| 摘要: |
| 设计初期，体形系数与建筑负荷的
相关性及负荷相关参数的敏感度排序有助
于提升建筑能耗表现。本文以北京气候为背
景，对A（体形规律变化）、B（体形非规律变
化）两组办公建筑负荷模型实施莫里斯莫里
斯（Morris）敏感性分析。全局样本检验下，
仅单因素（层数或长宽比）作用下的体形系数
与供暖负荷呈线性正相关且具备普遍性。总
负荷的高敏感参数中，供暖制冷设定温度、
室内负荷参数敏感度排序靠前但受建筑体形
影响微弱；垂直外围护结构及屋顶结构参数
的敏感度排序靠后，并分别与体形指标S 1 （平
面周长/平面面积）、S 2 （1/层数）表现出相似
的波动趋势，进而可根据已有案例中的莫里
斯敏感度对相似案例中的参数敏感度进行估
算，对于热活动以导热为主的参数，误差率可
控制在10%以内。在寒冷地区，单纯的层数或
长宽比变化更能保障体形系数对能耗表现评
价的准确性，而敏感度估算可快速获得热活
动以导热为主的外围护结构参数敏感度。 |
| 关键词: 莫里斯敏感性分析 供暖负荷 制冷负荷 体形系数 敏感度估算 |
| DOI：10.13791/j.cnki.hsfwest.20200409 |
| 分类号: |
| 基金项目:国家自然科学基金项目（51338006） |
|
| Effect of Building Shape on Building Load and the Sensitivity of Its Influential Parameters |
|
GAO Feng,ZHU Neng
|
| Abstract: |
| Due to the growing demand of energy conservation and emission reduction, building
energy performance（BEP）has become an important property for current building. The
optimization of BEP can be expressed at all stages of a building’s life circle. In early design
stage, the correlation analysis between building shape coefficient (SC) and building load as well
as the sensitivity analysis of important building loads parameters are helpful for improving BEP;
however, the former analysis lacks the test of universality in most studies and the latter analysis
is a troublesome process especially for architects. Therefore, in this article, the global sensitivity
analysis (GSA) of multi-parameters dominated by building envelop property was implemented in
building load models with different shapes in order to observe the correlation of sensitivity index
and SC; on the other hand, GSA can offer the universality test of the correlation between SC and
building load.
GSA is a big family, which contains various methods, such as standard regression
coefficient based method, variance-based method, screening method, and so on. So, before
starting the analysis, the characteristics of different GSA methods was considered, and Morris
method was a compromise selection for its relative low calculation cost and reliable result. Morris
method can not only present the importance of parameter by sensitivity index (μ * ) but also present
the susceptible level of one parameter’s sensitivity suffered from other parameters through
correlation index (σ/μ * ). Besides, study showed that the correlation of inputs and outputs (such as
linear, monotonic, near monotonic and non-linear non-monotonic) can be described by the value
of correlation index (σ/μ * ) on the basis of statistics analysis. As for its defect of unstable analysis
result, which can be solved by testing the samples with 10, 20, 40, 80 trajectories respectively in
the pre-analysis, and the result showed that the effect of trajectory number is weak on sensitivity
index (μ * ) but relatively strong on correlation index (σ/μ * ), and the final trajectory number could
be identified as 20 trajectories based on pre-analysis.
In next step, 20 building load models with different shapes carried out Morris sensitivity
analysis under Beijing climate by Jeplus and Energyplus. In total 20 shapes was considered,
which can be divided into group A (regular change of shape: square, a series of rectangle)
and group B (irregular change of shape: L-form, U-form, square-atrium, hexagon). A total of
43 parameters dominated by envelop parameters, such as thermal conductivity, construction
thickness, solar heat gain coefficient (SHGC), and window to wall ratio (WWR) were considered. And the sample was set as uniformly distributed value and created by Simlab.
The global sample test shows that under the simple change of storey number or the length-width ratio, SC has a significant and universal
linear positive correlation with heating load. Although there is also a linear correlation between cooling load and SC, the positive or negative
correlation is uncertain. When multiple factors act on the building form, the uncertainty between the SC and heating load will increase obviously,
and the universality shows decrease. However, there is no correlation with cooling load, so it is inappropriate to use SC to evaluate BEP. For the
total load, its correlation with SC depends on its dominant load and the uncertainty is enlarged on current basis. In Beijing, the dominant load is
the heating load, while in hot regions the dominant load should be cooling load. Therefore, only the single scale shape change in the design stage
can better ensure the accuracy of BEP evaluation by using SC.
Among the high-sensitivity parameters of the total load of office buildings in Beijing, the sensitivity of indoor heating and cooling
setting temperature and indoor load parameters rank high, and are slightly affected by the body shape. The sensitivity ranking of WWR,
heat transfer coefficient of window, orientation, thermal conductivity of external wall and roof was relatively low and was significantly
affected by body shape. Even though, only the correlation of orientation could not be determined because σ/μ * was greater than 1. In a
similar climate background, the building envelope is limited by energy conservation standards and has certain similarity in thermal
performance. In this case, the Morris sensitivity analysis results of existing cases can be used to estimate the sensitivity of the parameters
in a target building. For the parameters whose thermal activity was dominated by heat conduction, the error rate of sensitivity index
estimation can be controlled within 10%. This method is helpful to acquire Morris sensitivity index during the evolution of the scheme
design or similar cases, but it is not suitable for areas with abundant solar energy at high altitude or hot areas, because the parameters
whose thermal activity dominated by radiative transfer may have high sensitivity ranking, and the error rate of sensitivity index estimation
of such parameters is commonly more than 10%. |
| Key words: Morris Sensitivity Analysis Heating Load Cooling Load Shape Coefficient Sensitivity Estimation |
