Engine simulations based on RANS and LES depend on the mesh resolution and modeling assumptions (turbulence, combustion, wall flow, etc.) and thus a validation with experimental data is necessary. Direct numerical simulations (DNS) can be an alternative to the engine experiments, since the results are not dependent on model assumptions and therefore can be seen as “high-quality numerical experiment” (Moin & Mahesh 1998). The main advantage of the DNS data are the fully resolved fields (velocity, temperature and chemical species), which can be used for model validation or an improved understanding of the physical processes. In this work, a highly scalable parallel, spectral element, low Mach number code based on the flow solver nek5000 is employed to perform multiple-cycle simulations in the axisymmetric piston-cylinder assembly investigated experimentally by Morse et al. (1979).   Eleven cycles were computed and the comparison between the experiment and the DNS data for the mean and rms velocities show very good agreement for all measured times and locations. The flow field remaining at top dead center from the previous cycle and its effect on the jet breakup and the vortex ring dynamics below the valve are found to have a dominant effect on the observed significant cyclic variations. In a second step the valve is closed at BDC and the compression stroke is simulated. Of special interest is the unsteady wall heat transfer, since regions close to the walls are very challenging to access with experimental methods. In addition, the evolution of the thermal stratification is investigated since it is very important for ignition timing and pressure raise rate for HCCI operating conditions.

Bio:
Martin Schmitt is a scientist at the Aerothermochemistry and Combustion Systems Laboratory (LAV) at ETH Zurich. Martin did his diploma in Mechanical Engineering at the RWTH Aachen and wrote his Master thesis in the research department of Volkswagen. After completing his Masters, he began his doctoral studies at ETH Zurich and completed his Ph.D. in October, 2014. His research focus is on simulation of the gas phase in internal combustion engines based on RANS, LES and DNS simulations.