Prediction of dibenzothiophene conversion in the ultrasound assisted oxidative desulfurization process

ABSTRACT

In order to produce ultra-low sulfur diesel, ultrasound-assisted oxidation desulfurization of dibenzothiophene (DBT) was carried out with acetic acid and hydrogen peroxide. Due to its complexity, ultrasound-assisted oxidation process lacks a precise analytical solution. This paper explores the application of linear multiple regression and neural network for the prediction of dibenzothiophene conversion. Models were employed with respect to hydrogen peroxide dosage, temperature, reaction time, initial DBT concentration, and rate constant. The most accurate results were achieved by neural network model. Developed models facilitate future research in terms of better understanding the influence of process conditions of DBT conversion.

KEYWORDS:

• Desulfurization
• kinetics
• linear multiple regression
• neural network
• ultrasound