CORE Report No.

2005-04

SAFETY OF OFFSHORE STRUCTURES

by

TORGEIR MOAN
Professor Norwegian University of Science and Technology and Keppel Professor National University of Singapore

Centre for Offshore Research & Engineering National University of Singapore

Keppel Offshore and Marine Lecture

November 26. 2004

Safety of Offshore Structures

By

Professor Torgeir Moan

Director, Centre for Ships and Ocean Structures, Norwegian University of Science and Technology Keppel Professor, National University of Singapore

Foreword
The Second Keppel Offshore & Marine Lecture was delivered by Professor Torgeir Moan at the National University of Singapore (NUS) on 26 November 2004. This report is the written version of the lecture. The Lecture was supported by many professional

societies, including American Society of Mechanical Engineers (Singapore Section), The Institution of Engineers Singapore, The Institute of Structural Engineers, The Joint Branch of Royal Institution of Naval Architects and Institute of Marine Engineering Science & Technology, Society of Naval Architects and Marine Engineers Singapore, Singapore Shipping Association and Singapore Structural Steel Society. The Keppel Professorship in Ocean, Offshore and Marine Technology in NUS was launched officially by His Excellency, President S.R. Nathan of Singapore and Chancellor of NUS on 19 September 2002. The Professorship has been established

under the Department of Civil Engineering, and is part of the bigger umbrella of the Centre for Offshore Research & Engineering (CORE) in NUS. CORE has received seed funding from Keppel Offshore & Marine Ltd and Economic Development Board (Singapore). The Centre aims to be a focal point for industry participation and activities in Singapore, and promotes multi-disciplinary research by drawing on the expertise of various universities, research institutes, and centres for integrated R&D. Professor Moan, the first Keppel Professor, is an academic of international stature through his 30 years of close involvement in the international offshore oil and gas and marine fraternity. He brings a global perspective to promote R&D in Singapore, with a particular emphasis on the technological interests of Keppel Offshore & Marine. He will serve as the beacon of guidance and inspiration for academics as well as industry. CORE acknowledges the significant support of everyone in Keppel Offshore & Marine, especially Mr Choo Chiau Beng, Mr Tong Chong Heong and Mr Charles Foo, for their immense contributions. CORE is working closely in partnership with Keppel on R&D, Education and Training to draw young talents into the offshore and marine industry.

Safety of Offshore Structures

Torgeir Moan Centre for Ships and Ocean Structures Norwegian University of Science and Technology

Summary

An overview of important developments regarding safety management of offshore structures is given. Based on relevant experiences with accidents, the hazards and the means to control the associated risk are categorized from a technical-physical as well as human and organizational point of view. This includes considerations of the risk associated with fatigue, corrosion and other degrading phenomena. The risk can be controlled by use of adequate design criteria, inspection, repair and maintenance of the structures as well as quality assurance and control of the engineering processes. Such measures are briefly outlined, while the emphasis is placed upon a quantitative design approach for dealing with structural robustness. In this connection the inherent differences in the robustness of various structural concepts are pointed out. The application of reliability methodology to obtain quantitative measures of structural safety relating to ultimate failure as well as handle the combined effect of design, inspection and repair strategy on fatigue failure is highlighted. The application of risk analysis to establish robustness criteria corresponding to a certain risk acceptance level is briefly mentioned. The challenges of Quality Assurance and Control to new structures are briefly outlined, with particular reference to recent examples of new loading phenomena such as ringing and springing of platforms.

1. Introduction Oil and gas are the dominant sources of energy in our society. Twenty percent of these hydrocarbons are recovered from reservoirs beneath the seabed. Various kinds of platforms are used to support exploratory drilling equipment, and the chemical (production) plants required to process the hydrocarbons. Large production platforms, such as some of those in the North Sea, represent investment of billions of U.S. dollars and significant operational costs. Pipelines or tankers are used to transport the hydrocarbons to shore. This paper is limited to deal with offshore structures, see Fig. 1.

a) Offshore oil and gas exploitation

b) Platform for exploratory drilling operation

c) Platform for oil and gas production (chemical processing)

Fig. 1 Subsea oil and gas exploitation

The continuous innovations to deal with new serviceability requirements and demanding environments as well the inherent potential of risk of fires and explosions have lead to an industry which has been in the forefront of development of design and analysis methodology. Fig. 2 shows phases in the life cycle of offshore structures. The life cycle of marine systems is similar to that of other systems. In view of the ecological issue removal or clean-up need to be considered. In this aspect the recycling of the material is an important issue. In the life cycle phases of structures the design phase is particularly important. Offshore structures need to fulfil serviceability and safety requirements. Serviceability requirements depend upon the function of the structure, which is to provide a platform and support of equipment for drilling or for the production of hydrocarbons. Drilling units need to be mobile while production platforms are generally permanent. Production platforms are commonly designed to carry large chemical factories together with large hydrocarbon inventories (Fig. 1c). Safety re1

quirements are introduced to limit fatalities as well as environmental and property damages. The focus here is on the structural safety during the life cycle of the platforms. A rational safety approach should be based on: Goal-setting; not prescriptive Probabilistic; not deterministic First principles; not purely experimental Integrated total; not separately Balance of safety elements; not hardware
Design for - serviceability & - producability - safety Data, methods, criteria

Layout/ Scantlings

Fabrication - Fabrication plan - Inspection/repair Operation - Operation plan Inspection/ monitoring/ repair / maintenance Reassessment Removal and reuse

Fabrication & Operation data

Fig. 2: Life cycle phases of offshore structures

The safety management of structures is different for different industries depending on the organisation as well as regulatory contents. For instance the safety management of offshore structures differs from that for trading ships. One reason for this difference is the fact that safety management of trading vessels emerged for centuries through empiricism while off-shore structures have primarily come about in the last 50 years when first principles of engineering science had been adequately developed to serve as basis for design. Similarly the regulatory regime differs between ships and offshore structures in that offshore operation take place on continental shelves under the jurisdiction of the local government. Typically authorities in the continental shelf states have to issue regulations. Examples of these are MMS/API in USA, HSE in UK and NPD in Norway. However the provisions for stability and other maritime issues are based on those of IMO and their corresponding national organizations. Classification societies primarily provide rules for drilling units and services for production facilities, which primarily are handled by national authorities. Since early 1990s ISO has been developing a harmonized set of codes for offshore structures with contribution from all countries with major offshore operations. Over the time safety management of offshore structures has been developed, in parallel with the evolvement of the technology and the competence to deal with it. Initially civil engineering was the driving force for structural safety management. Later the

aeronautical and nuclear industries also played an important role. However in the last 20-30 years the developments in the offshore industry has had a significant impact on the development of safety approaches. This is partly because the offshore industry plays a key role in the worlds economy. Moreover the oil and gas represent energy with large potential accident consequences. Companies involved in structures and/or facilities that experience accidents may suffer loss of reputation and this may damage the publics trust on the companies. The rationalisation of safety management of offshore structures began in early 70s. Among the milestones is the introduction of Risk Analysis in 1981 and Accidental Collapse Design criterion in 1984 by the Norwegian Petroleum Directorate and the HSE Safety case approach in 1992, when the ALARP principle - as large as reasonably practicable was introduced for determining the target safety level. To limit the likelihood of fatalities and environmental and property damages, offshore structures should be designed, fabricated and operated in such a manner that the probability of the following failure modes is adequately small: - overall, rigid body instability (capsizing) under permanent, variable and environmental loads - failure of (parts of) the structure foundation or mooring systems, considering permanent, variable, and environmental as well as accidental loads Stability requirements for floating platforms affect the layout and the internal structure subdivision in compartments. Criteria to prevent progressive structural failure after fatigue failure or accidental damage would have implications on overall layout of all types of platforms. Otherwise a structural strength criterion affects the scantlings of the stiffened, flat, and cylindrical panels that typically constitute floating offshore structures. If the location is far off shore then evacuation and rescue will be difficult. On the other hand, this implies that accidents on offshore facilities affect the general public to a lesser extent than accidents on similar facilities on land. In the following sections accident experiences on offshore structures will be briefly explained. Then, an outline of various measures to manage the safety or ensure that the risk is within acceptable limits, is given. This includes: Design and inspection criteria as well as reliability methodology to calibrate the partial safety factors to correspond to a defined acceptable safety level. In particular it is explained how fatigue failures can be avoided by design as well as inspections and regular monitoring of the structure. Emphasis is placed on how structural robustness can be ensured by using so-called Accidental Collapse Limit State (ALS) criteria. Such criteria are exemplified in relation to fires, explosions and other accidental loads. Finally, Quality Assurance and/or Quality Control of the engineering process will be described with particular reference to dealing with unknown wave loading and response phenomena. 2. Accident Experiences 2.1 Accident experiences at large Safety may be regarded as the absence of accidents or failures. Hence the insight about safety features can be gained from detailed information about accidents and failures. To learn about the intrinsic nature of accidents, it is mandatory to study the detailed accounts provided from investigations of catastrophic accidents since the
3

necessary resources are then spent to investigate such accidents. Such studies include those of the platforms Alexander Kielland in 1980 (ALK, 1981, Moan and Holand, 1981b), Ocean Ranger in 1982 (OR, 1984), Piper Alpha in 1988 (PA, 1990), and P-36 in 2001 (P-36, 2003). See also Bea (2000a, 2000b). In addition, the statistics about offshore accidents, such as ones given in WOAD (1996), provide an overview. Global failure modes of concern are - capsizing/sinking - structural failure - positioning system failure the former two modes represent catastrophic events while the latter one is only critical for Tension-leg platforms. Global failures normally develop in a sequence of technical and physical events. However, to fully understand accidents it is necessary to interpret them in the view of human and organizational factors (HOF). This includes possible deficiencies in relevant codes, possible unknown phenomena that have materialized as well as possible errors and omissions made in engineering processes, fabrication processes or in the operation itself.

a) Alexander L. Kielland before and after capsizing in 1980

b) Model of Ocean Ranger, which capsized in 1982, during survival testing

c) Piper Alpha fire and explosion in 1988

d) P - 36 accident in 2001

Fig. 3: Examples of accidents which resulted in a total loss.

Let us consider an example: The platform shown in Figure 4a in the Gulf of Mexico. This is one of many platforms that were damaged during the passage of the hurricane Lilli. Physically there is no doubt that this accident was due to extreme wave forces. To explain from a human and organizational point of view why the platform was not strong enough to resist the wave forces, we have to look at the decisions that were made during the design phase regarding loads, load effects, resistance and safety factors. The explanation might be that design was based on an inadequate wave conditions or load calculation. The damage could also be due to the occurrence of a particular a wave phenomenon, such as an abnormal wave crest (see Fig. 4b) or another unknown wave phenomenon. In the case of the exceptional wave in Fig. 4b, the question is whether the extreme crest height of 18.5 m should be considered as the socalled freak wave or simply a rare wave. Alternatively, the reason could be inadequate air-gap provided in the design. Yet another explanation might be that an improper strength formulation was used (as was the case in design of early generation platforms). Finally, the safety factors might not have been sufficient to cover the inherent uncertainties. For each of these possible causes, two explanations need to be
4

considered, namely 1) The state of art in offshore engineering was inadequate at the time of design ; 2) Errors and omission were made during design or fabrication! Obviously, these two explanations have different implications on the risk reducing actions. In this connection it is noted that several types of environmental load phenomena, such as green water on deck and slamming (Fig. 4 c-d) are subjected to large uncertainties. In general, if the phenomenon is known but subject to significant uncertainties, the design approach taken is normally conservative.
18

b) Wave record from a platform site in the North Sea on January 1. 1995.

a)Severe damage caused on a jacket platform in the Gulf of Mexico by Hurricane Lilli c) Green water and deck slamming on FPSO d) Deck slamming on semisubmersible platform

