Knock has been studied by internal combustion engine researchers for well over a
century. It remains perhaps the main limit on spark-ignition engine efficiency
today. In an engine development environment, knock is typically described
through quantification of the high-frequency signal content of cylinder pressure
measurements. A cylinder pressure transducer gives a point measurement in the
combustion chamber volume. In non-knocking combustion cycles, there is little
pressure variation across the chamber; hence, this point measurement adequately
represents the average gas pressure acting on the piston. This is not the case
for knock where autoignition leads to strong pressure gradients and standing
wave behavior or even supersonic shock wave propagation. The resulting pressure
signal is complex to interpret. Knocking phenomena can be simulated in
Computational Fluid Dynamics (CFD), ideally using a combination of Large Eddy
Simulations (LES), chemical kinetics, moving meshes, and small timesteps. Such
approaches are computationally intensive, however, and may not be feasible to
apply over the wide range of conditions for which an engine must be calibrated.
A simpler model that could aid in cylinder pressure data interpretation would be
invaluable as a support to calibration engineers.
To this end, a new methodology is proposed. This again uses CFD but concentrates
on the pressure distributions in a combustion chamber following a localized
exothermic event. The model is applied to modern engine geometry of
high-performance Normally Aspirated (NA) and turbocharged Direct Injection (DI)
gasoline engines. Output from the CFD model is compared to experimental data at
high-load and beyond borderline knock conditions. It is shown that this approach
can give new insight into experimental results interpretation and allow firmer
conclusions to be drawn on the relationship between knocking cylinder pressure
measurements and the phenomena that are driving them.
