Automatic Generation of Event Trees and Fault Trees: A Model-Based Approach

Abstract

In the past few decades, the increasing complexity of modern engineering systems has been driven by the integration of a large number of components whose operations may involve many disciplines (e.g., thermal hydraulics, plant operations, cybersecurity). Most computational tools used by industry and regulators for system safety and reliability assessments are still based on the traditional fault tree (FT) and event tree (ET) approach, which may not be able to capture complex interactions among system constituents. The use of simulation tools has widely increased in the past few decades to improve the fidelity of the reliability and safety analyses. However, the direct use of simulation tools as part of dynamic probabilistic risk assessment (DPRA) methods is not getting traction since (1) modeling the whole system under consideration with DPRA methods may be computationally expensive and unnecessary, and (2) the manual integration of DPRA models into existing state-of-practice probabilistic risk assessment models (i.e., based on FTs and ETs) can be time consuming and prone to errors. In this paper we propose a procedure to overcome this limitation by presenting several algorithms designed to automatically construct subsystem ETs and FTs from DPRA methods for integration into an existing ET/FT system model.

Keywords:

• PRA
• dynamic PRA
• model integration

Acknowledgments

The U.S. government retains and the publisher, by accepting this paper for publication, acknowledges that the U.S. government retains a nonexclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this paper, or allow others to do so, for U.S. government purposes.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

^a Note that only the status of each unit (i.e., operating, failed open, failed closed) is sampled. No timing of state transitions is sampled.

^b When considering the simulations belonging to $C1$ in the first iteration, the stopping condition ${\mathit{M}}_{C1}={\mathit{M}}_{C2}$ (see Algorithm 3) was in fact reached.

^c As an example, a safety-related requirement would be the ability to withstand a loss-of-coolant accident scenario. The verification of such a requirement is modeled in the SysML model by linking other diagrams (e.g., parametric diagram) to the specific requirement.

^d In this paper, we show time series as simulation runs or histories.

^e This allows us to maintain generality by having time series with different time lengths.

^f As an example, $x_2^3$ is a vector of length $T_3$, which represents the temporal profile of variable 2 for scenario 3.

^g A third path currently under investigation is to reconstruct the major clustering algorithms available in the literature (e.g., K-Means, Mean-Shift) so that they can natively perform data analysis on the time series data set (i.e., without prior transformation, see path 1). The major challenge in this approach is the need to define an operator that, given a subset of time series, can generate a distance-based average time series. This average value can be challenging to obtain depending on the distance metrics employed.

^h Note that here we have relaxed the requirement $T_S=T_Q$.

Additional information

Funding

The research presented in this paper was partially supported by the U.S. Department of Energy (DOE) Nuclear Energy University Program award 17-12723. This research made use of the resources of the High Performance Computing Center at Idaho National Laboratory, which is supported by the DOE Office of Nuclear Energy and the Nuclear Science User Facilities under contract no. DE-AC07-05ID14517. This paper has been authored by Battelle Energy Alliance, LLC under contract no. DE-AC07-05ID14517 with the DOE.