Reinforcement learning at the Distillation Gym

AI proof of concept from Cambridge University showcases Cape Open simulation of hydrocarbon processing.

Speaking at the 2020 CAPE-OPEN 2020 Annual Meeting, Laurence Midgley (University of Cambridge) presented a paper on his ‘Distillation Gym’, an application of reinforcement learning (a branch of artificial intelligence) in chemical engineering. The Distillation Gym is an AI agent that designs processes using the COCO simulator*. A Python wrapper controls the computing.

Reinforcement learning is a novel approach to chemical engineering process synthesis with the potential to be applied to more open-ended design problems than conventional computer-aided techniques. In RL for process synthesis, the environment is the simulator (e.g. COFE, Aspen Plus), the RL agent is the process designer and the reward is the objective function (e.g. profit).

Midgley’s simple proof of concept covered a reinforcement learning agent that optimizes the design of a hydrocarbon distillation column train simulated with the CAPE-OPEN Flowsheet Environment, COFE and ChemSep. Also presented is the application of RL to simulation in general and ‘how CAPE-OPEN may facilitate such applications’.

* Cape Open to Cape Open simulation environment.

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.