Numerical study of auto-ignition propagation modes in toluene reference fuel–air mixtures: Toward a better understanding of abnormal combustion in spark-ignition engines

Abstract

Two main abnormal combustions are observed in spark-ignition engines: knock and lowspeed pre-ignition. Controlling these abnormal processes requires understanding how auto-ignition is triggered at the "hot spot" but also how it propagates inside the combustion chamber. The original theory regarding the auto-ignition propagation modes was defined by Zeldovich and developed by Bradley who highlighted different modes by considering various hot spot characteristics and thermodynamic conditions around the hot spot. Two dimensionless parameters (ε, ξ) were then defined to classify these modes and a so-called detonation peninsula was obtained for H 2 -CO-air mixtures.Similar simulations as those performed by Bradley et al. are undertaken to check the relevancy of the original detonation peninsula when considering realistic fuels used in modern gasoline engines. First, chemical kinetics calculations in homogeneous reactor are performed to determine the auto-ignition delay time τ i , and the excitation time τ e of E10-air mixtures in various conditions (calculations for a RON 95 TRF surrogate with 42.8% isooctane, 13.7% n-heptane, 43.5% toluene, and using the LLNL kinetic mechanism considering 1388 species and 5935 reactions). Results point out that H 2 -COair mixtures are much more reactive than E10-air mixtures featuring much lower excitation times τ e . The resulting maximal hot spot reactivity ε is thus limited which also restrains the use of the detonation peninsula for the analysis of practical occurrences of auto-ignition in gasoline engines.The tabulated (τ i , τ e ) values are then used to perform 1D LES of auto-ignition propagation considering different hot spots and thermodynamic conditions around them. The detailed analysis of the coupling conditions between the reaction and pressure waves shows thus that the different propagation modes can appear with gasoline and that the original detonation peninsula can be reproduced, confirming for the first time that the propagation mode can be well defined by the two non-dimensional parameters for more realistic fuels. KeywordsAuto-ignition, Deflagration, Detonation, Fuel Reactivity, LES first observations of low-speed pre-ignition (LSPI) date back to the beginning of years 2000 [2]. However, in both cases, the triggering and the development of the abnormal combustion process rely on the auto-ignition characteristics of the air/fuel mixture. In order to control these abnormal phenomena, it is necessary not only to better understand how and when an auto-ignition can be triggered by "hot spots", but also how it will propagate inside the combustion chamber since the auto-ignition intensity and the potential resulting engine damages are linked to both aspects. Different approaches such as the Livengood-Wu integral can be used to predict the autoignition temporal onset. However, advanced tools and methodologies are still being developed to better understand and predict the auto-ignition propagation modes. The original theory regarding the auto-ignition propagation mode was p...

Detonation Peninsula for TRF-Air Mixtures: Assessment for the Analysis of Auto-Ignition Events in Spark-Ignition Engines

“…2) around the hot spot. This behavior is related to the specific auto-ignition delay variation as a function of the temperature for TRF-air mixtures and was not reported so far in the case of H 2 -CO-air mixtures because the maximal  value for these mixtures is much higher (around 100 for the same initial conditions as Figure 3 of [15]).…”

“…This article aims at confirming the relevancy of the original detonation peninsula when considering realistic fuels used in modern gasoline engines based on similar simulations as those performed by Bradley et al. The detonation peninsula position has already been discussed during the last Conference on Modeling and Diagnostics for Advanced Engine Systems (COMODIA, Okayama, July 2017) and then published in the International Journal of Engine Research [15] with only an initial condition for P 0 and T 0 . This new study provides further analyses and presents the latest results of the authors.…”

“…This topic has already been discussed in [15] but Figure 3 shows for other initial conditions, the variations of these parameters as a function of the temperature increase at the center of the hot spot T 0 for three hypotheses: when the reference temperature is taken at the outer limit of the hot spot (assumption n°1, r = r 0 ), at the middle of the hot spot (assumption n°2, r = r 0 /2), and at the center of the hot spot (assumption n°3, r = 0). Even if thermal initial conditions are really different, conclusions are the same as in [15]. For the highest T 0 , the estimation of the two parameters varies a lot, meaning that for the same initial conditions, the position on the detonation peninsula is totally different.…”

“…This methodology is too CPU time consuming as the objective is to analyze a large number of operating conditions to precisely define the detonation peninsula. The tabulated model TKI-LES model [15][16][17][18][19][20][21][22] has thus been chosen to simulate autoignition as previous studies have already shown its ability to catch such phenomenon [23]. This model is based on a look-up table of  i and  e , obtained using a priori calculations for the same surrogate fuel in homogeneous reactors and considering the LLNL kinetic mechanism with 1388 species and 5935 reactions [24].…”

Special Issue on COMODIA 2017

“…The first two articles investigate the mechanism of auto-ignition by means of large eddy simulation (LES). The first paper by Robert et al 1 describes a numerical procedure to simulate autoignition originating from a one-dimensional thermal hotspot. By combining chemical kinetics calculations and the LES model, all the possible propagation modes can be reproduced with a toluene reference fuel (TRF) surrogate.…”