are scheduled according to inspection program. The objective objective and scope of of the internal tank inspections are defined. The access methods and data recording methods are chosen, and then the inspections are performed. The inspection results including corrosion gauging, cracking, status of coating and corrosion protection systems and other structure/equipment defects are updated into the corresponding database. Using the inspectional data, maintenance and repair strategies can be developed and the repair are finally carried out.
8
T. XU, Y. BAI, M. WANG, R. BEA
OTC 12949
REENGINEERING THE IMMR PROCESS
ACKNOWLEDGEMENT
FPSO in-service inspections and repairs are components in an Inspection, Maintenance, Monitoring and Repair (IMR) system that is intended to help disclose the presence of ‘anticipated’ and ‘unanticipated’ defects and damage to critical structural details [4,12].
The technical views expressed in this paper are those of the authors and are not necessarily those of the institutions they are affiliated with.
An IMMR system is a critical part of the maintenance of inservice quality of a FPSO. The IMMR process should be in place, working, and being further developed during the entire lifetime of the structure. The IMMR process is responsible for maintaining the quality of the structure during the useful lifetime of the structure. A fundamental and essential part of the IMMR process is knowledge. The IMMR process can be no more effective or efficient than the knowledge, data, and experience that form the basis for the process.
REFERENCES 1.
2.
3.
4.
Xu et al [4] indicates that organization of the FPSO inspection should be developed to:
• • • • •
Define inspection processes in which only that information that is absolutely necessary to assure acceptable quality in the FPSO is gathered, Determine how to minimize the people and man-hours required in the entire IMMR process, Define how to minimize the steps, interfaces, multiple processing, and paper required in the IMMR process, Determine how to minimize the checks, controls, reporting, and reconciliation required in the IMMR process, and Define how take full advantage of present computing, communications, and information technologies (CCIT).
5.
6.
7.
8.
CONCLUSIONS This paper addresses the development of ‘optimum’ inspection strategies for FPSO ′s. It details with the technical basis for 1) probability-based inspection method, 2) riskbased inspection method, and 3) ‘optimum ’ inspection method. The probability based inspection methods are recognized to be able to address only part of the potentials for damage in FPSO ′s. The risk-based inspection addresses the component/element’s damages/defects from the system risk point of view. The ‘optimum’ inspection method is a full-scope, life-cycle development of inspection system for the risk management. The ‘optimum’ inspection method not only includes use of expert observation and analysis (deductive) methods but also the use of structural information system. It is demonstrated that application of this comprehensive and integrated approach will result in better allocation of the resources used to develop effective inspection strategies.
9.
10.
11. 12.
13.
14.
15.
Tanker Structural Co-operative Forum (1986) Guidance Manuals for the Inspection and Condition Assessment of Tanker Structures. Tanker Structure Co-operate Forum, (1990) Inspection, Assessment and Experience of Old Tankers, Witherby Marine Publication. Marine Technology Directorate, (1989) "Underwater Inspection of Steel Offshore Installations: Implementation of a New Approach", Report 89/104. Xu, T., Bea, R. G., (1996) " Inspection of Marine Structures", Report for Joint Industry Research, Marine Technology and Management Group, Dept. of Civil Engineering, University of California at Berkeley, Berkeley, CA 94720. Demsetz, L.A., Cario, R., and Schulte-Strathaus, R., (1995) Inspection of Marine Structures, Report for Maritime Administration under Project Number DTMA91-93-G00040. Yang, J.N. and Chen, S., (1985) "Fatigue Reliability of Structural Components Under Scheduled Inspection and Repair Maintenance," Probabilistic Methods in Mechanics of Solids and Structures, edited by S. Eggwertz and N.C.Lind, Springer-Verlag, Berlin, pp. 103-110. Moan, T., (1993) "Reliability and Risk Analysis for Design and Operations Planning of Offshore Structures", Proc of the 6st Intl Conf on Struct Safety and Reliability, ICOSSAR'93. Itagaki, H., Akita, Y. and Nitta, A., (1983) "Application of Subjective Reliability Analysis to the Evaluation of Inspection Procedures on Ship Structures", Proceedings of the International Symposium on the Role of Design, Inspection and Redundancy on Marine Structural Reliability, National Academy Press, Nov. 13-16. Førli, O., (1990) "The Reliability and Cost-Effectiveness of Offshore Inspections", Proc. Intl Conf on Monitoring, Surveillance and Predictive Maintenance of Plants and Structures, Sicily, Italy. Shinozuka, M., (1990) "Relation of Inspection Findings to Fatigue Reliability", Ship Structural Committee Report, SSC-355 Marine Technology Directorate, (1994) "Review of Repairs to Offshore Structures and Pipelines, Report 94/102. Bea, R.G, (1993) Ship Structural Maintenance: Recent Research Results and Experience, Transactions, The Institution of Marine Engineers, London Jones, R. B. (1995). "Risk-Based Management - A Reliability-Centered Approach", Gulf Publishing Co., Houston, Texas. Platten, J. L., (1984) “Periodic (in-service) Inspection of Nuclear Station Piping Welds, for the Minimum overall th Radiation Risk”, Proceedings of the 5 International Meeting on Thermal Nuclear Reactor Safety, Vol. 1. 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
OTC 12949
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
9
Committee on Human Performance, organizational Systems, and Maritime Safety, National Academy of Engineering, National Research Council, Washington, D. C.
f A(a) Probability of Failure ac
f a(a)
f a(a) Experiment/Inspection
Experiment/Inspection Posterior
I d n i i t s i t a r i l b c u r t i a o c k n
Prior
Original distribution Mean Crack Size
N Inspection
Updated Number of Cycles distribution
Figure 1 Illustration of POD Updating
a
a
Figure 2 Illustration of Bayesian Estimation
10
T. XU, Y. BAI, M. WANG, R. BEA
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
OTC 12949
OTC 12949
RISK BASED 'OPTIMUM' INSPECTION FOR FPSO HULLS
11
CAUSE OF DAMAGE Operating error Design error C
o
r
r
o
Installation fault
A
Fabrication fault Dropped Object
B
C
o F
C
l a
0
0
l t
0
0
i i
0
.
1
.
FREQUENCY 0F PLATFORM DAMAGE (% / year) Consequences
Figure 5 Causes of Damage to North Sea Structures
Figure 4 Definition of Risk Index
1. Select Potential Inspection Strategies * Define Potential for Damage States * Define Potential for Inspection Damage * Define Reliability of Inspection Methods
Update
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 Appropriate IMR System
4. Later
(Sensitive Studies)
Now
(2) Implement IMR System
Update State Knowledge and Information
Figure 6 Development of In-Service Inspection Program
1
.
12
T. XU, Y. BAI, M. WANG, R. BEA
VESSEL
OTC 12949
ENVIRONMENT
Design
Inspection Performance
External - Weather - Location of Vessel
- Structural Layout - Size
Procedure
-Coatings
- Lighting - Cleanliness
Condition/Maintenance Cargo Defects
INSPECTOR Experience Fatigue Training Motivation
- Temperatures/Humidity - Ventilation - Access Method - Inspection Method - 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 participating 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 Copyright 2001, Offshore Technology Conference This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, 30 April–3 May 2001. This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
Abstract This paper describes the reliability methods for position mooring design and analysis with various degrees of complexity and with a particular emphasis on deepwater applications. The limit state design and safety factor formats for position mooring systems of various floating platforms are first addressed. This paper addresses technical issues of both passive and active mooring systems including winch or thruster assisted and dynamic positioning systems. The modeling uncertainties of numerical analysis, mooring components, environmental conditions and reliability methods are addressed. This paper presents a logical approach to reliability analysis for the calibration of partial safety factors for both passive and active mooring systems. Introduction As the development of offshore oil and gas is moving into deeper waters, floating platforms are used for drilling and/or production operations. For these operations, position mooring systems are required in order to keep the floating platforms on station under the design environmental criteria of wind, current and waves. Various types of floating platforms and mooring systems can be considered for offshore applications. For mooring strength limit states, partial safety factor formats have been proposed recently in the offshore industry. It is important, however, to fully understand how to use the reliability methods to calibrate the partial safety factors selected. The subject matter is significant to the offshore industry for enhancing the reliability of position mooring systems with cost-effective designs. This is especially important for deepwater applications because the consequences of position mooring system failure would be much more costly than in relatively shallow waters.
Limit States of Mooring Design Limit state design methods are first described with particular reference to the design of position mooring systems for floating installations. The limit state design philosophy may be used to provide a rational framework for the design of safe and serviceable structures or structural components, by accounting for uncertainties and variabilities in the basic 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 state formulation, including the deterministic and reliability analysis methods used in evaluating the probability that the limit state will be exceeded, to ensure that existing accepted safety levels are achieved. The performance of a structural system, or a component of the structure, may be described by a set of limit states (limiting conditions) beyond which the structure, or component, is no longer deemed to satisfy the design requirements. Limit states can be regarded as a discrete representation of a more general continuous loss function. In general, the failure of a structural system or component to satisfy the design requirements may be represented by an inequality of the form: (1) g X 1 , K , X n , C ≤ 0
(
)
Where, g is the limit state function, X 1 ,…, X n are the basic variables associated with the limit state, and C are constraints describing the limits of acceptable accelerations, offsets, clearances, etc. The acceptable performance of position mooring systems depends on system responses to environmental actions and the system’s strength, in intact and damaged conditions. Additionally, moored installations may depend on the availability of active systems, such as thrusters and winches, to control or reduce critical system responses. In general, the evaluation of mooring system limit states is more complicated than the evaluation of component limit states. In the analysis of position mooring systems, computer programs, databases, model tests, statistical models, empirical rules and assumptions are used in evaluating the complex nonlinear coupled behavior of the vessel, mooring lines, and risers. All of this is symbolized by the limit state function g in eqn. (1). For example, linear potential theory may be used to synthesize approximate first and second order vessel forces
2
KEN HUANG AND YONG BAI
and motions in irregular seas. Wind and current force coefficients may be derived from databases for generic vessel types or from wind tunnel tests on scale models. Characteristic and extreme values of response parameters may be calculated from statistical models and assumed distributions and calculated output parameters from one analysis program are often used to define input variables or coefficients in a second program. The basic variables that determine the behavior and responses of moored systems fall into one of the following categories. (1) Action or load variables - associated with intensity and spatial variations of actions or loads, such as; wind and wave intensities, spectra, and directions, current profile, velocity, and direction, mooring line pretensions, vessel draft/ballast, etc. (2) Component and material variables - associated with component and material properties, such as; mooring component break strength, fatigue endurance, density, elasticity, soil strength, etc. (3) Geometric variables - associated with the structural system or component geometry, such as; vessel shapes/dimensions, mooring line diameter, line segment length, line angles, water depth, etc. (4) State variables - associated with combined properties of the system or component and the environment, such as; wind, wave drift, and current force and moment coefficients, response amplitude operators (RAOs), wave frequency natural periods and damping, mooring line and vessel drag and added mass coefficients, low frequency damping, etc. (5) Active system variables - associated with the performance and mean time to failure and repair of the mechanical and electrical, distribution, monitoring, and control systems and the experience and training (human factors) of the personnel operating the active systems. (6) Modeling uncertainty variables - associated with the physical (hydrodynamic and structural theories) and statistical models and the reliability methods, which are used in calculating system responses, component and system strengths, endurance, etc. and the notional reliability levels of components and systems. For position mooring systems, strength limit states are usually associated with the maximum installation, survival, operational, disconnect, and reconnect criteria in their most extreme design environments. Depending upon the inspection and maintenance program, strength may also depend on damage accumulation; that is on the fatigue, corrosion, and wear endurance limit states, generally in the long-term environment. While, operational performance is usually related to offset and motion limit states in the maximum operational environment. These limit states are normally subdivided further. For example, in the maximum design environment individual limit states will address extreme line loads, total turret loads, anchor loads and line angles, vessel offsets, motions, and clearances etc. for intact, damaged, and transient cases. Design requirements, criteria, and the safety factors
OTC 13269
recommended in present codes developed and evolved to serve the needs of mobile offshore drilling units (MODUs). In particular, the methods of analysis and design criteria contained in API codes were originally developed based mainly on the operational experience of chain and wire catenary spread moored semisubmersible drilling units, [3 – 7]. Over the past twenty years these methods of designing mooring systems for MODUs have been used worldwide and have proved to provide adequate safety levels for conventionally moored semisubmersibles, [8]. More recently, spread moored semisubmersibles, spars, and turret moored ships, with or without thruster assistance, using taut or catenary mooring systems, and synthetic mooring line components as well as chain and wire, have been used as permanent production installations. API and ISO have developed codes that also address the design requirements of these types of installations, [9 – 12]. However, in these codes the combined effects of variability in environments and system responses and strengths, for different types of vessel, mooring systems, and environments, have not been fully accounted for. Consequently, safety levels inherent in today’s codes are largely arbitrary and cannot ensure either an optimum level of safety or economy in construction. Due to the historical experience on which code development has been based, the analysis methods and safety factors recommended by the present API and ISO codes, [10, 11], are most applicable to conventionally moored semisubmersibles. Safety Factor Formats When limit states are evaluated using deterministic methods, the basic variables ( X 1 ,…, X n) are replaced by their characteristic design values ( y1 ,…, yn), and a set of safety factors (γ 1 ,…, γ m) or safety elements are introduced which are chosen to ensure a minimum level of safety, [2]. This allows the limit state checking equation used in deterministic design methods to be represented by: (2) g ( y 1 , K , y n , C , γ 1 , K , γ m ) ≤ 0
For example in evaluating the mooring system’s strength limit state, the characteristic values of wind, wave, and current are usually chosen as the most probable maximum values associated with the design return period, and conservative ranges (safety elements) of peak periods and relative wind, wave, and current directions are considered. While the mooring line break strength is represented by the catalog break strength of the component, including a reduction in diameter (safety element) to allow for corrosion losses. In this traditional design method some safety elements are preapplied (spectral peak period and wind, wave, and current directions) while intact, damaged, and transient line tension safety factors are post-applied . API RP 2SK and the ISO station keeping standard, [10, 11], both use a total safety factor that is applied to the mooring system responses, post-applied , and contain some references to pre-applied safety elements. Generally, line strength postapplied safety factor formats for intact, damaged, and transient cases may be expressed as:
OTC 13269 BS
γ m
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
≥ γ µ T µ ( y1 , K , y n ) + γ LF T LF (y1 ,K
, yn
) + γ
T
WF WF
(y ,K , y ) 1
n
(3)
or
(
(
BS ≥ γ m γ µT µ y1 , K , yn
) + γ
T
LF LF
(y ,K , y ) + γ 1
n
(
T WF y1 ,K , yn
WF
))
