Improving risk and safety decision-making in the high-risk energy sector industries through cross-industry learning opportunities

Abstract

The overall objective of this thesis is to contribute new knowledge to the applied area of decision-making under uncertainty. More specifically, this research relates to improving risk and safety decision-making in the high-risk energy sector industries, by exploring cross-industry learning opportunities. The prevalence of common risk and safety issues faced by the energy sector industries presents opportunities for cross-industry knowledge transfer. Cross-industry learning requires fewer resources, to learn by experience. The commonality of accident causes, and high-level lessons make it a practical way to proceed towards achieving more effective safety management at the industrial level. In fact, these industries have adopted methods, principles, and tools from each other in the past. There is a trend towards developing more general holistic concepts for capturing the needs of assessing and managing decision problems in their industrial context. While the traditional safety and risk analysis tools and principles are still relevant for these industries, major learning opportunities that can prove useful for decision-support should not be left unexplored. Observing and understanding the decision-making processes followed by industries in the energy sector (oil & gas, nuclear and chemical processing industries) reveals commonalities. All of them broadly involve decision problem identification and alternative description, decision-analysis, decision-makers’ review and making the decision. A key feature of this process is the role of the stakeholder’s inputs, i.e., his goals, criteria and preferences. Since they heavily influence all elements in the decision-making process, they need to be actively accounted for when evaluating the usefulness of an improvement opportunity. Based on the evident commonality in risks and decision-making processes, several sources of learning opportunities for improving the decision-making process emerge, some of which have already been adopted. However, identifying other potential improvement opportunities, assessing them and finding a suitable criterion to evaluate them is not so straightforward. Currently, there is a gap in this area that this thesis strives to fill. During the research, several sources of learning became evident. Some of these learnings have been inspired by major accidents in the past. The accident mechanisms can reveal characteristics and conditions shared by other high-risk industries. Information on energy-related accident risks, such as containment barrier weaknesses, reliability of human and organisational barriers, weaknesses in safety performance systems, failure of monitoring and diagnosability systems, etc., can provide useful information to stakeholders with a critical decision-making role in the industry. A second source is the use of well-established assessment techniques for capturing risks in a difficult area (e.g., human performance). It can readily provide inspiration for adoption by other industries lacking it. Other areas to look for such learning opportunities are evident through the scientific works of the risk & safety community, tracking the developments in upcoming modern tools/techniques, etc. The thesis makes use of logical frameworks, rationality criteria, scientific reasonings and case studies, to evaluate the actual usefulness of a learning opportunity, when needed by that industry. Certain cases of incompatibility, and alignment issues with the adopting industry, were discovered. Papers I & II demonstrate this. Here, the Return on Investment (ROI) tool and the Human Reliability Assessment (HRA) method were adopted from the financial and nuclear industries, respectively, for the purpose of decision-support within the oil and gas domain. In particular, the need to align the human reliability assessment method with the risk perspective of the adopting industry has been evident. Both the papers recommend ways to overcome their corresponding limitations in capturing the industry-specific uncertainties and risks. Contributions have also been made regarding improving the analysis criteria for accepting/rejecting the adoption of safety principles that may prove useful for the decision-making process (Paper IV). This paper takes on a decision-maker’s broader perspective on the usefulness of a safety indicator within a portfolio of other indicators, not just on a stand-alone basis. To this end, improving the existing SMART acronym (‘specific’, ‘measurable & manageable’, ‘relevant’ and ‘timely’) to STAR, for evaluating the usefulness of indicators measuring safety performance has been suggested. This will assist in evaluating and selecting safety indicators that provide the decision-makers with a more useful risk trend. The thesis also found a case where a learning opportunity with limited usefulness was discovered. The Texas City accident highlighted limitations of the defence-in-depth safety principle. It was suggested that this principle should be used with another safety principle that advocates having superior monitoring and diagnosis (Paper V). While such a recommendation may seem to be useful immediately, on evaluation, such a recommendation did not seem to add significant value for decision-makers in the nuclear industry. The learning has also been in the direction of employing caution and determining a concrete rationale before adopting multiple safety principles. It is possible that just improving the implementation of existing safety principles may be sufficient. This means that, while there is a growing consciousness among energy sector industries regarding looking towards cross-industry learning opportunities, they also need to carefully consider gaps within their own systems and processes first. Lastly (Paper III), the thesis inspires us to not limit the learning horizon to only across the industries but also look into the emerging techniques for more complex decision-making needs in a high-risk operating environment, where wrong decisions can prove to be costly in the long run. For this, a novel decision-support technique was developed, since offshore and other industries were just beginning to explore the possibilities of modern data-based techniques for improving decision-support.