A Two-Stage Knock Model for the Development of Future SI Engine Concepts 2018-01-0855

At specific operating conditions, the auto-ignition in the unburnt mixture that precedes the occurrence of knock in conventional SI engines happens in two stages. In a previous publication, the authors demonstrated that the low-temperature heat release significantly influences the auto-ignition behavior of the mixture, thus severely impairing the prediction capabilities of the Livengood-Wu integral that the majority of the commonly used 0D/1D knock models are based on. Consequently, a new two-stage auto-ignition prediction approach for modeling the progress of the chemical reactions was introduced. It was demonstrated that the proposed auto-ignition model predicts the occurrence of two-stage ignition and accurately considers the significant influence of low-temperature heat release on the mixture’s auto-ignition behavior at various operating conditions. However, the correct prediction of local auto-ignition is not sufficient for the reliable calculation of the knock boundary, as the occurrence of this phenomenon does not necessarily result in knock. Based on the proposed two-stage auto-ignition prediction approach, this paper presents a new knock model for the development of future SI engine concepts in a 0D/1D simulation environment. In addition to not considering low-temperature ignition, the commonly used knock models based on the Livengood-Wu integral assume that no knock can occur after a pre-defined, constant MFB-point. The evaluation of measured knocking single cycles however has revealed that the latest possible MFB-point where knock can occur changes significantly with parameters such as engine speed and EGR rate. Hence, a cycle-individual criterion for the occurrence of knock accounting for the current operating conditions is needed. To this end, an approach based on the unburnt mass fraction in the cool thermal boundary layer at the time of auto-ignition is proposed. Besides the operating conditions, this knock occurrence criterion also considers the flame propagation and the cylinder geometry. Additionally, the already published submodels integrated into the two-stage auto-ignition prediction approach are expanded by the influence of injected water. Thus, as the new knock model also accounts for EGR, fuel composition and properties as well as air-fuel equivalence ratio effects, it fulfills all requirements for the simulation of future engine concepts. The approach contains no empirical measurement data fits and has just one calibration parameter that does not change with the operating conditions. Finally, model validation against measurement data on a handful of different engines at various operating conditions is performed. It is demonstrated that the new model can estimate the knock boundary very accurately with errors in the predicted center of combustion below 2°CA and thus contributes to the cost-effective development of future SI engine concepts in a 0D/1D simulation environment.