Yearly Performance Prediction of Thermosyphon Domestic Water Heating Systems with Neural Networks and Minimum System Testing

Abstract

The objective of this work is to use Artificial Neural Networks (ANN) for the yearly performance prediction of thermosyphonic type solar domestic water heating (SDWH) systems. Twenty-seven SDWH systems have been tested and modelled according to the procedures outlined in the standard ISO 9459-2 at three locations in Greece. These data were used for training and testing the networks. Three neural networks were trained to perform different tasks. The first one predicts the system performance characteristics on a daily basis whereas the other two predict the system long-term performance. The collector areas of the considered systems were varying between 1.81 m{sup 2} and 4.38 m{sup 2}. Open and closed thermosyphonic systems have been considered both with horizontal and vertical storage tanks. In this way the networks were trained to accept and handle a number of unusual cases. From the trained networks three dynamic library link (DLL) files were produced and used in different workbooks of the same spreadsheet, using the CALL function. The results from the first workbook, which are produced from the input data described above, are used in the other two to produce the annual results thus reducing drastically the experimental data required. The method was validated by using actual data for one system, which the networks have not seen before. The accuracy obtained indicate that the prediction of the solar energy output of the system for a draw-off equal to the volume of the storage tank is within 3.5% and for the same parameter and the average quantity of the hot water per month for the two demand water temperatures is within 7.8%. These results are considered satisfactory.