Hydrogen has received substantial attention because of its diverse application in the energy sector. Steam methane reforming (SMR) dominates the current hydrogen production and is the least expensive endothermic reaction to produce grey hydrogen. This technology provides the advantages of low cost and high energy efficiency; however, it emits an enormous amount of CO₂. Carbon capture storage (CCS) technology helps reduce these emissions by 47% to 53%, producing blue hydrogen. Methane pyrolysis is an alternative to SMR that produces (ideally) CO₂-free turquoise hydrogen. In practice, methane pyrolysis reduces CO₂ emissions by 71% compared to grey hydrogen and 46% compared to blue hydrogen. While carbon dioxide emissions decrease with CCS, fugitive methane emissions (FMEs) for blue and turquoise hydrogen are higher than those for grey hydrogen because of the increased use of natural gas to power carbon capture. We undertake FMEs of 3.6% of natural gas consumption for individual processes. In this study, we also explore the utilization of biogas as a feedstock and additional Boudouard reactions for efficient utilization of solid carbon from methane pyrolysis and carbon dioxide from biogas. The present study focuses on possible ways to reduce overall emissions from turquoise hydrogen to provide solutions for a sustainable low-CO₂ energy source.

In this study, we develop physics-informed neural networks (PINNs) to solve an isothermal fixed-bed (IFB) model for catalytic CO₂ methanation. The PINN includes a feed-forward artificial neural network (FF-ANN) and physics-informed constraints, such as governing equations, boundary conditions, and reaction kinetics. The most effective PINN structure consists of 5–7 hidden layers, 256 neurons per layer, and a hyperbolic tangent (tanh) activation function. The forward PINN model solves the plug-flow reactor model of the IFB, whereas the inverse PINN model reveals an unknown effectiveness factor involved in the reaction kinetics. The forward PINN shows excellent extrapolation performance with an accuracy of 88.1% when concentrations outside the training domain are predicted using only one-sixth of the entire domain. The inverse PINN model identifies an unknown effectiveness factor with an error of 0.3%, even for a small number of observation datasets (e.g., 20 sets). These results suggest that forward and inverse PINNs can be used in the solution and system identification of fixed-bed models with chemical reaction kinetics.

This review covers the scope of multiscale computational fluid dynamics (CFD), laying the framework for studying hydrodynamics with and without chemical reactions in single and multiple phases regarded as continuum fluids. The molecular, coarse-grained particle, and meso-scale dynamics at the individual scale are excluded in this review. Scoping single-scale Eulerian CFD approaches, the necessity of multiscale CFD is highlighted. First, the Eulerian CFD theory, including the governing and turbulence equations, is described for single and multiple phases. The Reynolds-averaged Navier–Stokes (RANS)-based turbulence model such as the standard k-ε equation is briefly presented, which is commonly used for industrial flow conditions. Following the general CFD theories based on the first-principle laws, a multiscale CFD strategy interacting between micro- and macroscale domains is introduced. Next, the applications of single-scale CFD are presented for chemical and biological processes such as gas distributors, combustors, gas storage tanks, bioreactors, fuel cells, random- and structured-packing columns, gas-liquid bubble columns, and gas-solid and gas-liquid-solid fluidized beds. Several multiscale simulations coupled with Eulerian CFD are reported, focusing on the coupling strategy between two scales. Finally, challenges to multiscale CFD simulations are discussed. The need for experimental validation of CFD results is also presented to lay the groundwork for digital twins supported by CFD. This review culminates in conclusions and perspectives of multiscale CFD.