Numerical modelling of pulsation and convection in cepheids

In order to simulate the pulsation convection coupling in a Cepheid the ANTARES-code was equipped with a polar and moving grid. The numerical cost of a fully parallelized, sufficiently large, and fully resolved section would be immense. Thus it was not only necessary to find a suitable model, but also save to costs for parallelisation and grid refinement. The equations governing the hydrodynamics were derived for this particular grid and implemented in the code. The grey short characteristics method for the radiative transfer equation was also adjusted. Different methods of parallelisation for the radiative transfer were tested. Abstract Within ANTARES shocks are treated with an essentially non oscillatory (ENO) scheme with Marquina flux splitting. As this method is only valid for grids that are equidistant or uniformly stretched in all directions two differnt sets of ENO-coefficients were implemented and tested. It was found that the traditional approach is indeed no longer valid and the system is not conservative when the original set of coefficients is used. In the upper or hydrogen ionisation zone the gradient of density, temperature etc. is very steep, therefore a finer resolution with a minimum of additional time steps is needed. In order to resolve these few points a co-moving grid refinement was developed. Simulations in one and two dimensions were performed, a comparison between them helps to better understand the effects of convection on the e.c. light curve. Analysis of the fluxes and the work integral was done for the helium ionisation zone. The effects of subgrid modelling were tested on the hydrogen convection zone and compared with a resolved simulation of this zone. (author)