Abstract
This research presents a novel framework for the surrogate-based dynamic optimization of control schemes within chemical separation and purification processes such as the biorefinery downstream processing. The current study investigated the downstream of an enzymatic bioreactor responsible for the synthesis of 5-hydroxymethylfurfural value-added derivatives, focusing on the critical balance between operational costs and productivity. Two high-fidelity long short-term memory neural network-based surrogate models were developed to predict energy consumption and economic gain, both achieving a coefficient of determination (R2) exceeding 0.97. These models were subsequently integrated into a multi-objective optimization architecture to address an operating efficiency testing scenario characterized by stepwise inflow parameter changes. By exploring the resulting Pareto front, an optimal set of operational (control) settings was identified and validated. The results demonstrate that while energy consumption remained nearly constant, the total economic benefit increased by approximately 20% over the whole studied timeframe. Critically, during the transient period between steady-states, the economic benefit surged by nearly 60%, highlighting the potential of surrogate-based dynamic optimization. The high computational efficiency of the developed models (optimal solution obtained in less than three minutes) suggests significant real-time process control application possibilities. Unlike computationally expensive first-principles models, the developed neural architectures allow for virtually instantaneous setpoint adjustments. This paves the way for their integration into model predictive control or real-time process control layers, enabling industrial plants to respond adaptively to feed fluctuations, minimize off-spec product during transitions, and maximize profitability.