| dc.description.abstract | Minimum Ignition Energy (MIE) is a critical dust hazard parameter guiding elimination of ignition sources in solids handling facilities. Partial inerting is an important but underutilized mitigation technique in which MIE of a dust cloud is increased through inerting, reducing the risk of an accidental dust explosion or more accurately, a dust deflagration. This dissertation has reported advances in MIE testing and prediction to prevent and mitigate dust explosions. In this work, a novel purge add-on device to the standard MIE test apparatus was designed which facilitated purging the Hartmann tube before MIE testing. Through experimentation and CFD modeling, this dissertation has attempted to refine the existing MIE testing standard for partial inerting applications by introducing purge time as an essential parameter. The effective experimental purge time required for partial inerting testing in the MIE apparatus was determined to be > 40 s and validated through the ANSYS Fluent CFD purging model. In addition, this work has demonstrated that purging the MIE apparatus Hartmann tube before experimentation significantly affected the measured values in partially inerted atmospheres (O2 < 21 vol. %). It is recommended through this research that purging should be an essential step while MIE testing and reporting. Using this improved methodology, an accurate MIE with changing oxygen concentrations for the combustible dusts Niacin, Anthraquinone, Lycopodium clavatum and Calcium Stearate was obtained and a mathematical equation for MIE-O2 was proposed.
Furthermore, Quantitative-Structure Property (QSPR) models for MIE prediction using machine learning algorithms such as Random Forests (RF) and Decision Trees (DT) were developed. A binary classification model was developed for predicting the MIE category of the combustible dusts. The results indicated good MIE predictability through the RF algorithm indicated by the Receiver Operating Characteristic – Area Under Curve (ROC-AUC) of 0.95. Additionally, RF algorithm was used to identify the molecular descriptors which most significantly affected the MIE prediction accuracy. Thus, through experimentation and modeling, this study aims to provide a scientific foundation for a partial inerting MIE test method to supplement existing testing standards (such as ASTM E2019-03) and provides a solid framework for MIE prediction of combustible dusts. | |
