Comparative Study of Structural Reliability Assessment Methods for Fixed Offshore Structures

The oil and gas sector has recognised structural integrity assessment of ageing platform for prospective life extension as a rising concern, particularly in encountering the randomness of the harsh ocean environments. This condition leads to uncertainty in wave-in-deck load estimates and a high load level being imposed on offshore structures. This emphasises the necessity of enhanced reliability, as failure might result in inaccessibility because of the uncertainties related to long-distance services, such as accuracy of predictions of loads and responses. Even though the established guidelines present a fundamental assessment, additionally, comprehensive rules are required. This paper performed a reliability analysis incorporating practical approaches that can more accurately represent time-dependent structural deterioration. The following two procedures have been adopted by a majority of significant oil and gas operators to monitor the safety and integrity of these structures: a) Ultimate Strength Assessment (USA) method and b) Reliability Design Assessment (ReDA) method. A comparison of these two reliability approaches was performed on selected ageing jacket structures in the region of the Malaysian sea. The comparative findings, namely, reserve strength ratio (RSR) at various years of the return period (RP) and ratio value for risk of failure regarding the probability of failure (POF), provided a check and balance in strengthening confidence in the results. The findings showed that the structural components might safely survive either using the USA and ReDA method in such conditions, as the reliability indexes were determined to be satisfactory compared to allowable values from ISO 19902 design specifications. Therefore, these evaluations were determined to control the risk level of the structure during the remaining of its lifetime and undertake cost-effective inspections or mitigation strategies when necessary.

determining the probability of failure (POF) over a certain reference period is important to ensure precision in estimating its structural integrity and remaining service life [14].
Some studies have been undertaken on the reliability of Malaysian jacket platforms [15,16] and other types of platforms from around the world in order to demonstrate the structure's capability and decide the valid mitigation measures [17]. Since the late 1990s, the Malaysian oil and gas industry has adopted a reliability strategy. Structure failure is foreseen when the strength capacity cannot withstand the maximum force (imposed loads) [18], which would need the evaluation of the possibility of extending the life of ageing platforms. Therefore, the implications of a failure might include interrupting production after the platform's prior life span has expired, substantial undersea modifications, and decommissioning [19,20]. In view of the continual production necessary after the first design life of these installations, the extension of life is unavoidable.
This article aimed to review the methods for the derivation of the POF on the structural reliability of an aged jacket platform in Malaysian waters. The two common methods used on an ageing platform were evaluated utilising Ultimate Strength Assessment (USA) and Reliability Design Assessment (ReDA) procedures to assess life extension possibilities. As studied by Mat Soom, et al. [21], both procedures determine an appropriate offshore platform as the primary basis of analysis. These methods were then applied to a case study of an ageing jacket structure regarding the selected region of existing platforms. The USA and ReDA analysis results are necessary to provide a high level of confidence in structural strength for a longer design life alongside additional years of production. Overall, Section 2 represents the risk analysis implementation and the fundamental review of SRA, Section 3 delivers a valuable comparison of USA and ReDA techniques, Section 4 provides a comprehensive explanation of the test structure criteria, Section 5 compares the findings of the USA and ReDA, and Section 6 describes the study's findings.

Overview on the Offshore Structural Assessment
A range of methods is utilised in the risk, reliability and safety disciplines. These disciplines are defined separately in another division; for example, Quantitative Risk Analysis, Probabilistic Risk Analysis, Probabilistic Safety Assessment, and Formal Safety Assessment are all fundamentally similar approaches [22]. The essential regulations in the offshore sector are commonly referred to as Quantitative Risk Analysis (QRA) and Structural Reliability Analysis (SRA).

