An Artificial Intelligence Method for Estimation of Optimum Operating Conditions of a Furnace under Uncertainty

Abstract

Furnaces have been known as the central preliminary units employed for hydrocarbon processing in petrochemical industries. The most significant energy consumption part in refineries is associated with furnace units. Therefore, achieving higher thermal efficiencies is the primary concern in designing and during operation. Due to significant energy consumption in such systems, a minor improvement in thermal efficiency would lead to considerable savings. Much work has been done in the literature on the design and optimization of the furnace using different optimization methods. However, no one has focused on the optimization of the furnace under uncertain process conditions. The current work developed an Integrated Framework of Artificial Intelligence and Genetic Algorithm for the furnace of a petroleum refinery to predict the optimum amount of excess air and mass flow rates of crude oil and fuel stream in the presence of uncertainty in process conditions. Using optimized industrial data, a furnace model was regenerated in Aspen EDR. The COM server was used to build the interface between Aspen HYSYS and MATLAB. The data set was generated by inserting the variation of ±1, ±2, ±3, ±4, and ±5 in the crude oil composition as well as in the inlet temperature and pressure of cold crude oil, fuel, and air stream. The optimum amount of excess air and mass flow rates for each variation was determined using a single objective genetic algorithm. A total of 360 data points were generated. 70% were used for the Feedforward neural network (ANN) and the remaining data points were equally divided for the validation and testing of the model. The proposed artificial neural network (ANN) model achieved a correlation coefficient value of 0.99984. The high accuracy and robustness of the ANN model make it suitable for real-time industrial applicationsto reduce energy consumption.