Advanced thermal-hydraulic modelling for nuclear reactor applications

Nuclear Engineering

Imperial College London

The thermal-hydraulic modelling requirements for modern nuclear reactor applications are many and varied.  Natural circulation phenomena occur in many modes of reactor operation and are design dependent e.g. propulsion reactors may be required to operate at inclined angles.  Loss of Coolant Accidents are characterised by complex flow behaviour during the reflooding of the core and the reactor coolant system.  The project objective is to improve thermal- hydraulic 3 dimensional, multi-channel and multi-phase modelling capabilities.

Whilst it is still the case that full 'whole systems’ studies are way beyond them, and ‘best estimate’ systems codes are necessarily employed, modern CFD codes are now getting to the point where they offer real opportunities for practical reactor analysis.  Our objective in this project is to use CFD methods to complement and reinforce best estimate systems codes.

Current such codes (e.g. 1D-RELAP5 and TRAC) contain many empirical correlations to achieve closure of their models.  Some systems codes (e.g. 1D-RELAP5) were explicitly based on 1-D pipe network assumptions. In the case of TRAC, although the code provides more quasi multi-dimensional modelling, it is not a three dimensional solution. All current such codes contain many empirical correlations in order to achieve closure of their models.

It is less than ideal that parameter choices in these empirical representations necessarily influence the predictions of the codes. It is proposed to analyse configurations and circumstances where such representations are important using full three dimensional CFD.

Typical (but just one example) of the circumstances we propose to study is the degree of cross-flow between adjacent sub-channels, and the consequent and inter-pin variations in thermal conditions the fuel experiences.  The combination of the physical insights the CFD offers, and the quantification of the parameters and responses it makes possible, will allow more reliable parametric representations to be incorporated in models like TRACE.

Note: There is no fixed date for applying and the project can be started either in January, April, July or October 2013.