The application of Markov chain Monte Carlo method for the estimation of hydrogeological parameters

• Main Authors: Chi-HuaChu, 朱紀樺
• Other Authors: Kuo-Chin Hsu
• Format: Others
• Language: en_US
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/76754601272629525432

碩士 === 國立成功大學 === 資源工程學系 === 103 === Nuclear waste deposit is commonly isolated by buffer material, such as bentonite, to prevent its leak from deposit cane. Therefore, the hydrogeological property of buffer material is the key issue for the success of nuclear waste deposition. Lee et al. (2013) performed an experimental work to explore the diffusion coefficient of Bentonite (MX-80) which is used as the buffer material of nuclear waste deposits. Scale effect was found in the diffusion coefficient. The result contradicts to the stochastic theory which states that the scale effect appears for the dispersion coefficient but not the diffusion coefficient. We reexamine the experimental data to explore the issue. Both analytical solutions of diffusion and advection-dispersion equations (ADE) were applied to estimate the parameters. Considering the micro-heterogeneity of bentonite, Markov chain Monte Carlo (MCMC) method is used to analyze the velocity, dispersion and diffusion coefficients of the breakthrough data from column tests. The results show that the experiment is influenced by the velocity. Diffusion model generates significant error in matching the breakthrough data. ADE model which considers velocity and dispersion performs better than the diffusion model. Scale effect is found in dispersion coefficient. Dispersion coefficient increases linearly with experimental lengths. We conclude that ADE model is better representative for the solute transport experiment of bentonite material.