A statistical model of the spatial variability of weather for use in building simulation practice
| Autor: | Godfried Augenbroe, Mayuri Rajput, Mostafa Reisi Gahrooei |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Environmental Engineering
Meteorology Mean squared error Homogeneity (statistics) Geography Planning and Development Urban morphology Statistical model Building and Construction Standard deviation Weather station Variable (computer science) Environmental science Spatial variability Civil and Structural Engineering |
| Zdroj: | Building and Environment. 206:108331 |
| ISSN: | 0360-1323 |
| DOI: | 10.1016/j.buildenv.2021.108331 |
| Popis: | Traditionally, statistically representative (e.g. TMY) weather data is used as boundary condition in building simulation practice and is deemed homogenous for a defined region, e.g., a city. However, this presumed homogeneity does not exist in most cases because due to the effect of variable urban morphology, local weather may be different in different parts of the region. This paper uses Multiple Tensor on Tensor Regression (MTOT) for creation of meso-scale weather for applications in building simulation practice, which is fast and efficient, provided that enough local weather station data and regional morphology data are available. The derived statistical weather model captures variation in local weather due to change in urban morphology with an ensemble mean of standardized mean square error 0.038, i.e, 96.2% of the variation in weather may be predicted by this method. Chicago is used as a demonstration region with a case study of single family dwelling to exemplify the impact of local weather on building performance indicators. A standard deviation of approximately 372 kWh for end use heating, 33.6 kWh for end use cooling and 4 h for hours of discomfort in winters is observed, highlighting the variation caused by the use of local weather in building performance simulation. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect