초록 열기/닫기 버튼
Thermal properties of building envelope for existing buildings may not be identical to the design specification and installed condition due to workmanship, onsite construction errors, and age of building. Therefore, it is very essential to accurately estimate the properties for improving energy performance and planning building retrofitting and renovation. This study suggests a novel approach to estimate key thermal parameters based on a numerical optimization algorithm, downhill simplex method, incorporated with indoor air temperature variance and change of thermal parameters. Insulation thickness, infiltration, U-factor, and solar heat gain coefficient of window are the estimated variables for a typical medium-sized office in US reference building models. The numerical model searches the variables to minimize the difference of the indoor temperature variance between the original parameter model and the estimated parameter model in minutes for three days of no air conditioning and occupants. The results show that individual parameter estimation for four variables has converged less than 20 iterations with 99.0% accuracy. Although the iteration extended to 201 when four variables are simultaneously unknown conditions, estimated values are very close to the original values under 0.01% of root mean square errors.
