Nuclear Techniques ›› 2020, Vol. 43 ›› Issue (4): 40009-040009.doi: 10.11889/j.0253-3219.2020.hjs.43.040009

• SPECIAL SECTION ON THE 11TH NATIONAL CONFERENCE ON NEW AND RESEARCH REACTORS (PART I) • Previous Articles     Next Articles

Uncertainty analysis of Gaussian plume model based on Bayesian MCMC method

Weijie CUI1,2,Bo CAO1,2(),Yixue CHEN1,2   

  1. 1.School of Nuclear Science and Engineering, North China Electric Power University, Beijing 102206, China
    2.Beijing Key Laboratory of Passive Nuclear Safety Technology, North China Electric Power University, Beijing 102206, China
  • Received:2020-02-09 Revised:2020-03-05 Online:2020-04-15 Published:2020-04-20
  • Contact: Bo CAO E-mail:caobo@ncepu.edu.cn
  • About author:CUI Weijie, male, born in 1995, graduated from North China Electric Power University in 2018, master student, focusing on nuclear environmental safety
  • Supported by:
    National Natural Science Foundation of China(11605059);Fundamental Research Funds for the Central Universities(2018MS042)

Abstract: Background

Appropriate atmospheric diffusion model is necessary to evaluate the consequences of hypothetical accidents at nuclear power plants. Compared with the traditional uncertainty analysis method, the Bayesian method fully considers the existing observation data. The Markov Chain Monte Carlo (MCMC) method can conveniently combine the Bayesian method and Gaussian plume model.

Purpose

This study aims to analyze the parameter uncertainty for improving the credibility of model prediction by using Bayesian MCMC based Gaussian plume model (GPM).

Methods

The Bayesian method and Gaussian plume model were combined to be adopted in the MCMC method. Firstly, the sensitivity of the GPM to several important parameters was analyzed by changing the value of one variable at a time, and then the two parameters with the highest sensitivity were selected using the Bayesian MCMC method for uncertainty analysis.

Results

By analyzing the MCMC sample sequences, the best fitting of observation data and confidence intervals of simulation results are obtained.

Conclusions

The Bayesian method provides a more reliable confidence interval, which can provide better reference data for emergency response after the accident.

Key words: Gaussian plume model, Uncertainty analysis, Bayesian MCMC method, Confidence interval

CLC Number: 

  • TL99