Molecular management in continuous catalytic reforming operations by enhanced aromatics production through transformer-driven entropy maximization reconstruction

Abstract

Catalytic Reforming (CCR) units are pivotal in refining industries for hydrogen and aromatics production. Traditional CCR optimization, focusing on operational conditions and lumping kinetics, fails to capture molecular interactions and feedstock variability, leading to inefficiencies. This research introduces a CCR optimization framework leveraging molecular reconstruction adaptable to minor feedstock changes. We propose the Transformer-driven Entropy Maximization (TREM) method, integrating historical data to improve naphtha composition accuracy. A molecular-level CCR kinetic model is then developed, linked to the TREM method, within a multi-objective optimization (MOO) framework for enhanced production adaptability. Furthermore, we introduce the Online Synchronized Learning Multi-Objective Optimization (OSLMOO) method, applying online learning to a meta-heuristic algorithm, reducing computational complexity and enhancing adaptability to property and market fluctuations. These methods collectively enhance the efficiency and profitability of CCR operations by overcoming the limitations of traditional optimization approaches.