Uncertainties in geochemical models of natural systems: Implications for performance assessments
- PDF / 75,156 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 93 Downloads / 131 Views
CC3.3.1
Uncertainties in geochemical models of natural systems: Implications for performance assessments Christopher S. Palenik1, Keld A. Jensen2 and Rodney C. Ewing1,3 1
Department of Geological Sciences, University of Michigan, Ann Arbor, MI 48109-1063, USA. National Institute of Occupational Health, Denmark, Lersø Parkallé 105, DK-2100 Copenhagen Ø, Denmark. 3 Department of Nuclear Engineering & Radiological Sciences, University of Michigan, Ann Arbor, MI 48109-2104, USA. 2
ABSTRACT Licensing a geologic repository for nuclear waste will be based primarily upon the results of a performance assessment (PA), which evaluates compliance of the repository with radiation exposure limits by modeling the engineered and natural barrier systems over extended periods (104 to 106 years). The properly completed PA must include an analysis of the sources of uncertainty, as well as the range of input parameter values, in order to determine whether the analysis is “robust”, “realistic” or “conservative”. This paper examines the sources of uncertainty that may be encountered in source-term and near-field models in a performance assessment of a repository for spent nuclear fuel by examining the uncertainties in modeling the geochemical behavior of the Bangombé natural reactor in Gabon. INTRODUCTION In U.S. regulations, a performance assessment (PA) “identifies the features, events, processes, and sequences of events (except human intrusion)” that might affect the geologic repository system and their probabilities of occurring during the 10,000 years after disposal [1]. In practical terms, the PA is composed of a cascade of coupled models that describe specific processes (e.g., source-term, near-field and far-field processes). Each of these individual models introduces uncertainties into the overall model through a variety of sources: conceptual flaws in the model, uncertainties in the value of parameters used by the model, and difficulties in applying the model at different scales. For individual models coupled within a PA, the total uncertainty is the multiplied uncertainty of the individual models propagated through the analysis [2]. To address uncertainty, EPA regulations require that the PA “account for uncertainties and variablilites in parameter values and provide for the technical basis for parameter ranges, probability distributions, or bounding values used [3].” While the magnitude of uncertainty in parametric values can be quantified through experimentation, other uncertainties are distinctly qualitative, particularly those pertaining to alternative conceptual models. To capture the scale of these types of uncertainties in the model, the PA requires support from field and laboratory tests, site monitoring, natural analogue studies and extensive use of expert judgment [3]. At present, it remains unclear exactly how these supporting studies are to be utilized, and there is no single or accepted means of characterizing the uncertainty in the analysis. This paper uses the analysis of a natural analogue site, the Bangombé n
Data Loading...
