Prediction of the dynamic behavior of a solar chimney by means of artificial neural networks

A strategy is described for the construction of a reference model for the design of solar chimneys that includes variations in geometry and materials of the chimney's components. The model will be developed from dynamic simulations in the transient state of the solar chimney under solar irradiation real conditions. The strategy is based on the artificial neural networks (ANNs) generalization properties allowing predictions for multiple geometries and materials of the solar chimney. The strategy can serve as a basis for cost optimization during the design stage by allowing selection of the best geometry and materials given the desired performance specification for the solar chimney, including operating and replacement costs.

Keywords: Artificial neural networks, mathematical model, thermal comfort.