A Deep Learning Approach on Surrogate Model Optimization of a Cryogenic NGL Recovery Unit Operation

Natural gas liquids (NGL) are utilized in nearly all sectors of the economy such as feedstock for petrochemical plants and blended for vehicles fuel. In this work, the operation of a cryogenic expansion unit for the extraction of NGL is optimized through the implementation of data-driven techniques. The proposed approach is based on an optimization framework that integrates dynamic process simulations with two deep learning based surrogate models. The first model utilizes a recurrent neural network (RNN) based surrogate model to disclose the dynamics involved in the process. The second regression model is built to generate profit predictions of the process. The integration of these models allows the determination of the process operating conditions that maximize the hourly profit. Results from two case studies show the capabilities of the proposed optimization framework to find optimal operating conditions and improve the process profits.