Verification of scaling criteria for bubbling fluidized beds by DEM-CFD simulation

Fluidized bed scaling rules have been proposed in the literature to preserve hydrodynamic similarity between systems at different scales. Experimental verification of such criteria requires maintaining several dimensionless groups constant across the scales. The procedure proves to be extremely complex due to the limited ability to adapt materials, geometry and operating conditions. The capability of DEM simulations to fulfil the criteria exactly is exploited in the present work to check the suitability of two procedures for scaling factors down to 0.5 and up to 10. Two reference systems are considered in order to assess the full set and the simplified set for inertial-flow conditions (L.R. Glicksman, M.R. Hyre, PA. Farrell. Int. J. Multiphase Flow. 20 (1994), 331). Pressure signals for pairs of reference and evaluation systems are evaluated, with investigation of the effects of the particle contact properties and integration time-step. Comparison in the statistical sense between pairs of scaled systems is carried out in terms of the distance between attractors reconstructed by time series of the pressure signals, by making use of a test on the S-statistic (C. Diks et al., Phys. Rev. E. 53 (1996), 2169). The results for the full set produce scaled systems preserving the similitude for all size factors investigated and two base-case systems. Adoption of the simplified set for inertial flow turns out to be valid only for a size factor of 2, while with 4 the similitude properties resulted lost, at least for the investigated systems. Tuning of the parameters of the statistical analysis was necessary to ensure reliable predictions of the test. In particular, the time window taken as a quarter of the cycle time proved more appropriate than the whole cycle time in identifying three systems possessing mismatched properties. As a final remark, it is shown how the potential advantage of scale-up in decreasing the computational demands of DEM simulations in practice is very limited. (C) 2013 Elsevier B.V. All rights reserved.