Validation of CFD simulations for three dimensional temperature distributions of a naturally ventilated multispan greenhouse obtained by wind tunnel measurements

In the last decade there has been an exponential increase in the number of research works using computational fluid dynamic simulations to study natural ventilation processes in greenhouses. The main reasons for this are that Computational Fluid Dynamics (CFD) simulations can provide a vast amount of information in a cost effective way in a reasonable time period compared to experimental techniques. It also provides designers and researchers a very fast and reliable tool to improve natural ventilation designs by studying the effects of multiple combinations of vent cases and external climatic conditions. However, the main limiting factor of the simulations is that validation studies must be performed with some source of experimental data, which are normally very scarce, due to high costs of time and money. In the present work, a CFD package has been used to perform simulations of a set of previous experiments conducted in a wind tunnel with naturally ventilated multi-span greenhouse with scaled models. The main focus of the current study was to compare three dimensional temperature distributions obtained for different ventilation cases from wind tunnel study with CFD simulations to validate the model. After a first set of simulations the quantitative agreement found for the temperature distribution in the middle vertical plane of the greenhouse for the four vent cases studied was only approximate, whereas for the mid horizontal plane the agreements was less good, with values in all cases lower than in the experiments. The qualitative comparison of the temperature distribution on both middle planes (vertical and horizontal) showed fairly good agreement for cases in which side vents were opened in combination with the roof vents whereas the agreements for those cases in which only roof vents opened was not similar. In view of this, two new turbulence models were used checking the effect on the agreement with experimental results, in order to have a better knowledge of the simulations set up for future validations.