Reinforcement learning for chemical engineering process design with Aspen Simulator. This is for the paper here by Stephan C.P.A. van Kalmthout.

Ultimately, we found Aspen simulator to be too slow to be a good option for RL. For other work on the topic of RL for chemical engineering process synthesis see the RL for process design repo and the Distillation Train Gym repo.

Our Aspen-Python API (see here) may be of interest for those who wish to control Aspen simulations via python code. But be warned - this is not recommended as it is slow.