W25 GIRS Seminars: SAFEST: A Probabilistic Risk Assessment toolchain based on Dynamic Fault Trees & Dynamic Event Trees

Falak Sher Vira, Ph.D., DGB Technoloiges LLC.

Abstract: SAFEST is a powerful tool for probabilistic risk assessment (PRA) of fail-operational and fault-tolerant systems. It has been built in collaboration with Prof. Joost-Pieter Katoen (RWTH Aachen), Prof. Marielle Stoelinga (Twente University), and Matthias Volk (Eindhoven University of Technology), published at ESREL 2023 and RAMS 2024.

SAFEST allows modeling and analyzing both static and dynamic fault trees. Dynamic fault trees (DFTs) extend standard fault trees by providing support for faithfully modeling spare management, functional dependencies, and order-dependent failures. The SAFEST tool provides an efficient and powerful analysis of DFTs via probabilistic model checking—a rigorous, automated analysis technique for probabilistic systems. The backbone of the analysis is based on efficient state space generation. Several optimization techniques are incorporated, such as exploiting irrelevant failures, symmetries, and independent modules. Probabilistic model checking allows analysis of the resulting state space concerning a wide range of measures of interest. In addition, an approximation approach is provided that builds only parts of the state space and allows to iteratively refine the computations up to the desired accuracy.

SAFEST extends classical event trees by equipping tree nodes with rewards/losses and non-deterministic choices, enabling decision-making at nodes. DFTs can be embedded into event trees, providing transition probabilities of nodes. By analyzing event trees (integrated with DFTs), one can determine 1) expected gains or losses, such as radioactive leakage, fatalities, etc.; 2) limits on the frequencies and probability of the outcomes in event trees; and 3) wise choices to, for example, lessen the unfavorable effects/outcomes.

The SAFEST tool provides a graphical user interface for creating, generating, simulating, and simplifying fault trees/even trees, as well as visualizing the results from the fault/event tree analysis. SAFEST is a state-of-the-art analysis tool, as demonstrated by an experimental evaluation and comparison with existing tools. In addition, it has been used in a variety of case studies, including vehicle guidance systems, train operations in railway station areas, and energy systems such as nuclear power plants.

Short Bio: Dr. Falak Sher Vira is a Ph.D. in probabilistic model-checking from RWTH Aachen University and a Postdoc from Fortiss GmbH, Germany, who heads the DGB Technoloiges LLC. Dr. Vira brings cutting-edge academic expertise to solve industry problems and has driven the development of innovative verification tools under his leadership. Under his direction and collaboration with RWTH Aachen and Twente University & Eindhoven University of Technology in the Netherlands, DGB has developed several formal verification tools, including STORM-GUI (Markov automata modeling and analysis tool) and SAFEST (static & dynamic fault and event trees modeling and analysis tool), published at the prestigious safety and reliability conferences ESREL 2024, 2023 (UK) and RAMS 2024 (USA). His research interests include stochastic verification, probabilistic model-checking, safety assessment in model-based systems engineering (MBSE), AI/ML, and safe and explainable AI, among others. He is a prolific author and has published in international conferences such as IAEA (2024), ESREL (2023, 2024), RAMS (2024), IEEE (2021), CS (2020), SACTI (2019), STTT (2014), CONCUR (2014), IC (2014), SMC (2014), TCS (2012), VMCAI (2011), and IEEE (2002).

When: Thursday, March 13, 2025
Time: 1:00-2:00pm PT
Where: UCLA Engineering 6 BLDG, Rm 580B