A FRAMEWORK FOR GENERATING STOCHASTIC METEOROLOGICAL YEARS FOR RISK-CONSCIOUS DESIGN OF BUILDINGS
Benjamin D. Lee, Yuming Sun, Huafen Hu, Godfried Augenbroe, Christiaan J. J. Paredis
Current performance-based design of buildings is predominantly based on deterministic simulation, but it is being increasingly recognized that the analysis of uncertainty and risk is important for project success, especially in the design of ultra-low energy buildings. As they represent the boundary conditions in energy simulation, weather data significantly affect model outcomes, so the uncertainty of the meteorological conditions should be taken into account in a risk- conscious design process. In the United States, Typical Meteorological Years are used extensively in the analysis of building energy performance, though they fail to account for any variation from typical weather conditions. This paper seeks to address this issue through the stochastic modeling of meteorological data as a Vector Auto-Regressive (VAR) process with seasonal non- stationarity. We present a framework that characterizes the VAR process using historical meteorological data for any given location. Once defined, the VAR process is able to generate any number of Stochastic Meteorological Years (SMY) for use in simulation packages. The framework is validated with a case study examining predictions of the energy-performance of a solar decathlon competition home.
- There are currently no refbacks.