Where, BS = catalog break strength of mooring component T µ = mean tension, nonlinearly dependent on ( y1 ,…, yn) T LF = low frequency tension, nonlinearly dependent on ( y1 ,…, yn) T WF = wave frequency tension, nonlinearly dependent on ( y1 ,…, yn) ( y1 ,…, yn) = characteristic design values of the basic variables on which line tensions depend γ m = mooring component partial safety factor = mean tension partial safety factor γ µ = low frequency tension partial safety factor γ LF = wave frequency tension partial safety factor γ WF Design methods that are based on a safety factor format of this type, eqn. (3), are known as load and resistance factored design (LRFD) methods. In the case of the API and ISO mooring codes only a single or total safety factor is specified. Where there is uncertainty or variability in the basic (input) variables, post-applied safety factor formats of the LRFD type will produce inconsistent safety levels for mooring systems that have different nonlinear response characteristics. Additionally, inconsistent levels of safety may also result where the variability in environmental conditions differs from site to site. For example, the coefficient of variation associated with 100-year return period wind speeds for Gulf of Mexico hurricanes and North Sea winter storms are approximately 13% and 5% respectively, and wind loads vary as the square of the wind speed. Consequently, mooring systems designed to the same post-applied (LRFD) safety factors in the Gulf of Mexico and the North Sea will be associated with different notional reliability or safety levels, this is true even for systems which display identical, linear or nonlinear, response characteristics. It may be possible to achieve more consistent designs, in terms of safety levels, by the use of more complicated safety factor formats. One possibility is to use a post-applied safety factor format, eqn. (3), with the values of the partial safety factors defined as functions of vessel/positioning system type and environment. Alternatively, a pre-applied partial safety factor format, eqn. (2), could be developed, where the values of the partial factors are related to the uncertainty (coefficient of variation) in the basic variables. For example, the characteristic design value of environmental variables could be defined as the value associated with the mean plus nstandard deviations, instead of the most probable (modal) value that is currently used. At present the industry has no experience with mooring codes that use pre-applied safety factor formats, and such design codes would represent a major departure from the design methods that have evolved with the offshore industry. Additionally, it has yet to be demonstrated that a reasonably
3
simple and practical pre-applied safety factor format that will reduce the variability in safety levels across the range of types of positioning systems and environments can be devised. Limit State Design Limit states define the limiting conditions beyond which the structural system or component is no longer deemed to satisfy the design requirements, consequently, there is a correspondence between limit states and traditional design criteria. By calibrating individual limit states, the limit state design method may be used to achieve more consistent position mooring designs than is presently possible by the use of the traditional (un-calibrated) design methods specified by today’s codes. In some cases it is not practical, or possible, to express design criteria in terms of a limit state function that allows numerical evaluation. For example, protection against corrosion and wear of mooring line components is normally provided for by increasing the component diameter, jacketing the line, using blocking compounds, providing galvanic protection etc. and specifying inspection and maintenance programs. Not all operational limit states depend upon the position mooring system, for example processing operations may be limited by wave frequency vessel motions and accelerations which in many cases are unaffected by the mooring system. In general, position mooring systems are redundant seriessystems, in which system responses, including line-loads, are nonlinearly dependent on the environmental (wind, wave, and current) actions. Redundancy is provided for in the number of mooring lines (load paths), and recognized in mooring codes by specifying intact and broken line factors of safety. Single anchor leg moorings (SALMs), and some portions of the load paths in single point moorings (SPMs) and catenary anchor leg moorings (CALMs) may not provide redundancy. These systems may be susceptible to single point failure and details of their design require special consideration. For example, in the design of SALMs API RP 2SK recommends that the level of structural redundancy be recognized in selecting an appropriate safety factor. Mooring codes, [10, 11], recognize redundancy in position mooring systems by specifying design requirements and safety factors for intact, damaged, and transient cases. However, these codes do not allow for differences in nonlinear system response characteristics, which depend upon the type of vessel, type of mooring system, and water depth. Or for differences in mooring line (series-system) strength, which depends on the type of mooring component (chain, wire, or synthetic rope) and the length of the line segment. Nor do these codes allow for differences in the variability of the environmental actions, which depend upon geographic region and season. In addition, position mooring systems may be entirely passive systems, or they may depend on active intervention. Active station keeping systems make use of thruster assistance or adjustment of mooring line lengths to reduce critical system responses. While some turret designs require power generation and functioning hydraulic equipment
4
KEN HUANG AND YONG BAI
to release brakes that hold the turret in its normally locked condition. Thrusters may be used to assist the mooring system by controlling the vessel’s heading, reducing the mean environmental forces, damping low frequency motions, or a combination of these functions. In high current environments adjustment of mooring lines may be used to position the vessel up current. Mooring lines may be adjusted in preparation for storm conditions. If the position mooring system depends upon the use of thrusters or line length adjustments to satisfy the design requirements, then the capability of the thruster system or the ability to perform winching operations, in the associated environmental conditions, must be assessed. In evaluating the capability of active systems, it is necessary to consider the equipment that support and control the thrusters or winches, their modes of failure, repair times, and the training and experience of the personnel operating the systems. Even for the most advanced mechanical and electrical systems there is a risk, albeit small, of a total loss of all electrical power (blackout). Consequently, wherever possible active systems should be designed to “fail-safe” and the consequences of a power blackout should be known for all high-risk operations. For passive mooring systems, structural reliability methods use models furnished by physics and probability theory to calculate the notional probability that a limit state will be exceeded. Evaluation of notional probabilities of failure for active position mooring limit states requires the evaluation of the probability, availability, that the active systems (mechanical and electrical, distribution, monitoring, control, operators, etc.) perform on demand in the relevant environmental conditions, in addition to the evaluation of the mooring system’s passive structural reliability. That is, the evaluation of notional safety levels for active position mooring systems requires both availability and reliability analyses of the active and passive systems on which station keeping depends. If limit state design methods are to be used to arrive at optimum designs for position mooring systems, through reliability based calibration, then the calibration of the limit states must allow for differences that exist in: • System redundancy, which depends on type of mooring system • System response characteristics, which depend on the type of vessel, type of mooring system and components, and water depth • Mooring line segment strength, which depends on mooring component material and construction and length of the line segment • Environmental variability, which depends on the geographic region and season • Passive or active system reliability, which depends on structural reliability, mechanical and electrical system availability, operator training, and human factors
OTC 13269
Reliability Analysis of Position-Mooring Systems Guidance and suggestions are discussed next to assist in the application of reliability methods to the design of position mooring systems. Methods of probabilistic structural analysis, reliability analysis, provide a rational means of dealing with the uncertainty and unpredictability in the basic variables on which the design of mooring systems depend, by means of the application of probability theory, statistics, and deterministic structural theory. Reliability analysis methods are concerned with quantifying the ‘measured chance’ that a structure will support the loads to which it is subjected, [1]. In particular, reliability methods provide approximate means for evaluating the notional probability of failure of limit states under defined actions (loads). In principle, the notional probability that the structure as whole will satisfy the design requirements over its design life may be synthesized from individual limit state notional probabilities of failure. Some examples of position mooring systems and environmental distributions are illustrated on Figure 1 and a general discussion encompassing other vessel types and station keeping systems is contained in API RP 2SK, [10]. The vessel and mooring system are subject to wind, wave, and current actions and mooring system responses are complex, coupled, and nonlinear. Mean, low, and wave frequency vessel offsets, motions, and line tensions are system responses that are nonlinearly dependent on the environmental actions and interactions (coupling) among the vessel, mooring system and environments. The degree and type of system nonlinearity is a function of vessel type, mooring system type and water depth, and the wind, wave, and current intensities and directions. These system responses enter into the limit state functions that define acceptable performance of the position mooring system, thus, the design of position mooring systems is concerned with the global analysis of nonlinear systems. Additionally, the variability in environmental actions depends upon the geographic region and season, which determine the type of storm that the system is exposed to. Absolute values of component or system probabilities of failure can only be calculated if the deterministic mooring analysis method, and the data and statistical models used to represent the vessel, mooring system, and the environment are completely known and the reliability analysis method is perfect. However, reliability methods provide a powerful tool for comparing the safety implications of different designs. Target safety levels for new designs, may be established by performing reliability calculations for old designs that have a history of satisfactory performance, in this way absolute values of failure probabilities are not required. For these reasons, probabilities of failure calculated using reliability methods are referred to as notional probabilities of failure, and target notional failure probabilities for limit states of new designs are established by calibrating to old designs for which satisfactory past experience exists. The results of reliability computations will depend upon: (1) The Deterministic Mooring Analysis – The hydrodynamic and engineering/analytical models and
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
their ability to predict the nonlinear coupled behavior of different system types. (2) The Basic Variables – The variables treated as random variables in the reliability analysis and their statistical descriptions, including modeling uncertainties, mooring component and line strength models, and the modeling of environmental conditions. (3) The Reliability Method – The methodology and assumptions used to simplify the reliability analysis and synthesize system notional probabilities of failure. If the resulting notional probabilities of failure, and comparisons between notional probabilities of failure across different types of moored systems, are to be meaningful then these aspects of the reliability analysis methodology and their influence on the final results must be understood and accounted for. Deterministic Analysis of Position Mooring Systems The results of reliability computations depend directly upon the deterministic analysis used to model the physical behavior of the position mooring system. For each position mooring system, a large number of deterministic mooring analyses (typically of the order of 103 to 105) are required to calculate intact and damaged notional failure probabilities using Level 2 or Monte Carlo reliability methods. As a consequence reliability analyses are usually based on frequency domain methods of analyzing the mooring system. Frequency domain analysis methods decompose the total response of the mooring system into mean, low frequency, and wave frequency responses. Statistical values of the low and wave frequency responses are then calculated and combined with the mean response to yield the maximum response in a specified storm duration. There are a number of ways in which system responses may be decomposed into mean, low frequency, and wave frequency responses and by which the statistical values of low and wave frequency responses may be calculated. For example, the mean equilibrium position of a spar may include the effects of pitch and roll on the wind, wave, and current force coefficients and allow for vessel setdown in calculating mean line tensions. The low frequency modal decomposition may use surge, sway, and yaw modes for all position mooring systems, or may use different modal decompositions for turret moored ships, spars, and spread moored semisubmersibles. The definition of characteristic low and wave frequency tension responses may be calculated based on a single vessel position, or may make use of different vessel positions for each mooring line. In this section details of the analysis methodology that have a bearing on the notional probabilities of failure calculated in the reliability analysis of position mooring systems are discussed. M ean response
The vessel’s mean offset position is calculated by requiring that the mean environmental loads and the net mooring system restoring forces and moments are in equilibrium. The mean environmental load is composed of the
5
mean wave drift and drag forces and moments, and the mean wind and current forces and moments. Most mooring analysis programs consider force and moment equilibrium for the surge, sway, and yaw degrees of freedom. In general, mean wave, wind, and current force and moment coefficients depend on the vessel’s mean pitch, roll, and yaw angles with respect to the environmental actions. For turret moored ships, the mean offset position depends on the mean yaw equilibrium heading and its stability, which must be established before the mean surge and sway forces are calculated. While for spars, mean environmental loads are also sensitive to the mean pitch and roll angles, requiring an iterative solution. Due to the small water plane area of spars, vessel set-down, heave, may also significantly effect mean system responses. For all vessel types, the mean current velocity and direction affect wave drift and drag forces and moments, and the relative importance of the wave forces and moments depend upon the type of vessel and environment, through the environmental force coefficients and relative magnitude of the environmental actions. Low and wave frequency system responses are dependent on the mean equilibrium position, consequently any error in the mean vessel position will also result in errors in the prediction of low and wave frequency responses. The capability of the mooring analysis method embodied in the computer program(s) used to calculate mean loads and system responses must be assessed for the types of vessel, mooring systems, water depths, and environments for which reliability analyses are performed. L ow f r equency modal r esponses
Ideally, low frequency responses of the vessel and mooring system should be decomposed into the system’s normal modes (eigen-modes), or an approximation of the normal modes. Generally, the normal modes will depend on the type of vessel and mooring system. For spread moored semisubmersibles, the surge, sway, and yaw modes provide a good approximation to the normal modes. Whereas for turret moored ships, sway and yaw responses are coupled, while for spread moored spars, the surge and pitch responses and sway and roll responses are coupled. The methods of decomposing low frequency responses recommended by API RP 2SK and the ISO code, [10,11], are most applicable to spread moored semisubmersibles. Low frequency motions about the vessel’s mean offset position are excited by the slowly varying components of the wind, wave reflection, wave drag, and current forces and moments. The calculation of the energy dissipated by the vessel, mooring lines, and risers, during the low frequency oscillations of the vessel (the low frequency damping) is one of the most difficult problems in the analysis of moored systems. In general, low frequency system damping consists of the following components: • Still water damping • Damping due to wind • Damping due to current • Damping due to wave reflection • Damping due to wave drag
6
KEN HUANG AND YONG BAI
•
Damping due to mooring line and riser motions
Low frequency damping is nonlinear and increases with the severity of the environment and the amplitude of the low and wave frequency motions of the vessel, mooring lines, and risers. Consequently, low frequency damping can only be calculated as part of a coupled analysis in which low frequency damping is solved for iteratively in time domain. From the system properties at the mean offset position, the low frequency force and moment spectra, and the low frequency modal damping, the statistical values of low frequency modal responses are calculated. API RP 2SK and the ISO code present formulae for a linear frequency domain method of calculating significant and maximum low frequency modal responses. However, linearization of the combined mean environmental and mooring system forces and moments with respect to offset, can introduce large errors in the prediction of significant and extreme low frequency offsets and line tensions. For a simple case, differences in the low frequency surge response predicted by linear and nonlinear methods are illustrated on Figure 2. The upper diagram on Figure 2 shows the 3x3 grouped mooring pattern with the vessel at the mean equilibrium position and defines the x-y coordinate system. The net mooring system restoring force, F x, increases more rapidly (stiffer) for negative offsets than for positive offsets, this is shown on the graph on Figure 2. Both equally spaced and grouped mooring patterns display non-linear force versus offset characteristics. Figure 2 also shows the mooring system potential energy, calculated by integrating the nonlinear mooring system restoring force, F x, with respect to offset, x, and the quadratic approximation to the potential energy calculated from the linearized system stiffness, K x. For both linear and nonlinear methods, the mean energy and the associated standard deviation, σ x, of the surge motion (rms response) is calculated from the low frequency surge force spectrum and the linearized system properties and damping at the mean offset position. For a given storm duration the maximum energy in the surge mode, E max, is related to the mean energy, E mean, by:
( )
E max = E mean2ln N x
where 1 2 mean energy of the low frequency surge res E mean = K xσ x = 2 K linear mooring system surge stiffness at the mean off x =
σ x = standard deviation (rms) of the low frequency surge r T , T is the storm duration, andT is the surge period N x = T N N
(4) The minimum and maximum low frequency offsets that are calculated from the (nonlinear) system potential energy and the (linear) quadratic approximation to the potential energy are also shown on Figure 2. The nonlinear method yields different results for minimum and maximum offsets in the stiff and soft offset directions, which are as expected.