Fig. 4 Structural damage due to environmental loads

The technical-physical sequence of events for the Alexander Kielland platform was: fatigue failure of one brace, overload failure of 5 other braces, loss of column, flooding into deck, and capsizing. For Ocean Ranger the accident sequence was: flooding through broken window in ballast control room, closed electrical circuit, disabled ballast pumps, erroneous ballast operation, flooding through chain lockers and capsizing. Piper Alpha suffered total loss after: a sequence of accidental release of hydrocarbons, as well as escalating explosion and fire events. P-36 was lost after: an accidental release of explosive gas, burst of emergency tank, accidental explosion in a column, progressive flooding, capsizing and sinking after 6 days. Table 1 shows accident rates for mobile (drilling) and fixed (production) platforms according to the initiating event of the accident WOAD (1996). Table 1 is primarily based upon technical-physical causes. Severe weather conditions would normally affect capsizing/ foundering as well as structural damage. In most cases there existed human errors or omissions by designers, fabricators or operators of the given installation was a major contributor to the accident. The most notable in this connection is, of course accidents caused by loads such as ship impacts, fires and explosions which should not occur but do so because of errors and omissions during operation. In general, accidents take place in sequences. For floating platforms, the loss of buoyancy and stability is commonly an important aspect of total loss scenarios. Structural damage can cause progressive structural failure or flooding. Progressive flooding at5

tributes to a greater probability of total loss of floating structures than progressive structural failure. Degradation due to corrosion and fatigue crack growth are gradual phenomena. However, if the fatigue life is insufficient to make Inspection, Monitoring, Maintenance and Repair (IMMR) effective or if there is lack of robustness, fatigue can cause catastrophic accidents, see Fig. 5. Both cases shown in Fig.5 occurred for statically determinate platforms. In other situations through-the-thickness cracks were detected by inspections before they caused catastrophic failures (Moan, 2004). Corrosion is not known to have caused accidents with floating offshore structures of significance. On the other hand, maintenance related events for floating structures is limited. We need to be aware of this problem, especially for structures with a low fatigue life. Table 1: Number of accidents per 1000 platform-years. Adapted after WOAD (1996).
World wide Type of accident Blowout Capsizing/ foundering Dropped object Explosion Fire Grounding Spill/release Mobile 18.8/ 11.4 24.0/ 19.5 Fixed 2.5/0.9 0.5/0.8 1.6/1.0 0.5/0.8 0.7/1.6 2.0/7.5 1.8/8.7 0.5/0.6 Gulf of Mexico Fixed 2.2/1.0 0.3/1.1 1.3/0.7 0.1/0.4 0.3/0.4 1.0/7.8 1.0/5.8 0.4/0.5 North Sea Fixed 1970-79 /80-95 2.6/1.6 2.6/0.5 5.1/6.3 10.3/10.6 2.6/8.3 18.0/42.5 23.1/98.3 10.3/6.0

1970-79 /80-95 1970-79 /80-95 1970-79 /80-95

Collision / contact 24.6/ 14.6 4.2/ 6.1 7.4/3.3 12.3/ 11.9 6.1/3.3 4.9/5.9

Structural damage 25.6/ 18.4

Column D

Missing braces that cause no redundancy

Ranger I, 1979

Alexander Kielland, 1980

Fig. 5: The total losses of Ranger I in 1979 and Alexander Kielland in 1981 were initiated by fatigue failure
6

An overall picture of the accident rate in an industry may be displayed by the socalled Frequency-Consequence diagram as shown in Figure 6. The horizontal axis is plotted the consequence, in this case in terms of fatalities, N. The vertical axis is shown the frequency of N or more fatalities per accident. We see that the accident rate for mobile drilling units is much higher than for fixed production platforms. Fixed platforms are mainly used as production facilities. Moan and Holand (1981b) explained the main reasons for the differences in safety levels between mobile and fixed platforms. Floating production platforms are not included because of the limited experience with such platforms. The risk is similar that that of passenger vessels and tankers.
100 Annual frquency of an event with N or more fatalities 10-1 10-2
Marginally acceptable Acceptable

Oil platforms Mobile Fixed

10-3 10-4 Passenger ferries (not ro-ro) Tankers Merchant vessels 1 10 100 1000 10000

10-5

Number of lives lost, N

Fig. 6: Comparison of experienced overall accident rates with respect to fatalities in the offshore and shipping industries

2.2 Human and organizational factors Basically, structural failure occurs when the resistance, R is less than the load effect, S as indicated in Fig. 7. From a Human and Organizational Factor (HOF) point of view this can be due to too small safety factors to account for the normal uncertainty and variability in R and S relating to design criteria. But the main causes of actual structural failures are the abnormal resistance and accidental loads due to human errors and omissions. Design errors materialise as a deficient (or excessive) resistance, which cannot be derived from the parameters affecting the normal variability of resistance. Fabrication imperfections (such as cracks, plate misalignment, etc.), which also affect the resistance, are influenced by human actions. The normal variability of welders performance, environmental conditions, and soon lead to a normal variability in the imperfection size. This is characterised by a smooth variation of the relevant imperfection parameter. Occasionally a deviation from normal practice does occur, for instance as an abnormality caused by using a wet electrode, or another gross fabrication error. The Alexander L. accident in 1980 was caused by a fatigue failure of a brace and design checks had not been carried out. The implied fatigue life was further reduced to 3.5 years - by a fabrication error (70 mm weld defect) as well as inadequate inspections (ALK, 1981). Although the fatigue failures that had been experienced in semi-submersibles in the period 1965-70 resulted in fatigue standards, these

standards were not properly implemented even for platforms built in the 1970s. Many platforms built in the 1970s had joints with design fatigue lives as low as 2-5 years. This fact was evidenced in the extraordinary surveys undertaken after Alexander Kielland accident. The same happened to the first purpose built FPSO and shuttle tankers put into service in the mid-1980s. However, ships are obviously more robust or damage-tolerant than mobile semi-submersible platforms. Man-made live loads also have a normal and an abnormal component. While some loads, notably fires and explosions, ship collisions, etc. do not have a normal counterpart, they are simply caused by operational errors or technical faults. The mobile platform Ocean Ranger capsized in the offshore of Newfoundland in 1982. The accident was initiated by control room window breaking due to wave slamming. The water entering the control room lead to the short circuit of the ballast valve system, thereby leading to a spurious operation of ballast valves. The resulting accidental ballast condition could not be controlled partly because of lack of crew training and partly because of inadequate ballast pumps, and open chain lockers (OR, 1984). The catastrophic explosion and fire on the Piper Alpha platform in 1988 was initiated by a gas leak from a blind flange of a condensation pump that was under maintenance but not adequately shut down (PA, 1990). The main issue that caused the initiation of this accident was the lack of communication between the maintenance team and the control room operators. The gas ignited and the initial explosion lead to damage of an oil pipe and subsequent oil fires and explosions. In 2001 the platform P-36 in Brazil experienced a collapse of the emergency drainage tank, accidental explosion and subsequent flooding capsizing and sinking. A series of operational errors were identified as the main cause of the first event and also the sinking (P-36, 2001). It is a well known fact that the gross errors dominate as the cause of accidents, and therefore appropriate control measures should be implemented. It is found that the gross errors cause 80-90% of the failure of buildings and bridges and other civil engineering structures (Matousek and Schneider, 1976). The same applies to offshore structures.
R&D

Risk reduction
Do the job properly in the first place Unknown material or load phenomenon

Causes

Do the job properly in the first place

QA/QC
of design

Design error error

QA/QC
of the asfabricated structure

Abnormally Failure low Fabrication resistance R<S

QA/QC of
Design error - oversight of load Operational error - accidental load design

QA/QC of
operation Event control (leak, etc)

ULS: RC/R > S1SC1 + S2SC2 FLS: D=ni/Ni Dallowable Inadequate safety factors for normal variability of R and S

ALS design check

Apply adequate safety factors in ULS/FLS design check

Fig. 7 Interpretation of causes of structural failure and risk reduction measures.

It has been observed that errors and omissions occur especially in dealing with novel materials and concepts as well as during periods with economic and time pressures. In some cases, accidents have been caused by inadequate engineering practice such as the lack of knowledge regarding new phenomena. Recently new phenomena such as ringing and spinning of TLPs, degradation failure mechanism of flexible risers, have been discovered. Nevertheless they were observed in time before any catastrophic accident could occur.

3. Safety Management 3.1 General Offshore drilling, production or transport facilities are systems consisting of structures, equipments and other hardwares, as well as specified operational procedures and operational personnel. Ideally these systems should be designed and operated to comply with a certain acceptable risk levels as specified for example by the probability of undesirable consequences and their implications. The safety management needs to be synchronised with the life cycle of the structure. Structural failures are mainly attributed to errors and omissions in design, fabrication and, especially, during operation. Therefore, Quality Assurance and Control (QA/QC) of procedures and the structure during fabrication and use (operation) is crucial. To do a truly risk based design, by carrying out the design iteration on the basis of a risk acceptance criterion, and to achieve a design that satisfies the acceptable safety level, is not feasible. In reality, different subsystems, like: loads-carrying structure & mooring system process equipment evacuation and escape system

are designed according to criteria given for that particular subsystems. For instance, to achieve a certain target level, which implies a certain residual risk level, safety criteria for structural design are given in terms of Ultimate Limit State (ULS) and Fatigue Limit State (FLS) criteria. Using appropriate probabilistic definitions of loads and resistance together with safety factors, the desired safety level is achieved. The implicit risk associated with these common structural design criteria is generally small! The philosophy behind the Accidental Collapse Limit (ALS) criteria is discussed below. The nature of human errors differs from that of natural phenomena and normal man-made variability and uncertainty. Different safety measures are required to control error-induced risks. A number of people maintain that gross errors are Acts of God and cannot be dealt with. However, weld defects and fatigue failures due to gross errors had occurred before the Kielland accident ballast errors had occurred before the Ocean Ranger accident fires and explosions had occurred before the Piper Alpha accident

and so on

The occurrence of gross errors have been avoided by adequate competence, skills, attitude and self-checking of those who do the design, fabrication or operation in the first place; and by exercising self-checking in their work. In addition, quality assurance and control should be implemented in all stages of design, fabrication and operation. While the QA/QC in the design phase is concerned with scrutinizing the analysis, design checks and the final scantlings arrived at, the QA/QC during fabrication and operation phases refers to inspection of the structure itself. As mentioned above, operational errors typically result in fires or explosions or other accidental loads. Such events may be controlled by appropriate measures such as detecting the gas/oil leakage and activating shut down valve; extinguishing of a fire by an automatically-activated deluge system. These actions are often denoted as Event Control. Finally, Accidental Collapse Limit State criteria are implemented to achieve robust offshore structures, that is to prevent that the structural damage occurring as fabrication defects or due to accidental loads, escalate into total losses (Moan 1994). Table 2 summarises the causes of structural failure from a risk management point of view, and how the associated risk may be ameliorated. Adequate evacuation and escape systems and associated procedures are crucial for controlling failure consequences in terms of fatalities. Table 2: Causes of structural failures and risk reduction measures
Cause Less than adequate safety margin to cover normal inherent uncertainties. Gross error or omission during - design (d) - fabrication (f) - operation (o) Unknown phenomena Risk Reduction Measure - Increased safety factor or margin in ULS, FLS; - Improve inspection of the structure(FLS) - Improve skills, competence, self- checking (for d, f, o) - QA/QC of engineering process (for d) - Direct design for damage tolerance (ALS) and provide adequate damage condition (for f, o) - Inspection/repair of the structure (for f, o) - Research & Development

3.2 Design and inspection criteria Adequate performance of offshore structures is ensured by designing them to comply with serviceability and safety requirements for a service life of 20 years or more, as well as carrying out load or response monitoring, or inspection and taking the necessary actions to reduce loads directly or indirectly, by, e.g., removal of marine growth, or to repair, when necessary. Serviceability criteria are introduced to make the structure comply with the functions required. These criteria are commonly specified by the owner. Production platforms are usually made to be site- specific, while drilling units are commonly intended for operation in specific regions or world wide.

10

Safety requirements are imposed to avoid ultimate consequences such as fatalities and environmental or property damages. Depending upon the regulatory regime, separate acceptance criteria for these consequences are established. Property damage is measured in economic terms. Fatalities and pollution obviously also have economic implications. In particular, the increasing concern about environmental well-being can cause small damages to have severe economic implications. While fatalities caused by structural failures would be related to global failure, i.e. capsizing or total failure of deck support, smaller structural damages may result in pollution; or property damage which is costly to repair such as the damages of an underwater structure. The current practice which is implemented in new offshore codes, issued e.g. by API (1993/97), ISO 19900 (1994-) and NORSOK (1998a, 1998b, 1999, 2002) as well as by many classification societies, and the most advanced codes are characterized by design criteria formulated in terms of limit states (ISO 19900, 1994) see Table 3

- semi-probabilistic methods for ultimate strength design which have been calibrated by reliability or risk analysis methodology - fatigue design checks depending upon consequences of failure (damagetolerance) and access for inspection - explicit accidental collapse design criteria to achieve damage-tolerance for the system - considerations of loads that include payload; wave, current and wind loads, ice (for arctic structures), earthquake loads (for bottom supported structures), as well as accidental loads such as e.g. fires, explosions and ship impacts - global and local structural analysis by finite element methods for ultimate strength and fatigue design checks - nonlinear analyses to demonstrate damage tolerance in view of inspection planning and progressive failure due to accidental damage Fatigue crack growth is primarily a local phenomenon. It requires stresses to be calculated with due account of the long-term wave conditions, global behaviour as well as the geometric stress concentrations at all potential hot spot locations, and suitable fatigue criteria (e.g. Miners rule). Fatigue strength is commonly described by SNcurves, which have been obtained by laboratory experiments. Fracture mechanics analysis of fatigue strength have been adopted to assess more accurately the different stages of crack growth including calculation of residual fatigue life beyond throughthickness crack, which is normally defined as fatigue failure. Detailed information about crack propagation is also required to plan inspections and repair.

