Methods and issues for the analysis of complex systems by numerical simulation at EDF
The needs to assess robust performances for complex systems and to answer tighter regulatory processes (security, safety, environmental control, health impacts, etc.) have led to the emergence of a new industrial simulation challenge: to take uncertainties into account when dealing with complex numerical simulation frameworks. For the French company EDF (Electricité de France), these may be computer codes simulating all physical phenomena of interest for the operation and safety of its installations (such as welding operations, examinations non-destructive components, accidental situations in its pressurized water nuclear reactors, etc.). Many attempts at treating uncertainty in industrial applications have involved different mathematical approaches and standards (sometimes domain-specific): metrology, structural reliability, variational approaches, design of experiments, global sensitivity analysis, Bayesian and machine learning approaches, etc. Based on several EDF industrial applications, this communication will introduce some research issues in the field of statistics and related to the design, analysis and modeling of computer experiments.