Numerical modelling of groundwater flow and radioactive waste migration: Sellafield, England

Wu, Kejian (1999) Numerical modelling of groundwater flow and radioactive waste migration: Sellafield, England. PhD thesis, University of Glasgow.

Abstract

Sellafield in NW England was proposed as a candidate site for subsurface disposal in the UK of Intermediate Level radioactive Waste (ILW). Part of the concept of such a site is that the geosphere functions as one of many barriers against serious leakage. Assessment of overall performance requires predictions of the groundwater flow, which may transport radionuclide into shallow aquifers, or to the surface. This thesis develops a 2-D model, which simulate subsurface fluid flow as a means to aid prediction into the far future. The software uses the finite element method, and is an adaptation of OILGEN (Garven 1989). This permits coupling of rock properties, water properties, and flow resulting from differences of potential- including head, density and heat. Mass transport computations are based on a random walk particle tracking model. Data for this thesis was derived from an extensive site investigation program undertaken by UK NIREX. Conceptually, the regional flow system consists of rainwater falling on 1,000m high ground of the Cumbrian mountains and percolating into the Borrowdale Volcanic Group (BVG), passing westwards by deep flow and returning upwards through the repository site (PRZ), before discharge into overlying sediments and into the Irish Sea. A regional cross section (WSW to ENE) was constructed 115km long and 7.5km deep. This is much longer and deeper than previous studies of hydrogeology in this area. The objective was to reproduce the regional flow and local flow at the repository site. Modelling was carried out progressively, investigating the effects of rock permeability, geometry, anisotropy, faults, salinity and mesh geometry. The approach was to perform a very extensive and prolonged series of sensitivity tests and to adjust each parameter independently to achieve the best fit of predicted groundwater head profiles to head profiles measured in Boreholes 3, 10A and 2. Two sets of best-fit parameters were derived from this calibration exercise. A second stage of model validation used the two calibrated models to predict the streamlines and residence ages of groundwater in the PRZ. These were compared to the in-situ measured chemistry and salinity of groundwater, and to the measured mean residence age of the groundwater. Only one suite of parameters in the modelling was compatible with both head and residence age measured data. This validated model is taken to be the best representation of the natural regional flow system and to simulate the release of radionuclides from the repository and their pathways towards the surface transported by moving groundwater. Modelling shows three flow regimes, similar to measured geochemical data: Shallow Flow (high flux); Inland Flow (small flux); Irish Basin (very small flux). Best fit model parameters are close to the median rock permeability measured in the field. BVG permeability is 0.12m/yr and flow rate at the repository site is 1.0m/yr. Recharge is 16km east of the repository. Critical controls on flow are >2km deep permeabilities of Eskdale Granite and Skiddaw Slate, which have not been sampled by boreholes. Water residence age at the repository is predicted as 0.14-0.15Ma by both streamline and dispersion methods; this compares well with measured 0.03-1.5Ma ages. Leachate from the repository reaches the Calder Sandstone 400m deep aquifer after 25,000yr and the Irish Sea bed at 50,000-80,000yr, 3km west of the repository. It is concluded that the geosphere at this PRZ does not greatly assist performance. Previous local models have not correctly considered very deep flow. A PRZ is better sited on the inflow end, not the outflow end, of a regional groundwater system.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Keywords: Hydrologic sciences, geological engineering.
• Colleges/Schools: College of Science and Engineering
• Supervisor's Name: Haszeldine, Dr. Stuart
• Date of Award: 1999
• Depositing User: Enlighten Team
• Unique ID: glathesis:1999-72824
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 11 Jun 2019 11:06
• Last Modified: 04 Aug 2021 14:20
• URI: https://theses.gla.ac.uk/id/eprint/72824