11

Table 3 Limit State Criteria for safety with focus on structural integrity
L im it s ta te s P h y s ic a l a p p e a r a n c e o f fa ilu r e m o d e R e m a rk s

U ltim a te (U L S ) - O v e r a l l r ig i d b o d y s t a b i lit y - U lt im a t e s t r e n g t h o f s t r u c t u r e , m o o r in g o r p o s s ib le f o u n d a t io n

C o lla p s e d c y lin d e r

D if f e r e n t f o r b o t t o m s u p p o rte d , o r b u o y a n t s tru c tu re s . C o m p o n e n t d e s ig n c h e c k

F a tig u e (F L S ) - F a ilu r e o f w e ld e d jo in t s d u e t o r e p e t it iv e lo a d s

F a t ig u e fra c tu re

C o m p o n e n t d e s ig n c h e c k d e p e n d in g o n r e s i d u a l s y s te m s tre n g th a n d a c c e s s f o r in s p e c t io n

A c c id e n ta l c o lla p s e ( A L S ) - U lt im a t e c a p a c it y 1 ) o f d a m a g e d s t r u c t u r e w it h c r e d i b le d a m a g e

J a c k -u p c o lla p s e d

S y s t e m d e s ig n c h e c k

An adequate safety against fatigue failure is ensured by design as well as by inspections and repairs. Fatigue design requirements depends upon inspect ability and failure consequences. Current requirements for fatigue design check in NORSOK are shown in Table 4. These values were established by the NPD code committee in 1984 by judgement. Table 4 Fatigue design factor, FDF to multiply with the planned service life to obtain the required design fatigue life (NORSOK N-001, 2002).
Access for inspection and repair No access or in the splash zone Substantial consequences Without substantial consequences
1)

Accessible (inspection according to generic scheme is carried out) Below splash zone 3 2 Above splash zone or internal 2 1

10 3

The consequences are substantial if the Accidental Collapse Limit State (ALS) criterion is not satisfied in case of a failure of the relevant welded joint considered in the fatigue check.

Traditionally we design for dead-loads, payloads as well as environmental loads. But, loads can also be induced by human errors or omissions during operation and cause accidental loads. They commonly develop though a complex chain of events. For instance hydrocarbon fires and explosions result as a consequence of an accidental leak, spreading, ignition and combustion process. Accidental Collapse Limit State (ALS) requirements are motivated by the design philosophy that small damages, which inevitably occur, e.g. due to ship impacts, explosions and other accidental loads, should not cause disproportionate consequences.

12

The first explicit requirements were established in Britain following the Ronan Point apartment building progressive failure in 1968. In 1984 such criteria were extended by NPD, to include such robustness criteria for the structure and mooring system. While robustness requirements to the mooring are generally applied today, explicit ALS criteria are not yet widespread. The World Trade Centre and other recent catastrophes have lead to further developments of robustness criteria for civil engineering structures. See Figure 8. ALS checks should apply to all relevant failure modes as shown in Figure 9. It is interesting in this connection to note that ALS-type criteria were introduced for sinking/ instability of ships long before such criteria were established for structural integrity as such. Thus, ALS were introduced in the first mobile platform rules (as described e.g. by Beckwith and Skillman, 1976). The damage stability check has typically been specified with damage limited to be one or two compartments flooded. According to NPD this damage should be estimated by risk analysis, as discussed subsequently. The criterion was formally introduced for all failure modes of offshore structures in Norway in 1984 (NPD, 1984).
Applied since early codes

Ronan point appartment building accident, 1968 Flixborough explosion, 1974 ECCS model codes, 1978 Alexander L. Kielland accident, 1980 NPD Regulations for Risk analysis, 1981 NPDs ALS criterion, 1984 HSE Safety Case, 1992 WTC, September 11., 2001

Motivation:
small damages, which inevitably occur, should not cause disproportionate consequences!
Flooded volume

a) Capsizing/sinking due to (progressive) flooding Explosion damage

Gaining acceptance

b) Structural failure e.g. due to impact damage,....

Failure of Dynamic Positioning System is handled in a similar manner

One tether failed

One mooring line failed

Generally applied

c) Failure of mooring system due to "premature" failure

Fig. 8: Historical development of ALS assessment of structures

Fig. 9: Accidental Collapse Limit State (ALS) requirements

The assessment of structures during operation is necessary in connection with a planned change of platform function, extension of service life, occurrence of overload damage due to hurricanes (Dunlap and Ibbs, 1994), subsidence of North Sea jackets (Broughton, 1997), explosions, fires and ship impact, updating of inspection plans etc (ISO 19900). Basically, the reassessment involves the same analyses and design checks as carried out during initial design. However, depending upon the inherent damage tolerance ensured by the initial design, the measures that have to be implemented to improve the strength of an existing structure may be much more expensive than ones for a new structure. This fact commonly justifies more advanced analyses of loads, responses, resistances as well as use of reliability analysis and risk-based approaches than in the initial design (Moan, 2000a).

13

3.3 Inspection, Monitoring, Maintenance and Repair Inspection, Monitoring, Maintenance and Repair (IMMR) are important measures for maintaining safety, especially with respect to fatigue, corrosion and other deterioration phenomena. To ensure structural integrity within the offshore sector in the North Sea, the regulatory body defines the general framework while the audit of the oil companies or rig owners defines: inspection and maintenance needs, reports planned activity, findings and evaluates conditions annually and every fourth or fifth year. Hence, the inspection history of a given structure is actively incorporated in the planning of future activities. The inspection and repair history is important for a rational condition assessment procedure of the relevant structure and other, especially for sister structures. The objective of inspections is to detect cracks, buckling, corrosion and other damages. Overload phenomena are often associated with a warning for which the inspection can be targeted, while degradation needs continuous surveillance. However, normally ample time for repair will be available in the latter cases. An inspection plan involves: - prioritizing which locations are to be inspected - selecting inspection method (visual inspection, Magnet Particle Inspection, Eddy Current) depending upon the damage of concern - scheduling inspections - establishing a repair strategy (size of damage to be repaired, repair method and time aspects of repair) Whether the inspection should be chosen to aim at detecting cracks by non-destructive examination (NDE), close visual inspection, detect through-thickness cracks e.g. by leak detection, or member failures would depend on how much resources are spent to make the structure damage tolerant. The choice again would have implication on the inspection method. The main inspection methods being the NDE methods consist of detection of through-thickness crack by e.g. leak detection, and visual inspection by failed members. The quality of visual inspection of NDE methods depends very much upon the conditions during inspection. A large volume offshore structure is normally accessible from the inside, while members with a small diameter such as TLP tethers and joints in jacket braces, are not. Permanent repairs are made by cutting out the old component and butt welding a new component, re-welding, adding or removing scantlings, brackets, stiffeners, lugs or collar plates. Typically major inspections of offshore structures (special surveys, renewal surveys) are carried every 4 - 5th year, while intermediate and annual inspections are normally less extensive. Further refinement of the inspection planning has been made by introducing probabilistic methods as described below. Inspection, monitoring and repair measures can contribute to the safety only when there is a certain damage tolerance. This implies that there is an interrelation between design criteria (fatigue life, damage tolerance) and the inspection and the repair criteria. Fatigue design criteria, hence, depend upon inspection and failure consequences as shown e.g. by Table 4. However, during the operation, the situation is different. The strengthening of the structure by increased scantlings is very expensive. The most relevant measure to influence safety relating to fatigue and other degradation phenomena is by using an im-

14

proved inspection method or increased frequency of inspections. The following section briefly describes how fatigue design and inspection plans (based on an assumed inspection method) can be established by reliability analysis to ensure an acceptable safety level. 3.4 Quantitative Measures of Safety Ideally the structural safety should be measured in a quantitative manner. Structural reliability methods are applied to determine the failure probability, Pf which is associated with normal uncertainties and variability in loads and resistance. Quantitative risk assessment can be used to deal with the probability of undesirable events and their consequences in general terms. This includes events induced by errors and omissions, see Fig. 10.
Structural reliability analysis
Deck

Prob. density function

Column

Load effect fS(s) R,S Resistance fR(r)

Wave pressure

PF=P[RS]

r,s
Quantitative risk analysis

Uncertainty in R and S can be modelled by probability density

End events
Critical event Event tree Consequences

Fault tree

Fig. 10: Methods for quantifying the risk or safety level

The quantitative safety approach is based on estimating the implied failure probability and comparing it with an acceptance level. This target safety level should depend upon the following factors (e.g. Moan, 1998): - type of initiating events (hazards) such as environmental loads, various accidental loads, .. which may lead to different consequences - type of SRA method or structural risk analysis, especially which uncertainties are included - failure cause and mode - the possible consequences of failure in terms of risk to life, injury, economic losses and the level of social inconvenience. - the expense and effort required to reduce the risk. In principle a target level which reflects all hazards (e.g. loads) and failure modes (collapse, fatigue, ... ) as well as the different phases (in-place operation and temporary phases associated with fabrication, installation and repair) is defined with respect to each of the three categories of ultimate consequences. The most severe of them governs the decisions to be made. If all consequences are measured in economic terms, then a single target safety level could be established. However, in practice it is convenient to treat different hazards, failure modes, and phases separately, with separate target levels. This may be reasonable because it is rare that all hazard scenarios
15

and failure modes contribute equally to the total failure probability. The principle of establishing target levels for each hazard separately was adopted by NPD for accidental loads; see e.g. Moan et al. (1993b). It was also advocated by Cornell (1995). In general it is recommended to calibrate the target level to correspond to that inherent in structures which are considered to have an acceptable safety. 3.5 Structural reliability analysis General Structural reliability methods for calculating the failure probability are readily available. If the uncertainty in the resistance R and load effect S are described by probability density functions. The failure probability can be calculated as P (R<S). It is important to recognize that there are different types of uncertainties used to determine the resultant uncertainties associated with loads and resistances. One type of uncertainty (Type 1) is natural or inherent; this type of uncertainty is information insensitive and random. A second type of uncertainty (Type 2) is associated with modelling, parametric, and state uncertainties; this type of uncertainty is information sensitive and systematic. Type 2 model uncertainties may be defined as the ratio of the actual or true value of the variable to the predicted or nominal (design) value of the variable. A variety of methods can be used to characterize the model uncertainty, including field test data, laboratory test data, numerical data, and expert judgment. Often it is not possible to develop explicit separations of Type 1 and Type 2 uncertainties and it is important not to include them twice. SRA is applied to determine the failure probability considering fundamental variability, as well as uncertainties due to the lack of knowledge in loads, load effects and resistance. The state of the art methods for calculating the failure probability are the numerical First Order and Second Order Reliability as well as Monte Carlo simulation methods (e.g. Melchers, 1999). However, analytical solutions exists for a few cases, for instance, when failure is expressed by g( ) =R S 0 and both the resistance R and the load effect S are lognormal random variables. The failure probability is expressed by:
Pf = P( g () 0) = ( ) or = 1 ( Pf )

(1)

where (-) is the standard cumulative normal distribution, with numerical values as shown in Table 5, and the reliability index, = LN can be exactly written as follows, see e.g. Melchers (1999):
1+V 2 S ln R S 1+V 2 ln ( R /S ) R = ' LN LN = 2 )(1 + 2 )] ln[(1 + V R VS V 2 +V 2 R S

(2)

This simple expression has turned out to be useful and was applied in the API reliability based code calibration (Moses, 1987). The analytical formulation can also conveniently be used to express the relationship between Pf and safety factors.

16

Table 5 Relation between and Pf.

