The application of Buckingham π theorem to Lattice-Boltzmann modelling of sewage sludge digestion. (15th September 2020)

• Record Type: Journal Article
• Authors: Dapelo, Davide; Trunk, Robin; Krause, Mathias J.; Cassidy, Nigel; Bridgeman, John
• Is Part Of: Computers & fluids. Volume 209(2020)
• Publication Date: 2020-09-15
• Subjects: Anaerobic digestion -- Grid independence -- Lattice-Boltzmann -- Euler-Lagrange -- Non-Newtonian -- OpenLB; Fluid dynamics -- Data processing -- Periodicals; 532.050285
• DOI: 10.1016/j.compfluid.2020.104632
• Languages: English
• ISSNs: 0045-7930

Abstract

Highlights: A Lattice-Boltzmann model is proposed to improve gas mixing in anaerobic digestion. The multiphase, non-Newtonian model is validated in the lab-scale. A scaling mechanism is proposed for Euler-Lagrangian model convergence. The model is much cheaper than other Lattice-Boltzmann and other CFD models. Abstract: For the first time, a set of Lattice-Boltzmann two-way coupling pointwise Euler-Lagrange models is applied to gas mixing of sludge for anaerobic digestion. The set comprises a local model, a "first-neighbour" (viz., back-coupling occurs to the voxel where a particle sits, plus its first neighbours) and a "smoothing-kernel" (forward- and back-coupling occur through a smoothed-kernel averaging procedure). Laboratory-scale tests display grid-independence problems due to bubble diameter being larger than voxel size, thereby breaking the pointwise Euler-Lagrange assumption of negligible particle size. To tackle this problem and thereby have grid-independent results, a novel data-scaling approach to pointwise Euler-Lagrange grid independence evaluation, based on an application of the Buckingham π theorem, is proposed. Evaluation of laboratory-scale flow patterns and comparison to experimental data show only marginal differences in between the models, and between numerical modelling and experimental data. Pilot-scale simulations show that all the models produce grid-independent, coherent data if the Euler-Lagrange assumption of negligible (or at least, small) particle size is recovered. In both cases, a second-order convergence was achieved. A discussion follows on the opportunity of applying the proposed data-scaling approach rather than the smoothing-kernel model.