Abstract
Diafiltration membrane processes are a promising technology for recovering critical minerals and materials (CMMs), but they also face challenges in that actual system performance during operation can deviate from design specifications. This requires risk reduction methods to aid in the deployment of these processes, including leveraging modeling and accounting for uncertainty during the design phase. To mitigate risks associated with membrane performance uncertainties, we applied a robust optimization approach using PyROS software to generate designs that maintain system performance under uncertain membrane sieving and flux through the membrane. We specifically show how the traditional use of a mere deterministic process model fails to protect against such uncertainties, leading to unnecessary overdesigns in membrane area and recycle pumping capacity. In contrast, we find that membrane systems designed with robust optimization achieve higher feasibility guarantees across uncertain scenarios while maintaining costs that are relatively similar to their deterministic counterparts.