0D/1D turbulent combustion model assessment from an ultra-lean spark ignition engine

This paper focuses on an assessment of predictive combustion model using a 0D/1D simulation tool under high load, different excess air ratio ?, and different combustion stabilities (based on coefficient of variation of indicated mean effective pressure COV_imep). To consider that, crank angle resolved data of experimental pressure of 500 cycles are recorded under engine speed 1000 RPM and 2000 RPM, wide-open throttle, and ?=1.0, 1.42, 1.7, and 2.0. Firstly, model calibration is conducted using 18 cases at 2000 RPM using 500 cycle-averaged in-cylinder pressure to find optimized model constants. Then, the model constants are unchanged for other cases. Next, different cycle-averaged pressure data are used as inputs in the simulation based on the COV_imep for studying sensitivity of the turbulent model constants. The simulation is conducted using 1D simulation software GT-Power. Firstly, a three-pressure analysis (TPA) model (intake, in-cylinder, exhaust) for experimental prediction and optimization of burn rate shape are studied. Boundary conditions such as the three pressure histories, intake/exhaust valves timings, boundary temperatures, and exhaust gas emissions are used as model inputs. Errors of indicated thermal efficiency, indicated mean effective pressure, and CA50 are within 3%. Predicted parameters from the TPA model such as air volumetric efficiency, trapped air/fuel vapor mass, trapped residual gas fraction, tumble, and surface temperature of the piston, head, and valves are used as initializations in the predictive combustion model. A built-in flame propagation model, termed as SITurb, is investigated whether it can replicate the in-cylinder pressure and burn rate shapes. A revised laminar flame speed correlation of five-component gasoline surrogate is incorporated in the combustion model via an encrypted dynamic link library file. The results show that thermodynamic histories of the combustion are reproducible under high load and stoichiometric-to-ultra-lean conditions. Under all cases, only turbulent flame speed multiplier needs to be calibrated.