Implementation of Risk Analysis
Risk analysis is a method of stimulating the inventive implementation of diverse approaches to perform systematic analysis methods and manage the uncertainty concerns associated with risk data [23]. The risk identification, assessment, monitoring, recommendation, and proper implementation are all part of this process. The different risk assessment procedures based on existing experiences have yielded more substantial advances and will provide more significant long-term advantages to the sector [24,25]. As a result, risk management must be perceived as essential to the effective completion of projects, even though it is frequently restricted by a lack of managerial practices and software tools [26]. A holistic comprehension of the many possible causes (risk) requires a clear definition of project performance objectives as significant considerations for successful project management and decision-making [27][28][29].
In comparison to other engineering fields, risk analysis is a relatively recent phenomenon. New disciplines require a considerable time for terminology to be accepted amongst professionals and even longer for widespread applicability [30]. QRA is determined via the entire risk to civilised health and safety [31], the ecosystem, and the property denoted by the installations. The followings are the significant elements in the analysis: • Identification of hazards • Assessment of probabilities /frequencies of initiation events • Development of accident (how a starting event might become several unintentional occurrences) • Consequence assessment (estimate of impacts of multiple accidents)

• Calculation of risks
As shown in Figure 1, there are two forms of risk assessment (QRA): qualitative risk assessment and quantitative risk assessment. In accordance with Bai [32],Bai and Jin [33], qualitative is a subjective concept based on a set of fundamental situation rankings. The qualitative risk assessment is based on risk-based inspection (RBI), and a risk assessment matrix is employed to rate the hazards [34]. On the other hand, the quantitative evaluation stresses the point confidence interval and the probability of distribution based on the risk-based assessment (RBA) [35]. RBA also relies on design, analysis, and the most recently subsea and topsides risk assessments. Regarding the RBA discipline, the risk assessment-based sequence of events can be accomplished in the Bow-Tie process [36]. It is part of the Health, Safety, Security, and Environment (HSSE) tools for As Low As Reasonably Practical (ALARP), usually utilised by oil and gas industries to assess and manage the risk. Bow-Tie is also a valuable tool for communicating about hazards and how to manage them properly [37]. Therefore, RBA is one of the main aspects of controlling barriers for avoiding the platform's collapse (Top event). The steps of Bow-Tie process are shown in Figure 2. In this study, four ageing platforms were considered: one with field subsidence and one platform without subsidence. According to Cameron and Raman [38],Khan, et al. [39], risk assessment has been deployed in various contexts to suggest that key elements of uncertainties, probabilities, or frequencies and implications have been addressed in some way. The assessment system in conventional scenarios is frequently imprecise due to various scenarios requiring history and historical data [40]. For that reason, the existing historical data is frequently precisely relevant to the subject analysed by SRA, which delivers a quantitative decision-making technique for evaluating the integrity of existing offshore platforms under Risk-Based Study (RBA) [41]. Moreover, evaluating the influence of more data, risk-reducing solutions, and modifications to the evaluated scenario is a significant aspect of a risk assessment [42].
SRA is regarded with assessing a structure's reliability or probability of failure [43,44]. Probabilistic models are used in the analysis [45]. SRA is also the planning and controlling behind relevant design calibration. It can be used in conjunction with legislature probabilistic-based mathematical models to recalculate design codes [46]. Thus, other legislature models were utilised to obtain the desired reliability level in conjunction with the proper code formulation.

Structural Reliability Assessment (SRA)
The load on offshore infrastructure arises primarily from two dimensions. According to Kharade and Kapadiya [47], the first is the self-weight of the jacket and deck structures, which is caused by gravity from semi-permanent and permanent equipment, storage commodities, human settlements, and operating loads. The majority of such loads do not change significantly over time and can be reliably predicted [48]. The second dimension of loading is environmental. It consists primarily of wind forces on the machinery, deck structures, opened members of the jacket, the impact of ocean waves and currents on the jacket structures, and, in rare situations, earthquake loading. These pressures change significantly over time, making it impossible to estimate the range of extreme climatic circumstances to which the structure will be subjected over its lifetime [49].
Predicting the mean wind velocity, wave height, wave period, and current velocity corresponding to the most severe storm is the initial stage in the in-place analysis and design of the intact structure [50,51]. However, a full description of the variance in sea state conditions over the structure's whole life is essential for fatigue damage evaluation [52]. This data was gathered through comprehensive oceanographic observations across several years prior to structural design and the application of statistical forecasting tools [53]. As mentioned in Section 2, this paper focused on fixed offshore structures located in Sarawak and Sabah regions. The five platforms were chosen based on the platform's global impact relating to substantial environmental loadings. The test substructure was defined as those employing a tripod, either a four-legged or an eight-legged structure with varying water depths reaching from 25 m to 130 m below mean sea level (MSL). Platforms range in age from 15 to 37 years old.
An offshore platform's primary design specification must fulfil the functionality demand for structural support of offshore oil and gas services, particularly both operational and extreme loadings [54]. At the design stage, a variety of loads must be considered, including permanent and operating loads, vibration, ship collisions, wind, wave, tide, current, fatigue, foundation responses, and seismic effects, amongst others [55]. Thus, substantial uncertainties of the environmental and material loads must be identified in order to appraise an existing offshore platform [56]. After a set period, material uncertainty may vary due to deterioration, particularly from a natural environment and corresponding fatigue [57]. The existing platforms are particular challenges, including damaged components, extra load from additional deck amenities, and extended design life from the initial design life [58,59]. Therefore, the strength of the structural system must be re-examined to observe whether operability and extensive safety regulations were complied with.
In the current rules of structural design, the code specifies a set of "nominal" loads, and material properties are based on the working stress for each form of component failure connected with a "factor safety" on the member strength [60]. As a result, load and resistance are considered singlevalued (and deterministic) numbers with no concern for variability. The defined safety factors are often obtained from experience and judgment rather than from any quantitative appraisal of the uncertainties associated with failure. Because numerous loads and material attributes are unpredictable (random) in nature, probabilistic approaches must be used in structural design [56]. It is now commonly accepted that sufficient safety is impossible to achieve in the context of uncertainty and that some hazard of inadequate structural performance must be permitted [61]. The primary goal of structural design is to assure, with a reasonable level of confidence, that a structure will not match improperly for its intended use at any time through its designated design life. Modern approaches of reliability-based design assist in achieving this goal.
Most measuring instruments are based on the nonlinear structural failure and structural reliability [62]. SRA is utilised to evaluate implications of uncertainties in load actions, resistances, and modelling of particular components of a structure. It can be carried out on individual structural components (local), as well as the entire structural system (global) [63]. Therefore, it might also benefit (re)calibrating partial action and resistance variables in uncommon or exceptional situations. Furthermore, SRA may be valuable for generating comparison data through the earliest design phase [64,65]. As a result, SRA provides a depth of confidence in the ability of a combination of components to perform as assigned, taking into account various uncertainties that arise with loads and resistances [66]. The 'probability of failure' or 'reliability', which is still widely acknowledged as a more realistic assessment of structural integrity, was used in these procedures to quantify the structure's safety [67].
The American Petroleum Institute API [68] has dedicated substantial work in recent years establishing a new "LRFD" design code in acknowledgement of the requirement for satisfactory estimation of uncertainties associated with the offshore structural design. This standard was recently made available for industry review. The load and resistance related to partial safety factors for system designs were generated from a reliability analysis assessing the numerous sources of uncertainties, and this code employs a "limit state design" principle [69]. In order to achieve reliability-based structural design, there are two types of suggested design techniques that may be used [13,70]. Both API-WSD working stress design practice and the API-LRFD load resistance factor design practice are granted by the API standard. A 'reliability-based format' is practised in the API-LRFD design approach [71,72]. Therefore, the new LRFD codes simplify regular design and convey a higher consistency in reliability between many members and under various load conditions [73].
Over a couple of decades, many studies have been conducted to understand ageing and its related mechanism better. Several initiatives have been exerted towards developing evaluation standards and a framework for extending the life of ageing structures [74]. The integration of evaluation techniques and frameworks for aged structures began in the mid-1990s with a section added to API RP 2A [75]. However, the input and approval criteria provided are related to US waters, the North Sea region, the Norwegian continental shelf, etc. Although several released standards were added to ISO 2394, ISO 13822, ISO 19902, NORSOK standard N-006 and DNV guideline [76][77][78][79][80], previous researches have minimum quantitative information associated with local regions [81]. Because various models, tools and equipment (software), and designer analysts are practised, there are typically differences in practises and hypotheses, which leads to the various mathematical models and statistical uncertainties being incorporated in the evaluation [82]. There is a legitimate requirement to better measure modelling uncertainty and acknowledge alternate approaches to integrate modelling uncertainty into reliability analysis.
Regarding the inconsistent assessments, there is a well-defined need to progress towards a set of principles, specifically in conjunction with structural integrity [83], in order to adopt a more reasonable approach. To address the issues mentioned above, this research reviewed two different methods for structural integrity evaluation to extend the life of aged offshore structures. The proposed framework included the theories and principles to anticipate remaining fatigue performance and assess structural capability in the Ultimate limit state (ULS), Serviceability limit state (SLS) and Accidental Limit State (ALS) during the whole prolonged service life (Nizamani, 2015). In this study, the recent analysis performed the quantitative risk-based assessment (RBA), which was used to calculate the structural reliability analysis (SRA) using the Ultimate Strength Assessment (USA) and Reliability Design Assessment (ReDA). The subsequent section explained the approaches' framework in further detail.

