Process Safety in the oil and gas industry is managed through a robust process safety management system that involves the assessment of the risks associated with a facility, throughout its lifecycle. Typical approaches for risk assessment of an accident scenario includes: the identification of hazards, the estimation of the frequency, the analysis of possible consequences, and the evaluation of the risk against a company's tolerance criteria. While current quantitative risk assessment methods (e.g.: Layer of protection analysis (LOPA), Bow Tie analysis (BT), etc.) have brought significant improvements in the management of major hazards, they tend to provide static values of risk at a given time (snapshot at the time of the assessment). In reality however, risk levels tend to fluctuate throughout the lifecycle of a process, due to several time-variant risk factors, which include: equipment aging (which impacts their reliability), variations in the integrity and vulnerability of safety barriers, plant activities (e.g.: maintenance, shut-down), health and efficiency of the process safety management system, process safety incidents or near misses, etc. Thus, to that effect, risk is dynamic in nature and risk assessment approaches should allow for continuous updating of risk values over time. Over the last decade, the petrochemical industry has put significant efforts in developing process safety indicators (PSIs) to continuously measure the health and efficiency of process safety management systems. This has increased the sources of information that are used to assess risks in real-time. Hence, there is an opportunity to leverage PSIs along with equipment/safety barrier performance data to estimate the quantitative measure of risk levels in a process facility on a time-variant basis. ExxonMobil Research Qatar (EMRQ) partnered with the Mary Kay O'Connor Process Safety Center – Qatar (MKOPSC-Q) to attempt development of a tool that uses Bayesian Belief Networks (BN) to capture any potential increase of risk levels in real-time as a result of pre-identified risk factors and reliability data of equipment and safety barriers. The tool is referred to as PULSE, which stands for Process Unit Life Safety Evaluation, and is intended to be used to support existing decision making practices. The work involves a phased approach that first included the development of a methodology to establish the framework for the tool. Then, implementation and testing of the framework was attempted using BN algorithms. The most recent phase involves application of the tool to maintenance and inspection planning. In this context, the work presented here demonstrates the feasibility of using PULSE to quantify changes in risk level for a process unit based on a case study from literature. Furthermore, the different aspects of PULSE development are described. These aspects include: translation of the Bow Tie into a BN, modification of the BN to include reliability data, and insertion of equipment failure incidents into the BN to perform dynamic modeling. The outcomes of the dynamic modeling with real time insertion of equipment failure evidence are discussed. Also, the application of dynamic modeling to support risk-based decision making with regards to inspection and maintenance planning is included.
