Evaluating the influence of turbulence models used in computational fluid dynamics for the prediction of airflows inside poultry houses

• Published in: Biosystems engineering 2019-07, Vol.183, p.1-12
• Main Authors: Küçüktopcu, Erdem, Cemek, Bilal
• Format: Article
• Language: English
• Subjects: Airspeed; k-ɛ; Poultry; Temperature; Turbulence
• ISSN: 1537-5110; 1537-5129
• DOI: 10.1016/j.biosystemseng.2019.04.009

There are various turbulence models in the computational fluid dynamics (CFD) literature but none has so far proven to be universally applicable. Accurate simulations require the proper choice of model appropriate for each particular situation. In this study, the performance of three types of k-ɛ turbulence model, the standard k-ɛ, renormalisation group (RNG) k-ɛ, and realisable k-ɛ, were evaluated for their ability to accurately simulate the internal turbulent flow of a poultry house. Each model's accuracy was analysed by comparing predicted and experimental results, and its performance was assessed using the coefficient of determination (r2), the root mean square error to the standard deviation ratio (RSR), and a Taylor diagram, which provides a concise statistical summary of how well the correlation (r) and standard deviation (SD) patterns match. The RSR values obtained for air temperature and airspeed were 0.57 and 0.19, 0.30 and 0.16, and 0.64 and 0.23 for the standard k-ɛ, RNG k-ɛ, and Realizable k-ɛ models, respectively, and showed that the RNG k-ɛ model predicted the airspeed and air temperature best. Other models also provided good results, particularly in predicting airspeed; however, their air temperature predictions were not as accurate as those of the RNG k-ɛ model. The results showed that RNG k-ɛ presented the best results overall, whilst realisable k-ɛ did not meet with our expectations. •3-D CFD models were developed using different k-ɛ turbulence models.•The models were evaluated for their ability to simulate poultry house's internal turbulent flow accurately.•Each model's accuracy was analysed by comparing predicted and experimental data.•RNG k-ɛ presented the best results overall.