Estimation of Temperature and Residence Time of Carbon Black Oil Furnace Industrial Reactors

Abstract

In the absence of powerful rigorous models, in this research a simple but practical method for calculating the temperature and residence time and carbon black yield in oil-furnace reactors is proposed. For this purpose an empirical formula of the form CxHy is assumed for the carbon black feedstock based on typical feedstock used in this industry. Based on the prevailing reactor conditions, thermodynamic considerations and available outlet tail gas from the reactor, a few representative reactions are considered to describe the entire reaction network. These include complete combustion for the fuel and, incomplete combustion and pyrolysis for the feedstock. Carbon black yield can be estimated from a mass balance. From the combined mass and heat balance, the reactor temperature as well as the residence time is predicted. The reactions can also be used to obtain the composition of the tail gas. Despite the simplicity of the model, in all cases studied, relatively good agreements were observed between the calculated values and the available industrial data as well as those obtained from computational fluid dynamics (CFD) simulation. An exception was the effects of the air/oil ratio on the reactor mean residence time in the CFD approach. The influences of air/oil ratio and inlet air temperature on reactor mean-temperature were studied. In both cases one observes good agreement between our results and CFD simulation data. They both show an increase in reactor mean-temperature as expected. The effects of process air flow rate on reactor residence time also were studied and a good agreement was observed. However, comparison the effects of the air/oil ratio on reactor residence time showed quite different trends. While the results of CFD simulation predicts insignificant change in reactor residence time versus air/oil ratio, our results show a decline in the residence time in this case. To explain this, prediction of carbon black particle size was considered. Utilizing our predicted residence time results for prediction of particle size showed a good agreement with industrial data, while in contrast the CFD simulation data show opposite trend with the industrial data and our work. Therefore, it can be said that one of the most important reasons caused poor prediction in referred CFD simulation in some cases could be considerable error in prediction of the residence time.