Ewen, J., Parkin, G., Sheffield, J., Chappell, N.A., Vaughan, M.D., Thorne, M.C., and Degnan, P.J., 1999. Validation testing of solute transport modelling using SHETRAN: The Calder Hollow Experiments. Nirex Report N/003. UK Nirex Ltd. (ISBN 1-84029-244-X).
Hill-slope plot experiments were conducted at 'Calder Hollow', West Cumbria, using lanthanum chloride (strongly sorbing) and strontium chloride (weakly sorbing). Water and solute were injected via boreholes into a sand layer in a sequence of Quaternary drift deposits and the resulting subsurface transport plumes studied using core extraction and laboratory measurement of lanthanum, and fluid conductivity and bulk resistivity monitoring of strontium in an array of observation boreholes.
Simulations of the field experiments were run using the SHETRAN hydrological river catchment modelling system. SHETRAN is a physically-based, spatially-distributed (PBSD) system. It was used to simulate the combined subsurface water flow (calibrated against field observations of piezometric head) and solute transport (uncalibrated, and run without any knowledge of the field observations for solute) taking place during the experiments. The results of the simulations were used to gauge the ability of SHETRAN to predict the subsurface transport of the lanthanum chloride and strontium chloride tracers in a validation exercise.
The validation approach is based on the 'blind' method of Ewen and Parkin (1996), and involved using SHETRAN as it would be used in a real application. Ranges were established for the measured values of the physical properties of the porous media, supplemented by values from the literature, and these were used in conjunction with an analysis of calibrated flow simulations to create 48 SHETRAN parameter datasets for the lanthanum experiment and 16 for the strontium experiment. The datasets each led to a physically plausible simulation that is consistent with all the available information. A number of 'tests' were decided on, with which to compare model simulation results against actual field measurements. The simulations were used as a starting point to derive upper and lower 'bounds' to the model generated results, in order to incorporate uncertainty.
Two teams produced two separate sets of bounds; a modelling team from Newcastle University (Water Resource Systems Research Laboratory) and a small group of assessment experts independent of the modelling team. The validation bounds set by each group were different as a result of the different applications that they are required for. The Newcastle University team derived their bounds by incorporating simulation and measurement uncertainty and modifying the ranges on the basis of their confidence in the underlying model. However, it is recognised that validation can also be considered against bounds set on the basis of how well the model is required to perform in order to constitute a useful assessment tool. That is, the bounds can be set in order to test a models 'fitness for purpose' and this was the approach of the performance assessment team.
Each set of bounds for each test was compared with the field observations, which were released by an independent referee only after all the bounds had been set. For the Newcastle University set of bounds, the simulation of the lanthanum experiment was successful, as was the simulation of the peak concentrations for the strontium breakthroughs at the observation boreholes, and the times to peak. The shapes of the strontium breakthrough curves were less well simulated. For the performance assessment related validation exercise, the SHETRAN system performed well, passing 5 out of 7 of the imposed tests with one failure predicting the 50% peak strontium concentrations and another failure in predicting lanthanum profiles for all the boreholes.
This, the first 'blind' testing of the solute transport modelling in SHETRAN, gives some insight into the whole process of applying SHETRAN, and other PBSD models, to real applications, and especially into the handling of uncertainty and the reliability of uncalibrated simulations. Given the rigorous nature of the validation tests undertaken, it is concluded that the SHETRAN system, which is still being developed further, is a powerful tool for application in post-closure radiological safety assessments.
To send me a mail message: