Resilience Analysis of a Remote Offshore Oil and Gas Facility for a Potential Hydrocarbon Release

Quantitative assessment of machine learning reliability and resilience

Advances in machine learning (ML) have led to applications in safety-critical domains, including security, defense, and healthcare. These ML models are confronted with dynamically changing and actively hostile conditions characteristic of real-world applications, requiring systems incorporating ML to be reliable and resilient. Many studies propose techniques to improve the robustness of ML algorithms. However, fewer consider quantitative techniques to assess changes in the reliability and resilience of these systems over time. To address this gap, this study demonstrates how to collect relevant data during the training and testing of ML suitable for the application of software reliability, with and without covariates, and resilience models and the subsequent interpretation of these analyses. The proposed approach promotes quantitative risk assessment of ML technologies, providing the ability to track and predict degradation and improvement in the ML model performance and assisting ML and system engineers with an objective approach to compare the relative effectiveness of alternative training and testing methods. The approach is illustrated in the context of an image recognition model, which is subjected to two generative adversarial attacks and then iteratively retrained to improve the system's performance. Our results indicate that software reliability models incorporating covariates characterized the misclassification discovery process more accurately than models without covariates. Moreover, the resilience model based on multiple linear regression incorporating interactions between covariates tracks and predicts degradation and recovery of performance best. Thus, software reliability and resilience models offer rigorous quantitative assurance methods for ML-enabled systems and processes.

The Resilience of Critical Infrastructure Systems: A Systematic Literature Review

Risk management is a fundamental approach to improving critical infrastructure systems’ safety against disruptive events. This approach focuses on designing robust critical infrastructure systems (CISs) that could resist disruptive events by minimizing the possible events’ probability and consequences using preventive and protective programs. However, recent disasters like COVID-19 have shown that most CISs cannot stand against all potential disruptions. Recently there is a transition from robust design to resilience design of CISs, increasing the focus on preparedness, response, and recovery. Resilient CISs withstand most of the internal and external shocks, and if they fail, they can bounce back to the operational phase as soon as possible using minimum resources. Moreover, in resilient CISs, early warning enables managers to get timely information about the proximity and development of distributions. An understanding of the concept of resilience, its influential factors, and available evaluation and analyzing tools are required to have effective resilience management. Moreover, it is important to highlight the current gaps. Technological resilience is a new concept associated with some ambiguity around its definition, its terms, and its applications. Hence, using the concept of resilience without understanding these variations may lead to ineffective pre- and post-disruption planning. A well-established systematic literature review can provide a deep understanding regarding the concept of resilience, its limitation, and applications. The aim of this paper is to conduct a systematic literature review to study the current research around technological CISs’ resilience. In the review, 192 primary studies published between 2003 and 2020 are reviewed. Based on the results, the concept of resilience has gradually found its place among researchers since 2003, and the number of related studies has grown significantly. It emerges from the review that a CIS can be considered as resilient if it has (i) the ability to imagine what to expect, (ii) the ability to protect and resist a disruption, (iii) the ability to absorb the adverse effects of disruption, (iv) the ability to adapt to new conditions and changes caused by disruption, and (v) the ability to recover the CIS’s normal performance level after a disruption. It was shown that robustness is the most frequent resilience contributing factor among the reviewed primary studies. Resilience analysis approaches can be classified into four main groups: empirical, simulation, index-based, and qualitative approaches. Simulation approaches, as dominant models, mostly study real case studies, while empirical methods, specifically those that are deterministic, are built based on many assumptions that are difficult to justify in many cases.

Understanding the Ecological Validity of Relying Practice as a Basis for Risk Identification

Understanding the reliability of hazardous organizations and their protective systems is central to understanding the risk they produce. Work on "high reliability organization" has done much to illuminate the conditions in which social organization becomes reliable in highly demanding conditions. But risk depends just as much on how relying entities do their relying as it does on the reliability of the entities they rely on. Patterns of relying are often opaque in sociotechnical systems, and processes of relying and being relied on are mutually influencing in complex ways, so the relationship between relying and risk may not be at all obvious. This study was an attempt to study relying as a social practice, in particular analyzing how it had ecological validity in a social organization-how practice was responsive to the conditions in which it took place. This involved observational fieldwork and inductive, qualitative analysis on an offshore oil and gas production platform that was nearing the end of its design life and undergoing refurbishment. The analysis produced four main categories of ecological validity: responsiveness to formal organization, responsiveness to situational contingency, responsiveness to information asymmetry, and responsiveness to sociomateriality. This ecological validity of relying practice should be a primary focus of risk identification, assessing how relying can become mismatched to reliability in certain ways, both when relying practice is responsive to circumstances and when it is not.

Reforming Health, Safety, and Environmental Regulation for Offshore Operations in China: Risk and Resilience Approaches?

Offshore drilling accidents have triggered regulatory reforms in China. The reforms aim to explore proper regulatory approaches to supervise offshore operations and improve their health, safety and environmental (HSE) performance. This study offers a review on the roles of risk and resilience in managing offshore operations and a well-defined analysis on their integrations with Chinese laws and regulations. The study finds risk and resilience approaches can promote the effectiveness of HSE regulation for offshore operations, while both are difficult to be transposed into legally binding rules in China. To fully develop and implement risk regulation for offshore operations, the study suggests to decentralize China’s command-and-control regulatory regime and encourage self-regulation in offshore petroleum companies. Transposing resilience thinking into legal practice is also highlighted so that various regulatory powers can keep proactive and flexible to any possible changes and uncertainties.