2017

Tyrväinen, Tero

Dynamic flowgraph methodology (DFM) is a method for the reliability analysis of dynamic systems with time-dependencies and feedback loops. A DFM model is a graph representation of the analysed system, and its components are analysed at discrete time points and they can have multiple states. The aim of DFM is to identify which conditions can cause a top event, e.g. the system's failure. The reason for the development of DFM is that traditional methods, such as fault tree analysis, can describe the system's dynamic behaviour only in a limited manner. DFM can more accurately represent the system's evolution in time. VTT has previously developed a DFM tool called Yadrat, which solves a DFM model by transforming it into a binary decision diagram. It has however not been possible to apply Yadrat to large systems. To improve Yadrat's computational capacity and to reduce computation times, this report studies an alternative way of solving a DFM model. A DFM model is transformed from Yadrat into the fault tree format of the probabilistic risk assessment (PRA) software FinPSA, then solved by the fault tree algorithm of FinPSA. To make this possible, FinPSA's fault tree algorithm has been developed to take into account the non-coherent logic of DFM. A simple DFM model has been solved correctly using FinPSA. However, with the preliminary implementation it has not yet been possible to solve more complex models correctly. Therefore, there has not been a possibility to evaluate the computational efficiency of the implementation yet. The next step is to resolve the issues conserning the computation of larger models. Then, FinPSA implementation can be benchmarked with Yadrat to find out whether FinPSA is faster and able to solve large models.