Ultimate Strength Assessment (USA) and Reliability Design & Assessment (ReDA) Procedures
The USA is an integrated approach designed to facilitate the re-evaluation of operations. The results of these analyses can support the interpretation of the failures of the structure and adequately identify the suitable preventive approach. In this article, two elements, Nonlinear Plastic Collapse (NPC) and Structural Reliability Evaluation (SRA) for global analysis, would be computed for global assessment. Since ReDA is an evaluation method for analysing structural reliability, guidelines for development and reassessment were generated from the necessary platform strength for the incident and associated with the probability of failure. Standards were established by decreasing risk to the lowest level (the ALARP system). Table 1 compared the structural reliability evaluation between USA and ReDA methods in terms of these two approaches. The USA utilises a simplified technique based on the computations of Lognormal Distribution, whereas ReDA utilises the integral solution. Regarding USA practice, it is restricted to Malaysian waters because of the local parameter of bias and the local Coefficient of Variance (COV) used. Shell's running industry across the world has effectively adopted the ReDA practice. The ReDA utilises up to three RPs of 100, 1000, and 10000 years, whereas USA utilises only one RP of 100 years. In order to get a high level of confidence, it is advisable to compute with a long-term distribution that the platform has a minimum of two environmental loads for 100 years as compared with its earlier design of RP. ReDA was nonetheless computed for the average probability of failure (POF). USA and ReDA methods were shown in the sequence diagram regarding the SRA development (refer to Figure 3). Both USA and ReDA procedures have a safety measure of standardised RSR and targeted reliability of POF. Then, nonlinear pushover analysis determined the RSR, base shear (BS), and ultimate loads. In conclusion, a computed POF associated with targeted reliability was compared with the USA (simplified approach), ReDA (convolution technique), and ISO 19902 regulation was achieved. Based on the Hs (significant wave height) hindcast data from the 1940s to the 2000s, the patterned data of bias and COV was generated on different locations, whereby Sarawak Operation and Sabah Operation regions related to USA procedure. The provided bias and COV values were initially obtained using the Weibull Distribution Graph Analysis. The computed POF was based on a return period of 100 years. In contrast, ReDA analysis contained three different returns periods of 100, 1000 and 1000 years as a value of POF. This approach is based on Type I and Type II uncertainty used to calculate the structure's POF over the remaining life of the structure. However, the calculated POF was based on the average return period. More information on the methods for USA and ReDA is available in Mat Soom, et al. [84].

