- Worldwide production of natural gas has surged in the last few decades leading to an economic windfall for many communities, but sometimes these new sources of natural gas are located in geographically isolated locations. These areas are so remote that it isn’t economically feasible to transport this stranded gas to market; instead, novel technologies are required to process it on site and convert it to more valuable products. One path to accomplishing this is to utilize micro-scale reactors which offer radically smaller footprints, intensification of mass and heat transportation, and large surface area to volume ratios. Microreactors can be designed to accommodate catalytic processes such as steam reforming of methane or Fischer-Tropsch synthesis (FTS) and plasma processes such as methane activation in corona discharge plasmas. These kinds of processes can convert methane into more valuable products such as longer chain hydrocarbons or olefins. A key to successfully designing these microreactors is to first model their fundamental processes and examine how different conditions affect the product dissemination.
This work primarily consists of two mathematical models, the first for a Fischer-Tropsch synthesis catalytic microreactor and the second a corona discharge plasma microreactor. Both models are described as laminar flow in a rectangular duct, and as ideal gases with isotropic diffusion in the bulk phase. Steady-state mass balances are performed for all reactants and products present in the reactor. Chemical reactions are broken down to elemental-steps, with each step described by first-order kinetics. Additionally, the FTS model treats catalytic sites as chemical species that participate in the reactions while the corona plasma model does the same for free electrons in the plasma phase. Reaction rate constants are modelled using Arrhenius kinetics, with the exception of the electron impact dissociation reaction rate constants in the corona plasma model which have a unique form. All work was done using the COMSOL Multiphysics® software which uses numerical approximation techniques to solve the coupled, partial differential equations which arise from evaluating the mass balances.
Results from both models were split into two sections consisting of a detailed analysis of a base or optimal case of operating parameters and a parametric sweep of important variables. The optimal conditions of the FTS model were selected to be 240 °C, 20 bar, feed ratio of 2:1 H2/CO, and a residence time of 10 seconds. For these conditions, the carbon monoxide conversion was found to be 86.8% and the hydrocarbon concentrations peaked with nonane/decane. Only a small amount of carbon dioxide was produced, which is consistent with the literature. The FTS parametric sweep revealed that residence time was strongly correlated with higher conversions and weakly correlated with a tendency to shorter hydrocarbon chains, that higher pressure was directly correlated with conversion up to 20 bar and had no correlation to hydrocarbon distribution, that higher H2/CO ratios had no correlation with conversion and was strongly correlated with shorter hydrocarbons, and that higher temperature was weakly inversely correlated with temperature and correlated with a tendency to longer hydrocarbon chains. All of these results were comparable to literature trends, with the important exception of the hydrocarbon selectivity at higher temperatures which was the opposite, most likely due to the model failing to capture the correct diffusion rates of adsorbed species on the catalyst surface. The optimal conditions for the corona plasma model were selected to be 1 bar, 5 mA of current, 2 kV across the gap, feed ratio of 1:3:6 N2/CH4/CO2, and a residence time of 2 milliseconds within the plasma flame. These conditions yielded a methane conversion of 18.5%, which compares favorably to a similar real world reactor’s conversion of 15.3 ± 4.2%. Selectivity to C2+ products was 0.875 in the model, much higher than the 0.446 ± 0.126 measured in the real world reactor. This difference was driven by the formation of carbon monoxide in the reactor, which the real world reactor had considerably more of, likely due to a catalytic reaction taking place at the tungsten electrodes. The corona parametric sweep showed that higher residence times correlated to higher conversions and increased selectivity towards ethylene and acetylene, but no correlation on C2+ selectivity more generally. Increasing the pressure in the corona from 1 to 2 bar led to modest increased in conversion and selectivity, but pressures beyond 2 bar showed no change. Higher methane feed concentrations were negatively correlated with conversion, but showed increases in selectivity indicating that methane was the limiting reagent at these conditions. Changes in the feed concentrations of carbon dioxide or nitrogen showed no significant correlation with either conversion or selectivity. Increasing current correlated to higher conversions up until 2.5 mA where it leveled off, and it also correlated to higher C2+ selectivities up until 1 mA after which is began to decrease. Higher voltages correlated to higher conversions and higher acetylene selectivity.