CFD simulation of air distributions in a small multi-layer vertical farm: Impact of computational and physical parameters

Computational fluid dynamics (CFD) simulations have been extensively used in designing air distribution systems for controlled environment agriculture (CEA). In recent years, more application studies using CFD simulations can be found for vertical farms due to the increasing interest in indoor vertical farming systems. However, it is well-known that CFD simulations are sensitive to many computational parameters and settings. The requirement of a crop response model in the CFD simulation for a vertical farm makes it even more complicated. Despite increased interest, guidelines for CFD simulations in vertical farms are scarce based on a literature study. Therefore, a systematic sensitivity analysis is conducted for a small generic multi-layer vertical farm with sole source lighting, which was the object of study in the literature before. The impact of a wide range of computational and physical parameters is investigated, including grid resolution, turbulence model, turbulence intensity, discretisation scheme, drag coefficient of the crops and computational time. The analysis shows that in this case (inlet Re = 46,923, Ar = 0.078, cultivated with lettuce), the RNG k-ε turbulence model outperforms other commonly used two-equation turbulence models. Compared to the experimental results from the literature, the simulation results from the first-order upwind scheme show large discrepancies, especially on the coarse grid. Although the influence of drag coefficient on the airflow inside the crop canopy is pronounced, little difference is observed in the air distributions in the vertical farm away from the crops.