Multi-speciation in shock tube experiments using a single laser and deep neural networks

Abstract

Chemical kinetic experiments involving the oxidation or pyrolysis of fuels can be complex, especially when multiple species are formed and consumed simultaneously. Therefore, a diagnostic strategy that enables fast and selective detection of multiple species is highly desirable. In this work, we present a mid-infrared laser diagnostic that can simultaneously detect multiple species in high-temperature shock-tube experiments using a single laser. By tuning the wavelength of the laser over 3038 – 3039.6 cm⁻¹ wavelength range and employing a denoising model based on deep neural networks (DNN), we were able to differentiate the absorbance spectra of ethane, ethylene, methane, propane, and propylene. The denoising model is able to clean noisy absorbance spectra, and the denoised spectra are then split these into contributions from evolving species using multidimensional linear regression (MLR). To the best of our knowledge, this work represents the first successful implementation of time-resolved multispecies detection using a single narrow wavelength-tuning laser. To validate our methodology, we conducted pyrolysis experiments of ethane and propane. The results of our experiments showed excellent agreement with previous experimental data and chemical kinetic model simulations. Overall, our diagnostic strategy represents a promising approach for detecting multiple species in high-temperature transient environments.