Three modelling phases have been carried out using CODE-BRIGHT. The first, for design, was performed using few real data, but was focused on understanding THM processes involved, determining heater power and defining the distribution of the sensors.

The second modelling phase was performed at the end of the pre-operational stage ("blind" prediction) using laboratory data for bentonite chracterisation.

The third modelling phase (operational modelling) was carried out to help understand deviation behind predictions and observational data.

Pre-operational thermo-hydro-mechanical modelling (CODE-BRIGHT code) of the "mock-up" test: contours of degree of saturation at different times.

Modelling of the possible effects of the initial flooding carried out before heating the "mock-up" test has been focusing on consideration of the inherent heterogeneities involved: joints between bentonite blocks, and the blocks between heater and the confining structure.

Operational thermo-hydro-mechanical modelling of the "in-situ" test with the CODE-BRIGHT code: computed and observed results of the variation of relative humidity with time.

Fairly good agreement is obtained on the evolution of relative humidity at different points in the buffer. Generally, the correspondence between computations and measurements is very good, although in some points near the heater the observed drying is stronger than computed.