Proposition of a simplified method for predicting hourly indoor temperatures in test cells

Abstract Test cells can be used for testing the thermal performance of different passive systems and building components. Predictive methods for estimating indoor air temperatures can further enhance the number of configurations tested without increasing the amount of test cells to be built. Thus, direct comparisons can be drawn for identical background climatic conditions. In its most basic form, formulas are generated by linear regression from relatively short data sets, which provide daily indoor temperature conditions. However, for more detailed analyses, daily indoor temperature predictions may not suffice. In this paper, a method for obtaining hourly indoor air temperature predictions is proposed. It is based on rising and decreasing rates of the indoor temperature fluctuation relative to outdoors, which translates to warming or cooling trends of indoor thermal conditions. The applicability of the method is for test cells. It is a simple method yet capable of predicting the thermal behavior of complex physical processes. The method was tested using measured data from experiments in a test cell, built with conventional building materials in Brazil. Results showed high performance with mean bias of 0.27 °C to measured data and Pearson's r 0.99. When compared to traditional regression-based models, the method proposed showed better results.