Testing the accuracy of synthetically-generated weather data for driving building energy simulation models

Weather data in formats required for annual energy simulations are not available at all locations where building designs are being evaluated. Synthetically generated weather data for these sites could be a viable option. This paper reports on the differences in weather and energy prediction results when using synthetically generated data vs. the use of recorded weather data for 50 cities worldwide. Energy simulations were performed on an office building placed in each city – first, by using recorded weather data and second, by using weather data synthetically generated by a statistically based model. Annual energy predictions were the main focus of the study; however, other parameters of interest were also observed. These included peak heating and cooling loads and other weather-related values, such as high and low dry-bulb temperatures, humidity, annual heating and cooling degree days, and global solar radiation.