A Parameter Study of the Performance of Atmospheric Water Generators and Their Forecasting Models

The purpose of this research was to analyze the parameters affecting the performance of two atmospheric water generators and to create and compare two forecast models that will predict how well the generators perform in varying geographic locations. As climate conditions vary throughout the different seasons of the year, so will the amount of water produced from atmospheric water generators. By running two atmospheric water generators over a period of seven months and recording the amount of water produced along with the varying climate condition values, each parameter affecting water production was analyzed and compared to the others in order to distinguish which parameter had the greatest impact on the performance of the generators. Relative humidity was found to be the most important parameter affecting water production, followed by temperature and air pressure. The forecast models created in this study include a neural network model and a multiple linear regression model. With R2 values of 86.51% and 58.73% for each respective generator, the neural network model outperformed the multiple, linear regression model whose R2 values were 76.19% and 55.79%, respectively. Additionally, simplified versions for each type of model were produced to provide a generalized solution for those who do not have detailed climate condition data readily available.