Evaluation of turbulence models to predict airflow and ammonia concentrations in a scale model swine building enclosure

The performance of five widely used turbulence models, the standard k–ε model (SKE), the renormalization group k–ε model (RNG), the realizable k–ε model (RKE), the standard k–ω model (SKW) and the shear stress transport k–ω model (KWSST), were evaluated for simulations of airflow velocities and ammonia concentrations in a 1:12.5 scale model swine building without a floor (100% floor opening) and with a slatted floor with 16.7% floor opening area. The 100% floor opening case was used as a reference. The turbulence models were evaluated by comparing the numerical results with experimental data at representative points inside the scale model. The RKE and RNG models required less elements for grid-independent results with the predicted airflow patterns agreeing well with the smoke tests. The velocities and concentrations predicted by the RNG model were closer to the measured values with a maximum velocity difference of less than 0.03 m s−1 (9.3%) and a maximum normalized concentration difference of less than 0.09 (12.3%) for the 100% floor opening. For the 16.7% floor opening, the maximum velocity difference in the main space was less than 0.02 m s−1 (6.8%) and the maximum normalized concentration difference was less than 0.2 (25%). Thus, the RNG model most accurately predicts the airflow velocities and ammonia concentrations in the scale model swine building enclosure.

► Five turbulence models were evaluated by comparing simulated airflow and concentration distribution with measured values.
► The RKE and RNG models require less elements to accurately predict the airflow pattern.
► Comparison of the predicted and measured values shows that RNG model give the best agreement.
► The RNG model can be used to predict non-fully turbulent airflow.