Development of a computationally efficient monolith reactor simulator: CFD-hybrid model analysis of methane oxidation monolith catalysed systems.

Abstract

The optimisation of complex geometries such as that of monolith reactors can be supported by computation and simulation. However, complex boundaries such as those found in multi-channel monoliths render such simulations of extremely high computational expense. Adding to the computational expense is the strong coupling among reaction kinetics, heat and mass transfer limitations in these channels. This severely limits the possibilities for geometric optimisation. In the first step toward developing a fast-solving hybrid simulation, a detailed CFD simulation was used to obtain the unsteady state, spatial temperature and concentration (and hence reaction rate) profiles for a range of input conditions. The results of the CFD simulation were then accepted as the benchmark to which faster-solving models were measured against to be considered as viable descriptions. A modified plug flow with effectiveness factor correction for wall mass-transfer was developed and evaluated as the first step towards the development of a multi-channel model. However, the modified plug model is only applicable to single channel monoliths and cannot account for heat transfer across high-density multi-channel beds. For multichannel simulations, the modified plug flow model is embedded into a hybrid-model framework. The hybrid model is based on the principle that, due to the high density of channels in a monolith, there will exist an equivalent homogeneous cylindrical model that approximates the behaviour of a bundle of channels acting as axial heat sources. This model entails the coupling of analytical solutions to single channel mass and momentum transfer with heat transfer across the single-shell extra-multi-channel space. Due to the application of effectiveness-factor type approaches, it is shown that the model can be represented by algebraic models that accurately represent the partial differential equations (PDEs) that describe monolith reactors. A close agreement between both temperature and species mole fraction profiles predicted from the modified plug flow model and a detailed CFD model was found with R2 values of 0.994 for temperature. The time needed to find a converged solution for plug flow model on an Intel(R) Core(TM) i5-5300U CPU @ 2.30GHz workstation was found to be 53 seconds in comparison to 1.3 hours taken by a CFD model. The hybrid model was itself validated against the CFD multichannel model. The hybrid model axial temperature and species concentration profiles at various radial positions were found to be in a close agreement with CFD simulations, with relative error found to be in the 0.35 % range. The clock time on an Intel(R) Core(TM) i5-5300U CPU @ 2.30GHz workstation was found to be 38 hours for a CFD multi-channel simulation which when compared with the 53 seconds clock time of the hybrid model implies the suitability of hybridisation for the application to geometric optimisation in the design of monolith reactors. The hybrid-model is developed to facilitate geometric optimization with the view of reducing hot spot formation, pressure drop and manufacturing costs. This is because monolith reactors applied in catalytic partial oxidation of methane are coated with precious metal catalysts, significantly contributing to capital costs. By isolating regions of high catalytic activity, it becomes possible to reduce the amount of precious metal coating required to achieve high conversion. The fast-solving hybrid model was used in the economic analysis of the catalytic partial oxidation of methane to syngas. Due to the low computational expense of the hybrid model, it was possible to investigate a wide range of design geometry and operating condition .It is shown that, for methane oxidation over a Platinum gauze catalyst, the channel diameter could be optimised to the 0.8 mm level resulting in the highest syngas revenue (R 65754.14 /day). The distribution of the catalytic material on the monolithic walls was found to influence the reactor performance hence the process profitability. The non-uniform distribution was found to significantly reduce the cost of fabrication while maintaining a high syngas productivity. In general, a method is proposed to optimise design and operation of catalytic monolith reactors through the application of fast-solving models.