Knock mitigation study on alternative fuel heavy duty engines

Abstract

With increasingly stricter emissions regulations and a growing demand for higher efficiency, alternative fuels are becoming more viable alternatives to fossil fuels. Methane, for example, can be used in conventional combustion engines with minimal modifications, meeting lower emission targets. However, engines running on alternative fuels still face issues common to conventional engines, such as “knock,” which limits their efficiency.

This thesis investigates the manifestation of knock in alternative fuel engines and proposes a method to prevent it while maintaining high efficiency. The method utilizes Model Predictive Control (MPC) and Artificial Neural Networks (ANNs). A detailed outline of the proposed controller is provided, along with the rationale behind its key structural characteristics. The use of ANNs to model the system’s state is also explored and evaluated. The controller is tested under various operating conditions and tuning settings to verify its effectiveness and identify optimal tuning.

The proposed control method performed well in the selected cases and setup. Further improvements to both the controller and neural network structure are suggested to enhance performance. This thesis establishes that MPC can improve engine performance and prevent unfavorable operating conditions. Initially useful for engine mapping, the controller has future potential for direct implementation on the ECU.