An Efficient Approach to the Simulation-Based Sensitivity Analysis of Building Performance: OPAT-Based LSA and Sobol-Based GSA

Abstract

This work presents a new efficient approach to the simulation-based sensitivity analysis (SBSA) of building performance. To this end, through a new initiative, the whole building energy simulation program EnergyPlus is combined with the local sensitivity analysis (LSA) and global sensitivity analysis (GSA) through the C++ programming language. The developed method is applied to a dwelling house in the hot semi-arid climate region of Iran. Hereupon, the building design parameters including BR, WWR, DSH, CSPT, HSPT, SA_{IN_W}, SA_{Ex_W}, Th_{Ex_W}, ST_win, VT_win, Th_win, and Th_gas-win are adopted as input variables. Moreover, four major building criteria including annual heating energy consumption (AHC), annual cooling energy consumption (ACC), annual lighting energy consumption (ALC), and predicted percentage of dissatisfied (PPD) index are adopted as output variables. The one-parameter-at-a-time (OPAT) as the LSA and Sobol's analysis as the GSA are carried out to explore the behavior of outputs versus inputs changes and to quantify the total sensitivity of outputs-to-inputs (S_T). In the LSA approach, a new sensitivity indicator called the Dispersion Index (DI) is proposed to define the influence of inputs on outputs. The results demonstrate that for our typical building under study, AHC is most sensitive to the HSPT and SA_{Ex_W}, with S_T of respectively 80% and 79%. While ACC is most sensitive to the CSPT and SA_{Ex_W} with S_T of respectively 72% and 63%. Besides, WWR, VT_win, and BR with S_T of respectively 33%, 25%, and 21% are the most influential inputs on the ALC. Furthermore, CSPT, HSPT, SA_{Ex_W}, and WWR with S_T of respectively 81%, 40%, 36%, and 21% are the most influential inputs on the PPD. ALC has no dependence on the CSPT and HSPT of VAV and thermo-physical traits of wall and window. Besides, the sensitivity results obtained by the proposed DI in OPAT-based LSA are in good accordance with the Sobol-based GSA ones.