(705a) Optimization of Sustainable Processes Incorporating Data Envelopment Analysis

Developing effective tools for process design is of paramount importance in the transition towards sustainable systems. In this work, we present a framework to incorporate sustainability principles in the design of chemical processes which combines a palette of tools, including life cycle assessment, surrogate modeling, objective reduction, multi-objective optimization, and multi-criteria decision analysis tools. In particular, we propose the use of Data Envelopment Analysis (DEA) to facilitate the post-optimal analysis of the Pareto front generated during the optimization. The use of DEA in the framework allows to narrow down the number of designs and rank them without the need to define weights in an explicit manner. Additionally, DEA also provides improvement targets for the sub-optimal alternatives that if attained would make them optimal. This characteristic of DEA can help to guide retrofit efforts when benchmark processes are compared against the best technologies available. The application of all the stages of the framework is demonstrated in the hydrogenation of CO₂ to produce methanol, where the process was initially modeled using a commercial process simulator to be further replaced by a surrogate model applying Artificial Neural Networks. The sustainability of the process was measured with the environmental categories included in the CML-2001 methodology and the total annualized cost. The objective-reduction stage narrowed down the number of total objectives to evaluate from 11 to five and the optimization was performed using a multi-objective Genetic Algorithm. The final application of DEA to analyze the solutions obtained from the optimization presented only 9 designs as the best options out of the 126 solutions that conformed the Pareto front. Finally, these 9 solutions were ranked and the best design identified. Overall, the methodology allowed an entire assessment of the process without compromising the performance of any sustainable indicator while reducing the computational resources used in the evaluation. The development of contributions such as the one presented in this work aim to facilitate the sustainable development of the chemical industry, not only by assessing new processes and technologies but also by identifying improvement targets for current sub-optimal ones.