Propylene is industrially produced in a mixture with propane and generally separated from the mixture via distillation. However, because distillation is an energy-consuming process, a more efficient separation process should be developed to mitigate both carbon dioxide (CO₂) emissions and production costs. In this study, a two-stage membrane-separation process was designed, and its CO₂ emission and production costs were evaluated. The separation processes were designed to minimize energy consumption using different membrane combinations (two recently developed membranes each). To evaluate the separation processes using various membrane combinations, two indicators, i.e., CO₂ emissions and total annual costs (TACs), were estimated based on the process simulation (Pro/II, version 10.1.1) results, including energy consumptions, operation expenditure, and capital expenditure. These results were compared to the distillation processes as benchmarks, and the advantages of the membrane-separation process were discussed. In the comparison, carbon taxes were implemented for assessing these two independent indicators as a single indicator, i.e., TAC with carbon tax. Furthermore, using the same scheme, model membranes were also employed in the two-stage membrane-separation process as case studies of technological forecasts.

Various factors need to be considered in process design optimization to implement the complex processes of CO₂ capture, utilization, and storage (CCUS). Here, bi-objective optimization of single-stage CO₂ membrane separation was performed for two evaluation indexes: cost and CO₂ emissions. During optimization, the process flow configuration was fixed, the membrane performance was set under the condition of the Robeson upper bound, and the membrane area and operating conditions were set as variables. Bi-objective optimization was performed using an original algorithm that combines the adaptive design of experiments, machine learning, a genetic algorithm, and Bayesian optimization. Five case studies with different product CO₂ purities in the constraint were analyzed. Pareto solutions were superior for case studies with lower product CO₂ purities. The set of Pareto solutions revealed opposite directions for optimization: either (1) increase the membrane area to reduce CO₂ emissions but increase costs or (2) increase power consumption and reduce costs but increase CO₂ emissions. The implemented bi-objective optimization approach is promising for evaluating the membrane CO₂ capture process and the individual processes of CCUS.

To effectively implement complex CO₂ capture, utilization, and storage (CCUS) processes, it is essential to optimize their design by considering various factors. This research bi-objectively optimized a two-stage membrane-based separation process that includes recycling, concentrating on minimizing both costs and CO₂ emissions. The implemented algorithm combined experimental design, machine learning, genetic algorithms, and Bayesian optimization. Under the constraints of a recovery rate of 0.9 and a produced CO₂ purity of 0.95, six case studies were conducted on two types of membrane performance: the Robeson upper bound and a tenfold increase in permeability. The maximum value of α*(CO₂/N₂), used as a constraint, was adjusted to three levels: 50, 100, and 200. The analysis of the Pareto solutions and the relationship between each design variable and the final evaluation index indicates that electricity consumption significantly impacts operating costs and CO₂ emissions. The results of the case studies quantitatively clarify that improving the α*(CO₂/N₂) results in a greater enhancement of process performance than increasing the membrane’s performance by increasing its permeability. Our bi-objective optimization analysis allowed us to effectively evaluate the membrane’s CO₂ separation and individual CCUS processes.