Method and Tool Integration for Rigorous Design and Optimization of Multiunit Processes

This research introduces a novel approach for process optimization, unlike conventional methodologies that often overlook extensive variable selection; this approach emphasizes identifying high-impact variables (HIV) and using parallel computing for efficient simulations. This work presents the implementation of a strategy based on the optimization of 4E G (energy, efficiency, economy, and environmental) analysis of GREENSCOPE for the optimal design of the combined acetone–butanol–ethanol (ABE) and succinic acid (SA) production process from lignocellulosic biomass. The strategy includes (1) rigorous kinetic modeling of metabolic reactions within subroutines in FORTRAN and MATLAB; (2) rigorous simulation of the ABE and SA reactors in Aspen Plus; (3) selection of eight indicators of GREENSCOPE to evaluate the 4E G metrics; for the economic indicators, machine learning was implemented to estimate the economic feasibility based on Aspen Economic, thus including rigorous calculations in an efficient manner; (4) the identification of HIV using the Latin hypercube sampling method followed by a sensitivity analysis; and (5) the multiobjective metaheuristic optimization, evaluating three scenarios: HIV sets correlated with objective functions greater than 5, 10, and 20%. All simulations were executed in a parallel computing environment with seven workers. The results of the HIV analysis identified that 12 of 31 variables were relevant as decision variables. The optimization shows that the best indicator values were achieved with HIV greater than 10% correlation (using 7 of the 12 identified variables). Therefore, this strategy enables the design of multiunit processes across varying levels of complexity and scales, offering a more comprehensive and efficient approach to process design and sustainability evaluation.