A Probabilistic-Driven Approach for Early Design Quality Risk and Crux Identification Using Non-Markovian Stochastic Petri Nets

Abstract Stochastic Petri Net (SPN) is a powerful and widely used approach for modelling, deadlock detection and safety analysis of stochastic processes in complex stochastic systems. However, applying this method is rarely seen in formal Risk Analysis, especially for quality risks, which refer to the system’s capability and performance to fulfil its objectives on time. This paper investigates the potential of applying Risk Analysis in the early conceptual design stages by modelling the design problem using non-Markovian SPN as a formal analytical tool. The use of SPN offers several advantages, such as high flexibility and strong compatibility with statistical methods such as performance analysis with semi-Markov models, as well as sensitivity analysis and uncertainty analysis. The proposed method enables designers to address quality risks quantitatively and focus explicitly on the design crux problem - the key quality issue directly affecting the design's success. Preferred high-level solutions obtained throughout the Risk Analysis are evaluated using Monte Carlo simulations as initial decision-making insights. A case study of the concept development of a remote maintenance system for the In-Bioshield area of the DEMO fusion power plant is presented to demonstrate the method’s applicability. Initial results showed potential in identifying quality risks, addressing key factors contributing to the design problem, and finding optimal design specifications in the early stages.