OTC 13269
While, the linear method incorrectly predicts the same minimum and maximum surge offsets for both stiff and soft offset directions. The example above illustrates the effect of nonlinear mooring system force response on minimum and maximum low frequency surge offsets. For turret moored vessels the mean environmental moment contributes to the rotational stiffness of the long period sway-yaw mode, and the nonlinearity of the mean environmental moment with respect to the vessel’s heading will generally result in large differences between the linearly and nonlinearly calculated minimum and maximum modal responses. Moored systems display different degrees of nonlinearity, where reliability analyses are used to compare safety levels for different vessel and mooring system types, the ability of the analysis method(s) to predict low frequency responses should be investigated and where possible calibrated against model test results. Combin in g low fr equency modal responses
API RP 2SK and the ISO code recommend that wave frequency responses are calculated at the mean plus significant low frequency offset position and at the mean plus maximum low frequency offset position. However, neither code specifies how the combined significant and maximum low frequency offsets are to be calculated from the low frequency modal responses, nor do the codes discuss the need to define these positions differently depending on the system response under investigation. The low frequency meanderings of the vessel will involve, at least, surge, sway, and yaw motions, which may be mapped in a three-dimensional space where the ( x, y, z ) axes represent surge, sway, and yaw offsets measured from the mean equilibrium position. In this space the significant and maximum low frequency offset positions referred to by API RP 2SK and the ISO code are represented by an infinite combination (a surface) of surge, sway, and yaw values. Mooring codes contain no guidance or discussion as to how the low frequency significant and maximum design values should be selected from the significant and maximum low frequency surfaces. The “worst” position on the significant and maximum low frequency surfaces will depend upon the system response under investigation. For example, in the calculation of maximum line tensions the “worst” position of the vessel associated with the mean plus maximum low frequency response will depend upon the line under investigation. In the case of surge only motions illustrated on Figure 2, the mean plus maximum low frequency tension for the bow lines will occur at the minimum extreme vessel offset (turn-around) point. While the mean plus maximum low frequency tensions for lines in the other two groups will occur at the maximum extreme vessel offset point. In reliability analyses of strength limit states, system notional probabilities of one and two line failures are synthesized from individual line notional probabilities of failure. If the individual mooring line notional probabilities of failure are to be calculated consistently, then the mooring analysis software must use the individual line’s “worst”
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
significant and maximum low frequency vessel position in calculating mooring line wave frequency tensions. This is particularly important where reliability analysis of grouped and equally spaced moored systems are being compared. Wave fr equency li ne tensions
Present codes, [10, 11], provide formulae for calculating significant and maximum wave frequency line tensions from the standard deviation of the line tension. These formulae assume a frequency domain method that is based on; transforming the vessel motion RAOs to the fairleads, synthesizing the fairlead motion spectra from the wave spectrum and the fairlead RAOs, and solving linearized dynamic equations for mooring line tensions in the frequency domain. The principle result of the frequency domain analysis is the standard deviation of the mooring line tension, from which significant and maximum line tensions are calculated assuming a Rayleigh distribution. Alternatively, the significant and maximum fairlead motions may be calculated from the fairlead motion spectra, assuming a Rayleigh distribution, and significant and maximum line tensions may then be calculated dynamically by imposing the significant and maximum fairlead motions. These are both frequency domain line dynamic methods of calculating wave frequency tensions, however due to differences in the way in which the linearization is performed the two methods will result in different predictions of wave frequency line tensions. The reliability analyst should be aware of the assumptions used in the deterministic analysis and their effects on the calculated wave frequency line tensions. Basic Variables and Probability Distributions In reliability analyses the uncertainty in the basic variables that enter into the limit state function are represented by probability distributions. Sensitivity studies may be used to assist in deciding which variables should be described as random and which may take fixed values in the reliability analysis. The results of reliability analyses are generally sensitive to the tails of a few key probability distributions and in most cases there is insufficient data available to accurately define the distributions and parameters of the basic variables. The basic variables and their distributions for the description of uncertainty in the reliability analysis of position mooring systems are further discussed below. M odeli ng uncertainty
Regardless of the sophistication of the deterministic tool used to analyze the system, it is necessary to define the modeling uncertainty, as the modeling uncertainty provides the link between the results of theory with reality. Neglecting to include the model uncertainty in the reliability analysis is identical to defining it as having a (mean) value of 1.0 with a standard deviation of zero. This is equivalent to making the two claims that; theory is perfect and cannot be improved upon and the experimental data show absolutely no scatter. In many cases of interest real responses are sensitive to boundary conditions or initial conditions and the experimental data on
7
its own will show considerable scatter for nominally identical experiments. If reliability calculations are used comparatively then the modeling uncertainty must be carefully investigated and defined for the cases being compared. For full scale mooring systems the definition of modeling uncertainty requires a comprehensive instrumentation package, monitoring environmental loading parameters and system response parameters, and the occurrence of severe weather to ensure that the responses of interest are captured. Alternatively, wave basin tests on scale models may be used to assist in defining the model uncertainty across a range of vessel and mooring system types, for different environmental criteria. For the mean, low frequency, and wave frequency components of mooring line tensions, their modeling uncertainties are post-applied, as factors, to the line tension components calculated by the deterministic mooring analysis, in which the distributions of the other basic variables are preapplied. The modeling uncertainty is meant to include the effects of conservative or non-conservative assumptions, bias, in the methods used to predict line tensions, as well as the naturally occurring scatter in the line tensions. Progressive fail ur e and envir onmental return peri ods
The definition of the basic variables describing the environmental parameters determine, the reoccurrence period that the notional probability of failure is associated with. When a mooring line fails it will be retrieved and replaced at the earliest possible opportunity, therefore the use of longterm environmental distributions is not sensible when evaluating damaged line limit states. Consider the case in which the i’th and j’th lines are the first and second lines to fail in a storm. The probability that this occurs is given by Pf (i)Pf (j | i), where Pf (i) is the intact notional probability of failure of the i’th line and Pf (j | i) is the conditional probability of failure of the j’th line, given that the i’th line has already failed. Clearly, Pf (i) and Pf (j | i) cannot be calculated based on the long-term environmental distribution, as both lines must fail within a short time interval of each other. Thus, the notional probability of failure of limit states associated with damaged mooring systems cannot be evaluated using longterm environmental distributions. However, notional probabilities of failure for damaged limit states may be synthesized from notional intact and damaged probabilities of failure calculated for individual storms, i.e. using the “shortterm” environmental distributions. If the reliability analysis is performed using the design (100-year) return period environmental distributions, then the calculated notional probability of intact and damaged failure is the conditional probability of failure given the occurrence of the design (100-year) return period storm. This has the advantage that it preserves a direct connection between the intact and damaged safety factors used in the deterministic design analysis and the notional intact and damaged probabilities of failure, as both are associated with the same (100-year) metocean conditions. However, for geographic regions with very different extreme environmental distributions, such as winter storm and tropical revolving
8
KEN HUANG AND YONG BAI
storm areas, comparisons of notional probabilities of failure conditional on the occurrence of the design return period storm may not be a true indicator of comparative safety levels. From the response-based point of view, the better approach to obtain annual notional failure probabilities is to perform reliability analyses over a range of return period weather bins, and summing the product of the conditional notional probabilities of failure with the annual probability of occurrence of the weather bins. Envir onmental basic vari ables
Gumbel (FT Type I) or Weibull (FT Type III) extreme value distributions are generally used to describe the intensities of environmental parameters. The 100-year return period distributions of wind and current velocity and the significant wave height are defined by Weibull distributions for maxima. A method for deriving consistent distributions of environmental parameters for arbitrary return periods is outlined below. Typical metocean design criteria for the Gulf of Mexico and North Sea, based on ISO 13819-1, [18], are summarized in the Tables on Figure 3. Parameters for distributions of annual maxima, which are consistent with the design criteria, are readily obtained by fitting the distributions to the 10 and 100-year (P = 0.9 and P = 0.99) return period design values. Weibull distributions for Gulf of Mexico and North Sea annual maxima are shown at the bottom of Figure 3. These distributions are consistent with the 10 and 100-year design criteria, note that the probability of exceedence (1 – P) is plotted on the y-axis. The Weibull distribution for maxima is given by:
F Wmax ( x )
exp − ε − α = 1
x
β
; x
;
x > ε
≤
ε , α > 0, β > 0
(5) The three parameters ε, α, and β are the threshold, scale, and shape parameters respectively. If F(x) is the distribution function of annual maximum values, and these annual events are independent and identically distributed, then the probability, F N(x), that in N-years the random quantity X does not exceed x is given by: N
N
F ( x )
=
∏ i
N
prob ( X
≤
x)
=1
=
∏ i
F ( x )
= [ F ( x ) ]
N
=1
(6) If the distribution for annual maxima is a Weibull distribution, then the distribution of N-year return values, F N(x), is given by:
N
F ( x )
=
[ F
Wmax
( x )
]
N
e − = exp − a
N x = b
exp
β ε − x − α 1 β N
(7)
OTC 13269
Hence, if the parent distribution is a Weibull distribution for maxima, the distribution for N-year return values is also a Weibull distribution for maxima with threshold ( ε N), scale (α N), and shape (β N) parameters given by:
ε N
=ε
α N
=
β N
=
α
(8)
1β
N
β
This self-locking property applies to all three of the extreme value distributions, [19], and is extremely useful in deriving consistent distributions for series systems. L in e segment strength basic var iabl es
Similarities in the maximum storm event (the maximum value in a series of N years) and mooring line strength (the minimum value in series of N test lengths), allow the same statistical methods to be used for both problems. A mooring line is a typical series system and the prediction of mooring line strength is one of the classic problems in extreme value theory. If F(x) is the distribution function for the break strength of test lengths, then [1 - F(x)] is the probability that the strength of a test length exceeds x. For a mooring line segment composed of N similar independent and identically distributed lengths the probability that the strength of all N lengths exceeds x is given by [1 - F(x)] N, and the probability that the strength of the line segment is less then x is 1 - [1 - F(x)] N. That is, the strength distribution, F N(x), of a line segment composed of N test lengths in series is given by: F N ( x )
= 1 − [1 − F ( x ) ] N
(9)
The Weibull distribution for minima is given by: β 1 − exp − x − ε ; x ≥ ε , α > 0, β > 0 α ( )= F Wmin x 0 ; x < ε
(10) If the distribution of test length break strengths, F(x) in eqn. (9), is a Weibull distribution for minima, then the distribution of strength for a line segment composed of N test lengths, F N(x), is also a Weibull distribution for minima, with threshold (ε N), scale (α N), and shape (β N) parameters given by eqn. (8). The formulae above allow break strength to be defined consistently for line segments of different lengths (number of test lengths N). Mooring lines may be made of chain, steel rope or 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 statistical analysis of break test data. As mooring line segments are typically of the order of 10 to 1,000 times longer than the lengths used in break tests the strength of a line segment will depend upon the left tail of the test break strength distribution. The method used to determine the “best-
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
fit” parameters for the break test data should be chosen carefully. Plotting the “best-fit” distributions on the same graph as the break data, see Figure 4, will allow the fit to the all-important left tail of the strength distribution to be examined.
distributions of basic variables, rather than the computing requirements of the Monte Carlo method. Monte Carlo methods are often used to validate other methods, [20]. However, due to the speed of present day computers, there is no reason why Monte Carlo methods should not be used for all the reliability analyses of position moored systems, rather than simply for validation, [8]. Level 3 methods evaluate eqn. (11) by “exact” methods. For Level 3 methods to be feasible, the joint distribution of the basic variables X1,…, Xn must be known. The failure probability Pf can then be written as the n-dimensional integral
Sensiti vity studi es and other basic var iabl es
In addition to modeling uncertainty, environment, and strength, other variables whose uncertainties are judged to be important, either by experience or by sensitivity studies, should be represented as random variables in the reliability analysis. Deterministic sensitivity studies which encompass the range of vessel and mooring system types, water depths, and environmental conditions over which reliability analyses are to be performed can be used to assist in identifying those variables that should be represented as random variables. The sensitivity plots on Figure 5 show the tension in the most loaded line for variations of ±15% in the 1-hour average wind velocity, pretension in line 2 and heave natural period. The probability distribution used for each basic variable should be based, as far as possible, on the statistical analysis of supporting background data. In general, normal, lognormal, Gumbel, or Weibull distributions should be used. When no detailed information is available, the three-parameter lognormal and Weibull distributions should be used to represent variables that are known to have lower or upper limits, while the normal and Gumbel distributions should be used for distributions that are unlimited.
∫ ∫ f ( x)dx K
Classification of Reliability Methods Reliability theory provides methods for the approximate evaluation of the failure probability, (11) P f = Prob g X 1 , K , X n , C ≤ 0
((
)
)
In order that eqn. (11) can be evaluated the joint probability distribution of the basic variables has to be known, or alternatively independence of X1,…, Xn is assumed and knowledge of the marginal distributions suffices. The choice of the marginal distributions arises from statistical studies analyzing available data for the respective basic variables. The Law of Large Numbers, [19], allows the following frequency interpretation of the probability of failure for large l , P f
=
1 l
×
number of indices
i = 1, K , l with g ( X i , C )
≤
(12) Where Xi, i = 1, 2,…, l is an independent sequence of random vectors each having the distribution of X=(X 1,…,Xn)T. Monte Carlo methods exploit eqn. (12) to arrive at an approximate value of Pf , [20], using independent random samples from the joint distributions, if they are known, of the variables, X1,…, Xn. As the failure probability of the limit state function decreases the computing time required by Monte Carlo methods increases. However, in practice the accuracy with which small probabilities of failure can be calculated will usually be controlled by the lack of data available to define the tails, extreme values, of the
9
0
x
1
K dx n ,
{x ∈ℜ n : g( x ) ≤ 0}
(13) Where f X denotes the density of the distribution of X=(X 1,…, Xn). Numerical integration may then be used for the evaluation of the integral above. However, this is not practical for multiple degrees of integration. These methods are often not feasible due to a lack of knowledge of the "true" joint distribution of the variables, or because of the complexity of the integration required, [1, 20]. Level 2 methods are approximate techniques for calculating Pf by replacing the limit state function, eqn. (11), by an approximating function, which renders the problem amenable to analysis. In Level 2 methods the original basic variables, (X1,…, Xn), are transformed into independent standard normal variables, (Z1,…, Zn), which implies a transformation of the original g-function to a new limit state function f. Working with the n-dimensional standard normal distribution is computationally convenient due to the rotational symmetry of its density and because the probability content of a half-space can be easily represented as Φ(-β), where Φ is the standard normal distribution and β denotes the distance of the separating hyperplane from the origin. Level 2 methods involve an iterative search for the minimum distance of the failure surface from the origin, in the n-dimensional hyperspace, [8], as shown on Figure 6. First order reliability methods (FORM) approximate the limit state surface, at the closest point to the origin, by its tangent hyperplane, and the probability of failure is approximated by the probability content of the half-space lying near the failure region, [22, 23]. Second order reliability methods (SORM) also utilize the curvatures of the failure surface in arriving at the approximation of the failure probability, [22]. Level 1 methods are design methods that use a number of partial safety factors, which are related to pre-defined characteristic values of the basic variables, to provide the necessary level of structural reliability. That is, Level 1 methods are not methods of structural reliability, but methods of safety checking used in design, [2]. The main limitations of level 2 methods, illustrated on Figure 6, are those associated with any optimization process which are: (1) Convergence of the processes does not guarantee that the global minima has been found. (2) Only one limit state may be evaluated at time, i.e.