1.0 1.4 1.8 0.036 2.2 0.014 2.6 0.47 10
-2

3.0 0.14 10
-2

3.4 0.34 10
-3

3.8 0.72 10
-4

4.2 0.13 10
-4

4.6 0.21 105

Pf 0.16 0.081

Reliability estimates are found to be sensitive to the distributions used for R and S. The failure probability should refer to a time interval, e.g. a year or the service life. This can be achieved by considering a load effect S that refers to an annual or service life time maximum value. We note that the results of code calibration depend upon the choice of reference period. Reliability based code calibration Reliability methods are increasingly used to make optimal decisions regarding safety and the life cycle costs of offshore structures (see e.g. ISSC, 1988-1994; Moan, 1994). In particular the efforts by Fjeld (1977); Lloyd and Karsan (1988), Moan (1988), Jordan and Maes (1991) to calibrate their codes to a certain reliability. An evaluation of previous efforts on calibration of offshore codes was provided by Moan (1995) in conjunction with the ISO effort to harmonize the safety level in codes for offshore structures across the variety of structural types (ISO, 1994). However, safety factors on loads are not properly varied to reflect the differences in uncertainty in load predictions for different types of structures. To illustrate the relationship between partial safety factors, the uncertainty in resistance and loads as well as Pf , consider the simplest design format, often used in code calibration,
R c / R SSc

(3)

where Rc and Sc are characteristic resistance and load effect, respectively. Let the (true) random load effect, S and resistance, R be defined by their mean value () and the coefficient of variation (V):
S = BSSC ,BS 1; VS = 0.15 0.30 R = BR R C ,BR 1; VR = 0.1

The BS reflects the ratio of the mean load (which refers to an annual maximum if the annual failure probability is to be calculated,) and the characteristic load effect (typically the 100 year value) as well as a possible bias in predicting wave load effects, e.g. due to model uncertainty. By inserting the design equation Eq.(3) into the approximate expression of Eq.(2)
LN
ln ( R /S )

VR2 + VS2

ln(BR R S /BS )

VR2 + VS2

or R S = (BR /BS ) exp(' L N VR2 +VS2 )

(4)

With R S = 1.5; a typical BS = 0.8 for wave-induced load effects; BR = 1.1 and VR = 0.1, it is found that LN is about 3.2 for a VS of 0.20. This reliability index corresponds to a Pf of 6 10-4. By decreasing BR/BS by 10 % reduces LN by 15%. It is noted that the Similarly, by increasing Vs by 10 % reduces LN by 8%. At the same time it is noted that the uncertainty in R has minimal influence on the safety level. Yet it is important to estimate the mean bias of the resistance, BR accurately. It is also possible to approximately express R and S by (BR, VR) and (BS, VS), respectively, and
17

hence to express partial factors by the relevant uncertainties. (e.g. Melchers, 1999). It is important to recognize that variables used in designing offshore structures are often conservative. Thus, there exists sources of bias that must be recognized quantitatively by the Bi's.
WSD: RC/ > DC + LC + EC Target Pf or LRFD: RC/R > DDC + LLC + EEC Load ratio, Ec/(Lc+Ec)

Goal: Implied Pf Pft

R resistance D, L, E load effects due to permanent live load environmental effects

Fig. 11: Schematic illustration on how the implied safety level in a design code for ultimate strength can be calibrated to be close to a given target level. Fatigue Reliability Analysis

Structural reliability methods can also be used to calculate the probability of fatigue failure. In Figure 12 the solid line with diamond symbol shows the fatigue failure probability in the service life as a function of the design criterion the fatigue design factor, FDF. It is shown that the cumulative failure probability in the service life varies from 10-1 to 10-4 when FDF varies from 1 to 10.
1.0E+00

Cumulative f ailure probability 1.0E-01 Cumulative, stdv (lnA )=0.15 Cumulative, stdv (lnA )=0.3 A nnual f ailure probability A nnual, stdv (lnA )=0.15 1.0E-02 Failure probability A nnual, stdv (lnA )=0.3

1.0E-03

1.0E-04

1.0E-05

1.0E-06 1 2 3 4 5 6 7 8 9 10 Fatigue de s ign factor

Fig. 12: Fatigue failure probabilities in the 20 year service period, as a function of the fatigue design factor and the uncertainty level. A is an equivalent constant stress range that represents the long term stress level (Moan, 2004).

A consistent fatigue safety level can be achieved, by varying the FDF versus the effect of an inspection program as well as the consequences of failure.

18

Reliability estimates by account of inspection The effect of the inspection on the reliability level can be illustrated by representing the crack depth using a random variable, A(t) which is a function of time t. The quality of the inspection in terms of the detectable crack size is also represented by a random variable, Ad. The distribution of Ad corresponds to the Probability of Detection (POD) curve for the inspection method in question. The failure probability at the time, t (N-cycles) can be formulated
Pf ( t ) = P(a f - a N 0 ) = P [ F ( 0, t )]

(5)

where af and aN are the crack size at failure and after N cycles, respectively. The outcomes of inspections are assumed to be no crack detection (ND) or crack detection (D) at time t after N cycles, which are described by:
I ND ( t ) : a N -a d 0
I D ( t ):a N -a d 0

(6a) (6b)

In general, it is difficult to determine the distribution of the crack size (A) explicitly when taking into account all uncertainties that affect the distribution as well as the effect of inspections. Based on the Paris crack propagation law, Eqs. (5-6) can be recasted into a convenient form for analysis as shown e.g. by Madsen and Srensen (1990). The effect of inspection may be viewed in two different ways depending upon whether it is assessed before inspections are done, e. g. during the design phase, or afterwards during operation. If the effect of inspections is estimated before they are carried out, two outcomes: D and ND are possible. The exact outcome is not known but the probability of the outcomes can be estimated based on the reliability method. At the design stage, the outcomes (e.g., crack detection or no detection) are not known. When a single inspection is assumed to be made at time tI and possible cracks detected are repaired, the failure probability in the period t tI can be determined by:
Pf (t) = P [ F(0, t I )] + P [S(0, t I ) and F(t I , t) | ID (t I )] P [ ID (t I )] + P [S(0, t I )and F(t1 , t) | I ND (t I )] P [ I ND (t I )]

(7)

where F(t1,t2) and S(t1,t2) are, respectively, mean failure and survival in time period (t1,t2). Equation (7) can be generalised to cover cases with several inspections with two alternative outcomes. Moan et al. (1993a) showed, based on reliability analysis, how the allowable cumulative damage (D) at the design stage can be relaxed when inspections are carried out. Such analyses served as basis for Table 4. On the other hand if no failure has occurred before time tI and it is known that no crack is detected at time tI, then the failure probability in the period t tI is
Pf (t) = P [ F(t I , t) | I ND (t I )]

(8)

The knowledge of survival up to time tI and no crack detection at time tI reduces the uncertainty and makes the failure probability drop. The reliability index increases at the time of inspection as illustrated by the example shown in Fig. 13.

19

Event tree analysis Basic case, No inspection

6

Upd, full inspection history Upd, ONLY last inspection

Reliability Index

Inspection during operation with No crack detection

No inspection

10-3 310-3 Effect of Inspection predicted at design stage 3.510-2 Pf

20

1 0 5 10 15

Time (years)

Fig. 13: Reliability index as a function of time and inspection strategy. Inspection Event Tree analysis is based on predictions at the design stage. The other curves are based on inspections with known outcome during the service life (Ayala-Uraga and Moan, 2002)

The updating methodology is useful in connection with extension of service life for structures with joints governed by the fatigue criterion (Vrdal et al, 2000). In such cases, the design fatigue life is in principle exhausted at the end of the planned service life. Nevertheless, if no cracks have been detected during inspections, then a remaining fatigue life can be demonstrated. However, it is not possible to bring the structure back to its initial condition by inspection only. This is because the mean detectable crack depth by NDE methods typically is 1.0 2.0 mm, while the initial crack depth is 0.1 0.4 mm. The calculation of the system failure probability after inspection may be approximated by independent system failure modes (Moan et al., 1999, 2002, 2004)
PFSYS|up = P [ FSYS | I]P [ FSYS(U)] + P Fj | I P FSYS(U) | Fj +....
j=1 n

(9)

This formulation is based on modeling the ultimate failure of the system by a single mode. Moreover, the formulation is limited to failure modes initiated by a single fatigue failure and followed by ultimate global failure. The failure probability in Eq. (9) is applicable when the inspection event I aims at detecting cracks before the failure of individual members, (i.e. before they have caused rupture of the member). Another inspection strategy would be to apply visual inspection to detect members failure and repair failed members after the winter season in which those particular members failed. In this case the Eq. (9) will have to be modified as follows: the individual fatigue failures of components (Fj ) does not depend on the inspection event, and, rather such an inspection and repair strategy will have implication on the time period, for which the failure probability P[FSYS(U)|Fj] should be calculated. A further simplification is to update the failure probability of each joint based on the inspection result for that joint. This is conservative if no cracks are detected, but nonconservative if cracks are detected. Inspections may be prioritized by using Eq. (9) for each joint separately by allowing a
20

certain target probability level, PfSYS(T) to each term in the sum of Eq. (9). The target fatigue failure probability for joint i, PFfT(i) is then obtained from
PfSYS(i) = P[FSYS | Fi ] PFfT(i) PfSYS(T)

(10)

where the system failure probability, PfSYS(i) is associated with a fatigue failure of member (i) followed by an ultimate system failure. PfSYS(T) is obtained by generalizing the acceptance criteria implied by Table 4. This approach has been implemented for template-space frame structures (Moan et al., 1999). Given the target level for a given joint, inspections and repairs by grinding or other modifications are scheduled to maintain the reliability level at the target level as shown in Fig. 14.
Reliability level,
No inspection Inspection at time t=8 with no crack detection

Target level for a given joint

12

16

20

Time (years)

1st inspection

2nd inspection

Fig. 14: Scheduling of inspections to achieve a target safety level of PFfT(i).

This methodology is used to calibrate fatigue design requirements. It is then found that the criteria in Table 4 are slightly non-conservative.

3.6 Safety implications of Ultimate and Fatigue Limit State criteria and Inspection, Monitoring, Maintenace and Repair The failure probability estimated by structural reliability analysis (SRA) normally does not represent the experienced Pf for structures. This is because the safety factors or margins normally applied to ensure safety are so large that Pf calculated by SRA becomes much smaller than that related to other causes. For instance when proper fatigue design checks and inspections have not been carried out, the likelihood of fatigue failures (through-thickness cracks) for platforms (e.g. in the North Sea), is large and cracks have occurred. However, with the exception of the Ranger I (1979) and Alexander Kielland failure (1980) such cracks have been detected before they caused total losses. As discussed above, errors and omissions in design, fabrication and operation represent the main causes of the accidents experienced. On the other hand, frequent occurrences of cracks provide a basis for correlating actual crack occurrences with state of art predictions for various offshore structures. Hence, the current predictions for jackets are found to be conservative (Vrdal and Moan, 1997), while for semi-submersibles and ships, the predictions seem to be reasonable, as summarized by Moan (2004). This agreement is achieved when the SN approach (or a calibrated fracture mechanics approach) is applied to predict the occurrence of fatigue failure (e.g. through thickness crack). Yet, if ULS and FLS design checks are properly carried out, Pf will be negligible within the current safety regime. This reserve capacity, implied by ULS and FLS requirements, provides some
21

resistance against other hazards like fires, explosions etc. However providing safety for the mentioned hazards in this indirect manner is not an optimal risk-based design. If more efforts were directed towards risk reduction actions by implementing ALS criteria, then current safety factors for ULS and FLS could be reduced without increasing the failure rate noticeably. As explained above, SRA does not provide a measure of the actual total risk level associated with offshore facilities. Yet, it is useful in ensuring that the ultimate strength and fatigue design criteria are consistent by calibrating safety factors. Moreover, SRA provides a measure of the influence of various parameters on the reliability and, hence, the effect of reducing the uncertainty on the failure probability. Finally, it is noted that the random uncertainties in the ultimate strength commonly have limited effect on the reliability compared to that inherent in load effects. On the other hand, the systematic uncertainty (bias) in strength and load effects has the same effect on the reliability measure. 3.7 Risk assessment Risk assessment (Qualitative Risk assessment or Formal Safety Analysis etc.) is a tool to support decision making regarding the safety of systems. The application of risk assessment has evolved over 25 years in the offshore industry (Moan and Holand, 1981b, NPD (1981)). The Piper Alpha disaster (PA, 1990), was the direct reason for introducing PRA, (or QRA), in the UK in 1992 (HSE, 1992). In the last 5 years such methods have been applied in the maritime industry, albeit in different directions (Moore et al. 2003). The offshore industry has focused on the application of risk assessment to evaluate the safety of individual offshore facilities. The maritime industry has primarily focused on the application of risk assessment to further enhance and bring greater clarity to the process of making new ship rules or regulations. The risk assessment methods is used because they provide a reliable direct determination of events probabilities e.g. probabilities as low as 10-4 per year. Up to now the accumulated number of platform years world wide is about 120 000, 15 000 and 1 200 for fixed, mobile structures and FPSOs, respectively. However, to determine probabilities as low as 10-4 per year requires about 23000 years of experiences to have a 90% chance of one occurrence. A further complexity is that the available data refer to various types of platforms and, not least, different technologies over the years. Application of a systems risk assessment is therefore attractive. The basis for this approach is the facts that : a)almost every major accidental events have originated from a small fault and gradually developed through long sequences or several parallel sequences of increasingly more serious events, and culminates in the final event b) it is often reasonably well known how a system responds to a certain event. By combining the knowledge about system build-up with the knowledge about failure rates for the elements of the system, it is possible to achieve an indication of the risks in the system (Vinnem, 1999; Moan, 2000b). The risk analysis process normally consists in the following steps (Fig 15): definition and description of the system identification of hazards analysis of possible causal event of hazards determination of the influence of the environmental conditions determination of the influence of active/passive safety systems (capacity; reliability, accident action integrity, maintenance system ..)
22

estimation of event probabilities/event magnitudes estimation of risk

Risk Analysis Planning Risk Analysis Planning System Definition System Definition Hazard Identification Hazard Identification Risk Risk Reducing Reducing Measures Measures

