Search

The results of the experimental analyses, described in Part 1, are here employed to build up an innovative numerical approach for the 1D modeling of combustion, cycle-by-cycle variations and knock of a high performance 12-cylinder spark-ignition engine. The whole engine is schematized in detail in a 1D framework simulation. Proper “in-house developed” sub-models are used to describe the combustion process, turbulence phenomenon, cycle-by-cycle variations (CCV) and knock occurrence. In a first stage, the engine model is validated in terms of overall performance parameter and ensemble averaged pressure cycles, for various full and part load operating points and spark timings. Then, the correlation regarding the maximum in-cylinder pressure distribution developed in Part 1 is here applied to predict representative faster-then-average and slower-than-average cycles, miming the effects of the experimentally observed CCV. A proper knock index is introduced and evaluated with reference to the above faster-than-average cycle. An automatic procedure is implemented to identify the Knock Limited Spark Advance (KLSA), based on the same threshold level utilized in the experimental knock analysis of Part 1. The numerical and experimental KLSA presents an excellent agreement, denoting the accuracy of the proposed combustion and knock modeling.

In this paper, a high performance V12 spark-ignition engine is experimentally investigated at test-bench in order to fully characterize its behavior in terms of both average parameters, cycle-by-cycle variations and knock tendency, for different operating conditions. In particular, for each considered operating point, a spark advance sweep is actuated, starting from a knock-free calibration, up to intense knock operation. Sequences of 300 consecutive pressure cycles are measured for each cylinder, together with the main overall engine performance, including fuel flow, torque, and fuel consumption. Acquired data are statistically analyzed to derive the distributions of main indicated parameters, in order to find proper correlations with ensemble-averaged quantities. In particular, the Coefficient of Variation (CoV) of IMEP and of the in-cylinder peak pressure (pmax) are correlated to the average combustion phasing and duration (MFB50 and Δθb), with a good coefficient of determination. In addition, a high-pass-filtering technique is used to derive the cycle-by-cycle scattering of the Maximum Amplitude of Pressure Oscillation (MAPO) index. A similar statistical analysis is carried out to derive the log-normal distributions of the MAPO index and a methodology to asses a proper knock threshold is applied to identify the presence of knocking combustion. The above data represent the prerequisites for the implementation and validation of an advanced 1D model, described in Part 2, taking into account cycle-by-cycle variations, and finally aiming to identify the knock-limited spark advance on a completely theoretical basis.