Mathematical modelling of groundwater flow at Sellafield, UK

Sellafield in West Cumbria was a potential site for the location of the UK's first underground repository for radioactive, intermediate level waste (ILW). The repository was to lie around 650m beneath the ground surface within rocks of the Borrowdale volcanic group (BVG), a thick suite of SW dipping, fractured, folded and metamorphosed Ordovician meta-andesites and ignimbrites. These are overlain by an onlapping sequence of Carboniferous and Permo-Triassic sediments. In situ borehole measurements showed that upward trending fluid pressure gradients exist in the area of the potential repository site, and that there are three distinct fluid types in the subsurface; fresh, saline and brine (at depth, to the west of the site). Simulations of fluid flow in the Sellafield region were undertaken with a 2D, steady-state, coupled fluid and heat flow simulation code (OILGEN). In both simplified and geologically complex models, topographically driven flow dominated the regional hydrogeology. Fluids trended persistently upwards through the potential repository site. The dense brine to the west of the site promoted upward deflection of topographically driven groundwaters. The inclusion in hydrogeological models of faults and variably saline sub-surface fluids was essential to the accurate reproduction of regional hydraulic head variations. Sensitivity analyses of geological variables showed that the rate of groundwater flow through the potential repository site was dependent upon the hydraulic conductivity of the BVG, and was unaffected by the hydraulic conductivity of other hydrostratigraphic units. Calibration of the model was achieved by matching simulated subsurface pressures to those measured in situ. Simulations performed with BVG hydraulic conductivity 100 times the base case median value provided the “best-fit” comparison between the calculated equivalent freshwater head and that measured in situ, regardless of the hydraulic conductivity of other hydrostratigraphic units. Transient mass transport simulations utilising the hydraulic conductivities of this “best fit” simulation showed that fluids passing through the potential repository site could reach the surface in 15?000years. Simple safety case implications drawn from the results of the study showed that the measured BVG hydraulic conductivity must be less than 0.03myear^-1 to be simply declared safe. Recent BVG hydraulic conductivity measurements showed that the maximum BVG hydraulic conductivity is around 1000 times this safety limit.