Responsive Building Performance: A Case Study of Electrochromic Building Envelopes

ConCave Ph.D. Symposium 2022 Proceedings, April 7-8, 2022. Georgia Institute of Technology, Atlanta, Georgia.Building envelopes play an important role in the building performance of energy efficiency, thermal insulation, and visual comfort. Controlling solar radiation and daylight through responsive building envelope systems is an emerging sustainable strategy to improve building performance. The effectiveness of responsive building envelopes depends on the dynamic properties of building envelope materials and control algorithms. Architects and researchers are exploring possible ways to integrate responsive electrochromic (EC) glazing materials in building envelopes and testing the dynamic impacts on building performance (DeForest et al. 2013; Hamidpour and Blouin 2018; Eleanor S. Lee et al. 2013). Up to now, the research has tended to focus on control logic, rather than on the responsiveness of the building envelope itself. The modeling of responsive behaviors of an electrochromic building envelope system is challenging due to the dynamic properties of the electrochromic materials and unpredictable behaviors. In this paper, we proposed a case study using four different electrochromic glazing materials to test the impacts of responsiveness on building performance in terms of visual comfort and energy saving for the climate conditions in Tampa, FL. We developed a novel approach, Dynamic Sequence Modeling (DSM), by which these responsive EC building envelope behaviors can be simulated. The simulation results are then used to feed our Supervised Machine Learning (SML) algorithms to enable prediction under changing weather conditions. The SML algorithms are promising avenues to solve this type of predictive learning problem (Murphy 2012). Our SML algorithms seek to optimize performance with altered responsiveness of our EC building envelopes, as a generally capable agent to predict effective responses given similar weather conditions to the learned representation of the climate model. We find that all three responsive building envelope variants demonstrate large improvements in both energy and visual comfort performance compared to the static building envelope. In three EC alternatives, where each has different tint responsiveness, the cooling and heating energy loads were reduced by 54.36% on average, and the illuminance measures had almost the same mean values close to the visual comfort threshold. The most responsive 4-mode EC had the least absolute deviations. On the other hand, the prediction accuracy of supervised machine learning models decreases as the complexity of tint responsiveness (tint mode) increases in electrochromic building envelopes. Our study demonstrates the impacts of responsive electrochromic materials on building performance. Moreover, we show that the complexity of responsiveness decreases the prediction accuracy for SML-based building control of dynamic materials