Predicting IQT Ignition Delay Time of Oxygenated Fuels Using Machine Learning

Accurately predicting ignition delay time (IDT) is vital toward the development of cleaner and more efficient burning fuels. This is particularly important for advanced oxygenated fuels that contain functional groups such as alcohols, ethers, ketones, and esters. In the present work, a machine learning (ML) model was built using artificial neural networks (ANN) in order to predict IDT obtained using an ignition quality tester (IQT). The dataset used for model development consisted of 557 fuel samples, which included pure chemical compounds, blended mixtures, and commercial gasolines. Each sample’s IDT was measured using standardized procedures with the IQT. The input to the ML model was derived from the chemical composition of the fuel in the form of ten distinct functional groups, along with two structural descriptors: molecular weight (MW) and branching index (BI). The ANN achieved high predictive accuracy (R² = 0.992, average error = 0.91 ms), capturing ignition trends across alcohols, ethers, ketones, and esters. Correlation analyses showed that MW and BI have a significant positive impact on IDT, while paraffinic CH₂ content has a negative correlation. The ANN model is an effective means of predicting the IDT of oxygenated fuels based on their molecular structure, thus aiding in fuel formulation and design.