Test Structure Specification
This analysis focused on the regions of Sarawak Operation (SKO) and Sabah Operation (SBO). For case studies, several sites and kinds of offshore platforms were considered. Deck level of test platforms were based on levelling survey installation(s) undertaken in previous years. These platforms consisted of subsidence platforms except P37D-4. Table 2 summarises the description of the platform specifications. In order to obtain a realistic result, the long-term analysis was analysed to predict the maximum load distribution in terms of RPs, i.e. 100, 1000 and 10000 years. According to PETRONAS Research and Scientific Services [85], the results were generated and given in this case study using Met-Ocean data provided by the Department of Petronas Carigali Sdn Bhd. The Met-ocean data have been produced from hindcast data and were based on the hydrodynamics of deep water. A minimum of eight directions of wave loadings were necessary for 4-legged and 8-legged platform types, however, for 3-legged platform types, a minimum of twelve directions, with an equal distance of 30° between them.

Comparison of the outputs from USA and ReDA procedures
In order to achieve a high confidence level, five platforms were analysed for reserve strength ratio (RSR), and POF values with a comparison between both methods, which were USA and ReDA. The pushover analysis can be performed to determine whether the ultimate platform resistance exceeds the platform's capacity, RSR and base shear (BS) values. An approximate reliability measure of the platform in terms of POF can be established by determining the return period of the environmental load that the structure can withstand with the (lowest) calculated RSR. In this section, the results from the two methods have been tabulated in detail.

Reserve Strength Ratio (RSR) and Base Shear (BS)
Reserve strength ratio can be determined by the first 'significant' peak or the highest peak of the load-deformation curve as the 'collapse' RSR [63] from the result of software of Ultimate Strength for Offshore Structure (USFOS) [86,87]. For this study, for each selected platform, the worst direction (in degrees) of more considerable base shear based on the minimum or maximum water depth calculated was identified and selected before.
The results of RSR alongside base shear from USA and ReDA methods were recorded regarding various return periods in Table 3 with the worst direction for each platform. The RSR for every return period has been evaluated together with BS values. RSR is a load factor for a number of waves attacking the platform until collapse. The different result between these three return periods was where the lowest RSR at 10,000-years return period, base shear was the highest, and vice-versa when the RSR was highest. RSR and BS results were computed by software push over USFOS at selected return periods. The RSR results for 100 years at minimum, 1000 years at medium, and 10000 years return period at abnormal/extreme conditions are presented in Figure 4. It shows that all RSR values (for all test platforms) associated 100 years were above or exceeded both acceptable safety criteria limit for USA and ReDA. In comparison, for 1000 years return period only two platforms were above and exceeded both acceptance safety criteria, which were P88P-8 and P37D-4 Platforms. Those for 10000 years return period showed only that the P37D-4 platform exceeded both acceptance safety criteria.

