Differences
This shows you the differences between two versions of the page.
Both sides previous revision Previous revision Next revision | Previous revision Next revision Both sides next revision | ||
participant_pages:lsce [2014/05/16 13:55] grenier |
participant_pages:lsce [2014/05/16 14:05] grenier |
||
---|---|---|---|
Line 7: | Line 7: | ||
**Numerical approach for the coupled set of Thermo-Hydrological equations within the benchmark** | **Numerical approach for the coupled set of Thermo-Hydrological equations within the benchmark** | ||
+ | The procedure solving coupled TH transfers with phase change was developed based on the core procedure TRANGEOL solving time step Eulerian transport (diffusive-dispersive & advective transport). TRANGEOL provides a variety of options: a VF (Le Potier 2004, 2005, 2010) or a MHFE (Mosé et al. 1994; Dabbene et al. 1998) numerical scheme, theta-methods (from implicit to explicit) for diffusion and advection, as well as various options for system matrix inversion and conditioning (www-cast3m.cea.fr). It was largely tested for nuclear waste storage applications showing the practical pros and cons of most of these options and approaches. | ||
+ | |||
+ | To resolve the TH benchmark cases the TRANGEOL procedure was used with the implementation of new development, | ||
+ | |||
+ | //Phase change resolution// | ||
+ | |||
+ | The phase change term is treated as a storage term controlling the velocity of the propagation. Phase change has indeed a strong local influence on the propagation of a cooling front for instance by slowing it down as compared with the same temperature front case without phase change. This is due to the energy required to bring the elementary volume of porous medium to the temperature of phase change and then provide the energy for phase change. This leads to steep fronts. Special effort was put to stabilize the resolution with an under-relaxation scheme. The basic idea is to compute some terms of the equation which are function of the temperature (unknown for the heat equation) by their estimation based on the linear combination of temperature at present calculated time and at the previous. This approach stabilizes the convergence of the iteration but under-relaxation is generally less required after a certain amount of iteration and should be stopped to accelerate convergence. So, to optimize this approach we used the procedure by (Durbin & Delemos 2007). The basic idea is there that the under-relaxation factor depends on the progress of the iterations: the further from convergence, | ||