Effects of thermal stratification and turbulent intensity on auto-ignition and combustion mode transition

Abstract

Hydrogen engines with zero-carbon emissions have recently drawn attention while they often suffer from pre-ignition and knocking issues. It is generally understood that knocking intensity is closely associated with auto-ignition development modes, which are affected by multiple variables and their interactions. In this work, the role of thermal stratification and turbulent intensity in auto-ignition development modes were numerically investigated, addressing combustion mode transition. Representative end-gas auto-ignition conditions were employed for hydrogen/air mixtures. Meanwhile, the support vector machine (SVM) algorithm was first applied in the predictions of detonation development. The results show that thermal stratification plays a decisive role in auto-ignition development modes, and supersonic auto-ignition deflagration, direct detonation, and subsonic auto-ignition deflagration can be sequentially observed with the increase of temperature gradient. However, affected by multiple hotspot interactions and pressure wave disturbance, deflagration to detonation transition arising from secondary hotspot auto-ignition becomes prevalent. It is noted that, unlike direct detonation that evolves at the entire circumferential directions, deflagration to detonation transition develops only along certain directions with intermediate temperature gradient, and such gradient steepness is much higher than previous results based on simplified models. When turbulence fluctuation is introduced, the distribution of thermal stratification is modified significantly and the local temperature field is broken into multiple smaller regions filled with mixed temperature gradients. Consequently, deflagration to detonation transition is promoted both temporally and spatially, especially at high turbulence intensity. Besides, the SVM classifier is trained using the above results of numerical simulations for detonation predictions. A diagram consisting of the most energetic length scale and temperature fluctuation is proposed, which shows good performance in predicting detonation occurrence. The present study demonstrates that thermal stratification and turbulent intensity can significantly affect the transition behavior of auto-ignition development mode, and machine learning approaches have the potential to achieve good predictions for detonation occurrence in multiple physical fields.