10
KEN HUANG AND YONG BAI
separate iterative searches for minima must be performed for each mooring line in intact and damaged conditions. The Monte Carlo method may be used to evaluate the reliability of position mooring systems. Monte Carlo methods do not involve an iterative search, consequently any number of limit states may be evaluated simultaneously. This is particularly useful for mooring systems in which a number of the mooring lines contribute significantly to the total notional intact and damaged probabilities of one and two line failures.
Conclusion The following can be concluded from this paper: (1) Where there is uncertainty or variability in the basic (input) variables, post-applied safety factor formats of the LRFD type will produce inconsistent safety levels for mooring systems that have different nonlinear response characteristics. (2) The use of pre-applied safety factor formats has yet to be demonstrated that a reasonably simple and practical format can be devised. (3) The ability of the analysis method(s) to predict low frequency responses should be investigated and where possible calibrated against model test results. (4) The reliability analyst should be aware of the assumptions used in the deterministic analysis and their effects on the calculated wave frequency line tensions. (5) Sensitivity studies can be used to assist in deciding which variables should be described as random and which may take fixed values in the reliability analysis. (6) Notional probabilities of failure for damaged limit states should be synthesized from notional intact and damaged probabilities of failure calculated for individual storms, i.e. using the “short-term” environmental distributions. (7) Finally, reliability methods can be a powerful tool for comparing the safety implications of different designs, and target safety levels for new designs, can be established by performing reliability calculations for old designs that have a history of satisfactory performance.
8.
9.
10.
11.
12.
13.
14.
15.
16. 17. 18.
19.
References 1. Rationalization of safety and serviceability factors in structural codes, CIRIA, Report 63, 1977. 2. General principles for the verification of the safety of structures, ISO 2394, International Standards Organization, 1986. 3. Recommended Practice for the Analysis of Spread Mooring Systems for Floating Drilling Units, API RP 2P, 2nd Ed, American Petroleum Institute, 1987. 4. Recommended Practice for In-Service Inspection of Mooring Hardware for Floating Drilling Units, API RP 2I, 1st Ed., American Petroleum Institute, 1987. 5. Specification for Mooring Chain, API Spec 2F, 6 th Ed., American Petroleum Institute. 6. Qualification Testing of Steel Anchor Designs for Floating Structures, API RP 2M, 2 nd Ed., 1996, American Petroleum Institute, 1997. 7. Recommended Practice for Design, Selection,
20.
21.
22.
23.
OTC 13269
Operation, and Maintenance of Marine Drilling Riser Systems, API RP 2Q, 2 nd Ed., American Petroleum Institute, 1993. Calibration of ABS, API, HSE, and NMD Mooring Design Codes for Drilling and Production Platforms, Mooring Code JIS, Noble, Denton & Associates, Inc. Rpt. No. H3687, 1995. Recommended Practice for Design, Analysis, and Maintenance of For Floating Production Systems, API RP 2FP1, 1st Ed., American Petroleum Institute, 1993. Recommended Practice for Design and Analysis of Stationkeeping System for Floating Structures, API RP 2SK, 1st Ed., American Petroleum Institute, 1997. Draft ISO Standard for Design and Analysis of Stationkeeping Systems for Floating Structures, Informative April 14, 1999, Normative Nov. 8, 1999, ISO/TC67/SC7/WG5/P5 Committee Correspondence. Recommended Practice for Design, Manufacture, Installation, and Maintenance of Synthetic Fiber Ropes for Offshore Moorings, API RP 2SM, Draft, American Petroleum Institute, 1999. ABS Guide for Building and Classing Floating Production Installations, American Bureau of Shipping, June, 2000. Recommended Practice for Design, Selection, Operation, and Maintenance of Marine Drilling Riser Systems, API RP 16Q, 2 nd Ed., American Petroleum Institute, 1993. Recommended Practice for Design and Operation of Subsea Production Systems, API RP 17A, 1 st Ed., American Petroleum Institute, 1987. Recommended Practice for Flexible Pipe, API RP 17B, 1st Ed., American Petroleum Institute, 1988. Comparison of Marine Drilling Riser Analyses, API Bull. 16J, 1st Ed., American Petroleum Institute, 1992. Draft ISO Standard for Fixed Steel Offshore Structures, General Annexes, Draft C ISO 13819-1, ISO/TC67/SC7/WG3 Committee Correspondence, 1997. Mann, N.R., Schafer, R.E., and Scigpunwalla, N.D., Methods for Statistical Analysis of Reliability and Life, Wiley, N.Y., 1974. Sweeting, T.J., and Finn A.F., A Monte Carlo Method Based on First- and Second-Order Reliability Approximations, Structural Safety, 11, pp. 203-212, 1992. Shinozuka, M., Basic Analysis of Structural Safety, Journal of Structural Engineering, ASCE, Vol. 109, No. 3, pp. 721-740, 1983. Madsen, H.O., First Order vs. Second Order Reliability Analysis Of Series Structures, Structural Safety, Vol. 2, pp. 207-214, 1985. Dolinski, K., First-order second-moment approximation in reliability of structural systems: critical review and alternative approach, Structural Safety, Vol.1, pp. 211231, 1983.
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
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
11
12
KEN HUANG AND YONG BAI
3x3 Grouped Mooring Pattern
Figure 2 Comparison of Linear and Nonlinear Frequency Domain Methods of Calculating Low Frequency Offsets
OTC 13269
OTC 13269
RELIABILITY METHODS FOR DEEPWATER POSITION-MOORING DESIGN AND ANALYSIS
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
13
14
KEN HUANG AND YONG BAI
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
OTC 13269
ABS BIOGRAPHIES KEN HUANG Ken Huang is a graduate of the Texas A&M University with a M.S. in Ocean Engineering and Hydrodynamics. 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 participating 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
Copyright 2001, Offshore Technology Conference This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, 30 April–3 May 2001. This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
Abstract The oil and gas production industry has expressed an interest in being able to utilize Floating, Production, Storage and Offloading (FPSO) units as a development option in the deepwater areas of the United States (U.S.) Gulf of Mexico (GOM) outer continental shelf (OCS). Operators will need regulatory approval from both the United States Coast Guard (USCG) and the Minerals Management Service (MMS) for a FPSO project. Neither the USCG nor MMS currently have regulations specifically for the design and operation of FPSOs on the OCS. A workgroup was formed under the Offshore Operators Committee’s Deepwater Subcommittee to assist MMS and USCG in reviewing the existing regulations and body of standards, specifications, recommended practices and classification society rules and guides concerning the design and operation of FPSOs on the OCS for the Gulf of Mexico. The effort also aimed to identify gaps in the regulations and industry standards. This paper provides a summary of the major findings of the workgroup. In addition to the items discussed in this paper, the workgroup report identified other areas where additional modifications to regulations or industry standards may be warranted. Introduction As operators have moved into the deeper waters of the GOM over the last several years, interest has been growing in potentially utilizing FPSOs as a development option to the floating production systems (tension leg platforms (TLP), spars, etc) and subsea tie backs to either floating production systems or fixed platforms that are currently being utilized. In discussions with MMS and USCG, it became apparent that several studies would need to be conducted to confirm the acceptability of these systems for the GOM. MMS indicated
that an Environmental Impact Statement (EIS) would be prepared for the first FPSO proposed to be utilized in the GOM. In working with industry, MMS agreed to do a programmatic EIS on the generic use of FPSOs in the GOM if industry would fund the study. The joint industry project DeepStar agreed to fund the EIS 1. The draft EIS on the Proposed Use of Floating Production, Storage and Offloading Systems on the Gulf of Mexico Outer Continental Shelf in the Western and Central Planning Areas 2,3 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 Decision to be published no earlier than 30 days from the publication of the final EIS. The second study effort was to do a comparative risk assessment to evaluate and compare oil spill and fatality risks for the FPSO with a spar, a TLP and a shallow-water jacket serving as a hub and host to deepwater production. The Offshore Technology Research Center completed that study for MMS4. The third step in the process was to identify any gaps in the existing regulations and to develop a regulatory model that could be used by MMS and the USCG in the review and approval of a FPSO project. On March 22, 2000, Mr. Chris Oynes, MMS GOM Regional Director, sponsored a meeting between MMS, USCG and industry to discuss the regulatory requirements for FPSOs in the GOM, should they be found to be an acceptable development option. Although it was recognized that MMS and the USCG would have to agree among themselves the appropriate regulations and regulatory split between the two agencies, they agreed that it would be beneficial and appropriate to have industry provide input on the model. It was decided that a workgroup would be formed under the direction of the OOC Deepwater Subcommittee and consists of industry representatives and classification society representatives along with personnel from MMS and the USCG. A report was prepared by the workgroup and submitted to MMS and the USCG in September 2000 for their consideration5.
Regulatory Model Workgroup Goal The overall goal of the workgroup was to review the existing regulations and industry standards covering the design, construction and operation of FPSOs in the GOM and identify
2
WANDA J. PARKER AND TODD W. GROVE
any gaps in either the regulations or standards that needed to be addressed prior to bringing FPSOs into the GOM. The workgroup was focused on the design and operational considerations for the FPSO. The work group did not address the design of shuttle tankers or operational considerations once they were disconnected from the FPSO. Participants The workgroup was formed under the direction of the OOC Deepwater Subcommittee and met five times following the initial meeting. It was agreed that a cooperative effort with open discussions between the regulatory agencies and industry was desired and would produce the best work product. Due to the broad scope of the discussions, it was necessary to include a large number of participants. MMS agreed to have Mr. James Regg, Section Chief, Technical Assessment and Operations Support serve in the workgroup along with support from other MMS personnel. LCDR Bill Daughdrill from the Eighth USCG District represented the USCG in the work group. Personnel from the MMS Headquarters and USCG Headquarters groups were kept informed of the workgroup’s activities through e-mail. Twenty-five persons representing 17 companies participated in one or more of the meetings. Since classification societies have historically played a large role in the approval of FPSOs worldwide, it was felt that it was important to have broad representation from the major classification societies who currently class FPSOs. Four classification societies; American Bureau of Shipping, Bureau Veritas, Det Norske Veritas and Lloyd’s Register of Shipping were represented in the workgroup. Tim Sampson represented the American Petroleum Institute (API). Wanda Parker agreed to chair the workgroup for OOC. A complete listing of all workgroup participants is in the workgroup report5. All of these individuals dedicated a considerable amount of time and expertise to this effort. FPSO Characteristics For consistency, the workgroup decided to use the FPSO characteristics used in the EIS and CRA studies as the FPSO model for this effort. The workgroup considered the regulations that would apply to a US flag FPSO or an undocumented FPSO that is designed to US flag requirements (similar to the existing GOM floating production systems). Limited discussions were held on the differences in permitting a US flag FPSO and a foreign flag FPSO. The FPSO was considered to be ship-shaped with limited discussions on the differences between a ship-shaped FPSO and a non shipshaped FPSO. Discussions were limited to a permanently moored FPSO for simplicity since a disconnectable FPSO introduces many complicating factors into the discussion. Finally, discussions were focused on systems that are unique to FPSOs with only limited discussion of systems that are common to fixed platforms or other floating production systems. This list of system includes: 1. In Hull Cargo Storage Systems 2. Cargo Piping and Transfer 3. Turret/Mooring/Stationkeeping/Swivel
4. 5. 6. 7. 8. 9. 10. 11. 12.
OTC 13170
Propulsion Stability Motions/Global Performance Risers Offloading Systems Layout Operational Considerations Discharges Manning
Regulations MOU It is recognized that both MMS and USCG have a large body of regulations that could be applied to FPSOs. In the Memorandum of Understanding (MOU) between MMS and USCG6 signed on December 16, 1998, the delineation of jurisdictions regarding floating production system components, operations and issues is addressed. The workgroup reviewed the MOU for completeness in coverage of systems on a FPSO and put together a table showing the applicable regulations, industry standards and classification society rules for each system and sub-system in the MOU. While this table was not exhaustive, it quickly pointed out areas where regulations and standards were well established and systems where either regulations or standards were lacking. The work group made some specific recommendations concerning the implementation of the MOU that will be discussed below. MMS MMS regulations that are applicable to FPSOs are primarily located in 30 CFR 250, Subparts H and I. In conjunction with their regulations for specific systems, MMS has said that they intend to utilize the Deepwater Operations Plan (DWOP) process in their review of a FPSO project7. The workgroup reviewed the MMS regulations for adequacy and made some specific recommendations that will be discussed below. USCG The USCG has said that FPSOs will be regulated as vessels and therefore will be required to meet specific vessel design and operational regulations8,9. USCG regulations that are applicable to FPSOs are primarily located in 46 CFR, Subchapters D and IA and 33 CFR Subchapter N. On Dec 7, 1999, the USCG published a Notice of Proposed Rulemaking for 33 CFR Subchapter N that includes proposed regulations applicable to FPSOs. The workgroup considered these proposed regulations in addition to the established regulations. The workgroup report was submitted to the record as a comment to the proposed regulations. The specific recommendations will be discussed below. Industry Standards Both USCG and MMS regulations incorporate by reference a large number of industry specifications and recommended practices, particularly API documents. Many of these are applicable to FPSO system design or operation. In addition,
OTC 13170
FPSO STANDARDS AND RECOMMENDED PRACTICES
there are a host of other international standards that may be applicable to FPSOs. Most of these standards cover individual systems or subsystems that may be used in conjunction with a wide variety of types of installations. It was outside of the scope of the workgroup to conduct a thorough review of the adequacy of the individual industry specifications and recommended practices. Rather, the workgroup focused on a few key standards and made recommendations of standards the agencies should consider incorporating by reference. Classification Society Rules and Guides Another body of standards applicable to FPSOs and recognized worldwide are those developed and applied by a number of the major classification societies. The rules and guides of the societies are utilized in assessing the fitness-for purpose of FPSOs and are focused on safety aspects. The classification requirements address design requirements as well as those for fabrication, installation, and commissioning. Unlike the snapshot nature of certification, classification is an ongoing process by which the societies survey a FPSO periodically during its operational life to ensure compliance with the rules. Given their unique independent role internationally, classification societies are also often delegated statutory responsibilities by flag and coastal state authorities to act on behalf of the administration. USCG has, to varying degrees, delegated to several classification societies approval responsibilities for existing GOM floating production units. Expectations are that similar delegations will be forthcoming as experience is gained in GOM FPSO applications. Several classification societies have also been active to varying degrees in the MMS Certified Verification Agent program for existing structures, which may also be extended to FPSOs. The relevant practical experience from classification and statutory responsibilities is utilized in maintaining the rules and guides in a current and relevant form. The four participating societies contributed citations to the regulatory matrix for the most prevalent rule and guide requirements applicable to the various aspects of the FPSO.
