Thorough assessment of delayed coking correlations against literature data: Development of improved alternative models

The predictability of the existing and some improved correlations were evaluated against the available largest dataset for the prediction of the product yields from delayed coking, such as, coke, liquid, gas, gas oil, naphtha, heavy gas oil, liquid sulfur content as well as API vs. CCR. Except for some cases where the relationships of Volk et al. and Castiglioni and the simplistic models of Maples, and Gary-Handwerk predicted somewhat appropriately, the existing models failed in most of the cases. The alternative models contained seven independent variables including three feedstock properties and four operating conditions. Overall, the developed correlations accounted for higher than 86% of the variances. The quality of the regressions followed the order of naphtha