Akpan, Michael Effiong; (2019) Advanced Modelling of Dense Polydisperse Fluidized Powders via Hybrid CFD and Population Balance Modelling. Masters thesis (M.Phil), UCL (University College London).

Preview

Text MPhil_Final_Copy - Michael Effiong Akpan.pdf - Accepted Version Download (1MB) | Preview

Abstract

Fluidization is used in several industrial processes, for example in waste disposal, food processing, pharmaceutical applications, energy conversions and so on. Although virtually every industrial plant contains units which treat multiphase polydisperse systems (MPS), designing them is still subject to great uncertainties; this is because MPS involve many physical and chemical phenomena that occur simultaneously and are hard to model: particles move in space and may shrink, grow, aggregate or break (in some cases new particles may also form). The unit behaviour and the product quality strongly depend on these competing phenomena, on the system fluid dynamics and on the unit geometry and size. This research project focuses on fluidized beds (FB) and aims to implement, test and validate an advanced computational fluid dynamics (CFD) model able to describe the behaviour of dense polydisperse fluidized powders and the evolution of their particle size distribution (PSD) in space by coupling CFD with population balance modelling (PBM) and by accounting for size-changing phenomena such as aggregation and breakage. To begin our work, we derived useful mathematical models including the generalized population balance equation, DQMOM and QMOM multifluid models and size-based aggregation and breakage kernels based on the kinetic theory of granular flow. In this work, we have implemented, in the commercial code Fluent, the QMOM Model, employing quadrature formula based on two, three and four nodes. Our model included convection in physical space,particle aggregation and breakage. We tested that the model had been correctly implemented by comparing model results with those reported in Fan et al. (2004). In terms of trend, results from our simulations showed good agreement with those reported by Fan et al. (2004). However, we noticed some quantitative differences in the results. These differences are due to the different constitutive equations used in CFD codes Fluent and MFIX, employed in this work and Fan et al. (2004) respectively. Furthermore, for simulations using four-node quadrature approximation, we were unable to replicate the trend as a result of the corruption of transported moment set. We observed similar corruption of the moments when we used higher order discretization schemes. We tackled the problem of moment corruption by implementing moment correction algorithms in Fluent. Nevertheless, we observed that the moment correction algorithm was effective in the QMOM model with the moment set transported with the same velocity compared with QMOM model with different velocities for the moment sets. We observed that in the latter the corrupt moment set is used to compute ’corrupt’ velocity fields for the quadrature classes which in turn complicates the solution. In other to further test the robustness of the implemented model we ran segregation tests using the QMOM model and compared them with experimental results. Our numerical results for the quadrature nodes and volume fractions as well as axial segregation profile was in good agreement with experimental results.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as JSONCSVXML

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item