An optimization model for the market-responsive operation of naphtha cracking process with price prediction

This study presents a new optimization-based framework for the optimal operation of the naphtha cracking process under uncertain market conditions. Two distinct decision models are embedded in the proposed framework: (i) product price prediction based on the machine learning techniques under market condition uncertainty for sales strategy, and (ii) mathematic model of naphtha cracking model for production planning strategy, both of which aim to maximize operating profit in the final optimization model. Based on the predicted price and the unit operation principles, the optimization model determines the major operation conditions of the naphtha cracking process. In particular, different price prediction models (multiple linear regression, artificial neural networks, and system dynamics) are compared to secure precise product prices. The rigorous mathematical model is developed by considering transport phenomena, thermal reaction, and coking effect and validated compared to real industrial data. As a result, the annual profit of the naphtha thermal cracking is expected to be 305,154 USD/year, which is 7.11% better than the reference case. This study could also be used as an auxiliary indicator for decision makers or businesses to identify production planning (i.e., optimal process operating conditions) and sales strategy (i.e., selling timing and quantity) in order to improve profits.