Development of a Method for Predicting the Quality of Petroleum Coke

The development of algorithms that ensure manufacture of products of the required quality under minimal involvement of maintenance personnel is a relevant task for the oil refinery industry. Such algorithms have proven valuable in technological schemes based on continuous flow production. However, due to operational specifics, creation of such models for delayed coking units is associated with certain problems. In this work, we develop a methodology for predicting the quality of petroleum coke based on the statistical data of an industrial delayed coking unit (DCU). The input data included the technological indicators of the DCU and the quality indicators of feedstock.