Novel Insight into Engine Near-Wall Flows and Wall Heat Transfer Using Direct Numerical Simulations and High-Fidelity Experiments

Abstract

This study combines advanced optical diagnostics and high-fidelity Direct Numerical Simulations (DNS) to deepen the understanding of wall heat transfer processes in Otto engines under motored and fired conditions. To this end, a combination of optical diagnostics was applied simultaneously: High-resolution Particle Image Velocimetry (PIV) and Particle Tracking Velocimetry (PTV) to resolve the velocity boundary layer (BL) above the piston, Thermographic Phosphor Thermometry (TPT) to measure the wall temperature spatially resolved and Laser Induced Fluorescence (LIF) of SO2 to track the evolution of the flame. For the complementing simulations, an entire workflow was developed that employs process calculations (GT-Power), multi-cycle scale-resolving simulations (SRSs), and DNS. Well-calibrated GT-Power models provided boundary conditions for the experimentally validated SRSs, which in turn yielded initial conditions for the DNS. Using initial conditions from the SRSs at intake valve closure, the first ever DNS of a real engine geometry was successfully performed for one motored and one fired compression/expansion stroke. It was seen that momentum and thermal BLs evolve differently: the former are affected by changes in the bulk velocity (large scale tumble motion and its breakdown), while the temperature gradients monotonically follow the increase in pressure/Reynolds number. Both the scaled momentum and thermal BLs do not exhibit a logarithmic region and the law of the wall does not hold. Several sources for deviations thereto, both in momentum and thermal BLs, are extracted. For the reactive case, it was found that the early flame kernel development is significantly affected by the strong convective flow due to tumble and only when the flame is strong enough to counter-balance the strong convection it can propagate against it. A criterion has further been developed, which allows for distinction between head-on and side-wall quenching. The vast amount of high-fidelity experimental and fully resolved numerical data generated in this project provides a comprehensive database for validation of existing computational fluid dynamics (CFD) tools and can be used for the development of improved wall heat flux models. A first attempt has been made towards this direction by developing an algebraic wall heat transfer model for LES using a data-driven approach.