Probability of Failure (POF)
In this section, the POF was discussed, where the probability of structural failure was then evaluated by examining a limited number of significant sequences of member failures that produce the collapse of the structures. The structure will eventually survive, given one or more of its members (Mat Soom et al., 2015). The evaluation concerning the probability of failure (POF) was only examined for beyond than 100 years return period (RP), where the potential for the platform to collapse exceeded at this return period (RP). As shown in Figure 5, the probability of failure (POF) of target reliability (TR) limit for USA and ReDA was highlighted following the line in red and purple. In contrast, the calculated probability of failure (POF) for USA and the probability of failure (POF) for ReDA are highlighted as a bar chart. In the line of structural study, only one platform (P37D-4) had significantly less value of probability of failure (POF) using simplified method (USA) as well as convolution method (ReDA) in comparison with the probability of failure (POF) and its corresponding target reliability (TR). This platform had a ratio of 0.000233 (USA) and 0.000000 (ReDA) times risk of failure compared to target reliability for USA and ReDA, as showed by the tabulated results in Table 4. Due to this reason, this platform was very unlikely to fail beyond 100 years return period (RP).  Table 4. Thus, both were still considered unlikely to fail beyond 100 years above the return period (RP). While two platforms, P88Q-4 and P88P-8, had a very high significant ratio value for risk of failure either USA or ReDA target reliability and potentially very likely to fail platforms at beyond 100 years return period (RP).  POF exposure level can be determined by referring to ISO 19902 Clause 6.6 (ISO 2007). All test platforms (P88Q-4, P88V-3, P132D-4 and P37D-4) were under the L2 exposure level, which was S2 or S3 life safety category and C2 medium consequences. However, P88P-8 was under L1 exposure level, which was the S1 life safety category and C1 medium consequences. Figure 6 showed that the probability of failure (POF) was in line with the limit of ISO 19902 constant values at 5.00E-4 @ POF exposure level L2 for all platforms, respectively. The subsequent target reliability (TR) was shown in Table 5. Only one platform significantly reduced POF for both USA and ReDA than ISO 19902 target reliability limit. Thus, this platform P37D-4 had a ratio of 0.0 times risk of failure for both calculated probability of failure (POF). Thus, this platform was Extremely Reliable and Clearly Acceptable from Global Assessment and Analysis, referring to ISO 19902 beyond 100 years return period (RP). Two platforms, P88V-3 and P132D-4, passed either probability of failure (POF) target reliability for USA and ReDA. However, P88Q-4 did not meet the target reliability for ISO 19902. In terms of ratio analysis, three of them had marginal ratio value times risk of failure comparison against ISO 19902. Thus, three of them were still considered Reliable and Clearly Acceptable from Global Assessment and Analysis beyond 100 years return period (RP). Finally, P88P-8's USA and ReDA based calculations had very high significant ratio value for risk of failure, and it was Extremely Not Reliable and Not Clearly Acceptable from Global Assessment and Analysis at beyond 100 years return period (RP).
The smaller number of jacket's legs contributed to less base shear and gave minor impact to the ratio between USA and ReDA in comparison against ISO 19902. Only one platform (P37D-4) without subsidence issue significantly impacted the result where no ratio value for risk of failure. However, water depth did not influence the probability of failure (POF and return period (RP).

Conclusion
The conclusions of the conducted research are as follows: • USA and ReDA methods have been accepted by most of the marine operators in the offshore industry to manage their structures' safety, integrity, and reliability. • It was noted that both methods were based on the design code for fixed offshore structures by utilising the limit state equation of probabilistic models. • The USA method was limited to Malaysian waters due to its use of the local parameters of bias and coefficient of variance (COV). The calculation was based on the simplified method as normal distribution and considers only a 100-year return period. • The ReDA method has been successfully implemented in the North Sea and worldwide facilities using the integral equation method. ReDA was applied for three different return periods (100, 1000 and 10,000 years) as long-term distribution response. • All tested platforms were designed based on 100 years return period during the initial design.
Based on the probability of failure result above, comparison at sections 5.2 and 5.3, all the tested platforms were reliable and acceptable for 100 years return period and below only. • Consequently, it was better to calculate with long-term distribution for the platform to experience at least two environmental loading at different return periods to get a high confidence level. However, the result of 100 years return period will be compared with the initial return period, which is of paramount importance. • According to the result obtained based on different offshore structures themselves, both procedures were proven to increase reliability. Thus, the safety of these assets was sustained • The precision and efficiency of these methods will benefit the industry, particularly operators, to make decisions and, more precisely, to describe action plans as part of managing their business risks. The findings of these analyses may assist comprehension of the structure failure mechanism accurately and effectively describe the main kind of mitigation.