A CFD-DEM Study of the Aerodynamic Characteristics of Non-Spherical Biological Particles

Abstract

The processes in growing, harvesting, and storing forage play a critical role in the quality and yield of the crop. Many manufacturers of forage harvesters use computer-aided design tools to virtually model the behaviour of the material during the harvesting process such that improvements can be made to the designs without incurring excessive costs in funds and time of physical prototyping. To provide reliable and accurate results by simulating the virtual forage harvester, a fundamental understanding of the aerodynamic properties, the lift and the drag coefficients, of the forage particles is necessary. In this thesis, a coupled Computational Fluid Dynamics (CFD) and Discrete-Element Method (DEM) procedure for the simulation of cylindrical stem particles and flat, rectangular blade particles from perennial ryegrass forage samples in a horizontal wind tunnel is presented. The experimental data performed in the horizontal wind tunnel was supplemented with measurements of the terminal velocity and calculation of the drag coefficient using a vertical wind tunnel. The methodology of the simulations closely followed that of experiments performed in the horizontal wind tunnel. The reliability of several lift models (Magnus and a customized model) and drag models (non-spherical, Gidaspow, and a customized model) were tested to compare the effect of the models on the average horizontal mass distribution, also referred to as the geometric mean distance, in comparison to experimental data. Five case studies are included in the simulations based on experimental data to reflect the minimum and maximum values of particle length and moisture content. The geometric mean distance values in simulations were significantly different than the experimental results for all lift and drag models tested. Thus, the case study includes a novel calibration and verification procedure to produce the coefficients for other experimental tests between the length and moisture content ranges of the five simulated cases. The results of the five cases revealed that the percent difference of the experimental drag coefficients and the simulation drag coefficients varied linearly with particle length and moisture content. The drag coefficients of the five cases were scaled to produce simulated results agreeing within 10% of the experimental results and an interpolation procedure was performed to yield predicted scaled drag coefficients for the remainder of the experimental tests. The interpolation of the scaled drag coefficients was verified with two additional simulations that agreed with experimental data within 10%. The scaled drag coefficients (on average 2.25 for stem particles and 1.63 for blade particles) necessary to match experimental data were significantly greater than the values calculated from the terminal velocity in the vertical wind tunnel (0.35 for stems and 0.74 for blades). The need for calibrating the drag coefficient may be due to simplifications or limitations of the virtual models (e.g., rigid particles, sensitivity to particle dimensions), experimental measurement uncertainties, or a lack of lift forces included the simulations. Lastly, the translation of vertical wind tunnel drag coefficients to the horizontal wind tunnel simulations using the discrete phase drag model may be a significant contribution to the discrepancy given the different particle Reynolds number regimes in each apparatus.