Development of a real-time two-stroke marine diesel engine model with in-cylinder pressure prediction capability

Abstract

The in-cylinder pressure is an important parameter for the diesel engine but it fails to be used in the common scenes, like the control, hardware in loop, and virtual reality, for its lack of real-time capability. In order to have this capability, the conventional diesel engine models are ameliorated (named as MG model, the merged diesel engine model). These parameters that exclude the in-cylinder pressure and indicator torque are calculated by mean value model, the widely used model for real-time applications, to keep on its speed. The other parameters are calculated by the 0D model. To improve the in-cylinder pressure calculation speed, the simplification and asychronization are used. The compression, combustion, and expansion processes are calculated the same as 0D assumption but the exhausting and scavenging processes are simplified by two linear functions. Its calculation time saves about 33.3% comparing to the conventional 0D approach. The boundaries of cylinder model are asynchronous with the scavenging and exhausting manifolds by abandoning cylinder cycles at reasonable intervals so that the time can be reduced further. In practical coding, the in-cylinder pressure needs to be calculated by a parallel thread to realize asychronization. In the case that abandons 4 cycles every 5 cycles the calculation time saves nearly 80% further. Only during the dynamic process, is the reduced time positively correlated with the number of abandoned cycles. The proposed model is calibrated against shop test data, whose predicting accuracy is comparable to the 0D model. The maximum relative errors of steady MG model and steady 0D model are 3.96% and −3.23%, and the mean relative errors are 1.38% and 2.18%. In the case that abandons 4 engine cycles, the maximum relative error of explosion pressure is 0.363% during the dynamic process. This model can be used in real-time HIL, controller design, engine analysis, and simulator.