Calibrating the Livengood–Wu integral knock model for differently sized methanol engines

Abstract

Experimental test campaigns have begun to demonstrate the potential of methanol as an alternative fuel for heavy-duty spark-ignited engines. However, there is no consensus yet on the scope of this solution in terms of maximum power and engine size. A zero-dimensional combustion model is therefore being developed outside the scope of this work. Its main objective will be to predict key performance parameters such as power and efficiency as function of engine size. Due to the high loads typically encountered in heavy-duty engines, knock will be the main constraint to maximize the engine's potential. This work therefore aims to find an accurate knock model that can be implemented in the modelling framework. The Livengood–Wu knock integral model is being considered as a good candidate, as it is computationally inexpensive and thus allows for a large number of engine configurations to be modelled within a reasonable time. Due to a lack of autoignition delay times of methanol at conditions relevant to heavy-duty engines, a large database was created using chemical kinetics calculations. A neural network model was trained with the tabulated data for fast data retrieval. To validate whether the knock integral approach is robust enough to be applied to a wide range of engine sizes, a calibration constant was added to match the knock predictions to experimental data. Its value was calculated for three different engines, a light and heavy-duty SI engine and a large-bore dual-fuel engine. They highlight a remarkable difference in calibration constant across the different engines investigated.