Gaps and Recommendations During the review process, the workgroup identified a number of recommendations either for ways to close existing gaps in the regulations or for additional work that should be considered. These recommendations were broken out into areas for consideration by MMS and USCG and for industry. MMS In 30 CFR 250, Subpart I, MMS has established a platform verification program requiring a third party certification of the platform design. These regulations were established for fixed platforms, but MMS has been applying them to floating production facilities. It is recommended that MMS revise these regulations to update them for floating facilities, including FPSOs. The workgroup recommended adding the turret, risers and mooring systems to the verification program.
In lieu of the agency writing prescriptive regulations for
3
FPSOs, the workgroup recommended that MMS consider incorporating by reference additional industry standards and recommended practices as they become available. The following were identified as candidates for MMS review and consideration for incorporation in their entirety 10,11,12,13,14: 1. API RP 2FPS, Planning, Designing and Constructing Floating Production Systems 2. API RP 2SM, Design and Analysis of Synthetic Moorings 3. API RP 2SK, Design and analysis of Stationkeeping Systems for Floating Structures 4. API RP 2RD, Design of Risers for Floating Production Systems (FPSs) and Tension-Leg Platforms (TLPs) 5. API Spec 17J, Specification for Unbonded Flexible Pipe USCG On December 7, 1999, a Notice of Proposed Making was published revising 33 CFR, Subpart N. In that revision, both Mobile Offshore Drilling Unit (MODU) regulations located in 46 CFR, Subchapter I-A and tank vessel regulations located in 46 CFR, Subchapter D were referenced. However, the regulation was not clear in many cases which regulation should be followed if both regulations cover the same system or subsystem. Also, in many cases, modifications to these regulations may be needed to address the unique circumstances of FPSO or floating system operations versus a tank vessel or MODU. The workgroup recommends that specific regulations for floating facilities should be written in lieu of pointing to regulations for various types of vessels, which may not be completely applicable to floating production facilities. The workgroup also recommended that their report be included in the comment record for the proposed rulemaking. The workgroup recommended that marine crew manning and qualification regulations should be codified for FPSOs and for other floating production systems. The Eighth Coast Guard District has issued a policy letter 15 for marine crew manning for floating production systems other than those storing oil in bulk, which could serve as an appropriate starting base for non-self propelled FPSOs. It was recognized that self-propelled FPSOs would need additional consideration. MMS has adopted API RP 500/505 16,17 for area classification while the USCG has prescriptive regulations in 46 CFR, Subpart J. The workgroup recommended that the USCG adopt API RP 500/505 for floating production systems including FPSOs. Adopting common standards would minimize confusion and duplication of effort for industry since both agencies have jurisdiction for area classification. Although the USCG has been given sole jurisdiction over fire fighting systems for floating production systems, including FPSOs, the USCG has not yet proposed regulations for fire fighting systems in the production area. It is recommended that the USCG adopt fire-fighting regulations for the production area.
4
WANDA J. PARKER AND TODD W. GROVE
Both MMS and USCG have voluntary safety management system programs that could be applicable to FPSOs. MMS has recognized API RP 75 18 as an acceptable basis for a safety management system for fixed and floating production systems on the US OCS. Most oil and gas production companies have based their safety management programs for operations on the US OCS on API RP 75. However, the USCG has recognized the International Safety Management Code (ISM) 19, which is applicable to vessels that must comply with Chapter IX of SOLAS. While both programs have merit, and individual companies may want to base their programs on either standard, a combination of the standards or some other standard, the workgroup recommends that the USCG recognize API RP 75 as an acceptable basis for a safety management program in addition to ISM. In the event that foreign flagged FPSOs are acceptable for use in the GOM, the USCG will issue a Letter of Compliance (LOC)8. Since a full design review is not normally conducted for a LOC, the workgroup recommended that the USCG develop a LOC checklist that could be either used with existing foreign flagged FPSOs or new-built foreign flagged FPSOs proposed for operations in the GOM. Although the USCG has general lightering regulations and additional operational regulations that apply to the designated lightering areas in the GOM, there are no specific operational regulations that apply to FPSOS or to tandem offloading. It is recommended that a work group be formed to review existing international standards to determine their adequacy for GOM operations. Joint MMS and USCG In the MOU, both MMS and USCG have been given jurisdiction for reviewing and approving the design of the turret and mooring system. The workgroup agreed that technology is rapidly evolving for these systems and that it would be burdensome on the regulatory agencies to have personnel fully knowledgeable about these systems as they change. It is recommended that a verification agent acceptable to both agencies be selected to review and certify the design. In the MOU, both MMS and USCG have been given jurisdiction for reviewing and approving various portions of the integrated monitoring and safety systems. It is recommended that a work group consisting of representatives of Industry, MMS and the USCG be formed to address the integration of these systems. In the MOU, both MMS and USCG have been given jurisdiction over piping systems. It is recommended that for cargo tank piping that the specification break between MMS and USCG jurisdiction occur at the 1 st valve downstream of the last processing vessel (and its control valves and safety system) prior to the oil entering the cargo storage tanks. It is recommended that a work group be formed consisting of representatives of industry, MMS and USCG to review other similar systems and agree to where the specification breaks between the systems should occur. These breaks should be codified into the MMS and USCG regulations. Alternatively, MMS and USCG should consider adopting consistent industry
OTC 13170
standards for piping systems. Neither MMS nor USCG regulations address integral hull tanks used as process vessels (such as wet/dry oil tanks). The work group recommended that all integral hull tanks be under USCG jurisdiction for structural design. For tanks used as process vessels, the safety system, control valves, and piping to and from the process vessels should be under MMS jurisdiction. It is recommended that piping specification breaks should occur at the first flange outside the tank. Industry Both the Environmental Protection Agency (EPA) and the USCG have regulations concerning discharges to the ocean that could occur from FPSOs. It is recommended that a work group under the OOC Deepwater Subcommittee be formed to review the overboard discharge regulations of both agencies to ensure that all discharges are adequately addressed for FPSO operations. Although the USCG has regulations that provide for the inspection of foreign flagged tank vessels, no formal determination has been requested or received from US Customs that indicates if foreign flagged FPSOs will or will not be allowed to operated on the OCS (i.e. Will Customs interpretation and policy for foreign flagged FPSOs be similar to their current interpretation and policy for foreign flagged MODUs?) The workgroup recommended that industry develop a strategy for obtaining a formal, written determination from Customs. It was recommended that the various API standards and recommended practices for mooring systems and riser design be reviewed to determine if the inspection guidelines given in those documents are adequate for floating production systems and updating the documents as needed. It was recommended that API RP 14C be reviewed and revised if needed for floating production systems, including taking into account the effect environmentally induced motions may have on the safety and monitoring systems. In addition, it is recommended that the safety and monitoring systems for swivels, integral hull tank process vessels and other unique systems to a FPSO be covered in the recommended practice. It was recommended that API RP 14E 20 be reviewed for adequacy and updated as needed for the effects of motion and piping support. It was recommended that the various API documents on composite materials should be reviewed for adequacy for floating facilities including FPSOS. It was recommended that API RP 75 be reviewed and updated as needed to ensure the document is an adequate basis for a safety management system for FPSO systems and operations. Applicable portions of ISO 9000/14000 could be incorporated, if desired. The work group recognized that standards for cargo tank cleaning for FPSOs might be different from trading tankers since inspections may occur when the FPSO is on location and in operation and not in dry dock. It is recommended that appropriate standards for cargo tank cleaning be confirmed.
OTC 13170
FPSO STANDARDS AND RECOMMENDED PRACTICES
Conclusions While neither MMS nor USCG has regulations that specifically apply to the design or operation of FPSOs, there is a multitude of regulations that are applicable to FPSOs and with the modifications indicated in the work group’s recommendations, the existing framework is adequate. A large body of industry recommended practices and standards consisting of classification society rules and guides, API standards, specifications and recommended practices and international standards exist that cover FPSO system design and operation. In some cases these need to be reviewed and updated, where needed, to ensure they are applicable to GOM FPSO operations. Many of the recommendations identified for FPSOs are also applicable to other types of floating facilities. The majority of the discussions by the workgroup focused on systems that are unique to FPSOs. The FPSO model used for the discussions was a ship-shaped permanently moored FPSO. It is recognized that a FPSO is not dependent on shape and some variations occur as you move between ship-shaped and non-ship-shaped FPSOs that will need additional attention. Likewise, a disconnectable, self-propelled FPSO has many aspects that are different from a permanently moored FPSO that will need to be considered. The review of the existing regulations was conducted without representation from the USCG Marine Safety Center or Headquarters groups. These groups need to be fully engaged before final determinations can be made on the appropriate regulations for FPSO design and operations. The representatives from the Eighth Coast Guard District participated fully in the discussions, but it was recognized by the workgroup that they were not authorized to speak definitively on USCG policy or regulation. OOC appreciated the opportunity to take the lead role in this cooperative effort between MMS, USCG and Industry. By working together and pooling our thoughts and ideas, regulations that meet the needs of all concerned can be put in place for FPSOs in the GOM. A large number of industry representatives and classification society representatives dedicated a considerable amount of time and expertise to this task. The efforts of Mr. Jim Regg and LCDR Bill Daughdrill were recognized for their active participation in the work group. As policies and rulemaking for FPSOs are drafted, continuing the cooperative effort between the regulators, industry and the classification societies will be beneficial. Acknowledgments The authors would like to thank the Offshore Operators Committee for allowing us to publish the results of the work group. References 1. 2.
Verret, Allen J., Hays, Paul R., “Deepstar’s Program Related to FPSO’s”, Offshore Technology Conference, 10703, May 1999. “Proposed Use of Floating Production, Storage, and Offloading Systems on the Gulf of Mexico Outer Continental Shelf,
3.
4.
5.
6.
7.
8.
9. 10.
11.
12.
13.
14. 15.
16.
17.
18.
19.
20.
5
Western and Central Planning Areas, Draft Environmental Impact Statement”, Minerals Management Service, Gulf of Mexico OCS Region, August 2000. George, J.E., Parker, W.J., Cranswick, D.J., “FPSO Environmental Impact Statement: What is Happening?”, Offshore Technology Conference, 10705, May 1999. Gilbert, R.B., Ward, E.G., Wolford, A.J., “A Comparative Risk Analysis of FPSO’s with Other Deepwater Production Systems in the Gulf of Mexico”, Offshore Technology Conference, 13173, May 2001. Letter to Carolita Kallaur, Minerals Management Service, from Allen Verret, Executive Director, Offshore Operators Committee, Regarding Regulatory Framework – Floating Production Storage and Offloading Systems, September 2000. Memorandum of Understanding (MOU) Between the Minerals Management Service and the United States Coast Guard”, Federal Register, Vol. 64, No. 10, January 15, 1999. Regg, James B., “Floating Production, Storage and Offloading Systems in the Gulf of Mexico OCS: A Regulatory Perspective”, Offshore Technology Conference, 10701, May 1999. Daughdrill, W.H., Brown, M.J., “The Regulatory Scheme Applicable to Floating Production, Storage, and Offloading Systems”, Offshore Technology Conference, 10702, May 1999. Letter to Carolita Kallaur, Minerals Management Service, from RADM North, United States Coast Guard, November 16, 1998. “API Recommended Practice for Planning, Designing, and Constructing Floating Production Systems”. American Petroleum Institute, 2000 Draft. “Recommended Practice for Design, Manufacture, Installation, and Maintenance of Synthetic Fiber Ropes for Offshore Mooring”. American Petroleum Institute, November 1999 Draft. “API Recommended Practice for Design and Analysis of Stationkeeping Systems for Floating Structures”. American Petroleum Institute, December 1996. “API Recommended Practice for Design of Risers for Floating Production (FPSs) and Tension-Leg Platforms (TLPs)”. American Petroleum Institute, June 1998. “API Specification for Unbonded Flexible Pipe”. American Petroleum Institute, July 2000. The Eighth Coast Guard District policy letter for marine crew manning for floating production system other than those storing oil in bulk “API Recommend Practice for Classification of Locations for Electrical Installations at Petroleum Facilities Classified as Class I, Division 1 and Division 2”. American Petroleum Institute, November 1997. “API Recommend Practice for Classification of Locations for Electrical Installations at Petroleum Facilities Classified as Class I, Zone 0, Zone 1 and Zone 2”. American Petroleum Institute, November 1997. “API Recommend Practice for Development of a Safety and Environmental Management Program for Outer Continental Shelf Operations and Facilities”. American Petroleum Institute, July 1998. “International Management Code for the Safe Operation of Ships and for Pollution Prevention”, International Maritime Organization, Assembly Resolution A.741(18), 1993. “API Recommend Practice for Design and Installation of Offshore Production Platform Piping Systems”. American Petroleum Institute, October 1991.
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 Divisional 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. Copyright 2001, Offshore Technology Conference This paper was prepared for presentation at the 2001 Offshore Technology Conference held in Houston, Texas, 30 April–3 May 2001. This paper was selected for presentation by the OTC Program Committee following review 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 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, 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 print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.
Abstract This paper describes a study to compare the risks of Floating Production Storage and Offloading Systems (FPSO’s), which have never been used in the Gulf of Mexico, with the risks for existing deepwater production systems in the Gulf of Mexico. The major conclusion is that the expected risks for fatalities and oil spills associated with FPSO’s are comparable to those for already accepted alternatives for deepwater production. In addition, the oil spill risks are dominated by spills that occur during transportation of oil from the production facility to the shore with either pipelines or shuttle tankers. The major recommendation is to periodically update these results so that they serve as a baseline for future analyses of risk in the Gulf of Mexico. This study was undertaken for the Minerals Management Service to provide information for their use in developing a policy for FPSO’s in the Gulf of Mexico. Introduction To date, deepwater (more than 3,000-foot water depth) reserves in the Gulf of Mexico have been developed primarily with the following types of production systems: Spars; Tension Leg Platforms (TLP’s); and Subsea Well Systems tied back to these floating systems or to shallow water jackets that may also serve as hubs for other deepwater production systems (Hub/Host Jacket). All three of these types of systems rely on pipelines to transport oil to shore. A potentially attractive alternative to these systems is a tanker based Floating Production Storage and Offloading (FPSO) system with oil transportation to shore via shuttle tankers. Floating Production Storage and Offloading systems have been used in many areas of the world, but not the Gulf of Mexico.
