Unsteady flows in IC engines are being extensively simulated with quasi-lD models. Such models are, in particular, well suited for multi-parameter optimization. Unfortunately, their fundamental assumptions often lead to unsatisfactory accuracy, and often there are no rational ways to improve models without usage of measured data. Hence there is a strong need in techniques that rationalize ways to utilize measurements results to ensure better model accuracy.
Main principle to build such a technique upon is the possibility to calibrate model by adjusting its free parameters. Rational quasi-1D model's free parameters are found only in the empirical relations used in sub-models for closure. In class of two-stroke engines, sub-model of scavenging affects both flow rate and indicated power output. Through parametric identification of this sub-model the overall predicting capability of quasi-1D model can probably be substantially improved for a range of design parameters (those not sensibly affecting the scavenging flow pattern).
In this study such a technique is applied to naturally aspirated crankcase scavenged two-stroke engine with tuned pipe exhaust. At first the measured cylinder pressure is used to identify model parameters mainly for a combustion phase of process. Computed indicated power output and especially air flow rate at this step substantially deviate from the values measured at full load. At the next step the parameters of scavenging characteristic are optimally adjusted. Compared to the previous step, accuracy has been substantially improved. Finally, the predictive capability of calibrated model is verified. Simulation results for the same engine with differently sized tuned pipe revealed essentially the same level of deviations from measured data.
These results lead to conclusion that the proposed simulation technique offers substantial improvement in accuracy of quasi-1D models for tuned pipe two-stroke engines.