Project Description:

Optimizing chemical processes is crucial to maximize yield, minimize environmental impact, and ensure cost-effective operations. Aspen Plus is the reigning process simulation software used throughout industry to design chemical process plants. However, finding the input parameters that result in the optimal outputs is challenging even for Aspen’s native optimization techniques, which struggles with complex, multi-objective problems.

To address this, we integrate Python-based algorithms like NSGA-II (for multi-objective optimization) and Bayesian Optimization (for efficient exploration of design spaces) to automate variable adjustments, run simulations, and retrieve results. Using these algorithms, we reduced the heat duty of a pressure swing system by 60,229 kcal/hour compared to previous work. At scale, this would represent a substantial amount of energy saved.

Our work has implications for simulation software beyond Aspen Plus and in other fields such as electrical and mechanical engineering by providing a framework for integration with optimization techniques in Python.

Project Photo:

The left image shows NSGA-II optimization on the Rosenbrock 2-D function, highlighting the optimal point in lime green. The right diagram illustrates our feedback loop between ASPEN Plus and Python, with which we applied NSGA-II and Bayesian optimization to minimize heat input in a pressure swing distillation simulation.