Risk Risk Acceptance Acceptance Criteria Criteria

Frequency Analysis RISK ESTIMATION RISK ESTIMATION

Consequence Analysis

Risk Picture Risk Picture RISK ANALYSIS RISK ANALYSIS Risk Evaluation Risk Evaluation Tolerable Acceptable Acceptable

Unacceptable

Fig. 15: General approach for risk based decision making

In most cases an Event-Fault Tree technique (Figure 16) is the most appropriate tool for systematizing and documenting the analyses made. Although the Event-Fault tree methodology is straightforward, there are many problems. An important challenge is to determine the dominant of the (infinitely) many sequences. Events are not uniquely defined in a single sequence but appear in many combinations. Moreover, human factors are difficult to account for in the risk assessment. However, operational errors that result in accidental loads are implicitly dealt with by using data on experienced releases of hydrocarbons, probability of ignition etc. Explicit prediction of design and fabrication errors and omissions for a given structure is impossible. However, it is possible to rate the likelihood of accidents as compared to gross errors (Bea, 2000a-b, Lotsberg et al., to appear). The risk analysis methodology currently applied in offshore engineering is reviewed in detail by Vinnem (1999). In connection with accidental loads, the purpose of the risk analysis is to determine the accidental events which annually are exceeded by a probability of 10-4.
End event

Critical event Fault tree Event tree Consequences

Fig.16 Schematic sketch of the event fault tree method.

23

3.8 Failure probability implied by Accidental Collapse Limit State Criteria The initial damage in the ALS criterion, (e.g. due to fires explosions, ship impacts, or, fabrication defects causing abnormal fatigue crack growth), corresponds to a characteristic event for each of the types of accidental loads which is exceeded by an annual probability of 10-4, as identified by risk analyses. The (local) damage, or permanent deformations or rupture of components need to be estimated by accounting for nonlinear effects. The structure is required to survive in the various damage conditions without global failure when subjected to expected still-water and characteristic sea loads which are exceeded by an annual probability of 10-2. In some cases compliance with this requirement can be demonstrated by removing the damaged parts and then accomplishing a conventional ULS design check based on a global linear analysis and component design checks using truly ultimate strength formulations. However, such methods may be very conservative and more accurate nonlinear analysis methods should be applied, as described subsequently. The conditional probability of failure in a year, for the damaged structure, can be estimated by Eqs. (1-2), assuming that the system failure can be modelled by one failure mode and that the design criterion is fully utilized. The design checks in the ALS criterion is based on a characteristic value of the resistance corresponding to a 95% or 5% fractile, implying a BR = 1.1. The characteristic load effect due to functional and sea loads are 1.0-1.2 and 1.2-1.3 of the corresponding mean annual values, respectively. The safety (load and resistance) factors are generally equal to 1.0 for both checks. For environmental loads, this conditional failure probability will be of the order of 0.1. The intended probability of total loss implied by the ALS criterion for each category of abnormal strength and accidental load would then be of the order of 10-5 (Moan, 1983). Obviously, such estimates are not possible to substantiate by experiences.

3.9 Design for damage tolerance Introduction The current regulations for offshore structures in Norway are based on the following principles: Design the structure to withstand environmental and operational loading throughout its lifecycle. Prevent accidents and protect against their effects Tolerate at least one failure or operational error without resulting in a major hazard or damage to structure Provide measures to detect, control, and mitigate hazards at an early time accidental escalation.

Accidental Collapse Limit State criteria can be viewed as a means to reduce the consequences of accidental events (Fig. 17). The NORSOK N-001 code specifies quantitative ALS criteria based on an estimated damage condition and a survival check. The robustness criteria in most other codes, however, do not refer to any specific hazard but rather require that progressive failure of the structure with one element removed at a time, is prevented. Hence, no performance objective for a real threat is created.

24

The weakness with such a criterion is that it does not distinguish between the differences in vulnerability In a risk analysis perspective the ALS check of offshore structures is aimed at preventing progressive failure and hence reduce the consequences due to accidental loads, as indicated in Figure 17. Beside progressive structural failure, such events may induce progressive flooding and hence the capsizing of floating structures.

Risk control of accidental events

A
A

P, F

Estimate the damage due to accidental loads (A) at an annual probability of 10 -4 - apply risk analysis to establish design accidental loads

Reduce probability

Reduce consequences "known events" "unknown events" Indirect design - robustness - redundancy - ductility
Fault tree
P, F

Critical event Event tree End events: Accidental loads Survival check of the damaged structure as a whole, considering P, F and environmental loads ( E ) at a probability of 10 -2

Reduce errors & omissions

Event Control

Direct ALS design - Abnormal resistance - Accidental loads

Risk Analysis, or, Prescriptive code requirements

E

Target annual probability of total loss: 10 -5 for each type of hazard

Fig. 17: The role of ALS in risk control

Fig. 18: Accidental Collapse Limit State (NPD, 1984)

The relevant accidental loads and abnormal conditions of structural strength are drawn from the risk analysis, see e.g. Vinnem (1999) and Moan (2000b), where the relevant factors that affect the accidental loads are accounted for. In particular, the risk reduction can be achieved by minimizing the probability of initiating events: leakage and ignition (that can cause fire or explosion), ship impact, etc. or by minimizing the consequences of hazards. The passive or active measures can be used to control the magnitude of an accidental event and, thereby, its consequences. For instance, fire loads are partly controlled by sprinkler/inert gas system or firewalls. Fenders are commonly used to reduce the damage due to collisions. ALS checks apply to all relevant failure modes as indicated in Table 6. An account of accidental loads in conjunction with the design of the structure, equipment, and safety systems is a crucial safety measure to prevent escalating accidents. Typical situations where direct design may affect the layout and scantlings are indicated by Table 7 for different subsystems: - loads-carrying structure & mooring system - process equipment - evacuation and escape system

25

Table 6 Examples of accidental loads for relevant failure modes of platforms.

Structural concept Fixed platforms Floating platforms Failure mode Structural failure Structural failure Instability Relevant accidental load or condition All All

Collision, dropped object, unintended pressure, unintended ballast that initiate flooding Collision on platform Abnormal strength Accidental actions that initiate flooding Collision on platform Dropped object on tether (Abnormal strength)

Mooring system strength Tension-leg platforms Structural failure Mooring - slack system - strength All

Table 7 Design implications of accidental loads for hull structure

Load Fire Structure Columns /deck (if not protected) Equipment Exposed equipment (if not protected) Exposed equipment (if not protected) Possibly exposed risers, (if not protected) Equipment on deck, risers and subsea (if not protected) Passive protection system Fire barriers Blast / Fire barriers Possible fender systems Impact protection

Explosion Topside (if not protected) Ship impact Dropped object Waterline structure (subdivision) (if not protected) Deck Buoyancy elements

Design accidental loads The characteristic value of accidental loads is defined as the load which annually is exceeded by a probability of 10-4 and should be determined by risk analysis. For each physical phenomenon (fire, explosions, collisions, ..) there is normally a continuous spectrum of accidental events. A finite number of events have to be selected by judgement. These events represent different load intensity at different probabilities. The characteristic accidental load on different components of a given installation can be determined as follows (Moan, 2000b): establish exceedance diagram for the load on each component allocate a certain portion of the reference exceedance probability (10-4) to each component determine the characteristic load for each component from the relevant load exceedance diagram and reference probability.

If the accidental load is described by several parameters (e.g. heat flux and duration for a fire; pressure peak and duration for an explosion) design values may be obtained from the joint probability distribution by contour curves (NORSOK N-003, 1999).
26

However, in view of the uncertainties associated with the probabilistic analysis, a more pragmatic approach is sufficient. Yet significant analysis efforts are involved in identifying the relevant design scenarios for the different types of accidental loads. For each design accident scenario, the damage imposed on the offshore installation needs to be estimated followed by an analysis of the residual ultimate strength of the damaged structure in order to demonstrate survival of the installation. To estimate damage, (permanent deformation, rupture etc of parts of the structure), the nonlinear material and the geometrical structural behaviour need to be accounted for. While in general the nonlinear finite element methods are applied, simplified methods (e.g. based on plastic mechanisms) are developed and calibrated using more refined methods, to limit the computational effort required. The risk analysis of novel structures and systems, is found to be useful, in that they provide insight which results in systems that have significant increase in safety at the same expense. This applies in particular to the topside system. However, for mature systems, the outcomes of risk analyses tend to confirm the results of previous analyses. This fact together with the desire to simplify design practice suggests using specific, generic values for such cases. Examples of typical values for some accidental loads are given in subsequent sections. Analysis tools for estimating the initial damage and survival Current ultimate strength code checks of marine structures are commonly based on load effects (member and joint forces) that are obtained by a linear global analysis. Experiments or theory which accounts for plasticity and large deflections are used to obtain resistances of the members and joints. Hence, this methodology focuses on the first failure of a structural component and not the overall collapse of the structure, which is of main concern. The advent of computer technology and the finite element method have made it possible to develop analysis tools that account for nonlinear geometrical and material effects, and, therefore, make it possible to account for redistribution of the forces and subsequent component failures until the systems collapse. By using such methods a more realistic measure of the overall strength of structures is achieved. Recently, Skallerud and Amdahl (2002) prepared a state-of-the-art review of methods for nonlinear analysis of space frame offshore structures. Paik and Thayamballi (2003) gave an overview of methods for ultimate strength analysis of steelplated structures. Simplified methods for calculating the hull girder strength are based on considerations of the intact longitudinal elements and beam theory, essentially based on Smiths work (1977), and reviewed by e.g. Yao et al. (2000). Such an approach has also been extended to estimate the ultimate capacity of the damaged hull girder (Smith, 1977). However, it is necessary to further investigate the implication of an initial damage that involves rupture and, hence, represent an initial crack type damage which could cause rupture before reaching the ultimate capacity obtained by calculation models based upon ductile material behaviour. Fires and explosions effects The dominant fire and explosion events are associated with hydrocarbon leak from flanges, valves, equipment seals, nozzles etc. As indicated in Fig. 19 fire and explosion events are strongly correlated. Commonly the effect of 40 60 scenarios needs to be analyzed. This means that the location and magnitude of relevant hydrocarbon

27

leaks, the likelihood of ignition, as well as the combustion and temperature development (in a fire) and the pressure-time development (for an explosion) need to be estimated and followed by a structural assessment of the potential damage.
No damage No Ignition Damage to Personnel and Material Immediate Ignition Release of Gas and/or Liquid Fire

Formation of Combustible Fuel-Air Cloud (Pre-mixed)

Fire Ignition (delayed)

Gas Explosion Fire and BLEVE

Fig. 19: Fire and explosion scenarios.

