Optimization of microclimate uniformity in a multi-layer container-type strawberry smart farm using CFD and the Taguchi method
Abstract
Vertical container-type smart farms offer an effective way to improve productivity in space-limited environments, but their multi-layer arrangement often creates stagnant airflow and nonuniform microclimates. In this study, computational fluid dynamics (CFD) and the Taguchi method were combined to optimize the layout of air-conditioning components in a container-type strawberry smart farm. Three spatial design factors were considered: inlet position, outlet position, and carbon dioxide (CO2) nozzle position. Nine cases were constructed using a three-level L9 orthogonal array, and spatial standard deviation was used as the uniformity index for temperature, CO2 concentration, and local mean age of air (LMA). The baseline model showed low-velocity and recirculation regions around the cultivation beds and structural frame, producing substantial nonuniformity in CO2 distribution. The middle layer had the highest mean CO2 concentration, whereas the lower layer had the lowest concentration, with a maximum interlayer difference of 53.7 ppm. The Taguchi analysis showed that outlet position was the most influential factor, with a delta value of 8.74 and a contribution ratio of 57.78%. The optimal configuration placed the inlet at 1,200 mm, the outlet at 1,500 mm, and the CO2 nozzle at 1,200 mm. Compared with the baseline model, this configuration reduced the standard deviation of CO2 concentration from 75.85 to 7.08 ppm and improved the standard deviation of LMA from 5.63 to 4.98 s. However, temperature uniformity slightly decreased as the analysis used steady-state conditions, fixed operating flow rates, and a simplified crop geometry; the optimal layout should be revalidated when the container size, crop canopy density, or operating strategy changes. These results demonstrate that spatial microclimate uniformity in multi-layer container-type strawberry smart farms can be improved efficiently by optimizing component layout.