A Numerical Study Of Uncertainty Of Shaft Smoke Model Parameters Based On Bayesian Approach

The fire appears brought the light and warmth to human beings, but the fire also brought incalculable losses to mankind. In recent years, economic development and progress of science and technology become a booster to prosperity of human society. Meanwhile the urban architecture become more complex and tall, these tall buildings make the city modernization, at the same time, there are also huge fire hazard in buildings.Faced such a fire hazard, scholars from various countries are constantly studied, in order to be able to control the occurrence and development of fire. In this paper, Bayesian inference method applied to analysis the parameters of shaft model, using the probability distribution to quantify the uncertainty of shaft model parameters and the influence on neutral plane, hoping to provide a reference to design of smoke control in buildings. Bayesian method as a theoretical method which keep pace with classical statistical methods, and it has been applied to biostatistics, actuarial, image analysis, decision-making events, criminology incomplete count and other fields, and have achieved significant effect. While in the field of fire safety engineering, the application of the Bayesian approach is still less, and lack the corresponding research papers. Shaft structure is an important part of high-rise buildings. The neutral plane is the key issues to buildings smoke control, because of it will impact the smoke movement when the building in fire.This paper selects the shaft structure of high-rise buildings and neutral plane as the basic model, the focus of research in two parts. The first part is the establishment of high-rise building shafts model by CFD (Computational Fluid Dynamics) software and calculate the neutral plane position distribution under the parameters changes. In order to verify the applicability of the model, an example is provided, that the smoke temperature similarity in the shaft as the standard, the flow similarity theory as the basis, using FDS software to establish three different scale shaft models, and conducted calculation under different conditions, to obtain the requirements of similarity in small scale shaft models, the Froude number should be preserved, and the turbulence level should always be ensured by compared the simulated results. Meanwhile, other configuration parameters also can influence the results of reduced-scale experiments, like the area of fire source, fire position, ventilation conditions and so on. And then we calculate the position distribution of neutral plane under the variation of top vent area and heat release rate, the data will be applied to further analysis in next part. In the second part, introducing the Bayesian method and uncertainty theory, using the data of neutral plane to study the influence of the parameter uncertainty of shaft model. In general fire dynamics models, the input parameters tend to have a certain degree of uncertainty, quantifying these parameters uncertainty have important significance to design fire safety engineering scientifically and authoritatively. In this part, the data of neutral plane as the sample value, using Bayesian method to quantify the uncertainty of top vent area and heat release rate, and then obtained the corresponding PDF(Probability Distribution Function) and CCDF(Complementary Cumulative Distribution Function), and discussed the impact of these two parameters uncertainty on the position of neutral plane distribution. Aimed to provide some reference for fire safety engineering design, shaft model design and the smoke-control design in buildings. In this paper, Bayesian methods and neutral plane model combined organically, for the position of the neutral plane problems, breaking the exact location of using empirical formula by the deterministic method. Such an approach can provide recommendations for improve the designing and make the final design results more scientific credibility.