Evaluation of Building Envelope Performance Including Uncertainty and Degradation Within a Multi-Objective Optimization Framework PDF Download

Author: Daniel Haeyoung Chung
Publisher:
ISBN:
Category : Buildings
Languages : en
Pages : 337

Get Book Here

Book Description
Predicting building envelope behavior over time can be difficult due to the uncertainty and complexity of the environment, materials and processes that influence heat, air, and moisture (H.A.M.) transport through the assembly. While the variability in outdoor weather's impact on H.A.M. analysis for envelopes has been studied over one-year prediction periods, long-term multi-year predictive simulations have not been well studied. Similarly, degradation and material variability are known to impact long-term envelope H.A.M. behavior, yet there has been limited documentation on predictive studies that integrate these areas into a useful framework to inform a design process. Thus, the primary objective of this dissertation is to create a novel method of determining probabilistic distributions of envelope moisture behavior outcomes that also includes variation from degradation, so that the results can augment decision frameworks for building envelope design. To fulfill the objective the research work explored four tasks. (1) The investigation of available probabilistic input distributions for building envelope simulations. Here inputs include weather conditions and material properties. (2) The stochastic investigation of the impact of variable multi-year weather on H.A.M. analysis outcomes. (3) The integration of time-dependent loadings and degradation modeling coupled with H.A.M. analysis (i.e. multi-year weather, indoor climate variation and insulation aging). (4) The creation of envelope metamodels (based on the probabilistic analysis using H.A.M. and degradation models) to expedite optimization studies. These are derived from the Monte Carlo studies via regression analysis. The investigation of available data resulted in the determination that while stochastic weather data is available, important aspects of rainfall and longwave radiation (from the sky) may be missing for many locations. Datasets regarding stochastic material properties are limited and require compilation from multiple studies. These deficiencies should be addressed by the industry to improve predictive capabilities and enhance the understanding of uncertainty in envelope analysis. The stochastic investigation of the analysis duration for building envelopes determined that variable weather for 10-year simulation periods had significant impact on the H.A.M. behavior. In comparison to frequently used one-year simulations, the 10-year results demonstrated much greater risks of undesirable moisture related outcomes (such as mold growth and corrosion). Currently ASHRAE standard 160 recommends that H.A.M. simulations be performed over either a reference year or a 10-year period. While ASHRAE provides reference year data, no guidance is given on selecting a 10-year weather period. Given the significantly higher risks demonstrated in this dissertation when using 10-year periods, this should be addressed in future research to better characterize long-term envelope behavior. The results from integrating time-dependent loadings and degradation coupling with H.A.M. analysis showed mixed results. Of the five coupled models studied, the two with the greatest influence on the H.A.M. behavior were longwave radiation exchange and wood decay. Wood decay while statistically infrequent had showed significantly accelerated mass loss when strongly coupled in the simulation model. Longwave radiation exchange greatly impacted moisture transport. Out of all the models studied, those that included longwave radiation exchange head the highest levels of moisture content. Further research including validation studies are needed to improve the use of longwave radiation exchange in envelope analysis. The results from the regression analyses and the optimization studies showed that the general framework for creating the metamodels is fairly robust. Metamodels were created with reasonable R-squared and error values from Monte Carlo trials to predict moisture related outcomes (i.e. moisture content, mold growth, and corrosion). These metamodels were demonstrated in a multi-objective optimization process that revealed in the assemblies studied that there is a general inverse relationship between moisture optimization and thermal transmittance. As minimizing moisture content and U-values are common goals for envelope design, discovering that the two objectives are inversely related through the optimization studies was significant. This outcome demonstrates how the methodology of the dissertation can be used as a framework to evaluate envelope performance to include uncertainty and degradation for multi-objective optimization and should help inform the envelope design process.