Robust optimal design of zero/low energy buildings considering uncertainties and the impacts of objective functions

Robust design optimization has received great attentions in aerospace/structural engineering fields for few decades. Recently, a few studies have been conducted on robust design optimization in building energy field mostly by directly adopting the objective functions used in these fields. However, the applicability of these objective functions have not been analyzed when applied for buildings while some researcher even questioned this fundamental issue. Furthermore, very few study is concerned with robust design optimization of building envelopes for zero/low energy buildings. In this study, the applicability of commonly-used objective functions for robust design optimization of buildings is analyzed and compared. The robust optimal design for envelopes of zero/low energy buildings in subtropical regions is studied. Major uncertain design inputs, which have significant effects on building performance in subtropical regions, are identified. An artificial neural network model is used for building performance evaluation to reduce computational time. Results show that it is necessary to consider uncertainties for zero/low energy building design as they have significant impacts. The commonly-used objective functions in pioneer fields are not suitable if applied in building energy field directly without proper revision. Revisions to objective functions, particularly the involvement of variance of performance indicator, are proposed for robust design optimization of buildings.