UNCERTAINTY ANALYSIS

The main goal of the project was to add the Parametrical Uncertainty to SCAIS. To this end, an analysis of the tools developed to this end was performed.

The most interesting was DAKOTA (Design Analysis Kit for Optimization and Terascale Applications) that was coupled to TRACE through SNAP. This is a free software tool ready to perform uncertainty analysis with Monte Carlo and other techniques.

For this reason we developed also an executable that used DAKOTA as a library to generate SCAIS inputs with parametrical uncertainty.

The outputs analysis performed by DAKOTA was not as evolved as we expected and no enough analysis could be done through it. For this reason we developed our own set of R (statistical program) scripts to analyze the whole information desired with SCAIS.

Uncertainty analysis with DAKOTA-SCAIS coupled


The sequence of Loos Of Coolant Acciendet (LOCA) has been analyzed. A selection of twenty input parameters and their Probability Density Functions (PDFs) are considered  in order to performance a random sampling by means of Monte-Carlo techniques.

Afterwards, the corresponding simulations (approximately 4,000) are conducted by coupling DAKOTA-SCAIS-MAAP.  released. The result of this sampling, allows assigning a probability to the final state of damage for each path, fulfilling the criterion of Wilks’s 95/95.

For this propose, the maximum value of the Peak Clad Temperature (PCT) is selected like the security variable in order to classify a transient (path) as successful (PCT < 1500 K) or damage (PCT > 1500 K).

Simulation results (Mote-Carlo sampling parameters)