Analysis of Cyclic Variability in a Bi-Fuel Engine by Means of a ‘Cycle-Resolved’ Diagnostic Technique

The paper investigates cyclic variability in a fast-burn engine running on both gasoline or CNG by applying a new diagnostic technique based on a quasi-dimensional multizone model. Two different procedures were proposed for the ‘cycle-resolved’ calibration of the heat transfer correlation in the multizone model. The first procedure relates the cycle-resolved unreleased energy of the charge at the end of the flame propagation to the combustion efficiency determined from the average exhaust gas composition. The second procedure evaluates the coefficient in the heat transfer correlation through the application of the overall energy balance to the ensemble-cycle combustion and keeps them unchanged for all cycles. Both methods gave similar results, though the second procedure showed to be more physically consistent and in better agreement with the experimental results reported in the literature. The experimental matrix covered different engine speeds (n = 2000–4600 rpm), loads (bmep = 200–790 kPa), relative air-fuel ratios (RAFR = 0.80–1.60) and spark advances (SA ranging from 8 deg retard to 2 deg advance from MBT), for both CNG and gasoline operations, 100 consecutive in-cylinder pressure cycles were analyzed for each point in the test matrix and the sensitivity to cyclic variability of pressure, burn-rate and flame front position related parameters was analyzed. Main results showed that maximum pressure derivative, delay from SA of detected combustion start, NO exhaust concentration and maximum burning speed were the most sensitive parameters to cyclic variability. Strong correlations were found to hold between PFP and burned-gas temperature peak value, as well as between peak values of HRR and burning speed. On the contrary, some seemingly reasonable correlations were not assessed: for example, delay from SA of detected combustion start is related neither with PFP value nor with combustion duration. Finally, the results from mean cycle and cycle-resolved calculations were compared. Though they were usually in good agreement, in the case of NO emission and combustion interval calculation. cycle-resolved approach results in improved accuracy.