The work presented in this publication complements a previous study that has been done (publishing in progress) by the Priority Research Center for Energy at the University of Newcastle, Australia on evaluating the thermal performance of typical walling designs used in Australian residential buildings. Four different walling systems (i.e. brick veneer, BV, cavity brick, CB, insulated cavity brick, ICB, and lightweight, LW) have been considered and their thermal performance has been studied numerically by predicting the room air temperature using a Neuro-Fuzzy model. The numerical predictions were obtained systematically at different heights and for various climatic conditions using a predictive tool called Adaptive Neuro-Fuzzy Inference System (ANFIS) of Sugeno-type. The ANFIS model was first extensively trained using experimental data collected from four test house modules which were built on the University of Newcastle campus for this purpose as part of a broader study of the thermal performance of Australian masonry housing. The model predictions were then compared with experimental data collected for later years. Results from experimental modelling, comparisons of actual experimental data and numerical predictions and analysis of prediction errors show that the ANFIS model is well capable of predicting the thermal behavior of the walling systems studied.
2nd Canadian Conference on Effective Design of Structures. CCESD-II: 2nd Canadian Conference on Effective Design of Structures: Sustainability of Civil Engineering Structures (Hamilton, Canada 20-23 May, 2008) p. 651-661