Enhancing Autoignition Characteristics: A Framework to Discover Fuel Additives and Making Predictions Using Machine Learning

TL;DR

This paper introduces a hybrid framework combining chemical kinetics and machine learning to predict the autoignition delay time of biofuel n-butanol with various fuel additives, aiming to accelerate fuel additive discovery.

Contribution

The work presents a novel hybrid approach that leverages machine learning and chemical kinetics for accurate prediction of autoignition properties of biofuels with additives.

Findings

High prediction accuracy for Ignition Delay Time (IDT) with unseen additives

Demonstrated potential of ML to explore chemical mechanisms for fuel additive development

Framework effectively integrates chemical kinetics and machine learning for autoignition prediction

Abstract

Combustion process can become more energy efficient and environment friendly if used with appropriate fuel additive. Discovery of fuel additive can be accelerated by applying hybrid approach of using of chemical kinetics and Machine Learning (ML). In this work, we present a framework that takes the robustness of Machine Learning and accuracy of chemical kinetics to predict the effect of fuel additive on autoignition process. We present a case of making predictions for Ignition Delay Time (IDT) of biofuel n-butanol ($C_4H_9OH$) with several fuel additives. The proposed framework was able to predict IDT of autoignition with high accuracy when used with unseen additives. This framework highlights the potential of ML to exploit chemical mechanisms in exploring and developing the fuel additives to obtain the desirable autoignition characteristics.