The fire thermal flux may be calculated on the basis of the type of hydrocarbons, its release rate, combustion, time and location of ignition, ventilation and structural geometry, using simplified conservative semi-empirical formulae or analytical/numerical models of the combustion process. The heat flux may be determined by empirical, phenomenological or numerical method (SCI, 1993; BEFETS, 1998). Typical thermal loading in hydrocarbon fire scenarios may be 200- 300 kW/m2 for a 15 minutes up to a two hours period. The structural effect is primarily due to the reduced strength with increasing temperature. An A-60 fire protection wall may be applied for a heat load of 100kW/m2 and less, while H-rated protection walls are needed for higher heat loading. In the case of explosion scenarios, the analysis of leaks is followed by a gas dispersion and possible formation of gas clouds, ignition, combustion and the development of overpressure. Tools such as FLACS, PROEXP, or AutoReGas are available for this purpose (Moan, 2000b; Czujko, 2001, Walker et al., 2003). The variability of conditions is accounted for by using a probabilistic approach. The results from the gas explosion simulations are the pressure time history. If the pressure duration is short compared to the natural period, the pressure impulse governs the structural response. Figure 20 compares the predicted impulse by state of a state of the art CFD method with measured values in large scale tests for deterministic explosion scenarios. The vertical axis is a logarithmic plot of the ratio of the predicted and measured value. The scattering is seen to be significant. The pressure peaks would obviously be even more uncertain.

28

10.0

Ratio of predicted and measured Impulse

1.0

0.1

Experiment No

Fig. 20: Comparison of predicted and measured pressure impulse for deterministic explosion scenarios, obtained by the computer code FLACS.

The typical overpressures for topsides of North Sea platforms are in the range 0.2-0.6 barg, with duration of 0.1-0.5s., while an explosion in open air at the drill floor typically implies 0.1 barg with duration of 0.2s. The explosion pressure in a totally enclosed compartment might be 4 barg. The damage due to explosions may be determined by simple and conservative singledegree-of freedom models (NORSOK N-004). In several cases where simplified methods have not been calibrated, nonlinear time domain analyses based on numerical methods like the finite element method should be applied. A recent overview of such methods may be found in Czujko (2001). Fig. 21 shows an explosion panel with deformations as determined by an experiment and finite element analysis. The calculated and measured deflections of the specimen are compared in Figure 21c. Fire and explosion events that result from the same scenario of released combustibles and ignition should be assumed to occur at the same time, i.e. to be fully dependent. The fire and blast analyses should be performed by taking into account the effects of one on the other. The damage done to the fire protection by an explosion preceding the fire should be considered.
0.7 0.6 0.5 0.4 0.3 PRESSURE [N/ 2] 0.2 0.1 0 Experiment Analysis

20

40 60 DISPLACEMENT [mm]

80

100

a) Experiment

b) FE analysis

c) Loadresponse histories

Fig. 21: Explosion response of an explosion wall (Czujko, 2001).

29

Pmax > 5bar

Fig. 22: Survival analysis of a deck suffering explosion damage (Amdahl, 2003). Deformations in the lower figure are not to scale.

Fig. 22 shows results from an analysis of a deck structure in a floating platform (Amdahl, 2003). The upper left figure in this slide illustrates the deck structure of a floating production platform. The design pressure on the East Wall is also indicated. In this case it is assumed that the panels are badly damaged that they can be removed. The lower figure shows the deformation pattern of the damaged deck. Ship impacts Significant efforts have been devoted to ship-ship collisions, as reviewed by the ISSC Committee on Collision and Grounding (Paik et al., 2003). The analysis of ship impacts on offshore structures follows the same principles but the collision scenarios and consequences are different; see e.g. NORSOK N-003 and -004 as well as Amdahl (1999). All ship traffic in the relevant area of the offshore installation should be mapped and should account for possible future changes in the operational pattern of vessels. The impact velocity can be determined based on the assumption of a drifting ship or on the assumption of erroneous operation of the ship. Ship traffic may therefore for this purpose be divided into categories: trading vessels and other ships outside the offshore activity, offshore tankers, and supply or other service vessels. Merchant vessels are often found to be the greatest platform collision hazard which depends upon the location of the structure relative to shipping lanes. Fig. 23a indicates situations where offshore structures are operating in close proximity. For the scenario in Fig. 23b the stern impact on the FPSO by the shuttle tanker is a challenge (Chen and Moan, 2004 )

30

a) Semi- submersible and jacket b) FPSO and shuttle tanker

Fig. 23: Special offshore collision scenarios.

Impact scenarios are established by considering bow, stern, and side impacts on the structure as appropriate. While historical data provides information about supply vessel impacts, risk analysis models are necessary to predict other types of impacts, such as incidents caused by trading vessels (see e.g. NORSOK N-003(1999) and Moan (2000b)). A minimum accidental load corresponding to 14 MJ and 11 MJ sideways and head-on impact, respectively, is required to be considered. The impact damage can normally be determined by splitting the problem into two uncoupled analyses. They are the external collision mechanics dealing with global inertia forces and hydrodynamic effects and internal mechanics dealing with the energy dissipation and distribution of damage in the two structures (Fig. 24). The external mechanics analysis is carried out by assuming a central impact and applying the principle of conservation of momentum and conservation of energy. The next step is to estimate how the energy is shared among the offshore structure and the ship. Methods for assessing the impact damage are described by Amdahl (1999), based on simplified load-indentation curves or direct finite element analysis. For the general case where both structures absorb energy, the analysis has to be carried out incrementally on the basis of the current deformation field, contact area and force distribution over the contact area.
External mechanics The fraction of the kinetic energy to be absorbed as deformation energy (structural damage) is determined by means of: Conservation of momentum Conservation of energy

External mechanics
Rs Ri

Es,s dws Ship

Es,i

FPSO

dwi

Internal mechanics Energy dissipated by vessel and offshore structure Equal force level Area under force-def. curve

Internal mechanics

Fig. 24: Simplified methods for calculating impact damage (NORSOK N- 004)
31

The recent advances in computer hardware and software have made nonlinear finite element analysis (NLFEM) a viable tool for assessing collisions. A careful choice of element type and mesh is required. It is found that a particularly fine mesh is required in order to obtain accurate results for components deformed by axial crushing. A major challenge in NLFEM analysis is the prediction of ductile crack initiation and propagation. This problem is yet to be solved. The crack initiation and propagation should be based on fracture mechanics analysis using the J-integral or Crack Tip Opening Displacement method rather than simple strain considerations. While the main concern about ship impacts on fixed platforms is the reduction of structural strength and possible progressive structural failure, the main effect for buoyant structures is damage that can lead to flooding and, hence, loss of buoyancy. The measure of such damages is the maximum indentation implying loss of water tightness. However, in the case of large damage, reduction of structural strength, as expressed by the indentation, is also a concern for floating structures. A ship impact involving the minimum energy of 14 MJ will normally imply an indentation of 0.4 1.2 m in a semi-submersible column. A 75-100 MJ impact may be required to cause an indentation equal to the column radius (Moan and Amdahl, 1989). The effect is highly dependent upon the location of impact contact area relative to decks and bulkheads in the column. Moan and Amdahl (2001) considered supply or merchant vessels with a displacement of 2000- 5000 tons which caused bow impacts on a typical side structure of an FPSO with displacement of about 100 000 t. For this case the limiting energy for rupture of the outer ship side was found to be ~10 MJ (700mm bow displacement). The energy at 1500mm bow indentation was estimated to be ~43 MJ while the limiting energy for rupture of the inner side is ~230 MJ, at 3700 mm bow displacement. This corresponds to a critical speed of 18.5 knots for a 5000 tons displacement vessel. We recall that rupture of plate material is very uncertain such that the quoted energy level should be used with cautiously. If the FPSO side structure is assumed to be sufficiently rigid to ensure that all energy is absorbed by the bow, then the maximum collision force (peak force) is 30 MN. However most of the time the force oscillates between 15-25 MN. Another situation dealt with by Moan and Amdahl was stern impact on the FPSO caused by a shuttle tanker, see Fig. 23b and 25. A 70 kdwt shuttle tanker was found to penetrate the machine room of a 140 kdwt FPSO after an energy dissipation of 38 MJ. On the other hand, it was demonstrated that it was feasible to strengthen the stern of the FPSO corresponding to ice-strengthening dimensions so that all dissipation could occur in the shuttle tanker bow.
Shuttle U p p er tanker d ec k
F o r ca s tle d ec k

FPSO

Fig. 25: Shuttle tanker in ballast condition impacting FPSO stern (Moan and Amdahl, 2001).
32

Environmental events The ALS criterion is supposed to be applicable to any abnormal wave loading as well. Obviously, the two-step ALS procedure then becomes a survival check based on a load event which annually is exceeded by a probability of 10-4. Consider a wave loading, F, which for simplicity is characterised only by the wave height, H. Assume that F is a smooth function of wave height. H i.e. F=cH and that the wave height follows a Weibull distribution with a shape parameter B. If it is assumed that there are 5108 waves in 100 years, the ratio between the loads at a 10-4 and 10-2 annual probability level, respectively, becomes Fc(-4)/Fc(-2) = (H(-4)/H(-2)) = (1.23)/B. The ratio of the required strength would then be Rc(-4)/Rc(-2) = Fc(-4)/ (T Fc(-2) ) = 1.23/B/T where T is the total ULS safety factor while the safety factor for ALS is 1.0. This implies that ULS will be governing for North Sea and other conditions (B=0.9-1.0) if < 1.8-2.0. In benign conditions (e.g.with B=0.6) ULS will be governing if < 1.2.

Courtesy: Statoil, Marintek

Fig. 26: Wave in deck load for a fixed platform

However, there are two issues that need to be observed in this connection. The first issue is the occurrence of possible abnormal waves, with high crest or other unusual shape which is not a simple extrapolation of the 10-2 event (Haver, 2000, Prevost and Forristall, 2002). The second issue is the possible sudden change in force, F, at a certain wave height. The most interesting case is when the wave reaches the platform deck, implying that the wave force will increase very fast with the wave height. By ensuring a deck clearance such that the 10-4 crest does not reach the deck, the ULS criterion would normally be governing. Otherwise, the ALS criterion based on the 104 wave event may govern the dimensions. Abnormal resistance It is not possible to determine the abnormal resistance (e. g. due to fabrication defects), using risk analyses. Up to now, abnormal resistance has been explicitly specified by generic values for specific types of structures based on some considerations of the vulnerability of the structural components. For instance, the ALS check is carried out for platforms with slender braces by considering the damage in terms of severed individual braces. This condition was established in the aftermath of the Alexander Kielland accident and was initially aimed to cover the effects of frequently occurring ship impacts relating to supply vessels as well as abnormal fatigue cracks. This damage condition is also applicable to the tether and other mooring systems.

