Diesel engines have been demanded to further increase the thermal efficiency through precise engine control under transient driving conditions, especially, it is essential to optimize the fuel injection timing and quantity cycle-by-cycle. Conventionally, fuel injection have been controlled by control maps, which resulted in large numbers of experiments and increase in cost. In order to overcome these problems, the present study focused on the model-based control and developed the onboard gas flow model, because the heat loss is affected by the turbulent intensity. Firstly, a validation of the CFD simulation is evaluated. The CFD simulation was used to validate the developed models and to determine unknown parameters used in the model. Secondly, modeling of in-cylinder gas flow is presented. To estimate the injection timing within 0.5 deg. against the target value, the heat loss must be estimated within the error range of 7.6%. Finally, as results, the error of heat loss obtained from gas flow model was 1.6%, and gas flow model fully met the requirement of tolerance range. From the viewpoint of calculation time, the calculation time of the model was 50.6 s per cycle, and thus the model is capable of the use of on-board applications.

Diesel engines are required to reduce exhaust emissions during real-world operations. In this regard, a new control concept called model-based control has been explored. Unlike the conventional method of relying on steady-state measurements, model-based control allows cycle-bycycle optimization of control inputs based on physical principles. Existing models for combustion control have been using empirical equations to predict polytropic index for the compression stroke for estimation of in-cylinder pressure and temperature at fuel injection. Therefore, in this study, a polytropic index prediction model was developed in MATLAB to maintain the engine performance under transient conditions and to reduce the required number of experiments. The model includes a heat loss model and a gas flow model to consider the effect of wall heat transfer and gas flows inside the cylinder. The computational load of the model was reduced through discretization of a single engine cycle into several calculation points. The model was validated against numerical simulation results under steady conditions first, and then applied to transient conditions for more realistic operational conditions. The model estimated the polytropic index with average errors under steady and transient conditions with 0.22% and 0.37%, respectively. Finally, the calculation time of the model was evaluated to be 50.6 μs. It was concluded the model can be implemented on a model-based controller in the future.