Deterministic and stochastic dependability analysis of industrial systems using Coloured Petri Nets approach

Industrial systems dependability analysis is a two-fold complex task. From one hand, it consists in quantitative reliability, maintainability and availability assessment and involves stochastic modelling of system behaviour. From the other hand, it requires deterministic modelling to capture the control system behaviour and to verify its safety properties. Generally two different models of system behaviour are used to achieve these two tasks, requiring different mathematical models: probabilistic and timed models for stochastic analysis and deterministic non-timed models for qualitative analysis. The purpose of this work is to use one mathematical model for both dependability tasks. The Coloured Petri Nets tool (CPN), which is a high level Petri Net, is used in this paper. The model allows the stochastic simulation of system behaviour and dependability assessment by means of Monte Carlo simulations. The safety analysis is performed by means of state space analysis and model checking techniques. Main addressed issues are related to abstraction and model transformation in order to adapt the CPN model to the stochastic or deterministic context of the dependability analysis. The described approach is tested on a case study, which is a part of a nuclear power plant sub-system developed by EDF company (Electricité de France). The considered system is characterised by components redundancy, different distribution laws (not only exponential) of failure and reparation times and control laws aiming to switch between configurations according to functional or dysfunctional purposes. Proposed approach appears to be efficient for evaluation of stochastic dependability indicators (availability, MTTF, MTTR, etc.) as well as for safety analysis (reachability of critical states, dead-locks, proof of control behavioural properties, etc.) of a concurrent controlled redundant system.