A Review of Intelligent and Predictive Control Strategies for Process Intensification in Nonlinear Acid Gas Removal Units

Acid gas removal units (AGRUs) are complex industrial processes characterized by strong nonlinearities, multivariable interactions, operational constraints, and high energy intensity, which pose significant challenges for advanced process control. Conventional propotional-integral-derivative (PID) based regulatory control often fails to maintain optimal performance under disturbances, tight product specifications, and energy efficiency demands. Although substantial research has been conducted on solvent chemistry, thermodynamics, and steady-state optimisation, control-focused developments remain fragmented and largely algorithm centered, with limited emphasis on system-level architectures and industrial deployment. This review examines intelligent and predictive control strategies for AGRUs within a hierarchical framework comprising regulatory control, model predictive control (MPC), data-driven intelligent methods such as artificial neural network (ANN) and adaptive neuro fuzzy inference system (ANFIS), and emerging hybrid model-learning approaches. Unlike previous reviews that primarily address solvent performance or process optimisation, this study critically evaluates how absorber-stripper coupling, operational constraints, and economic objectives drive the need for advanced control. The review critically evaluates the strengths, limitations, and practical barriers of these methods, supported by quantitative evidence, such as the fast model predictive control (FMPC) achieving 74.8% faster settling times and 59-fold reduction in integral absolute error (IAE) compared to classical MPC in monoethanolamine (MEA)-based CO ₂ capture plants. The review also discusses the enabling implementation ecosystem, including dynamic simulation in Aspen Dynamics, external controller development in MATLAB, and real-time integration via open platform communication (OPC). By linking process dynamics, control algorithms, and practical deployment frameworks, this study offers a unified control-engineering perspective and outlines future directions toward constraint-aware, digital twin-enabled, and self-optimizing AGRU systems for enhanced process intensification.