Fault Tree Analysis on R

Fault Tree Analysis on R

An R package has been developed to build fault trees as traditionally used for risk analysis.
Probabilistic Risk Assessment (PRA) and Reliability, Availability, and Maintainability (RAM) fault tree models are supported for related analyses. By our definition PRA models are often time dependent and are well suited for systems that cannot be repaired within a given operational mission, such as aerospace systems. PRA is also the premiere model for analyzing catastrophic risks in complex systems such as nuclear power generation. On the other hand RAM models are characterized by steady state, or time independent, systems. RAM analysis is well suited for systems that are routinely repaired to achieve continuous steady state performance. RAM models might be considered the stepping off point for more complex stochastic analysis involving fixed capability for stand-by, or backup services as often found in chemical process systems.
A fault tree is constructed by building a script such that each node of the tree is described by command line entry. Visualization of the tree can be made continuously during tree development. While this method is not as simplistic as a typical GUI driven front end, it is believed that it enforces more thought to be placed on the tree construction. Scripts are ideal for version control methodologies such that the editing process for a tree can be exposed and controlled for collaboration among a team of contributors.
Considerable growth in the capabilities of the FaultTree package for R are being explored through collaboration with the SCRAM project. SCRAM is a Command-line Risk Analysis Multi-tool. It is a free and open source PRA tool that implements the Open-PSA Model Exchange Format. Collaboration with SCRAM is a two-way win. As the FaultTree package for R leverages the advanced calculations of SCRAM, the SCRAM program gains relatively easy front end tree generation capabilities in R.

FaultTree package released on CRAN

Recently capability has been added to FaultTree to generate the binary decision diagram (BDD) for qualified trees enabling enhanced, indeed exact, probability calculation in the presence of multiply occuring events. Probabiity calculation by BDD is now available at each gate node of the tree, rather than simply the top event as previously provided by SCRAM connection. In the view of this now renders the FaultTree package as worthy of CRAN support. We are pleased to report that as of 3/20/2020 FaultTree has been accepted on the CRAN repository.
Cutset analysis has also been enhanced with a robust mocus algorithm and new prime-implicants capability from the BDD.
This latest package now requires compilation as the newest capabilities have been implemented in C++ for performance. However for Windows and OS X users packaage binaries are maintained on CRAN for ease of installation.
Development packages for FaultTree continue to be maintained on R-Forge. All attempts are made to keep the main branch of GitHub source repositories for FaultTree, FaultTree.widget, and FaultTree.SCRAM consistent with R-Forge.


FaultTree User's Tutorial

A user tutorial is under developement, but falling behind latest capabilities of the package. Check back periodically for additions and changes. Please also provide any comments on content via the Contacts page.