ESTIMATION OF FUEL QUALITY USING STATISTICAL REGRESSION-BASED ANALYSIS OF LEIDENFROST DROPLETS

摘要

Accurate measurement of fuel quality is critical for automotive applications as it impacts engine performance and emissions. A number of techniques have been proposed to measure fuel quality including acoustic wave speed sensors, chemometric modeling, near-infrared spectrophotometry, Raman spectroscopy etc. All of these techniques are complex in nature and require expensive equipment. In contrast, we propose a novel, simple and rapid method for estimating fuel quality in field environments. This involves measuring the evaporation time of fuel droplets in the Leidenfrost state, and using the results of statistical data analysis conducted on an experimental data bank comprising evaporation time data for similar droplets. The Leidenfrost state refers to a liquid droplet hovering on its own vapor layer on a superheated surface. To showcase our approach, evaporation time was measured for droplets consisting of isopropyl alcohol (IPA) and water blends with varying parameters such as IPA fraction (0-1), droplet volume (20-1 00μL) and surface temperature (200-340 °C). The resulting data bank (96 data points) was used to train and evaluate the performance of a polynomial regression (using a semi-logarithmic transformation) model in predicting the evaporation time as a function of the above-mentioned parameters. R~2 accuracies of 97.34% (training data), 96.82% (test data), 97.46% (total) and a relative error within ± 0.25% for the entire dataset was obtained using the regression model, which highlights the predictive capabilities of our approach.