Energy consumption in buildings accounts for ca. 40% of the world's total energy consumption, yielding a call for attention to their performance and improvement of their behaviour. The concept of performance testing has been developed as an approach to control, potentially lower, and bring awareness to buildings' energy consumption. In this paper we propose a methodology for continuous performance testing of smart buildings as a service. We discuss the automatic discovery and instantiation of performance tests, as well as usage of results for discovery of potential faults and ways to improve buildings' performance. The methodology that we present is sufficiently flexible to be extended to other cyber-physical systems, given minor adjustments. Finally, our proposed methodology has been partially implemented and deployed to a smart building at a university campus in Denmark, delivering promising results.

The general trend in building commissioning is that the commissioning is often either insufficiently comprehensive in its execution, or it is not conducted at all. As such, it has been observed that buildings often will contain faults that have not been discovered during the commissioning phase. In this paper, we present a framework for continuous verification and commissioning of a building with a set of performance tests that are able to monitor the performance and functionality with a daily granularity. The performance tests are run on a newly constructed campus building in Denmark, over the course of three months, and the results of the performance tests are presented within a dashboard application, which helps verify that the building is performing accordingly to its original design intent.

Buildings contribute to approx. 32% of the world’s energy consumption, and as such, are one of the prime producers of CO 2 emissions. Given documented discrepancies between the designed and operational behaviour of buildings, concepts such as continuous commissioning and performance testing have emerged. Performance testing utilises metadata schemas and generic libraries of tests to increase applicability and customizability of real-time performance monitoring. In this paper we present LEAF: a Live building performance EvAluation Framework. LEAF has been developed as a result of an extensive survey with building management professionals, who gave insight into the preferences for management of various buildings. We present LEAF’s microservice-style architecture, to increase reusability of its processes by other building intelligence applications, and use it to develop a performance monitoring application for a case study building. Furthermore, the live streaming processing of building operational data has posed significant challenges to the application of LEAF. We discuss these challenges and offer directions for their solutions.