The environment of a solar greenhouse is influenced by external temperature, humidity, sunlight intensity, wind speed, the structure of the greenhouse, and the crops. Optimal control of the greenhouse's environmental conditions is possible through the correct placement of sensors to accurately measure the greenhouse's environmental parameters. Currently, the placement of sensors is determined based on the experience of greenhouse operators and designers. The main objective of this research is to select the optimal location for temperature sensor installation based on computational fluid dynamics analysis. In this study, a greenhouse, measuring 52.5 meters in length, 21 meters in width, and 7.75 meters in height, was examined, containing three air temperature sensors, one humidity sensor, one soil temperature sensor, one upper room temperature sensor, and one sunlight intensity sensor. Initially, air temperature data from 60 points in the greenhouse were recorded over three consecutive days: March 13, 14, and 15, 2023, from 8 AM to 5 PM. The studied greenhouse was then modeled in SolidWorks and loaded into ANSYS Fluent for simulating heat transfer processes within the greenhouse. The meshing of the model was performed using ANSYS Meshing, and boundary conditions, including the greenhouse walls (floor, ceiling, and side walls), as well as air inlets and outlets, were defined in the software. The quality of the mesh was eva‎luated using orthogonality, skewness, and aspect ratio quality indices. The independence of the solution from the mesh was examined, and a model with 324,414 cells was deemed suitable. During the data collection period, heat transfer simulations in the greenhouse were conducted for six different scenarios where the cooling system was off and the exhaust vents were closed, and for seven other scenarios where the cooling system was on. Additionally, turbulent flow was considered, and the RNG k-ε model was selected for solving the problem. Temperature and air velocity data were extracted at 15 nodes at a height of 1.5 meters from the greenhouse floor for all simulation scenarios. The calculated Z index, using standardized temperature and air velocity data and the absolute temperature difference of each node from the average, was estimated to determine the optimal location for sensor installation. The method based on entropy and K-means clustering was used to determine the appropriate installation location for the sensor. A comparison of experimental and simulation data showed that the accuracy of the simulation was at an acceptable level, with the average recorded errors in all cases being less than 5%. The normalized mean square error (NMSE), root mean square error (RMSE), and mean absolute percentage error (MAPE) for the cooling system turned on were estimated to be 0.07 degrees Celsius, 0.38 degrees Celsius, and 2.01 percent, respectively, while for the cooling system turned off, they were estimated to be 0.08 degrees Celsius, 0.42 degrees Celsius, and 2.13 percent, respectively. In all simulation cases, the maximum and minimum temperatures for the greenhouse ceiling and floor were recorded. The results indicated that the difference between the maximum and minimum temperatures when the cooling system was off was greater than when the cooling system was on. Using the absolute temperature difference relative to the average, if a single air temperature sensor is to be used in the greenhouse, position number 4 (near the pad) was determined as the best location for sensor installation using the entropy method, while position number 8 (the center of the greenhouse) was determined using the K-means method. By concidering the Z index, the optimal sensor placement positions were selected as nodes 4 and 5 for the entropy and K-means methods, respectively. Additionally, the optimal number of sensors in the greenhouse was determined to be 3 using the Elbow method.

مهرنوش جعفري , رامين كوهي كمالي

مرتضي صادقي , علي نيك بخت

علي قاسمي , الهام فاضل