Rheology-based CFD modeling of magnetite medium segregation in a dense medium cyclone

• Dense medium cyclone is modeled with rheology-based CFD models with extra forces.
• Granular, modified Newtonian and non-Newtonian viscosity-based CFD models are tested.
• Predicted density contours are compared with experimental gamma ray tomography data.
• Medium segregation is well predicted by Herschel–Bulkely and Granular viscosity models.
• Herschel–Bulkely viscosity-based CFD model predicts high viscosities.

In this work, an improved multiphase computational fluid dynamics (CFD) model is developed for simulating medium segregation in a dense medium cyclone. This model approach uses the modified mixture model with the granular option and Reynolds stress model (RSM) to resolve the turbulent mixing of the particles. Rheological behavior of the medium is considered through various forms of viscosity models such as granular viscosity, Newtonian and non-Newtonian model corrected with particle loading and fraction of ultra-fines. Multiphase simulations using the granular viscosity option although predict the overall slurry volume split, product medium densities better than Newtonian models, but the mixture viscosity values are restricted to substantially lower values close to water viscosity levels. Simulations using the Newtonian viscosity model corrected with particle loading and fine fraction below 53 μm size are predicted the overall medium viscosity levels well above water viscosity levels but underpredicted the cyclone underflow medium density to the experimental data. Multiphase simulations with the non-Newtonian Herschel–Bulkley viscosity model are able to predict the medium segregation close to the gamma ray tomography data, and the predicted medium viscosity levels are increased up to 18 cp. Overall predicted product medium densities, underflow volume split and radial medium segregation levels are compared to the experimental data. Lagrangian particle tracking (LPT) is super imposed on converged medium simulations used for coal partitioning predictions. This model can be further used in developing new designs in the future.

Figure optionsDownload as PowerPoint slide