Towards Data-driven Building Retrofit

Due to the relatively low construction rate of new buildings in Europe, the energetic performance of the building stock can mainly be improved by retrofitting existing buildings. Current methods to derive and assess retrofit measures require either on-site inspection by an expert consultant and code based approaches, or numerical simulations using a building energy model that needs to be first developed. Since the former is rather inaccurate and prone to overestimating the retrofit measures (prebound effect), and the latter is time and cost intensive, in this work we propose a data-driven-retrofit (DDR) approach to derive optimal retrofit measures. The approach uses wireless sensor networks (WSN) for the determination of building characteristics, energy consumption and occupancy patterns. Building models in the broadest sense can be extracted from the collected data. Retrofit scenarios can be incorporated in these buildings models in order to estimate the future energy consumption of the building, and to choose the optimal measures. The approach is presented as a chain-like framework. The deployed WSN can remain installed to verify the retrofit execution, and to perform continuous long term monitoring for fault-detection and optimal performance of the installed building systems. This work extends our previous research in this area, and presents the DDR framework, and results from a case study on a listed building in Zurich, Switzerland.