The Study Of Gas Liquified Recovery Plant From Exergy Point Of View: ANN-ACOR Optimisation Method

This study introduces a hybrid modeling approach combining ant colony optimization (ACOR) with back propagation neural networks (BP ANN) to predict the exergy efficiency of a natural gas liquids (NGL) recovery plant under various operating conditions. Siri Island Gas Gathering and NGL Recovery Plant in Iran serves as the case study. Thermodynamic properties, mass, and energy balances of the process streams were determined using the ASPEN Plus process simulator, and the results were validated with industrial data. Exergy destruction and exergetic efficiency were calculated for both the main components and the entire system. Data from recorded and simulated cases were then utilized as input for the neural network. The hybrid ACOR-BP approach integrates the global search capabilities of ACOR with the local optimization strengths of the gradient descent method. Each initial point for the neural network is chosen using a standard ant colony optimization process, while the ACOR's fitness values are evaluated via the neural network. The findings demonstrate the effectiveness of ACOR in optimizing neural network performance, significantly enhancing the prediction of overall exergy efficiency.