The Minerals Management Service (MMS) funded the Offshore Technology Research Center (a National Science Foundation engineering research center located at Texas A&M University and The University of Texas at Austin), with EQE International, Inc. as a subcontractor, to conduct a Comparative Risk Analysis (CRA). The purpose of this study was to assess and compare the system risks for FPSO’s with those for existing deepwater production systems, specifically 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 Mexico 1. Information from both the Comparative Risk Analysis and the EIS will be used by the MMS in developing policies concerning the use of FPSO’s in the Gulf of Mexico. The primary objectives of the Comparative Risk Analysis were the following: 1. Assess and compare the system risks for FPSO’s with those for existing deepwater production systems, specifically Spars, TLP’s, and Hub/Host Jackets; and 2. Understand the contributions to system risk by subsystems and phases of operation.
Approach The approach used to conduct the Comparative Risk Analysis was developed with the following goals in mind: 1. Provide the MMS with information that can be used for a consistent and objective comparison of the risks associated with the four production systems; 2. Provide the MMS with a level of detail necessary to compare and understand overall system risks for typical production systems; and 3. Incorporate industry data, experience and expertise to the greatest extent possible into evaluating the risks. The approach used to achieve these goals involved teams of experts and a series of workshops. Participation of Technical Experts. Historical data on actual failures, particularly those very infrequent failures with large consequences that tend to drive overall risks, are scarce, and the risks must be estimated by other means. For this study, we chose to directly involve the expertise and experiences of engineers involved in the design and operation of these
2
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
production systems. The Deepstar consortia facilitated and coordinated the participation of industry engineers in this project. There was active participation by experienced engineers representing all segments of the industry, including oil companies, consultants, manufacturers, contractors, classification societies, as well as the regulatory agencies. They brought a detailed understanding of the nature of these risks as well as design and operational options to manage these risks. Many of the industry engineers who were involved in this study had also participated in risk studies within their companies, which are often undertaken either in the design of a specific system or to compare several systems in selecting the most appropriate system for a given project. The practical experience and perspective that these engineers brought to the study was deemed critical to the success of this study. Separate teams were formed for each of the four production systems, the Spar, the TLP, the Hub/Host Jacket, and the FPSO. These teams were made up of invited participants from industry and representatives from the MMS and the U. S. Coast Guard (USCG), the government agencies responsible for regulating the deployment and operation of deepwater production systems. The teams were designed to include engineers with expertise and experience in the design, construction and operation of the overall systems as well as the subsystems and components that make up the systems. There was an average of about ten members per team. The companies that provided one or more participants to these system teams are listed in Table 1. It is worth noting that these companies represent a large measure of the offshore industry’s deepwater experience and expertise. They have been very active in the design, operation, and/or certification of deepwater production systems in the Gulf of Mexico and elsewhere. Of particular importance is their direct involvement and experience with the deepwater production systems used in this study: Spars, TLP’s, Hub/Host Jackets, and FPSO’s. The teams were balanced to include members with overall systems expertise as well as those with expertise in various sub-systems, components, and operations, including: • Platform and subsea well systems; • Drilling and well intervention operations for both platform and subsea wells; • Topsides (processing facilities, equipment); • Production operations; • Pipelines and flowlines; • Tanker and FPSO design and operations; • Structures (hulls, decks, mooring systems, riser systems); • Helicopter operations (personnel transport); • Supply boat operations (material & personnel transport); and • Diving operations. Thus the teams were able to focus on risks at the sub-system, component, and operational levels as well as to focus on
OTC
overall system risks. Additional contributions from industry included input on detailed hazard identifications through participation in formal specialty interview sessions, and various other interactions in which individuals provided data, input, or advice. Technical experts from the companies and organizations listed in Table 2 as well as from most of those listed previously in Table 1 contributed in these areas. In all, over 100 of the industry’s more experienced engineers directly participated in the study either through the system teams and workshops, or the specialty interview sessions. The names and affiliations of these participants are summarized in Gilbert et al. 2. The average experience level for these experts was approximately 20 years. The level of participation by the industry experts was substantial. Their direct involvement in the workshops (preparation, participation, and review) and the specialty interviews involved an estimated 5,000 man-hours. Further, these experts often sought additional input and review from colleagues in their companies, and gathered additional relevant information for the study. Workshop Process. A flowchart for the workshops is shown in Fig. 1. Individual, one-day workshops were conducted for each system during the first three phases (Workshops #1 to #3). The final workshop was held collectively over a two-day period. The activities conducted between workshops are also indicated on Fig. 1. The objective of Workshop #1 was to develop conceptual system descriptions for the four production systems. This work included establishing study boundaries in space and time and describing the physical and operational features for each system. Draft system descriptions were distributed to the workshop participants ahead of the workshop and then used as the starting point in the workshop. The objective of Workshop #2 was to perform hazard identifications for each system. A list of possible adverse events (or initiating events) that could contribute to risk was developed for each study system and organized by sub-system or activity. Detailed hazard identifications were developed through a series of specialty interviews with technical experts before Workshop #2. The participants and topics for these interview sessions are summarized in Gilbert et al. 2. These detailed lists were then reviewed during Workshop #2 and used to develop a framework for the quantitative risk assessment. The objective of Workshop #3 was to elicit quantitative information about frequencies and consequences for oil spills and fatalities. A preliminary quantitative risk assessment based entirely on raw data was distributed to the workshop participants before the workshop. This preliminary risk assessment was then refined during Workshop #3 and needs for additional information and studies were identified. The objective of Workshop #4 was to review and refine the risk assessment. Information from additional studies conducted between Workshops #3 and #4 was incorporated into this review.
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
The general work processes used for the workshops was as follows. Preliminary information that had been distributed to the participants was reviewed and refined through an open forum process that included time for discussion and developing a consensus regarding the input on risks. The open forum approach encouraged an iterative and synergistic discussion of risks and information from different perspectives. The participation of both the industry and the regulatory agencies helped to provide balance and objectivity in the discussions and input. Consensus was generally readily achieved, but when significant disagreement occurred between participants, votes were taken to achieve a consensus and dissenters’ opinions were recorded. The phased and progressive nature of the workshops provided the opportunity to seek and incorporate additional expertise and information as the study progressed and additional needs became apparent. Interim reports summarizing information from each of the first three workshops were distributed to participants after each workshop. These reports provided participants with opportunities for ongoing review and a means to ensure consistency in assumptions and approaches among the different systems. Finally, evaluations were conducted at the completion of the first three sets of workshops to continually improve the process. Risk Measures. Risk measures for the study systems were developed using the following criteria: • The measures of risk should provide relevant and useful input to MMS in their decision making process; • The measures of risk should be tractable and quantifiable; and • The measures of risk should be measures that are currently tracked and recorded so that (i) available data can be used to support the results of this risk analysis and (ii) future data can be used to validate and calibrate the results of this risk analysis. From these criteria, the risk measures listed in Table 3 were adopted for this study. The total number of fatalities is intended to measure the human safety risk. The volume of oil spilled is intended to measure the environmental risk. The environmental effects of an oil spill are not considered directly in this study because (1) there is a correlation between the magnitude of environmental damage and the volume of oil spilled; (2) environmental effects are difficult to measure and quantify; and (3) the environmental impacts of oil spills from FPSO’s are included in the scope of the EIS 1. The total volume of oil spilled in the 20-year lifetime is intended to measure chronic environmental risks. The maximum volume of oil spilled in a single incident is intended to measure acute environmental risks. The risk measures in Table 3 are practical simplifications that are intended to approximately capture the multitude of risks present. These measures of risk were not discounted with time. In addition, each measure was treated separately in comparisons and no attempt was made to combine them into a single
3
measure, such as equivalent cost. Descriptions of Study Systems. The following criteria, in order of decreasing importance, were used to develop conceptual descriptions for each of the representative study systems: 1. The study systems for the Spar, TLP, and Hub/Host Jacket should be typical of existing systems and technologies that are currently being used in the Gulf of Mexico because these systems and technologies have been approved and therefore represent acceptable risks. 2. The study system for the FPSO should be comparable to that already developed for the base case in the EIS1 in order to capitalize on the substantial effort devoted to developing this study basis. 3. The study systems should be as comparable to one another as possible so that differences in risks among them represent realistic differences among these types of systems and are not an unintended artifact of the study system descriptions. As an example of how these criteria were applied, consider a Spar. In order for the study Spar to be as comparable as possible to the study FPSO (criterion 3), which has oil transport via shuttle tankers consistent with the EIS 1 (criterion 2), the Spar would also have oil storage and oil transport via shuttle tankers. However, while this type of a Spar is possible, it is not typical of existing Spars in the Gulf of Mexico (criterion 1). Therefore, the description for the study Spar did not include oil storage and has oil transport via pipeline. The first step in the system description process was to establish a time frame for the risk assessment. The intent was to assess risks covering all aspects of offshore production including oil and gas production and processing offshore; drilling and well intervention during production; export of the oil and gas to shore; and transport of personnel to and from shore. The “lifetime” for a study system was defined to start when oil flows through the first production riser and end when the last well is shut in. For this study, a 20-year lifetime was used. Other phases in the actual lifetime for a system, such as construction, system installation, commissioning, decommissioning and system removal, were not included in this risk analysis. The second step in the system description process was to establish physical boundaries for the risk assessment. The study boundaries included the production facility, the pipelines and shuttle tankers used to transport oil to a shore terminal, and the supply vessels and helicopters used to support the production operations. These physical boundaries are shown schematically on Fig. 2. The third step in the system description process was to define the physical and operational attributes for each system. Detailed descriptions for these attributes are contained in Gilbert et al. 2, and the major attributes are summarized in Table 4. For the Spar, TLP and Hub/Host Jacket, operating experience from the Gulf of Mexico was directly drawn upon in developing the system descriptions. For the FPSO,
4
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
experience with tanker operations in the Gulf of Mexico was used together with operating experience for FPSO’s in other parts of the world, such as the North Sea and the South China Sea. It is important to note that these study systems represent typical or generic systems. Therefore, the range of risks associated with possible variations in hardware and operating practices is not captured in the results of this project. Quantitative Risk Assessment. The objective of the quantitative risk assessment was to quantitatively assess the risk measures listed in Table 3. These risk measures were quantified by estimating representative or average values for each study system. As an example, consider the total volume of oil spilled during the operational lifetime. If a fleet of Spars similar to the one defined in this study were installed and operated for 20 years in the Gulf of Mexico, then the total oil spill risk associated with this type of system would be the average value for the total volume of oil spilled from each Spar (that is, the sum of all the oil spilled from each Spar in its 20 year lifetime divided by the total number of Spars). Likewise, the average values for the maximum volume of oil spilled in a single incident from each Spar and the total number of fatalities on each Spar would represent the other measures of risk. Since there is an extremely limited experience base in the Gulf of Mexico for the types of production systems being analyzed in this study, it is not possible to obtain average values directly for the total number of fatalities, the total volume of oil spilled, and the maximum volume of oil spilled in a single incident. The goal of this study was to predict what the average values would be (the expected value) if each study system were hypothetically installed and operated in the future in the Gulf of Mexico. As with any prediction, there is uncertainty that the actual average value for each risk measure (obtained many years in the future) will be equal to the predicted value in this study. The range of possible values for the actual average was represented in this study by two quantities: the expected value and the standard deviation. The expected value represents the predicted value for the average, while the standard deviation represents the magnitude of uncertainty in the prediction. The expected value and standard deviation can be used to calculate confidence intervals for the prediction. For example, the 90 percent confidence intervals indicate that there is a ninety percent probability that the actual average will be within this interval. This section describes how the quantitative risk assessments were conducted through a process of developing preliminary assessments and then refining those assessments using the input of the technical experts. Preliminary Risk Assessments. Preliminary (or preworkshop) quantitative risk assessments played a very important role in this project because they were used to elicit quantitative information from the technical experts during Workshops #3 and #4 (Fig. 1). These preliminary risk assessments were developed to be objective, consistent and complete in order to maximize the value of the information
OTC
obtained from the technical experts during the workshops. The philosophy adopted in developing the preliminary risk assessments was to extrapolate directly from historical experience in the Gulf of Mexico to predict future performance. The primary data sources were the MMS3 and the USCG4. The methodology used to develop the preliminary risk assessments had the following steps: 1. Summarize Data for Sub-Systems: The data sets were first divided into sub-systems based on the hazard identification work in Workshop #2. These subsystems are listed in Table 5. The data for fatalities were then summarized as the total number of fatalities in the data record for each sub-system. The data for oil spills were further sub-divided into categories by the size of the spill, and then the number of incidents in the data record that had occurred for each spill-size category was compiled. The data for oil spills were divided into categories to facilitate the assessment since the range of spill volumes per incident covered five to six orders of magnitude and the frequency distribution for spill sizes was highly skewed. 2. Select Exposure Factors for Sub-Systems: The exposure for a risk is an indicator of the factors that influence the risk. In this way, the data can be extrapolated to each study system based on the exposure to the risk for that study system. The factors used to express the exposure for each subsystem category were selected based on the hazard identification information work in Workshop #2. These factors are listed in Table 5. 3. Estimate Frequencies of Occurrence for SubSystems: Estimates for the frequencies of occurrence for incidents (from Step 1) relative to the exposure factors (from Step 2) were developed using statistical methods that are described in Gilbert et al. 2. Both the expected value and the standard deviation for these frequencies were calculated. For the oil spill frequencies, it was assumed that a spill could occur in the next largest spill-size category above the maximum spill size observed in the historical data set. 4. Determine Sub-System Exposures for Study Systems: The exposure for each sub-system was determined from the system descriptions. 5. Assess Sub-System Risks for Study Systems: The estimated frequencies from the historical data (from Step 3) were then combined with the exposures for the study systems (from Step 4) to assess the subsystem risks. Both an expected value and a standard deviation were calculated for each risk measure (see Gilbert et al. 2 for details). 6. Assess System Risks from Sub-System Risks: The final step was to combine the information for the subsystem risks (from Step 5) to assess the total system risk (see Gilbert et al.2 for details). F inal Risk Assessments. The preliminary risk assessments
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
were then refined through the workshop process to develop final risk assessments. This process involved the following steps: 1. Start with data-based estimates that are as complete as possible (the preliminary risk assessments). 2. Evaluate the data sources and refine raw data sets as necessary so that they are relevant for predicting future performance of the study systems. As an example, the data set for oil spills from tankers in the Gulf of Mexico was limited to years after 1990 to account for the positive effects that the Oil Pollution Act of 19905 (OPA ’90) has had on recent performance and is anticipated to have on future performance. 3. Extrapolate predictions of future performance from the data set, applying corrections to the data-based estimates if necessary. As an example, the frequencies for small spills from subsea well systems were increased from the data-based estimates to account for differences between subsea well systems and the platform well systems that dominate the data set. 4. Account for all sources of uncertainty in the estimates, including the following: • the limited quality and quantity of relevant data records, especially for rare events; • the sometimes limited information available on the exposures corresponding to the data sets; and • the extrapolation of future performance from historical performance. 5. Document the whole process clearly and thoroughly. The detailed quantitative risk assessments for fatalities and oil spills are contained in Gilbert et al. 2.
