Large-Eddy Simulations with Conjugated Heat Transfer of an Internal Combustion Engine

Abstract

Engine heat transfer affects the internal combustion engine’s (ICE) efficiency, performance, and emissions. Approximately 20 to 30% of the energy from the combustion process is lost through convective heat transfer across the engine walls. To improve the predictive capability of engine simulations, engine heat transfer needs to be simulated accurately. This requires that we capture the temporal, spatial, and cycle-to-cycle variations (CCV) in engine heat transfer. In conventional engine CFD, uniform and constant surface temperature boundary conditions are often applied. To improve these boundary conditions, conjugate heat transfer (CHT) can be used which couples the heat transfer solution between fluid and solid domains. In this work, CHT was integrated with large-eddy simulations (LES), with moving valves and pistons, to improve heat transfer predictions for motored and fired operating conditions using a commercial CFD software. The quality of these simulations was evaluated against bulk flow, near-wall flow, and near-wall temperature measurements performed by the Quantitative Laser Diagnostics Laboratory (QLDL) group at the University of Michigan. Such comparisons to the near-wall flow and temperature fields have not been available to date and are described here for the first time. By using CHT, improved heat transfer predictions were obtained compared to baseline LES with uniform temperature boundary conditions.

In the first part of the study, the motored operating condition was simulated with uniform temperature boundary conditions, and modeling methods were analyzed, including turbulence models, wall models, near-wall mesh resolution, and thermal boundary conditions, which improved heat transfer predictions. A statistical convergence criteria was developed using the well-known LES quality index, and statistical convergence was demonstrated for the motored LES after 10 cycles. However, more LES cycles are likely needed to capture the measured level of CCV. Bulk flow analysis shows that the simulation of the intake jet largely impacted the predicted vortex center locations and heat transfer. Near-wall flow analysis shows that improvements in the wall models are needed.

In the second part of the study, CHT was integrated with LES for motored and stoichiometric fired operating conditions, and validated with measured surface temperature within 1.4% and 3% error for the motored and fired conditions, respectively. Spatial, temporal, and CCV in the surface temperatures were predicted. The results show the impact of surface temperature on the predicted flow and temperature fields. For example, the spark plug surface temperature experienced a large spatial temperature range from 350 to 1000 K, which significantly impacts the heat transfer at the spark plug and the early combustion process. The LES CHT method improved the predicted level of CCV in the surface heat flux which compared better with the measurement. This work was performed in a parallel effort in near-wall PLIF temperature field measurements to assess the LES CHT predictive capability in the near-wall temperature field. Length scale analysis was performed in the wall-normal direction to provide insights into the spatial scales that are important in engine heat transfer. These length scales decrease towards the wall, indicating that temperature gradients increased towards the wall leading to increased heat transfer. Their distribution also became more homogeneous towards TDC, and therefore, heat transfer becomes more spatially uniform with increased compression.