On Parallelization Strategies for Multiple Representative Interactive Flamelets Combustion Models

A large fraction of today’s greenhouse gas emission is produced from burning fossil fuels. Within the transportation energy sector, liquid fossil fuels are still the major energy source. Even though huge effort has been directed to alternative propulsion systems, such as electrified powertrains, it is expected that conventional combustion engines cannot be substituted within the next decades. Liquid fuels that are synthesized using renewable energy can improve the overall \(\mathrm {CO_2}\) balance on a short time scale, if they are usable in conventional combustion systems. Furthermore, it has been shown in engine experiments that some synthetic fuels might simultaneously reduce pollutant emissions. However, the exact reasons are not entirely understood and more fundamental knowledge is required in order to design synthetic fuels for technical applications. Therefore, the simulation framework CIAO is being developed for studying the entire process chain of fuel injection and combustion with highly accurate numerical methods and physical models. Due to the broad range of scales that occur in such combustion systems, the simulations can only be performed on modern supercomputers. In this study, the performance and scalability of the Multiple Representative Interactive Flamelets (MRIF) combustion model is analyzed and two different parallelization strategies are discussed.