Thermal hydraulics simulation for nuclear reactors, historically based mainly on system and component scale modelling, is evolving towards a more intensive use of computing codes at small scales, typically CFD or DNS scales. Indeed, thanks to improved computer performances, industries and research organizations involved in nuclear reactor safety focus on the development of modelling at local scales. The objectives are two-fold: progressing in the small scale flow process modelling to get more confidence in the results obtained with system or component codes and a better understanding physical situations of interest.

 

However, it is a long way before CFD becomes an integral part of the safety demonstration. Among the obstacles to be overcome to get “good” computer simulation software, a better understanding of the numerical schemes (which must be accurate and robust) and of the physical modelling that closes the system must obtained. In addition, it seems that progress on any of these two item cannot be done independently.

 

The aim of this two-day workshop is to gather researchers from different background: researchers that are motivated by more pragmatic considerations and those which aim is to design potentially more sound approaches but are less worried by validations on “real life” cases.  We believe that both approaches have their merit and that an in-depth discussion between these communities could potentially lead to a better understanding and hopefully new collaborations and better science and hence better software.

 

We have identified the following subjects to be discussed:

• The different choices for the basic model, typically one-fluid model, two-fluids model or multi-fluids models;

• Treatment of interfaces: the possible choices being a deterministic treatment where the position of the interface in space and time is tacked or a statistical description of interfaces in which the presence of interfaces is described through average parameters such as the volume fraction;

• Numerics versus physical modeling? Is the better, more realistic or enough to achieve hyperbolicity thanks to a kind of mathematical regularization or to treat ill-posedness issue with numerical regularization?

• Feedback on the use of the single-pressure two-fluid model;

• The 7 equations model for an interface diffuse method: what are the issues for real configurations? Is this model really more difficult to discretize than a 6 equations model with pressure equilibrium? Why?

• Comparison with experimental results: role of data bases, how to interpret experimental result versus numerical ones?