Effects of Standard k-e and LES Turbulence Models on a Full Scale Numerical CFD Simulation for a Naturally Ventilated Pig Barn Prototype

Abstract. Computational Fluid Dynamics (CFD) has proven to be an important tool to unravel ventilation patterns within livestock barns, especially aiding the issue of natural ventilation (NV). This research study aimed at developing a numerical model of a full scale prototype that represents a section of a fattening pig barn typically found in the Region of Flanders, Belgium. The numerical model features a SW-NE oriented 5.5 W A— 12.0 D A— 4.8 H m section of a pig barn with openings 3.0 W A— 0.5 H m at inlet side and 1.0 W A— 0.5 H m at outlet side, surrounded by an air volume of 100 W A— 100 D A— 25 H m. A logarithmic vertical wind profile was used to represent the atmospheric boundary layer, calibrated to wind speed data measured at a 10 m H meteorological station placed nearby the prototype. Two turbulence models were tested: (a) the Standard k-Iµ (SKE) and (b) the Large Eddy Simulation (LES). The model was validated against wind speed and inside air velocity data monitored with 2D ultrasonic anemometers in a homogeneous grid of 8 points placed at a horizontal plane (H = 2.5 m) inside the prototype and two 3D ultrasonic anemometers placed at the openings. Steady state simulations revealed good agreement between model and experimental data. Both tested turbulence approaches yielded velocity profiles across the pig barn prototype that were comparable to the measurements. Mean velocities estimated by the model implemented with LES and SKE turbulence models were (0.29 ± 0.08) mI‡s -1 and (0.23 ± 0.07) mI‡s -1 , respectively, while the average velocity measured experimentally was (0.26 ± 0.09) mI‡s -1 . The outcomes from this study can help explaining how changes in wind direction affect air exchange rates and may assist on gas sampling strategies for emission measurements in practical conditions.