Results The results from the quantitative risk assessment for fatalities and oil spills are presented, analyzed and discussed in this section. Risks for Fatalities. Results for the average total number of fatalities are shown on Fig. 3 for each study system. The results indicate that the fatality risks are very similar among the four study systems (Fig. 3). The expected contributions to the total fatality risk are shown on Fig. 4. Production and drilling and well intervention activities dominate the total fatality risk for all of the study systems. This result occurs because these activities require the bulk of the man-hours over a 20-year lifetime. The estimated frequencies of fatalities per man-hour worked for production and for drilling and well intervention activities (Gilbert et al. 2) are comparable to those reported for common industrial activities (AIChE6) and for the oil and gas industry throughout the world (OGP7). The contribution of drilling and intervention activities to the total fatality risk for the FPSO is not as large as for the other systems because all of the wells on
5
the FPSO are subsea wells that are subjected to less frequent well intervention compared to platform wells. Risks for Oil Spills. The results for the oil spill risks are presented and discussed in this section. First, the frequencies for different spill sizes are addressed. Next, the average total volume of oil spilled and the maximum volume spilled in a single incident over the lifetime are addressed. Frequencies of Spills. The frequencies of spills from production and transportation are first presented and discussed, and then the frequencies of spills from all sources are addressed. Frequencies of Production Spills. The annual frequencies for spills from production (Table 5) are shown on Fig. 5 for each of the study systems. Note that the frequency of spills tends to decrease as the spill size increases. Also, note that the relative magnitude of uncertainty in the estimated frequency increases as the spill size increases. This relative increase in uncertainty occurs because large spills are rare events, so there are few occurrences available from which to estimate frequencies. The information on Fig. 5 highlights the similarities and differences among the systems regarding oil spills from production. First, the Spar and the TLP are indistinguishable. This result is reasonable in that the elements of the designs on both systems that affect the potential for oil spills from production are nearly identical on these two study systems. Second, the Hub/Host Jacket tends to have smaller spill frequencies from production than the Spar, TLP and FPSO for spill sizes less than 1,000 bbl (Fig. 5). This difference is due to the smaller indigenous production rate on the shallow-water Hub/Host Jacket versus the deepwater floating production systems (Table 4). Third, the frequency of very small spills (less than 10 bbl) on the FPSO is less than that on the Spar and TLP, even though the production rates are similar on all of these study systems. This difference is due to the large deck area and the solid decking that exist on an FPSO; the deck would contain most small spills. Fourth, the frequency of spills between 100 and 10,000 bbl is slightly larger for the FPSO versus the other systems. This relative difference is because the FPSO has more subsea wells than the other systems; subsea wells were considered to have a higher leak frequency than platform wells because of a greater potential for sand erosion and cutouts due to high flow rates and detection difficulties for sand. In addition, the FPSO has a greater number of flowlines and flowline risers, which both contribute to the frequency of spills between 100 and 10,000 bbl. Fifth, the Spar and the TLP have the potential for very large spills (greater than 10,000 bbl), although the frequency for these spills is very small (Fig. 5). The potential source for these very large spills on the Spar and TLP is the dry tree risers. This risk does not exist on the FPSO study system because the trees that control the reservoir pressure and flow are on the seafloor (wet trees) rather than at the surface (dry trees), and it is negligible for the Hub/Host Jacket study
6
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
system because of the lack of movement for this non-floating system. A comparison with published information for the frequency of large spills from production is shown on Fig. 6. Anderson and LaBelle8 report a frequency for spills greater than 1,000 bbl in size. Their frequency was estimated using spill data from offshore platforms operating in the United States between the years 1974 and 1992. They report their frequency on the basis of the volume produced. In order to develop Fig. 6, this frequency has been converted to an annual frequency for the study systems using the total volume of oil produced in the 20-year lifetime for each system. The estimated frequencies for the study systems are less than the values obtained from Anderson and LaBelle8 (Fig. 6). There are two reasons for this result. First, two different data sets have been used. In the CRA project, data before 1990 were discarded due to the implementation of new regulations in 1990 (API RP14C9), which improved operating procedures on platforms. The Anderson and LaBelle data set extends back to 1974. Second, the CRA study systems are not representative of conventional, shallow-water platforms in the United States, which dominate the population of platforms in the Anderson and LaBelle data set. Note that the agreement between the CRA and Anderson and LaBelle is best for the Hub/Host Jacket study system (Fig. 6), which is most similar to the platforms in the Anderson and LaBelle data set. Frequencies of Transportation Spills. The annual frequencies for spills from transportation (Table 5) are shown on Fig. 7 for each of the study systems. The results highlight the similarities and differences among the systems regarding oil spills from transportation. First, compare the systems with pipelines. The Spar and the TLP are indistinguishable. This result is reasonable in that the elements of the designs on both systems that affect the potential for oil spills from transportation are nearly identical on these two study systems. The Hub/Host Jacket has slightly smaller spill frequencies from its pipeline than the Spar and TLP (Fig. 7). This difference is because there is a shorter length of pipeline exposed for the Hub/Host Jacket due to the shorter distance to the shore (Fig. 2). In addition, there is relatively less uncertainty in the estimated spill frequencies for the Hub/Host Jacket for spills less than 1,000 bbl (Fig. 7). The greater uncertainty for the Spar and TLP is mainly due to the potential for spills from the more flexible steel catenary export pipeline risers versus the more rigid risers on fixed jackets. The uncertainty for the Spar and TLP reflects that there are limited data concerning the performance of these risers in deepwater applications. Second, compare the two different types of transportation systems. There are notable differences between the pipelines for the Spar, TLP and Hub/Host Jacket and the in-field storage and shuttle tanker system for the FPSO. For very small spill sizes (less than 10 bbl), the frequency of spills for the FPSO is greater than from pipelines due to the potential for spills during offloading from hoses and valves. For spill sizes between 1,000 and 100,000 bbl, the annual frequencies of
OTC
spills for the shuttle tanker are smaller than the annual frequencies for pipelines (Fig. 7). One reason for this difference is that the potential for spills from the pipeline remains a constant as long as there is oil in the pipeline, regardless of the production rate. However, the potential for spills from the shuttle tanker will go down as the production rate decreases since fewer offloading events are required. Lastly, very large spill sizes (greater than 100,000 bbl) are not considered possible for pipelines due to operational and physical constraints (Gilbert et al. 2), while they are possible although infrequent for the FPSO. A spill between 100,000 and 500,000 bbl represents a major loss from the shuttle tanker due to a collision or explosion. A spill greater than 500,000 bbl represents a major loss from the FPSO due to a collision or explosion. A comparison with published information for the frequency of large spills from transportation is shown on Fig. 8. Anderson and LaBelle8 report frequencies for spills greater than 1,000 bbl in size from pipelines and tankers. Their frequencies were estimated using spill data from offshore operations in the United States between the years 1974 and 1992. They report their frequency on the basis of the volume produced. In order to develop Fig. 8, this frequency has been converted to an annual frequency for the study systems using the total volume of oil produced in the 20-year lifetime for each system. The estimated frequency for the Hub/Host Jacket is comparable to that from Anderson and LaBelle8 (Fig. 8). This result is reasonable since the pipeline from the Hub/Host Jacket is representative of the conventional, shallow-water pipelines that are in the Anderson and LaBelle data set. However, the estimated frequencies for the Spar and TLP study systems are less than those obtained from Anderson and LaBelle (Fig. 8). The primary reason for this difference is that the Anderson and LaBelle frequency for pipeline spills is proportional to the volume produced. However, the potential for spills from pipelines in the CRA study was related to the length of the pipeline and the time of exposure, not the volume of throughput. Therefore, the higher production rates for deepwater systems do not necessarily lead to proportionally higher spill frequencies for pipelines. The estimated frequency for the FPSO is lower than that from Anderson and LaBelle8 (Fig. 8). This result is due to the different data sets used to estimate the frequency. The Anderson and LaBelle data set extends back to 1974, and includes data from all U. S. coastal and offshore waters. In the CRA project, data before 1992 were discarded due to the implementation of the OPA ’903, which improved operating procedures on tankers and probably reduced the frequency of spills. Data for crude oil tankers in the Gulf of Mexico are summarized in Table 6 to support the hypothesis that data prior to the passage of OPA ’90 are not representative of existing conditions. In addition, data from outside of the Gulf of Mexico were not applied directly in the CRA project to estimate the shuttle tanker risk in the Gulf of Mexico. An analysis of tanker spills from 1992 to 1999 indicates that frequencies of spills between 50 and 5,000 bbl and of spills
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
greater than 5,000 bbl in the Gulf of Mexico are approximately 40 percent of those for the rest of the world (Gilbert et al. 2). Tanker spills are considered to be less likely on average in the Gulf of Mexico than in the rest of the world for the following reasons, in order of importance: 1. The regulatory environment in the Gulf of Mexico is more restrictive; 2. The environmental conditions in the Gulf of Mexico are less severe; 3. The consequences of grounding are significantly less due to the lack of rocky coasts in the Gulf of Mexico; 4. Shuttle tankers used in the Gulf of Mexico have a smaller parcel size on average; 5. The Gulf of Mexico has less congested waterways on average; and 6. Newer vessels are used in the Gulf of Mexico due to recent federal regulations. Frequencies of Spills from All Sources. The annual frequencies for spills from all sources, including production and transportation, are shown on Fig. 9 for each of the study systems. The frequencies for spills are generally dominated by production-related spills for spill sizes up to 1,000 bbl and by transportation-related spills for spill sizes greater than 1,000 bbl. Therefore, the similarities and differences among the study systems are related to those for production for spills less than 1,000 bbl (Fig. 5) and to those for transportation for spills greater than 1,000 bbl (Fig. 7). Note that the Spar and TLP are indistinguishable for all spill sizes. Total Volu me of Oil Spill ed over L if etime. Results for the average total volume are shown on Fig. 10 for each study system. These results indicate that the systems provide very comparable risks. The risk for the Hub/Host Jacket is slightly smaller than the risks for the other systems because it has a smaller production rate and a shorter transportation distance to the shore. The risks for all of the deepwater systems (Spar, TLP and FPSO) are nearly identical even though the frequencies for different spill sizes are not identical (Fig. 9). This result occurs because the risk is a measure of both frequency and consequence (spill size). While very large spills (greater than 100,000 bbl) are more likely with the FPSO than with the Spar or TLP, the annual frequencies are still small. Furthermore, the frequencies for spills less than 100,000 bbl for the FPSO are generally smaller than those for the TLP or Spar (Fig. 9). Therefore, the risks for the Spar, TLP and FPSO are comparable. In order to facilitate interpretation of the results on Fig. 10, the relative contribution of each spill-category to the total volume spilled is shown on Fig. 11. Note that the chronic environmental risk is dominated by large spills (greater than 1,000 bbl), which are low frequency but high consequence events. Therefore, most of the systems in a fleet of study systems will have small volumes of oil spilled. Occasionally, one of the systems may have a large spill and this large spill will dominate the average for the fleet. To emphasize this point, Table 7 summarizes the expected time between spills of different sizes for each of the study systems. Note that most of the risk for the Spar and TLP study systems comes from
7
spills between 10,000 and 100,000 bbl, which are only expected approximately once every 600 years of operation. Furthermore, most of the risk for the FPSO study system comes from spills between 100,000 and 500,000 bbl, which are only expected once every 4,500 years of operation. Table 7 and Fig. 11 show how the contributions to the risks for the Spar and TLP versus those for the FPSO are different even though the resulting risks are comparable (Fig. 10). One effect of the spill risk being dominated by rare, high consequence events is that the confidence intervals in the predicted average oil spill volumes range over nearly an order of magnitude (Fig. 10). This uncertainty reflects the typically limited quantity and quality of historical data available to estimate frequencies for rare events. Note that the confidence interval for the FPSO is wider than those for the other systems (Fig. 10) because there are relatively fewer data available for FPSO’s in the Gulf of Mexico and because the FPSO risk is dominated by very rare events with expected return periods of approximately 4,500 years. The contributions to the total oil spill risk from different sub-systems are shown on Fig. 12. Production, which dominates the smaller spill sizes (Fig. 5), does not contribute substantially to the total risk (Fig. 12). The main contributor to oil spills from production are related to the processing facilities (topsides on Fig. 12). Transportation, which dominates the larger spill sizes (Fig. 7), is the main contributor to the total oil spill risk (Fig. 12). M aximum Single Oil Spill in Li fetime. Results for the average single maximum spill are shown on Fig. 13 for each study system. The results indicate that the risks for the different study systems are comparable. Furthermore, these results emphasize that the maximum spill volume from a single incident dominates the average total spill volume. More than 70 percent of the total is expected to come from a single incident. The wide confidence intervals on Fig. 13 reflect the uncertainty inherent in estimating frequencies for rare events.
