Architecture and field test of a RESTful web service for the optimization of space heating in a commercial office

The optimal operation of many building services (including HVAC) requires computational resources that are not necessarily available in commercial building management systems (BMS). Having this computational power available in a dedicated data center will ease the deployment of these algorithms, but raises several issues having mostly to do with getting the data out of the BMS and back to it. In this work we report on the architecture and field tests of Neurobat Online (NOL), a RESTful web service that implements the Neurobat heating control algorithm developed at EPFL. It has been controlling the space heating of a commercial building in Winterthur (Switzerland) for the second half of the 2014–2015 heating season. The original controller operated during the first half of the same heating season. By comparing the energy performance with and without NOL we derive estimates for the relative energy savings that such a system can achieve. INTRODUCTION In spite of the availability of advanced heating control systems, most commercial buildings in the developed world are still managed using the same principles as residential homes, e.g. by weather-compensated controllers. These controllers pump a heating fluid heated to a certain temperature whose setpoint is almost always a simple, monotonous function of the outdoor temperature. This scheme, while simple to setup and configure to ensure that the users are warm enough, tends to ignore the physics of the building and, more critically, makes no provision for the inclusion of any weather forecast. However, more and more buildings are being equipped with building management systems (BMS) that can, through a graphical interface, be configured by the facility manager to obtain values for this flow temperature setpoint from other sources than the heating curve. In particular, it has become possible to provide this value to the BMS either directly over the internet or indirectly, through the building automation and control system used in that building. Controlling the building services of a building over the internet appears to be a relatively unexplored topic in the academic literature. For example, [1] describes a set of web services for the “smart home”, i.e. a central computer in the house connected to sensors, actuators and HVAC systems, and that exposes a set of web services to the public internet. Through these services, the users can monitor their energy consumption or set their preferred setpoints, while utility companies can facilitate demand response or sell excess energy back to the grid when it is economical to do so. The same system has been extended in [2] to improve the demand response aspect and the energy management algorithms. CISBAT 2015 September 9-11, 2015 Lausanne, Switzerland 969 B A C ne t/ IP REST/HTTPS Internet Neurobat Online Optimal flow temperature Gateway