Section: Application Domains

Scalable numerical schemes for scientific applications

Participants : Rached Abdelkhalek, Olivier Coulaud, Yohann Dudouit, Luc Giraud, Guillaume Latu, Alexis Praga, Jean Roman, Pablo Salas Medina, Xavier Vasseur.

We are also collaborating with application research group to design or improve numerical schemes in the view of large scale parallel simulations.

Seismic wave propagation in heterogeneous media requires to properly capture the local heterogeneity and consequently requires locally refined meshes. In close collaboration with TOTAL we study new parallelizable schemes for the solution of the elastodynamic system with local spatial refinments based on discontinuous Galerkin techniques. The objective is to design novel parallel scalable implementations for large 3D simulations. A second work is currently carried on with TOTAL for Seismic modeling and Reverse Time Migration (RTM) based on the full wave equation discretization. These tools are of major importance since they give an accurate representation of complex wave propagation areas. Unfortunately, they are highly compute intensive. To address this challenge we have designed a fast parallel simulator that solves the acoustic wave equation on a GPU cluster.

Thermoacoustic instabilities are an important concern in the design of gas turbine combustion chambers. Most modern combustion chambers have annular shapes and this leads to the appearance of azimuthal acoustic modes. These modes are often powerful and can lead to structural vibrations being sometimes damaging. Therefore, they must be identified at the design stage in order to be able to eliminate them. However, due to the complexity of industrial combustion chambers with a large number of burners, numerical studies of real configurations are a challenging task. Such a challenging calculations performed in close collaboration with the Computational Fluid Dynamic project at CERFACS.

The chemistry and transport models (CTM) play a central role in global geophyscal models. The solution of the CTM represents up-to 50 % on the computing ressources involved in global geophyscal simulations. Therefore, the availability of efficient scalable parallel numerical schemes on emerging and future supercomputers is crucial. The purpose of this research activity is to study, design and implement novel numerical schemes following the work initiated by D. Cariolle in the framework of the ANR Solstice project. Alexi Praga, PhD hired by CERFACS, is conducing this research action under the joint supervision of HiePACS and the Aviation and Environment project at CERFACS in close collaboration with CNRM/Meteo-France.