Conclusions and Recommendations A quantitative risk analysis was performed to assess and compare oil spill and fatality risks for four representative deepwater production systems in the Gulf of Mexico. Three of the study system types have already been operated successfully in the Gulf of Mexico: two floating production systems in deepwater with oil pipelines, a Spar and a Tension Leg Platform (TLP); and a shallow-water jacket serving as a hub and host to deepwater production. One of the study system types has not been used in the Gulf of Mexico: a tanker-based Floating Production Storage and Offloading (FPSO) system with oil transportation to shore via shuttle tankers. The objective of this analysis was to understand and compare the risks of the FPSO with those for currently acceptable alternatives for deepwater production. Conceptual system descriptions that are representative of existing and typical technology in the Gulf of Mexico were developed for the four systems. The scope of these descriptions included the entire production systems and
8
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
operations from the wells through the transport of product to the shore. Three risk measures were assessed and analyzed for each system: the total number of fatalities in a 20-year production life as a measure of the human safety risk, the total volume of oil spilled in a 20-year production life as a measure of the chronic environmental risk, and the maximum volume spilled in a single incident in a 20-year production life as a measure of the acute environmental risk. The process of developing the conceptual descriptions for the systems and then evaluating the risks has drawn on expertise from all facets of oil and gas production, including operators, contractors, manufacturers, class societies and regulators. Conclusions. The following major conclusions have been drawn from the results of this analysis: 1. There are no significant differences in the fatality risks among the four study systems. 2. There are no significant differences in the oil-spill risks among the four study systems. 3. The average total volume of oil spilled during the facility lifetime will be dominated by rare, large spills rather than frequent, small spills. 4. The major contribution to the oil spill risks for all systems is the transportation of oil from the production facility to the shore terminal by either pipelines or shuttle tankers. Spill risks for pipelines and shuttle tankers are comparable, although the frequencies and sizes of possible spills are different for pipelines versus shuttle tankers. The spill risks for pipelines are dominated by the possibility of spills between 10,000 and 100,000 bbl in size that are expected to occur once every 600 years on average. The spill risks for shuttle tankers are dominated by the possibility of spills between 100,000 and 500,000 bbl in size that are expected to occur on average once every 4,500 years. 5. The confidence intervals in predicted oil spill volumes range over about an order of magnitude, reflecting the limited quantity and quality of historical data available to estimate frequencies for rare events. Therefore, the expected risks associated with the FPSO are comparable to those for already accepted alternatives for deepwater production, including a Spar, a TLP and a shallowwater jacket serving as a hub and a host to deepwater production. Recommendations. The following recommendations have been developed from this work: 1. The results from this study should be periodically updated because they provide a valuable baseline for future analyses of risk in the Gulf of Mexico. The three measures of risk used in this study can all be readily measured and tracked for new and existing deepwater production facilities in the Gulf of Mexico.
OTC
2.
The quality of existing data sets for the Gulf of Mexico should be improved so that they are of greater value in future risk analyses. First, the type and quality of data that are currently collected should be evaluated, and any changes recommended from this evaluation should be implemented in a timely manner. Second, single agencies should be responsible for tracking and compiling similar types of data. Third, all data records should be reviewed annually by the industry and regulators to improve the clarity, quality and usefulness of the information in these records. Finally, the data should be published annually in a clear and an easily accessible format. 3. Additional information about the populations of offshore facilities and operations in the Gulf of Mexico should be collected on an annual basis. Specifically, the following information from federal and state waters in the Gulf of Mexico would be valuable: the length of active pipelines operating per year, the number of tanker on-loading and offloading events in ports and lightering zones per year, and the number of man-hours in production-related activities, supply vessel operations and tanker operations per year. 4. Uncertainty in the predicted performance for these four study systems should be considered carefully in drawing conclusions from and applying the results from this study. 5. The process used on this project to assess risks has been effective in obtaining valuable technical information from industry and regulators, and should be considered in supporting other analyses of new technology in the Gulf of Mexico.
Acknowledgments The authors wish to acknowledge the Minerals Management Service for funding this project. Specific individuals at MMS who have been instrumental in this study are Paul Martin, Charles Smith, James Regg and Cheryl Anderson. In addition, the following individuals have played key roles in the project: Professor Larry Lake and Research Assistant Jihad Jaber, The University of Texas at Austin; Joe Gebara, EQE International; Lieutenant Commander Bill Daughdrill and Joe Myers, United States Coast Guard; and Allen Verret, Private Consultant. Finally, the Deepstar Consortium has been instrumental in coordinating and facilitating the involvement of industry experts on workshop teams. References 1. MMS (2000), “Proposed Use of Floating Production, Storage, and Offloading Systems On the Gulf of Mexico Outer Continental Shelf, Western and Central Planning Areas, Draft Environmental Impact Statement,” Prepared Under MMS Contract 1435-01-99-CT-30962, Minerals Management Service, Gulf of Mexico OCS Region. 2. Gilbert, R. B., Ward, E. G. and Wolford, A. J. (2001),
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
3.
4.
5. 6.
“Comparative Risk Analysis for Deepwater Production Systems,” Final Project Report, Prepared for Minerals Management Service, Washington, D. C. MMS (2000), “MMS OCS Spill Database,” Minerals Management Services, Available: [email protected], Accessed: January, 2000. USCG (1999), “Marine Casualty and Pollution Database,” CD ROM, Subscription Order No. 5441INC, Available: National Technical Information Services, Springfield, Virginia. Oil Pollution Act of 1990 (OPA ’90), 33 USCA Sec. 2701-2761. AIChE (1989), Chemical Process Quantitative Risk Analysis, Center for Chemical Process Safety, American Institute of
Chemical Engineers, New York, New York. 7. OGP (1999b), “Safety Performance of the Global E&P Industry, 1998,” Report No. 6.80/295, International Association of Oil and Gas Producers, London, England. 8. Anderson, C. M. and LaBelle, R. P. (1994), “Comparative Occurrence Rates for Offshore Oil Spills” Spill Science and Technology Bulletin, Vol. 1, No. 2, 131-141. 9. API RP14C (1998), “Analysis, Design, Installation and Testing of Basic Surface Safety Systems on Offshore Production Platforms,” Sixth Edition, American Petroleum Institute, Washington, D.C.
Table 1 - Industry sources for workshop participants. Oil Companies BP Amoco Chevron Conoco Elf ExxonMobil Marathon Oxy Shell Statoil Texaco
Consultants & Contractors EQE ABB Atlantia FMC Paragon McDermott Navion
Classification Societies ABS Lloyd’s Register DNV
Table 2 - Additional industry sources for technical expertise. Skaugen Petrotrans SBM IMODCO Global Maritime Aker R&B Falcon Transocean SedcoForex Edison Chouest Tidewater Marine HSAC Air Logistics PHI ERA Aviation
Association of Diving Contractors Oceaneering Cameron Mentor Bay Ltd. Spirit Energy Horizon Engineering Kerr McGee Mathews Daniels LOOP State of Louisiana
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. Water Depth (ft) Peak Production Oil (bopd) Gas (scfpd) Export Oil (bopd) Gas (scfpd) Wells Platform Subsea (MODU) Manning Production Marine Drilling – Platform Drilling MODU
9
Spar 4,000
TLP 4,000
Hub/Host Jacket 600
FPSO 5,000
150,000 200,000
150,000 200,000
50,000 50,000
150,000 200,000
150,000 200,000
150,000 200,000
250,000 550,000
150,000 200,000
6 3
6 3
6 3
0 9
30-45 6 65 65
30-45 6 65 65
30-45 0 50 65
30-45 10 0 65
10
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
OTC
Table 5 - Sub-system categories used in risk assessment. Risk Measure
Exposure Factor
Sub-System Category Production Drilling Supply Vessels Helicopter Transport Tanker Operations Major Accident Well Systems – Platform (or Surface) Well Systems – Subsea Dry Tree (or Production) Risers Flowlines Import Flowline Risers Topsides Supply Vessels Drilling and Intervention Pipelines Export Pipeline Risers Shuttle Tanker (Offloading in Field and at Port) FPSO Cargo Tank
Fatalities
Production System Oil Spills
Transportation System
man-hours man-hours docking calls passengers docking calls platform-years bbl produced bbl produced riser-years mile-years riser-years bbl processed docking calls man-hours mile-years riser-years docking calls platform-years
Table 6 - Summary of Oil Spills from Crude Tankers in Gulf of Mexico 1
Number of Spills Year
1-10 bbl
10-100 bbl
100-1,000 bbl
1,000-10,000 bbl
10,000-100,000 bbl
1991
0 2 4 5 3 7 5
1 1 0 1 1 2 0
0 0 0 0 2 1 0
0 0 0 0 0 0 0
0 0 0 1 0 0 0
30 28 5 15,401 1,146 266 17
Sub-Total
26
6
3
0
1
16,893
1992
0 2 2 0 0 1 1 1
0 1 0 0 0 0 1 0
0 1 0 1 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 191 8 179 0 2 22 9
7
2
2
0
0
411
1985 1986 1987 1988 1989 1990
1993 1994 1995 1996 1997 1998 1999 Sub-Total 1
Volume Spilled (bbl)
4
Note: Data from USCG .
Table 7 - Expected return periods for spills. System
1 – 10 bbl
10 – 100 bbl
Spar TLP Hub/Host Jacket FPSO
0.8 0.8 3 3
3 3 8 3
Expected Return Period between Spills (years) 100 - 1,000 1,000 - 10,000 10,000 - 100,000 100,000 - 500,000 bbl bbl bbl bbl 15 60 580 Not Credible 15 60 580 Not Credible 35 91 920 Not Credible 12 110 2,500 4,700
500,000 - 1,000,000 bbl Not Credible Not Credible Not Credible 300,000
1
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
Draft System Descriptions
Workshop #1 System Definition Develop System Descriptions
Develop Preliminary Event/Outcome Tables Gas Flange r (Port Terminal)
Workshop #2 Hazard Identification Elicit Event/Outcome Information Conduct Preliminary QRA
r
l Hub/Host
Workshop #3 Quantitative Risk Analysis Elicit Frequency/Consequence Input
Workshop #4 Review Review and Refine QRA
Gas Pipeline
Oil Pipeline
Refine QRA & Perform Additional Studies
Fig. 2 - Physical layout for study systems (plan view).
3.0
0.5
Shuttle Tanker
l l TLP FPSO Spar
Prepare Final Report
Fig. 1 - Flowchart for workshop process .
s 2.5 e i t i l a t a F 2.0 f o r e e m b i t m e u f i 1.5 N L l n a i t o T 1.0 e g a r e v A
Oil Flange (Port Terminal)
90% confidence intervals
0.0 Spar
TLP
Hub/Host Jacket
FPSO
System
Fig. 3 - Average total number of fatalities in lifetime .
11
12
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
OTC
Spar/TLP
Production Drilling and Intervention Supply Vessels Helicopter Transport Tanker Operations
Drilling and Intervention (52%)
Major Accident
FPSO
Hub/Host Jacket
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. ) r 1.0E+01 a e y r e p 1.0E+00 ( s l l i p S 1.0E-01 n o i t c 1.0E-02 u d o r P r 1.0E-03 o f y c n 1.0E-04 e u q e r F 1.0E-05 l a u n n A 1.0E-06
Spar TLP Hub/Host Jacket FPSO
Expected value
90% confidence interval
0 1 1
0 0 1 0 1
0 0 0 , 1 0 0 1
0 0 0 , 0 1 0 0 0 , 1
0 0 0 , 0 0 1 0 0 0 , 0 1
0 0 0 , 0 0 5 0 0 0 , 0 0 1
0 0 0 , 0 0 0 , 1 0 0 0 , 0 0 5
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.
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
0 0 0 , 1 n a h t r e t a e r G s l l i p ) S r a n e y o i t r c e u p d ( l o r b P b f o y c n e u q e r F l a u n n A
13
1.0E+00
1.0E-01 Anderson and LaBelle (1994)
1.0E-02
1.0E-03
CRA (90% confidence intervals)
1.0E-04
1.0E-05 Spar
TLP
Hub/Host Jacket
FPSO
System
Fig. 6 - Comparison of production spill (>1,000 bbl) frequencies with published data. CRA denotes the results from this study. ) r a 1.0E+01 e y r e p ( 1.0E+00 s l l i p S n 1.0E-01 o i t a t r 1.0E-02 o p s n a r T 1.0E-03 r o f y c 1.0E-04 n e u q e r 1.0E-05 F l a u n 1.0E-06 n A
Spar TLP Hub/Host Jacket FPSO
Expected value 90% confidence interval
0 1 1
0 0 1 0 1
0 0 0 , 1 0 0 1
0 0 0 , 0 1 0 0 0 , 1
0 0 0 , 0 0 1 0 0 0 , 0 1
0 0 0 , 0 0 5 0 0 0 , 0 0 1
0 0 0 , 0 0 0 , 1 0 0 0 , 0 0 5
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.
14
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
n a 1.0E+00 h t r e t a e r G 1.0E-01 s l l i p S ) n r o a e i t y a r 1.0E-02 t r e o ( p p l s b n b a r T 0 0 1.0E-03 f 0 , o 1 y c n e u q 1.0E-04 e r F l a u n n A 1.0E-05
OTC
Anderson and LaBelle (1994)
CRA (90% confidence intervals)
Spar
TLP
Hub/Host Jacket
FPSO
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 TLP Hub/Host Jacket
1.0E+00 ) r a e y r e p ( y c n e u q e r F l a u n n A
FPSO
1.0E-01 1.0E-02 Expected value 1.0E-03 90% confidence interval
1.0E-04 1.0E-05 1.0E-06 0 1 1
0 0 1 0 1
0 0 0 , 1 0 0 1
0 0 0 , 0 1 0 0 0 , 1
0 0 0 , 0 0 1 0 0 0 , 0 1
0 0 0 , 0 0 5 0 0 0 , 0 0 1
0 0 0 , 0 0 0 , 1 0 0 0 , 0 0 5
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.
OTC 13173 A COMPARATIVE RISK ANALYSIS OF FPSO’S WITH OTHER DEEPWATER PRODUCTION SYSTEMS IN THE GULF OF MEXICO
10000 90% confidence intervals 9000 8000 d ) e l l l i b p b ( S e l i m O i f t e o f i e L m r u e v l o o V
Average Total
7000 6000 5000 4000 3000 2000 1000 0 Spar
TLP
Hub/Host Jacket
FPSO
System Fig. 10 - Average total volume of oil spilled over lifetime – all sources.
n i d 100% e l l i p S e 80% m u l o V l a t 60% o e T i m e t e g f i a r L 40% e v A o t n 20% o i t u b i r t n 0% o C
Spar
0 - 1 1
0 0 - 1 1 0
0 1 0
TLP
Hub/Host Jacket
FPSO
0 0 0 0 0 0 0 0 0 , 0 , 0 , 0 , 0 0 0 0 0 0 5 0 0 - 1 , - 1 0 1 0 0 0 0 0 0 0 0 0 1 , , , 0 0 0 1 0 0 , 1 0 0 5
0 0 0 1 , -
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.
15
16
R.B. GILBERT, E.G. WARD AND A.J. WOLFORD
Well Systems - Platform Well Systems - Subsea Dry Tree Risers Flowlines Import Flowline Risers Topsides Export Pipeline Risers
Spar/TLP
Pipelines Shuttle Tanker FPSO Cargo Supply Vessels Drilling and Intervention
Pipelines (72%)
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 9000
90% confidence intervals
8000 d ) e l l l i b p b ( S e l i m O i f t e o f i e L m r u e v l o o V
Average Total
7000
Average Single Maximum
6000 5000 4000 3000 2000 1000 0 Spar
TLP
Hub/Host Jacket
FPSO
System
Fig. 13 - Average maximum volume spilled from a single incident in the lifetime .
OTC
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 transportation 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.
