Customized risk assessment in military shipbuilding. (May 2020)

• Record Type: Journal Article
• Authors: Crispim, José; Fernandes, Jorge; Rego, Nazaré
• Is Part Of: Reliability engineering & system safety. Volume 197(2020)
• Publication Date: 2020-05
• Subjects: Project management -- Shipbuilding projects -- Risk network model -- Delphi -- Bayesian network; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2020.106809
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: The proposed framework for risk assessment is customizable. The visual diagrams that support Delphi ease the design of the risk network. The use of Bayesian Network provides quantitative measures of risk. Leaky Noisy-OR is used to decrease the effort of expert opinion elicitation. The application to two real projects shows the suitability of the framework. Abstract: This paper describes a customized risk assessment framework to be applied in military shipbuilding projects. The framework incorporates the Delphi method with visual diagrams, Bayesian Networks (BN) and the expression of expert opinions through linguistic variables. Noisy-OR and Leak Canonical models are used to determine the conditional probabilities of the BN model. The approach can easily be adapted for other shipbuilding construction projects. The visual diagrams that support the Delphi questionnaire favor the comprehensive visualization of the interdependencies between risks, causes, risks and causes, and risks and effects. The applicability of the framework is illustrated through the assessment of risk of two real military shipbuilding projects. This assessment includes a sensitivity analysis that is useful to prioritize mitigation actions. In the two cases studies, the risks with higher probability of occurrence were failures or errors in production, of the contracted, in the requirements, and in planning. The results of the sensitivity analysis showed that a set of mitigation actions directed at relatively easily controllable causes would have achieved important reductions in risk probabilities.