Abstract The Ansys Fluent GPU solver is a native GPU-powered solver, which uses the high performance of graphics processing units (GPUs) to facilitate the computation of complex simulations. More than CPUs, GPUs are optimized for parallel processing and have high bandwidth memory, resulting in remarkable increase in performance of flow calculations with respect to the traditional multi-core CPU-based computations. The recent advancements in GPU hardware make the native GPU solver unique and attractive for applications with computationally intensive flow solutions like aeroengines where dense meshes, complex multiphysics and turbulent interacting phenomena coexist. The current work examines a general-purpose finite rate chemistry solver development in the native GPU solver. Specifically, we explore the capability of the finite rate chemistry solver in simulations of experimentally well-characterized fully premixed and partially premixed turbulent flames, ranging from hydrogen flames to blended fuels. We showcase good agreement between the computed resolved mean statistics and experimental measurements. Each case is used to validate the accuracy of the GPU solution against experimental measurements. The objective of this work is to understand and optimize computational efficiency for detailed chemistry solutions in GPU framework. Following the establishment of the baseline results, scaling and benchmarking studies are conducted over three different mesh densities.
Litrico et al. (Mon,) studied this question.