Preliminary uncertainty and sensitivity analysis for basic transport parameters at the Horonobe Site, Hokkaido, Japan.
- Gram Incorporated, Albuquerque, NM
Incorporating results from a previously developed finite element model, an uncertainty and parameter sensitivity analysis was conducted using preliminary site-specific data from Horonobe, Japan (data available from five boreholes as of 2003). Latin Hypercube Sampling was used to draw random parameter values from the site-specific measured, or approximated, physicochemical uncertainty distributions. Using pathlengths and groundwater velocities extracted from the three-dimensional, finite element flow and particle tracking model, breakthrough curves for multiple realizations were calculated with the semi-analytical, one-dimensional, multirate transport code, STAMMT-L. A stepwise linear regression analysis using the 5, 50, and 95% breakthrough times as the dependent variables and LHS sampled site physicochemical parameters as the independent variables was used to perform a sensitivity analysis. Results indicate that the distribution coefficients and hydraulic conductivities are the parameters responsible for most of the variation among simulated breakthrough times. This suggests that researchers and data collectors at the Horonobe site should focus on accurately assessing these parameters and quantifying their uncertainty. Because the Horonobe Underground Research Laboratory is in an early phase of its development, this work should be considered as a first step toward an integration of uncertainty and sensitivity analyses with decision analysis.
- Research Organization:
- Sandia National Laboratories (SNL), Albuquerque, NM, and Livermore, CA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- AC04-94AL85000
- OSTI ID:
- 918366
- Report Number(s):
- SAND2003-3156; TRN: US0805360
- Country of Publication:
- United States
- Language:
- English
Similar Records
Simplified predictive models for CO2 sequestration performance assessment
Mach Conference, 2022.