33

Crack control Most degradations of the structure are due to corrosion and crack growth. The effect of corrosion is ameliorated by corrosion allowance or a protection system, which makes the corrosion development gradual and, hence, be easy to control. The crack growth is more critical because cracks can result is a sudden rupture. Moreover, cracks are hard to detect because they are small for a significant part of the time service life. Abnormal defects, i.e. defects much larger than those implicit in fatigue design curves, are also of concern. As mentioned by Moan (2004), observations with jackets show that 2-3 % of cracks found in inspections can be attributed to abnormal defects. This also occurs in other offshore structures. Therefore, the crack control strategy, in general needs to include a combination of the following safety measures: design for adequate fatigue life and critical crack size design for robustness in relation to member failure plan inspection of the as-fabricated structure as well as during the service life, possibly using the Leak Before Break principle

An adequate design fatigue life gives ample time to detect cracks. For instance, if the fatigue life from the occurrence of a through thickness cracks to rupture is 25 % of the fatigue life determined by SN-curves, a 20-year characteristic life implies a characteristic value of the time to failure of 5 years and a mean time to failure of 15 years after a possible leak. The implementation of the Accidental Collapse Limit State criteria obviously provides a safety barrier with respect to the system failure given the fatigue failure of a member in a framed structure. When inspections are prioritised, the potential of gross fabrication defects (e.g. because of difficult access), should also be considered. Since inspections after fabrication on shore can be carried at less costs and with higher reliability than during operation offshore, it is worthwhile to emphasise such inspections, at least for critical components. Different strategies may be relevant for different types of offshore structures. This is because the existing structures possess different robustness and because inspection, repair and failure costs vary significantly. As an example of structural components with particular safety focus, consider tethers in TLPs. They are designed with a Fatigue Design Factor of 10 and the ALS criterion is implemented by requiring survival of any tether failed. Moreover, the tethers in the Heidrun platform in the North Sea, are tubular members which were joined by butt welds ground flush and were inspected twice with respect to surface defects on the outside and the inside of the tubular wall. Furthermore a 60-70 % X-ray examination for internal defects after fabrication was carried out in the yard. The first service inspection is now taking place after 10 years in service. The crack control in semi-submersibles with slender braces is based on a balanced fatigue design criterion and ALS according to Table 4 as well as leak detection during operation. A major difference between a trading tanker and an FPSO is that the routing, the speed reduction and the heading angle toward wave can be used for the tanker to re-

34

duce wave loads, but not for the FPSO. Also, a dry-docking for inspections is more complicated for FPSOs because it needs to be stationary at the offshore site. The fatigue criteria for FPSOs were established before they became common practice for tankers (Bach-Gansmo et al., 1987). Normally, the required cumulative damage is D = 1.0 for a 20-year service life for production ships. The thousands of welded joints of similar type and location encountered in the ships imply a high probability of fatigue cracking which suggests application of a more restrictive fatigue criterion. On the other hand, the significant residual strength of damaged ships makes it possible to detect cracks using the leak-before-break detection and a close visual inspection. However, it is desirable to clarify the crack propagation and critical crack length of large cracks. The critical crack length estimated by current methods, (e.g. BS 7910, 1999) is found to be much smaller than crack lengths experienced in ship hulls. The treatment of residual stresses and constraint seems to be important factors in this connection (Bjrheim et al., 2004). However, this issue is still open for debate. However, for economical reasons, it may be advantageous to apply more restrictive fatigue criteria when the consequences are high. This may be the case when the crack causes a leak from the cargo tank into the ballast tank, leading to an explosion hazard. For that matter, the number of potential crack sites in FPSOs and tankers emphasises such a consideration. Novel structures, for which there is no or limited service experiences, need a more rigorous monitoring and inspection, until adequate confidence is gained. This is because new structures involve a high utilisation of static strength, new structural details, possibly with high stress concentration, as well as large uncertainty in the response. 3.10 Quality assurance and control of the design process The quality assurance and control of the engineering process have to address two different situations, which require different type of attention, namely: detect, control and mitigate errors made in connection with technology that is known in the engineering community as such identify possible unknown phenomena, e.g. associated with load, response and resistance, and clarify the basis for accounting for such phenomena in design

Offshore structures are developed in several stages: conceptual, engineering and detailed engineering phases. QA/QC needs to be hierarchical, too, with an emphasis of the latter QA/QC process in the conceptual and early design phases. Errors can occur due to individual errors and omissions, inadequate procedures, software and lack of robustness of the organizations. Errors of omission and commission, violations (circumventions), mistakes, rejection of information, and incorrect transmission of information (communication errors) have been the dominant causes of failures. The lack of adequate training, time, and teamwork or back-up (insufficient redundancy) have been responsible for not detecting and correcting many of these errors (Bea, 2000b). With the advent of computers and their integration into many aspects of the design, construction, and operation of oil and gas structures, software errors are also a concern. Newly developed, advanced, and frequently very complex design technology applied in the development of design procedures and the design of the offshore structures has not been sufficiently debugged and failures have resulted.

35

Software errors, in which incorrect and inaccurate algorithms were coded into computer programs, have been at the root-cause of several recent failures of offshore structures. Guidelines have been developed to address the quality of the computer software for the performance of finite element analyses. An extensive benchmark testing is required to assure that the software performs as it should and that the documentation is sufficient. One particular importance is the provision of independent checking procedures that can be used to validate the results from analyses. It is found that errors are often made by individuals in organizations with a culture that does not promote quality and reliability in the design process. The culture and the organizations do not provide the incentives, values, standards, goals, resources, and controls that are required to achieve adequate quality. The loss of corporate memory in companies responsible for structural safety also has been a factor contributing to many cases of structural failures. Knowledge of the painful lessons in the past was lost and the lessons were repeated with generally even more painful results. Such loss of corporate memory is particularly probable in times of down-sizing, outsourcing and mergers (Bea, 2000a-b). QA/QC of the engineering process QA is the proactive process in which the planning is developed to help preserve desirable quality. QC is the interactive element in which the planning is implemented and carried out. QA/QC measures are focused both on error prevention and error detection and correction (Harris and Chaney, 1969). There can be a real danger in excessively formalized QA/QC processes. If not properly managed, they can lead to generation of paperwork, a waste of scarce resources that can be devoted to QA/QC, and a minimum compliance mentality. It is important that the QA/QC is hierarchical, in other words it should be performed by designers and others doing the work, their colleagues, and third parties. While selfchecking is very important, Matousek and Schneider (1976) found that 87% of the errors causing accidents in the construction industry, could have been detected either by the person next in line or by properly organized additional checks. Therefore, an additional QA/QC is necessary. A good support in organization by experienced managers, who have daily responsibilities for the quality of the project organizations and processes, is crucial. In the same way as the structure should be damage tolerant, the design organizations also need to be robust. It is when the organization or the operating team encounters defects and damage and is under serious stress, that the benefits of robustness become evident. Robust organizations have extensive auditing procedures to help spot safety problems and they have reward systems that encourage risk mitigating behaviours. Nevertheless, knowledgeable, trained, experienced, and sensitive third parties can help, encourage, and assist the owners of the concept to improve. The third-party QA and QC checking measures which are an integral part of the offshore structure design process provide an independent review. This checking should start with the basic tools (guidelines, codes, computer programs) of the structure design process to assure that standardized errors have not been embedded in the design tools. The checking should continue through the major phases of the design process, with a particular attention given to the loading analysis. Moreover the plans for fabrication and operation (manuals) also constitute an important part of the QA and QC process. The provision of adequate resources and motivations is also necessary, particularly the willingness

36

of management and engineering to provide integrity to the process and to be prepared to deal adequately with bad news. The true value of QA and QC lies in the disciplined process. The main objective of QA/QC is detection of errors and omissions and not their prediction. Yet the attempt made by Lotsberg et al. (to appear) is an interesting effort to assess the risk associated with gross errors and omissions. Unfortunately, sometimes the results of formal risk analysis approaches are only used to justify a compliance with regulatory targets and, in some cases the implementation is not clearly justified and needs improvements in the reliability of an engineered system. The intensity and the extent of the design checking process need to be matched to the particular design situation. Repetitive designs that have been adequately tested in operations to demonstrate that they have the required quality do not need to be verified and checked as closely as those that are first-offs and new designs that may push the boundaries of the current technology. New technologies compound the problems of latent system flaws (Reason, 1997). Identification of new phenomena The early phases of design are particularly important for the QA and QC of novel concepts. Novel concepts that could imply new physical phenomena relating for instance to loads and response need particular attention. Examples of phenomena discovered in the recent two decades include the ringing experienced in connection with the Draugen mono-tower shown in Fig. 27. In this case the extreme loading increased by about 30% due to a particular combination of hydrodynamic excitation on large diameter columns and a natural structural period in the range of 2-8s. The phenomenon was discovered when most of the manufacturing process had been completed. So, the safety was ensured by operational restrictions. Another example is the high frequency springing and ringing response of TLPs. It is noted that these phenomena, like many others in the history of offshore structures, became important due to a combination of nonlinear wave load excitation at certain natural structural frequencies. Model testing is crucial in this connection. Another recent, but more obvious, issue is the possible ship impact scenario associated with the tandem off-loading.
Offshore structural engineeringtoday - partlym ature (m anyaspectsof fixedplatformtechnology) - partlyinnovative technologyem erging (e.g. in relatingto floatingplatform s/ risers/offloadingof gas)

16.4m

267m

44.5m

30 20 Moment (kNm) 10 0 -10 -20 -30

QA/QC of novel conceptsrepresent particular challenges - requiresrobust control, i.e. indepent reviews - InnovationdependsuponR&D

Linear analysis Line beregning r Ikkeline beregning r -40 Nonlinear analysis 1860 1865 1870 1875 1880 1885 1890 1895 1900 Tim (s) e

e.g. Ringing in the Draugen platform

Fig. 27: Ringing in Draugen monotower

Fig. 28: Springing and ringing in TLP

37

From a safety point of view, the most important issue is to identify a new phenomenon that represents a new or altered hazard. When it is identified, the uncertainties associated with the hazard are commonly accounted for by conservative design approach. However, there is a strong incentive to reduce the uncertainties to limit the design conservatism for economical reasons. To a large extent the offshore industry has matured. Yet, the development of new offshore facilities are still expected in the future, for instance in connection with new environmental conditions- i.e. involving new combinations of wind waves and swell or ice or use of novel concepts in connection with production and transport of liquefied natural gas. Then sloshing in FPSOs with several slack tanks as well as possible interaction between fluid in tanks and rigid body motions of the vessels, need attention. In general, the aging of the existing offshore structures, especially due to crack growth, needs attention in the years to come.
S afe ty issu es C o m plex and com pact p rocess facility (fire/e xplosion hazards) C o nsequence of LN G leakage P rocess near sto rage facility (m ight im ply e scalation of fire..)

F loa ting production of LN G

C a rgo transfe r in open sea s S loshing of LN G in partly filled tanks (m ight im pair ta nk integrity) O p e ration of vessels close to p roduction/term inal fa cilities (collision hazard)

O ffsho re off-loa ding term ina ls (a w a y from den sly populate d a rea s and busy ports)

Fig. 29: Safety assessment of transport system for Liquefied Natural Gas

4. Concluding remarks Various measures to ensure an adequate safety in offshore structures have been reviewed based upon relevant accidents. A design criteria as well as load and structural analysis methods have been briefly presented. It is demonstrated how structural reliability analysis can be used to establish consistent design criteria for ultimate resistance and fatigue, and especially how refinement of analysis methods and additional information reduce uncertainty and hence the necessary safety factors. On the other hand, it is shown that the failure probability implied by current ultimate and fatigue limit state criteria is small and does not show up in the accident statistics. The main cause of accidents is human and organizational errors and omissions. Therefore, to achieve an acceptable safety level, QA and QC of the engineering process are required. This includes inspection, monitoring and repair of the structure, as well as design for structural robustness. The QA and QC tasks, to possibly identify new phenomena, especially associated with the loading and dynamic response, are particularly challenging in connection with novel concepts for new environmental conditions or new functions. In this paper, a particular emphasis is placed on the Accidental Collapse Limit State design check related to accidental loads and abnormal strength. The

38

philosophy behind this robustness criterion is described and it is shown how information has been established for a proper implementation of the criterion. In the view of the aging of offshore structures, crack growth and rupture are particularly addressed. It is shown how fatigue and robustness design criteria, as well as inspection strategy can be combined for different types of offshore structures, to yield an acceptable safety level. Acknowledgement I would like to acknowledge the invitation to serve a Keppel Professor at the National University of Singapore as well as the cooperation with its Department of Civil Engineering and the Centre for Offshore Research and Enginneering. I would also like to thank the many people I have been working with in carrying out the research as well as code development for the Norwegian Petroleum Directorate, ISO and other regulatory bodies that are reported in this paper. They particularly include J. Amdahl, S. Fjeld, S. Haver, I. Holand (deceased), D. Karsan, J. Lloyd and J.E.Vinnem. The opinions expressed, however, are those of the author.

