Reliable prediction of the flash point of organic compounds containing hazardous peroxide functional groups as compared to the best available methods

Abstract

This work introduces a simple method for reliable prediction of the flash point of various types of organic compounds containing hazardous peroxide functional groups that are significant for safety measures in industrial processes. Different types of 116 organic peroxides including hydroperoxides, dialkyl peroxides, α-oxygen substituted alkyl hydroperoxides, and dialkyl peroxides, primary and secondary ozonides, peroxy acids, diacyl peroxides (acyl and organosulfonyl peroxides), and alkyl peroxyesters (peroxycarboxylates, peroxysulfonates, and peroxyphosphates) are used to derive the new correlation. The new model is derived by considering three important different types of effective structural parameters, which contain the number of carbon and hydrogen atoms as well as the number of hydrogen bonding groups under certain conditions and the existence of some specific molecular fragments. Various types of statistical assessments are done to confirm the high reliability of the new model. Internal and external validations of the new model confirm that it is robust. High reliability of the new model is compared with outputs of the new software (TEST software) including a variety of Quantitative Structure-Activity Relationship (QSAR) methodologies, which are based on the methods of Food and Drug Administration (FDA), as well as one of the best structural group contribution (SGC) methods. The values of Mean Absolute Percent Error (MAPE) of the new model, TEST software and SGC are 5.18, 8.54 and 12.34, respectively.