High Performance 3D CFD Codes for Complex Piston Engine Applications

Economic (fuel price), political (energetic independence), health (cancerous pollutants) and environmental (green house effect) concerns have greatly contributed to the evolution of pollutant emission regulations and fuel consumption regulations in the last decade. A better understanding of the combustion process for operating conditions near the stability limit will probably be the next step towards cleaner engines in order to use more aggressive control strategies. 3D CFD tools for engine modeling evolve to account for systems with increased complexity and to provide insight on the underlying physics of the stability limit. To compete with rapid engine prototyping, predictive simulation (therefore information) return times must be small, which is only possible with the efficient parallelization of 3D CFD codes with advanced physical sub-models. In the present paper, the recent effort for parallelizing IFP-C3D (RANS code) using MPI paradigm is described. This allows the use of larger meshes for reduced elapsed times. Combined with an improvement of the auto-ignition modeling, this allows to predictively compute a full cycle in the CAI combustion mode within 1.5 days. With a similarly linear speed-up, AVBP (LES code) was upgraded to multi-cylinder capabilities opening the way towards the study, understanding and control of cylinder-to-cylinder and cycle-tocycle variability, which greatly influences engine efficiency.