Failure modelling and reliability assessment of repairable systems has been receiving a great deal of attention due to its pivotal role in risk and safety management of process industries. Meanwhile, the level of uncertainty that comes with characterizing the parameters of reliability models require a sound parameter estimator tool. For the purpose of comparison and cross-verification, this paper aims at identifying the most efficient and minimal variance parameter estimator. Hierarchical Bayesian modelling (HBM) and Maximum Likelihood Estimation (MLE) approaches are applied to investigate the effect of utilizing observed data on inter-arrival failure time modelling. A case study of Natural Gas Regulating and Metering Stations in Italy has been considered to illustrate the application of proposed framework. The results highlight that relaxing the renewal process assumption and taking the time dependency of the observed data into account will result in more precise failure models. The outcomes of this study can help asset managers to find the optimum approach to reliability assessment of repairable systems.

@en

Processing facilities handling large amounts of hazardous substances are attractive targets for terrorists. Thus, these work sites are exposed not only to accidents but also to intentional threats. Some research has separately studied risk caused by either potential accidental events or terrorist acts. However, studies focusing on integrated risk assessment and management (dealing with both safety and security issues) are lacking. This paper proposes an approach to assess and manage integrated risks. This method is based on an influence diagram which incorporates safety and security-related factors into one framework. It considers the effects of management actions on both accidental and intentional risks. This method can help to detect hidden risk (i.e., the risk not recognized during design and operation stages) and ensure to reduce the real risk to an acceptable level by guiding the selection of management actions. The effectiveness of the proposed method is demonstrated using the overfilling risk management of an oil tank.@en

Conventional risk assessment of chemical plants considers process accident related causal factors. In the current geopolitical situation, chemical plants have become the target of terrorism attacks, making security concerns as important as safety. To protect the public and environment from undue risks, security related causal factors need to be considered as part of the risk analysis of chemical plants. This paper presents an integrated approach to dynamically assess the occurrence probability of abnormal events. The abnormal event is a state of a process plant arrived either due to a process accident or an intentional (terrorist) threat. This approach considers both safety and security related risk factors in a unified framework. A Bayesian network is used to model specific evolution scenarios of process accidents directly initiated from security related factors and the interaction of causal factors. This model enables to dynamically analyze the occurrence probabilities of abnormal events and causal factors given evidence; it could also capture the impacts of interaction among safety and security related causal factors on these occurrence probabilities. The proposed approach is applied to an oil storage tank to demonstrate its applicability and effectiveness. It is observed that the effect of dependency between correlative accidental and security related factors significantly change the occurrence probability of abnormal events in dynamical assessment.@en

The process industry is confronted with terrorism threats. Effective security management demands the ability to defend facilities against different intrusion scenarios. This study first presented various intrusion scenarios to explain the corresponding intrusion process using graphical barriers. Subsequently, this work dynamically analyzed the successful intrusion probabilities and security potentials of barriers using a Bayesian network considering the dependency of barriers and interaction of different intrusion scenarios. It was observed that successful intrusion probabilities and security potentials are strong functions of intrusion scenarios. Therefore, extensive intrusion scenarios must be considered while assessing and designing the security systems of process facilities.

@en

The reliable prediction and diagnosis of abnormal events provide much needed guidance for risk management. The traditional Bayesian network (traditional BN) has been used to dynamically predict and diagnose abnormal events. However, its inherent limitation caused by discrete categorization of random variables degrades the assessment reliability. This paper applied a continuous Bayesian network (CBN)-based model to reduce the above-mentioned limitation. To compute complex posterior distributions of CBN, the Markov chain Monte Carlo method (MCMC) was used. A case study was conducted to demonstrate the application of CBN, based on which a comparative analysis of the traditional BN and CBN was presented. This work highlights that the use of CBN can overcome the drawbacks of traditional BN to make dynamic prediction and diagnosis analysis more reliable.

@en