Physically Based Modelling for Knock Prediction in SI Engines

Abstract: The high demand for an increase in performance and at the same time loweringthe emissions is forcing the automotive industry to increase the efficiency of thevehicles. This demand lead to a problem called knock, which often is the limitingfactor when increasing the efficiency of the engine. Knock occurs when theunburned gases inside the combustion chamber self-ignites due to the increasingpressure and temperature.This thesis investigates if it is possible to predict knock with a physicallybased knock model. The model consist of several sub-models such as pressuremodel, temperature model and knock model. The models are built by using measureddata and the goal is to get an independent knock prediction model that canfind the limited ignition angle that will cause knock.The results shows that an analytic pressure model can simulate a measuredpressure curve. But when it comes to predicting knock, there is an uncertaintywhich can be improved by changing the modelling strategy and making the modelsmore accurate.