References
ALK (1981) The Alexander L. Kielland Accident, (in Norwegian English translation available), NOU 1981:11, Oslo. Almar-Nss, A. (ed.) (1985). Fatigue Handbook for Offshore Steel Structures, Tapir Publ., Trondheim. Amdahl, J. (1999) Structural Response to Accidental Loads, Lecture Notes, Dept. of Marine Technology, Norwegian University of Science and Technology, Trondheim. Amdahl, J. (2003) FABIG API (1993/1997) Recommended Practice for Planning, Designing and Constructing Fixed Offshore Platforms, API RP2A-WSD July 1993 with Supplement 1 with Sect., 17.0, Assessment of Existing Platform, February 1997. American Petroleum Institute, Dallas. Ayala-Uraga, E. and Moan, T. (2002) System Reliability issues of Offshore Structures considering Fatigue Failure and Updating based on Inspection, First Int. ASRANet Colloquium, Glasgow, Scottland. Bach-Gansmo, O., Carlsen, C.A. and Moan, T. (1987) Fatigue Assessment of Hull Girder for Ship Type Floating Production Vessels, Proc. Conf. on Mobile Offshore Units, London: City University. Bea, RG (2000a) Achieving step change in Risk Assessment & Management (RAM), Centre for Oil & Gas Engineering, http://www.oil-gas.uwa.edu.au, University of Western Australia, Nedlands, WA. Bea RG (2000b) Performance Shaping Factors in Reliability Analysis of Design of Offshore Structures, J. OMAE, Vol. 122, ASME, New York, NY. Beckwith, I. and Skillman, M. (1975) Assessment of the Stability of Floating Platforms, North East Coast Inst. of Naval Architects. BEFETS (1998) Blast and Fire Engineering for Topside Systems, Phase 2, SCI Publication No. 253, Ascot, UK. Bjrheim, L., Berge, S. and Skaret, H. (2004) Damage Tolerance of FPSOs with Fatigue Cracks, Proc. ISOPE Conference. Broughton, P. (1997) "Engineering Challenges of Ekofisk, the First North Sea Oil Field, Proc. 8th BOSS Conf., Pergamon Press, New York, Vol. I, pp 41-70. BS 7910 (1999) Guidance on Methods for Assessing the Acceptability of Flaws in Fusion Welded Structures, BS 7910:1999, British Standard, London, UK. Chen, H. and Moan, T. (2004) Probabilistic Modeling and Evaluation of Collision between

39

Shuttle Tanker and FPSO in Tandem Offloading, J. Reliability Engineering and System Safety, 84, 2, 169-186. Cornell, C.A. (1995) Structural Reliability- Some contributions to Offshore Technology, Paper No. OTC 7753, Proc. 27th Offshore Technology Conference, Houston. Czujko, J., ed. (2001) Design of Offshore Facilities to Resist Gas Explosion Hazard', Engineering Handbook, Offshore Design, Oslo. Dunlap, W.A., and Ibbs, C.W. (eds) (1994) Proc. Int. Workshop on Assessment and Requalification of Offshore Production Structures, New Orleans, Offshore Technology Research Center, Texas A&M University and University of California, Berkeley, CA. Fjeld, S. (1977) Reliability of Offshore Structures, Paper No. OTC 3027, Proc. 9th Offshore Technology Conf., Houston, Vol. 4, 459-472. Harris, D. and Chaney, F. (1969) Human Factors in Quality Assurance, John Wiley & Sons, London, UK. Haver, S., ( 2000) Evidences of the Existence of Freak Waves, Proc. Rogue Waves 2000, Brest, France, 129-140. HSE (1992) Safety Case Regulations, Health and Safety Executive, HMSO, London. ISO 19900 (1994) Petroleum and Natural Gas Industries Offshore Structures Part 1: General Requirements, (1994), Part 2: Fixed Steel Structures, (2001), Draft,. Int. Standardization Organization, London. ISO 2394 (1998) General Principles on Reliability for Structures, International Standards Organisation, London., Draft for approval. ISSC (1988-1994) Proceedings of the International Ship and Offshore Structures Congress in 1988 (Lyngby, Denmark); 1991 (Wuxi, China); 1994 (St.Johns, Canada). Jordaan, I.J. and Maes, M.A. (1991) Rational for Load Specifications and Load Factors in the New CSA Code for Fixed Offshore Structures. C.J. Civil Engng. 18, 3, 454-464. Lloyd, J.R. and Karsan, D.I. (1988) Development of a Reliability-based Alternative to API RP2A, Paper No. OTC 5882, Proc. 20th . Offshore Technology Conf., Vol. 4, Houston, 593-600. Lotsberg, I. et al. (to appear) Risk Assessment of Loss of Structural Integrity of a Floating Production Platform due to Gross Errors, J. Marine Structures. Madsen, H.O. & S.D. Srensen (1990). Probability-Based Optimization of Fatigue Design Inspection and Maintenance. Presented at Int. Symp. on Offshore Structures, University of Glasgow. Matousek, M. and Schneider, J. (1976) Untersuchungen Zur Struktur des Zicherheitproblems bei Bauwerken, Institut fr Baustatik und Konstruktion der ETH Zrich, Bericht No. 59, ETH, Zrich. Melchers, R.E. (1999) Structural Reliability Analysis and Prediction, John Wiley & Sons, Chichester, UK. Moan, T.(1983) Safety of Offshore Structures, Proc. 4th ICASP Conference, Firenze, Pitagora Editrice. Moan, T. (1985) The Progressive Structural Failure of the Alexander L. Kielland Platform in Case Histories in Offshore Engineering, G. Maier (ed.) Springer-Verlag: Berlin. Moan, T. (1988) The Inherent Safety of Structures Designed According to the NPD Regulations. Report No. F88043 , SINTEF, Trondheim. Moan, T. (1994) Reliability and Risk Analysis for Design and Operations Planning of Offshore Structures. Keynote lecture, Proc. 6th ICOSSAR, Structural Safety and Reliability, Vol. I, Balkema, Rotterdam, 21-43. Moan, T. (1995) Safety Level Across Different Types of Structural Forms and Material Implicit in Codes for Offshore Structures, SINTEF Report STF70 A95210, Trondheim, Prepared for ISO/TC250/SC7. Moan, T. (1998) Target levels for structural reliability and risk analysis of offshore structures, in Risk Reliability in Marine Technology, Guedes Soares, C. (ed), A.A. Balkema publishers, 351368. Moan,T. (2000a) Recent Research and Development Relating to Platform Requalification J. OMAE, ASME, Vol. 122, New York, 20-32.

40

Moan,T. (2000b) Accidental Actions. Background to NORSOK N-003, Dept. of Marine Technology, Norwegian University of Science and Technology, Trondheim. Moan, T. (2004) Reliability-based Management of Inspection, Maintenance and Repair of Offshore Structures, J. Structure and Infrastructure Engineering, Moan, T., Berge, S. and Holthe, K., (1981a) Analysis of the Fatigue Failure of the Alexander L. Kielland, ASME Winter Annual Meeting, Washington D.C. Moan, T. and Holand, I. (1981b) Risk Assessment of Offshore Structures Experiences and Principles. Proc. 3rd ICOSSAR, Elsevier, Amsterdam, 803-820. Moan, T and Amdahl, J. (1989) Catastrophic Failure Modes of Marine Structures in Structural Failure, T.Wierzbicki and N. Jones (eds), John Wiley & Sons, Inc, New York. Moan, T. et al. (1993a) Reliability-Based Fatigue Design Criteria for Offshore Structures Considering the Effect of Inspection and Repair, Paper No. OTC 7189, Proc. 25th Offshore Technology Conference, Vol. 2, Houston, 591-599. Moan, T., Karsan, D. and Wilson, T (1993b) Analytical Risk Assessment and Risk Control of Floating Platforms Subjected to Ship Collisions and Dropped Objects. Paper No. OTC 7123, Proc. 25th Offshore Technology Conf., Vol. 1, Houston, 407-418. Moan, T., Vrdal, O.T. and Johannesen, J.M. (1999) Probabilistic Inspection Planning of Fixed Offshore Structures, Proc.8th ICASP, Applications of Statistics and Probability, A.A. Balkema, Rotterdam, 191-200. Moan, T. and Amdahl, J. (2001) Risk Analysis of FPSOs, with Emphasis on Collision Risk, Report RD 2001-12, American Bureau of Shipping, Houston. Moan, T. Amdahl, J. Wang, X. and Spencer, J. (2002) Risk Analysis of FPSO, with particular emphasis on collisions, Trans SNAME., Vol. 110, 307-339. Moore, W. et al (2003) Report of Specialist Committee V.1 Risk Assessment Proc. 15th ISSC Congress, Vol. 2, Elsevier, Amsterdam. Moses, F. (1987). Load and Resistance Factor Design-Recalibration LFRD Report API PRAC 87-22. API, Dallas, Texas. NORSOK N-001 (2002) Structural Design, Norwegian Technology Standards, Oslo. NORSOK N-003 (1999) Actions and Action Effects, Norwegian Technology Standards, Oslo. NORSOK N-004 (1998b) Steel Structures, Norwegian Technology Standards, Oslo. NORSOK Z-013 (1998a) Risk and Emergency Preparedness Analysis, NORSOK Standard Z-013, NTS, Oslo, Rev. 1. NPD (1977) Regulation for the Design of Fixed Structures on the Norwegian Continental Shelf. Norwegian Petroleum Directorate, Stavanger. NPD (1981) Guidelines for Safety Evaluation of Platform Conceptual Design. Norwegian Petroleum Directorate, Stavanger. NPD (1984) Regulations for load-carrying structures for extraction or exploitation of petroleum, The Norwegian Petroleum Directorate, Stavanger. NPD (1991) Regulations Relating to Implementation and Use of Risk Analysis in the Petroleum Activities, Norwegian Petroleum Directorate, Stavanger. OR (1984) Royal Commission on the Ocean Ranger Marine Disaster: Report One: The Loss of the Semi-submersible Drill Rig Ocean Ranger and its Crew, Edifice Fort William, St. Johns Newfoundland. Paik, J.K. et al (2003) Report of Committee V.2 Collision and Grounding, Proc. 15th ISSC, Vol. 2, Elsevier, Amsterdam. Paik, J.K. and Thayamballi, A.K. (2003) Ultimate Limit State Design of Steel-Plated Structures, John Wiley & Sons, Chichester. P-36 (2001) Workshop on the Accident with the P-36 Platform, Petrobas and Coppe (eds), Rio de Janeiro, Brazil. PA (1990) The Public Inquiry in the Pipe Alpha Disaster, Inquiry Commission HMSO, London. Prevosto, M. and Forristall, G.Z. (2002) Statistics of wave crests from models vs. measurements, Paper No. 28443, Proc. 21st OMAE Conference, Oslo. Reason, J. (1997) Managing the Risks of Organizational Accidents, Ashgate Publishers, Aldershot, UK.

41

Sarpkaya, T. and Isaacson, M. (1981) Mechanics of Wave Forces on Offshore Structures, Van Nostrand Reinholdt, New York. SCI(1993) Interim Guideance Notes for the Design and Protection of Topside Structures against Explosion and Fire, Steel Construction Institute, Document SCI-P-112/503, London. Sigurdsson, G., Lotsberg, I., Myhre, T. and rbeck-Nilssen, K. (2000) Fatigue Reliability of Old Semi-submersibles, OTC Paper no 11950, Proc. Offshore Technology Conference, Houston. Skallerud, B. and Amdahl, J. (2002) Nonlinear Analysis of Offshore Structures, Research Studies Press. Smith, C.S. (1977) Influence of Local Compressive Failure on Ultimate Longitudinal Strength of Ships hull Proc. PRADS, Tokyo, 73-79. Vrdal, O.T. and Moan, T. (1997) "Predicted versus Observed Fatigue Crack Growth. Validation of Probabilistic Fracture Mechanics Analysis of Fatigue in North Sea Jackets", Paper No. 1334, Proc. 16th OMAE Conference, Yokohama, Japan. Vrdal. O.T. , Moan, T. and Bjrheim, L.G. (2000) "Applications of Probabilistic Fracture Mechanics Analysis for Reassessment of Fatigue Life of a Floating Production UnitPhilosophy and Target Levels", Paper No. 00-2078, Proc. 19th OMAE Conference, New Orleans. Vinnem, J.E. (1999) Offshore Risk Assessment, Kluwer Academic Publishers, Doordrecht. Walker, S. et al. (2003) New Guidance on the Design of Offshore Structures to Resist the ExplosionHazard Paper No. 203-37120, Proc. 22nd OMAE Conference, Cancun, Mexico. Wirsching, P.H. (1983) Probability-based Fatigue Design Criteria for Offshore Structures Report API-PRAC Project No. 81-15, Dept. of Aerospace and Mechanical Engng. University of Arizona, Tucson. WOAD: Worldwide Offshore Accident Databank, Det Norske Veritas, Oslo, 1996. Yao, T, et al. (2000) Report of ISSC Committee VI.2. Ultimate Hull Girder Strength, Proc. 14th ISSC Congress, Nagasaki.

Acronyms ALS - Accidental Collapse Limit State API - American Petroleum Institute FDF Fatigue Design Factor (to multiply with the service life to get desired characteristic value of fatigue life) FLS - Fatigue Limit State HOF - Human and Organizational Factors HSE - Health and Safety Executive, UK IMO - International Maritime Organization ISO - International Standardization Organization MMS-Mineral Management Services, USA NPD - Norwegian Petroleum Directorate (now: Petroleum Safety Authority, Norway) TLP - Tension Leg Platform WOAD World Accident Data Bank, issued by Det Norske Veritas

42