Modeling and Simulation for Control of Lean-Burn Gas Engines

Abstract

In this thesis a lean-burn gas engine, produced by Rolls-Royce is studied. The goal is to make an improved simulation model in Simulink for the Bergen B-gas engine by including turbocharger dynamics. And also test and investigate if an MPC controller can improve the engine's behavior compared to the current controller.First the general internal combustion engine is presented. Then a literature study concerning mathematical modeling of engines is conducted. Further, a mean value engine model is derived, and a turbocharger model is implemented in Simulink. The turbocharger model is then combined with the complete engine model developed by Rolls-Royce. This extended model is compared to the original Rolls-Royce model, but is not found to amend the original Rolls-Royce model. The extended model adds new variables, such as the turbocharger angular speed and exhaust temperature, which is not included in the original model, but the precision of these variables should be better before using the extended model.Some basic theory on Model Predictive Control (MPC) is described and an MPC controller is designed and implemented for the original Rolls-Royce model. The response with an MPC controller is then compared to the former response with a PID controller. The MPC controller shows an improved behaviour for the intake manifold pressure, but for the other variables compared, such as engine rotation speed and engine load, the results are worse. But there seems to be a great potential in MPC control of engines. However, more work needs to be done before replacing the PID control.