Large-Eddy Simulation of in-cylinder flows in motored reciprocating-piston internal combustion engines

Open Access
Liu, Kai
Graduate Program:
Mechanical Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
March 08, 2012
Committee Members:
  • Daniel Connell Haworth, Dissertation Advisor
  • Daniel Connell Haworth, Committee Chair
  • Andre Louis Boehman, Committee Member
  • James Gordon Brasseur, Committee Member
  • Stephen R Turns, Committee Member
  • Turbulence
  •  Large-eddy simulation
  •  Internal Combustion engine
  • Proper orthogonal decomposition
  •  Cycle-to-cycle variations
Two key bottlenecks prevent engines from reaching their performance, efficiency, and emissions potential. The first bottleneck is limited understanding of turbulence hydrodynamics for in-cylinder flows including cycle-to-cycle variations (CCV), and the second one is the absence of an objective approach for making quantitative comparisons between simulation and experiment, beyond ensemble averaging. In this thesis, the CCV phenomenon in IC engines and its effects on IC-engine performance are introduced. Previous studies of CCV, its root causes, and its influences on engine performance are reviewed. The limitations of current practices for IC engine simulation and analysis are discussed. Large-eddy simulation (LES) has shown promise in internal combustion (IC) engine applications, and proper orthogonal decomposition (POD) has been proposed as an objective way to analyze complex turbulent flows and to make comprehensive comparisons between simulation and measurements. In the research performed here, LES and POD have been performed for two simplified motored IC engines: the Imperial College piston-cylinder assembly with and without swirl and the Transparent Combustion Chamber (TCC) engine. For the first configuration, the sensitivity of LES to key numerical and physical model parameters has been investigated. Results are especially sensitive to mesh and to the subfilter-scale (SFS) turbulence models. Satisfactory results can be obtained using simple viscosity-based SFS turbulence models, although there is room for improvement. No single model gives uniformly best agreement between model and measurements at all spatial locations and at all times. Compared to Reynolds-averaged Navier-Stokes (RANS) modeling, LES shows advantages in accuracy and in capturing more details of the complex in-cylinder flow dynamics. In particular, LES is able to capture CCV using computational meshes that are comparable to those that are used for RANS, in that case, the high computational cost of LES is mainly due to the need to compute multiple engine cycles. POD is then used to study the dynamics of the in-cylinder turbulent flow. Systematic parametric studies are performed, including two-dimensional (2-D) POD versus three-dimensional (3-D) POD, phase-dependent POD versus phase-invariant POD, and sensitivities of POD mode structure and mode convergence rate to spatial and temporal resolution. The use of POD to identify and quantify CCV is explored, and the ability of POD to distinguish between organized and disorganized flows is demonstrated. The LES and POD experience from the piston-cylinder assembly is then extended to a more realistic engine configuration (TCC engine) with full four-stroke motored cycles, where detailed particle image velocimetry (PIV) measurements are being made. The complex in-cylinder flows, including characterization of CCV, are analyzed by using LES and POD. Initial quantitative comparisons with PIV measurements are also performed. It is found that many of the key conclusions that were drawn from the POD analysis of the piston-cylinder assembly carry over to the more realistic engine. This suggests that the POD tools that have been developed will be useful in analyzing real engine flows.