Comparison between time domai n and freq uency domai n computer program for building energy analysis F Haghighat and A Athienitis
Two approaches to computerized building energy analysis are compared: the discrete Fourier series thermal network method ( a frequency domain approach) and the response factor approach ( a time domain technique). Two computer programs, one time domain and one frequency domain, are compared together and with experimental data. Both are shown to be sufficiently accurate for design. Their computational efficiency and weather data requirements are compared, and the frequency domain program is shown to be significantly more efficient and flexible and to require less storage space and computation time.
building thermal analysis, thermal network, simulation, frequency analysis, response factors
Computer simulation is becoming an increasingly popular tool in building thermal design. Design and analysis tools include detailed computer simulation programs such as Blast 1, DOE 2, and TARP 3, which were developed mainly for mainframe computers. These simulation programs are based on the time domain approach and use hourly weather data to predict the hourly profiles, an expensive process. The inputs to these programs are historical weather data that are available on magnetic tapes. A problem with these simulation procedures is that the year selected for energy analysis may not represent an average or typical year and, therefore, the output can be misleading in many cases. To ensure that the chosen data set represents a typical year, data for several years are required. Thus considerable data handling is needed. For example, one weather data file requires about 500 kbyte of disk storage. A better solution is to create a data file from many years of available data that represent a typical meterological year cs.
There have been several attempts to eliminate the direct use of hourly weather data to reduce the cost of data handling. Statistical methods have been developed to generate synthetic data 6,7.