Validation of an open source CFD code to simulate natural ventilation for agricultural buildings

To study natural ventilation for agricultural buildings, OpenFOAM, an open source CFD code, enables to develop tools that automate complex simulation procedures for nonprofessional CFD users. Prior to developing such tools, this paper investigates the accuracy of simulated air velocity in three test buildings according to different mesh sizes and turbulence models. The computed results, that were obtained using 13 different mesh sizes and 4 different k-ε turbulence models, were qualitatively and quantitatively compared to wind tunnel measurements by two validation metrics. The results showed that the smaller mesh size leads to slightly higher simulation accuracy, but an increase in the number of meshes above a certain level did not improve the simulation accuracy due to the problem of solution convergence. The mesh size larger than 0.25 m at building walls was appropriate for the standard k-ε model and the RNG k-ε model, while a small mesh size of 0.125 m was preferable for the low Re k-ε model. The simulation results, especially obtained from the RNG k-ε model, were highly influenced by whether the solution converged showing the relative error of on average 6.4% higher in the oscillatory solutions, while those from the low Re k-ε model were more dependent on the mesh size.