(652a) Multi-Scale Optimization of a Fixed-Bed Multi-Tubular Reactor for CO2 Methanation

The power-to-gas (PtG) technology has been known as one of the most viable solutions for storing intermittent renewable energy sources such as solar and wind to the existing gas grid. The chemical CO₂ methanation in a fixed-bed multi-tubular reactor was studied via a multi-scale optimization (MSO) approach covering an one-dimensional plug-flow reactor (1D PFR) model, a computational fluid dynamics (CFD) model, and a physical-informed neural network (PINN) model. The PINN model was composed of a feedforward neural network (NN) and physical-informed constraints where the mass and energy balances were introduced to the NN loss function using the automatic differentiation (AD) technique. The data for reaction kinetics, hydrodynamics, and heat transfer simulated by the CFD model were used for estimating the model parameter of 1D PFR via PINN. Taking the advantages of (1) 1D PFR on calculation speed, (2) CFD model on the rigorousness, and (3) PINN on the knowledge extraction from data, the MSO showed a good potential for fast and robust optimization of the fixed-bed multi-tubular CO₂ methanation process.