Event sequences encountered in nuclear reactor safety may be studied with numerical simulations that are safer and less costly than actual experiments. State-of-the-art computational means have been developed to support the exploration of open issues which are hard to investigate experimentally. On the other hand, accurate numerical models may provide guidelines for designing safety related experiments.  In this framework, nuclear reactor safety activities take advantage of the sustained evolution in information technology to develop realistic models of the complex physical phenomena taking place in nuclear reactor facilities with the intention to improve the overall system safety and performance.  According to the Strategic Research Agenda elaborated by the EU Sustainable Nuclear Energy Technology Platform (SNETP) modeling and simulation activities contribute to all phases of a reactor lifecycle i.e., design, safety assessment, licensing and issue resolution during operation.  It has become industry practice to:

The Research Reactor Laboratory (RRL) that hosts GRR-1 is a suitable environment to foster nuclear reactor safety activities since it may accommodate both approaches listed above. Experiments are symbiotic with numerical computations and the performance of a computational tool needs to be verified and validated against experimental data. Thus, successful GRR-1 upgrade will render the facility available to support the developments in nuclear engineering modeling and simulation.  Data availability will accommodate analysis and validation of computational tools.  Along this line, GRR-1 best-estimate models have already been developed to simulate the system response under various transient conditions. 
The SNETP Deployment Strategy in 2010 has identified key R&D topics for Gen III reactors in the areas of upgraded human system interfaces and simplified operation of reactor systems, while for Gen IV it embraces improved safety as well as advanced instrumentation and in-service inspection capabilities.  Thus, development of physical models using energy, momentum and mass balance equations has been a motivation for the Research Reactor Laboratory activities to expand into the realm of data modeling. In this area, empirical data sets are used to estimate the underlying system function with little or no knowledge of the function form.  Applying sophisticated machine learning algorithms in a nuclear facility, data modeling has been proven a powerful tool for monitoring, diagnostics and control tasks. Advanced tools – from the computational intelligence realm – are developed at RRL for (i) identifying non-linear system dynamics, (ii) predicting rapidly evolving transients, (iii) signal validation, (iv) enhanced man-machine interfaces and (v) plant-wide monitoring.

Back to Nuclear Technology main page