Reduced-order Surrogate Modeling as Foundation for Catalyzing Digital Twins in Process and Chemical Engineering
Please login to view abstract download link
Digital twins (DTs) can provide advanced, reliable solutions for monitoring and controlling chemical reactors (the "physical twins") under variable conditions, enabling safer and optimized reactor performance in real-time [1]. Our work's approach relies on computing fast, accurate models to predict optimal coolant temperature and prevent overheating. Our goal is to identify promising candidates for constructing a DT of the CO2 catalytic methanation reactor, which will be used to optimize its operation and facilitate its integration into renewable energy systems. We investigate and compare several surrogate modeling techniques: a graph attention network (GAT), operator inference (OpInf) with stability guarantees, and sparse regression with greedy sampling (SINDy) following [2]. These methods are evaluated for their ability to model the dynamic operation and control of the reactor. We compare these surrogate models' accuracy and computational efficiency to identify models that can improve renewable energy conversion processes.