Journal
BUILDING AND ENVIRONMENT
Volume 205, Issue -, Pages -Publisher
PERGAMON-ELSEVIER SCIENCE LTD
DOI: 10.1016/j.buildenv.2021.108185
Keywords
Building performance optimization; Building form; Building energy efficiency; Robustness analysis; Stochastic optimization; Office building
Funding
- National Science Foundation of China [52078294, 51825802and 51638003]
Ask authors/readers for more resources
Building performance optimization is effective but may lack robustness with uncertainties in input parameters. Stochastic optimization shows promise in handling this issue, but its impact on improving robustness is uncertain. Reducing uncertainty in input parameters and aligning assumed distribution with real distribution are more effective strategies.
Building performance optimization is effective in searching for optimal design solutions, but its result may be unrobust with uncertainties in input parameters, especially for form design. Stochastic optimization is a promising approach to handle this problem, but its effectiveness in improving robustness is unclear. In this study, stochastic optimization is compared to the traditional deterministic method under holistic uncertainties in 13 input parameters. Energy optimizations of two office building forms, a shoe-box and an irregular quadrilateral, were inspected. The results indicate that robustness problems exist in both deterministic and stochastic optimization with uncertainties, but utilizing optimization results is still a good strategy, as the average chance of the optimized form outperforming random designs exceeds 0.7 for all cases involved. Using stochastic optimization helps obtain more robust results, but the expected increase in the average chance is only 3.2 % and 1.2 % for the two forms, and thus its effect should not be overstressed. We suggest that it be used only in major decisionmaking events considering its hefty demands in time and computation power. Reducing the uncertainty in input parameters is a more effective strategy. By reducing the variance in input parameters by 44 % compared with the blind guess scenario, the average chance is expected to increase by 7.2 % and 1.4 %, respectively. Making the assumed distribution closer to the real distribution of uncertain parameters is also helpful, with expected increases of 2.2 % and 0.7 %, respectively, if the variance difference between the two distributions is reduced by 44 % relative to a blind guess.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available