Three-Dimensional Computations of Combustion in Premixed-Charge and Direct-Injected Two-Stroke Engines 920425
Combustion and flow were calculated in a spark-ignited two-stroke crankcase-scavenged engine using a laminar and turbulent characteristic-time combustion submodel in the three-dimensional KIVA code. Both premixed-charge and fuel-injected cases were examined. A multi-cylinder engine simulation program was used to specify initial and boundary conditions for the computation of the scavenging process.
A sensitivity study was conducted using the premixed-charge engine data. The influence of different port boundary conditions on the scavenging process was examined. At high delivery ratios, the results were insensitive to variations in the scavenging flow or residual fraction details. In this case, good agreement was obtained with the experimental data using an existing combustion submodel, previously validated in a four-stroke engine study. However, at low delivery ratios, both flow-field and combustion-model details were important, and the agreement with experiment was poor using the existing combustion submodel, which does not account for the effect of residual gas concentration.
To improve the agreement between modeling and experimental results, a modified combustion submodel was introduced that includes the effect of residual gas concentration on the laminar characteristic time. With the new submodel, agreement with the experiment has been improved considerably for all cases considered in this study. These levels of agreement between experiment and computations are similar to those found in previous applications of the laminar and turbulent characteristic-time combustion submodel to four-stroke engine combustion. Further improvement of the combustion submodel was made difficult by the observed coupling between the in-cylinder flow-field and the combustion-model details at low delivery ratios.
Three-dimensional computer models have been applied to predict combustion in internal combustion engines. For example, computations of spark-ignited premixed-charge combustion in research and production-type four-stroke engines have been presented by Kuo and Reitz *. Spark-ignited, premixed-charge and direct-injected rotary engine computations have been presented by Abraham and Bracco [2,3]. The latter computations suggested design changes that led to approximately 6 percent improvement in efficiency in a direct-injection stratified-charge rotary engine.
The above studies were conducted using a characteristic-time combustion submodel originally proposed by Abraham et al. . In this model, chemical species approach their thermodynamic equilibrium with a rate that is a combination of the turbulent-mixing time and the laminar chemical-kinetics time. The combination is formed in such a way that the longer of the two times has more influence on the conversion rate. An additional element of the model is that the laminar-flame kinetics strongly influence the early flame development following ignition.
Comparisons with experimental engine pressure measurements indicate that the model predictions agree reasonably well with measurements under normal engine operating conditions. The causes of discrepancies were discussed in detail by Kuo and Reitz [1,5] who observed that the level of agreement between the predictions and the experiments is consistent with the levels of uncertainty in the input parameters to the computations (e.g., there is some uncertainty about the gas temperature, turbulence intensity and length scale existing in the combustion chamber at the start of combustion, and also in the wall-heat-transfer and turbulence model constants).
There is current interest in engines that operate under dilute conditions either with air, internal residuals, or recirculated exhaust gas. Najt and Kuo  have successfully applied the laminar and turbulent characteristic-time combustion submodel to engines diluted with air. A major goal of the present study was to investigate the performance of the combustion submodel for engines diluted with residual gas. This condition is of interest in variable-valve-actuation engines with large valve overlap, and in premixed-charge and direct-injected two-stroke engines at low delivery ratios, which have high residual gas concentrations in the combustion chamber. For this study, the model was applied to crankcase-scavenged two-stroke engine combustion for which experimental data is available.
This paper is organized as follows. First, the experiments used to assess the performance of the combustion submodel are described. Then, details are given of the initial and boundary conditions, and the combustion and spray submodels. The results are divided into two parts. The first part explores the sensitivity of the results to the in-cylinder flow field and combustion model details in the premixed-charge engine. The second part presents a parametric study of the model performance in the premixed-charge and direct-injected engines.