A Hybrid Compute Engine Implemented for Dynamic Reliability and Risk Analysis

Abstract

In the paper, we present a hybrid compute engine integrating Boolean and analytical models for dynamic reliability and risk analysis of complex safety-critical industrial systems with dynamic interactions and multiphase mission consideration. The hybrid compute engine focuses on three aspects: i) Discrete-time Dynamic Event Tree (DDET) models generation and analysis; ii) risk-based reliability modeling and failure analysis of complex digital industrial process systems under uncertainties using Markov/CCMT approach; iii) dynamic mission reliability analysis of safety-critical systems and emergency response planning for mission success and safety management by GO-FLOW method. The DDET models implemented based on graph-based search and sequence diagram refactoring can be consistently linked to Markov/CCMT and GO-FLOW modules for branch probability estimation. The versatile multi-way search solver for computationally efficient Markov/CCMT analysis and supplementary success-oriented path tracing and planning are briefly illustrated with simplified case studies. It shows that the system integration solutions can provide the comprehensive probabilistic modeling toolkit with the connectivity to overcome drawbacks of any single methodologies when facing the challenges for dynamic reliability and risk analysis.