基于敏感性筛选及全局因子耦合的高层楼梯井烟气参量分析研究

发布时间：2018-04-18 16:16

本文选题：高层建筑 + 楼梯井　；参考：《中国科学技术大学》2017年硕士论文

【摘要】：高层楼梯井作为火灾发生时高层建筑唯一的逃生通道,是火灾研究中的一类重要结构。对这种特定结构火灾进行的敏感性分析可以得出影响因素对特定输出量的敏感性排序、分析各因素的作用、筛选重要因素并简化模型。敏感性分析中的筛选方法可以计算出反应系统内各影响因素大致的敏感性指数,可用于前期筛选出重要的因子。敏感性分析中的全局敏感性方法可以分别得出系统内影响因素的一阶敏感性和总敏感性指数。前者包含了该因素自身对系统输出不确定性的影响,其数值可为控制系统不确定性提供参考。后者包含了此因子变动给输出响应带来的所有影响,其与前者的相对差值也可用于分析此因子与其它因子耦合作用。另一方面,通过数值模拟实验可以大大减少研究成本。但是,高层楼梯井结构的精确模型使对其的研究即使是数值模拟仍需要很大的计算量,而使用计算量小的模型则会对结果造成较大偏差。基于蒙特卡洛方法的敏感性分析对实验重复次数要求较高,因此高层楼梯井结构的复杂结构使对其的敏感性分析研究大大受限于其计算成本。本文基于方差分析的全局敏感性分析方法,在不影响敏感性结果的前提下,减小了原计算量接近50%。针对敏感性分析的前期筛选过程和正式分析过程,本文分别建立了敏感性筛选平台及多因子耦合的全局敏感性分析平台。本文通过MATLAB实现两个平台的各个功能,结合CFD计算,完成了针对高层楼梯井火灾烟气各个特征量的敏感性筛选及多输入因子耦合的全局敏感性分析。通过筛选平台,经过70次工况计算,筛选掉影响较小的输入因子。在筛选掉几乎无影响的环境压强后,本文对高层楼梯井结构火灾进行了全局敏感性分析。全局敏感性分析平台选用了基于方差的全局敏感性分析方法。在此平台中,经过312次工况计算,得出建筑尺寸、热释放速率、环境参数对高层楼梯井内烟气湍流特性及输运特性的一阶敏感性和全局敏感性。结果显示火源热释放速率在各个所考察特征量中均具有较大影响,且在火灾发展阶段时的湍流温度均值、法向速度均值等响应中与其它输入因子相互耦合作用较小。而针对温度湍流脉动强度、标量速度均值等输出响应中,热释放速率与火源房间楼梯井连通口等因子之间相互耦合作用强。针对不同输出响应,敏感性分布不同,应具体分析。此外环境参数对本文所考察8个输出响应均影响很小。
[Abstract]:As the only escape way for high-rise buildings, high-rise stairwells are one of the most important structures in fire research.The sensitivity analysis of the fire of this particular structure can get the order of sensitivity of the influencing factors to the specific output, analyze the role of each factor, screen the important factors and simplify the model.The screening method in sensitivity analysis can calculate the general sensitivity index of each influencing factor in the reaction system and can be used to screen out important factors in the early stage.The global sensitivity method in sensitivity analysis can obtain the first order sensitivity index and the total sensitivity index of the influencing factors in the system.The former includes the influence of this factor on the system output uncertainty, and its value can be used as a reference for controlling the uncertainty of the system.The latter includes all the effects of the variation of this factor on the output response, and the relative difference between the latter and the former can also be used to analyze the coupling between this factor and other factors.On the other hand, the research cost can be greatly reduced by numerical simulation experiments.However, the precise model of the high-rise stairwell structure requires a large amount of calculation even for numerical simulation, while the use of a small computational model will cause a large deviation to the results.The sensitivity analysis based on Monte Carlo method requires a high number of repeated experiments, so the complex structure of the high-rise stairwell structure greatly limits the sensitivity analysis to its computational cost.In this paper, the global sensitivity analysis method based on ANOVA is presented, without affecting the sensitivity results, and the original computational complexity is reduced to 50%.In this paper, a sensitivity screening platform and a multi-factor coupling global sensitivity analysis platform are established for the pre-screening process and the formal analysis process of sensitivity analysis.In this paper, the functions of the two platforms are realized by MATLAB, and combined with CFD calculation, the sensitivity screening for each characteristic quantity of fire smoke in high-rise stairwell and the global sensitivity analysis of multi-input factor coupling are completed.Through screening platform and 70 working conditions, the input factors with less influence are screened out.After screening out the almost unaffected ambient pressure, the global sensitivity analysis of high-rise stairwell structure fire is carried out in this paper.The method of global sensitivity analysis based on variance is used in the global sensitivity analysis platform.In this platform, the first order and global sensitivity of building size, heat release rate and environmental parameters to the characteristics of flue gas turbulence and transport in high-rise stairwells are obtained after 312 working conditions.The results show that the heat release rate of the fire source has a great influence on each characteristic quantity investigated, and the interaction between the turbulent temperature mean and the normal velocity mean and other input factors in the development stage of the fire is less than that of the other input factors.For the temperature turbulence intensity, scalar velocity mean and other output response, the heat release rate and the fire source room stairwell connectivity and other factors are strongly coupled.According to the different output response, the sensitivity distribution is different, should be analyzed concretely.In addition, the environmental parameters have little effect on the eight output responses investigated in this paper.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TU998.1

【参考文献】