Conference Year

July 2018


model predictive control, weather forecast, building automation systems, solar radiation


Although the potential of model predictive control (MPC) for the operation of buildings is widely recognized, as of today its adoption has been rather limited. This is partly due to the lack of user-friendly software tools for MPC, such as tools to facilitate the incorporation of forecast information in building automation systems. In view of this, CanmetENERGY, a research centre of Natural Resources Canada, has developed CanMETEO, a software tool free of charge aimed at obtaining weather forecast data and make it available in a useful and practical format for building operators. CanMETEO, which was released officially in August 2017, uses raw data produced by the Meterological Service of Environment Canada. This data, with high spatial resolution (e.g., 2 km x 2 km grids, and even denser for urban areas) enables the possibility of obtaining forecasts for very specific locations in the Canadian territory. Hundreds of weather variables (such as temperature, humidity, wind speed, cloud cover, among many others) are available for each point, which can be selected by the user via a graphical interface. The data is converted from GRIB files (a standard binary format used by meteorologists) into comma-separated value (CSV) files, which can be easily accessed. New forecasts become available every 6 hours, with a prediction horizon of 48 hours at hourly time steps; the retrieval of new weather forecasts can be setup in order to be performed automatically. These continuously updated CSV files may then be easily incorporated into building operation algorithms or simple optimization routines. Once the basic variables are obtained, post-processing calculations are applied in order to estimate solar irradiance on any given plane required by the user, for example, building façades and building-integrated photovoltaic panels. This feature also makes it possible to estimate the effect of solar gains on the thermal response of a building, and to estimate the output of photovoltaic panels. A preliminary evaluation of the tool, based on on-site measurements, is presented in this paper. It is expected that CanMETEO (currently used by Canadian research centre and universities) will provide one further step to the widespread adoption of predictive control as a viable, popular solution in building operation.