Numerical modelling of solute transport at Forsmark with MIKE SHE. Site descriptive modelling SDM-Site Forsmark

The Swedish Nuclear Fuel and Waste Management Company (SKB) is performing site investigations at two different locations in Sweden, referred to as the Forsmark and Laxemar areas, with the objective of siting a final repository for high-level radioactive waste. Data from the site investigations are used in a variety of modelling activities. This report presents model development and results of numerical transport modelling based on the numerical flow modelling of surface water and near-surface groundwater at the Forsmark site. The numerical modelling was performed using the modelling tool MIKE SHE and is based on the site data and conceptual model of the Forsmark areas. This report presents solute transport applications based on both particle tracking simulations and advection-dispersion calculations. The MIKE SHE model is the basis for the transport modelling presented in this report. Simulation cases relevant for the transport from a deep geological repository have been studied, but also the pattern of near surface recharge and discharge areas. When the main part of the modelling work presented in this report was carried out, the flow modelling of the Forsmark site was not finalised. Thus, the focus of this work is to describe the sensitivity to different transport parameters, and not to point out specific areas as discharge areas from a future repository (this is to be done later, within the framework of the safety assessment). In the last chapter, however, results based on simulations with the re-calibrated MIKE SHE flow model are presented. The results from the MIKE SHE water movement calculations were used by cycling the calculated transient flow field for a selected one-year period as many times as needed to achieve the desired simulation period. The solute source was located either in the bedrock or on top of the model. In total, 15 different transport simulation cases were studied. Five of the simulations were particle tracking simulations, whereas the rest were performed with the MIKE SHE advection-dispersion module. Sensitivity analyses were made in order to study the effect of the number of computational layers, i.e. the relation between numerical dispersion and the vertical grid resolution, the influence of dispersion in the saturated zone, and the influence of sorption in the saturated zone. In the advection-dispersion simulations, input concentrations were given either as a pulse source or as a constant source in the bedrock layer at c. 140 m.b.s.l. (metres below sea level), or as a constant infiltration source in the top layer. In the particle tracking simulations reported here, particles were initially introduced in the top layer or in the deeper bedrock layer at approximately 140 m.b.s.l. In some simulation cases, particles were introduced through a concentration source in the deeper bedrock layer corresponding at c. 140 m.b.s.l. or as an infiltration source in the top layer. These scenarios were modelled primarily to enable comparisons with the corresponding advection-dispersion cases. The simulation results show that the solute introduced at a level of 140 m.b.s.l. was transported both upwards and downwards, although the main transport direction was upwards. The solute that was transported towards the ground surface was mainly transported through deformation zones below the lakes and the watercourses. In some of these zones, transport was rather fast and the solute was transported to the surface in only a few months. However, the solute was also transported towards the sea, but the transport towards the sea was much slower than that in the deformation zones. In large parts of the area, the transport towards the sea was very slow, although in some highly water-conductive parts the process was faster. In these zones with higher horizontal hydraulic conductivity in the bedrock, the solute was mainly transported in the horizontal direction towards the sea, cutting off the vertical transport upwards. Furthermore, results from the transport simulations illustrate that the applied infiltration source produces solute mass in recharge areas only, although a horizontal solute transport from higher altitude recharge areas to lower-lying discharge areas (e.g. lakes) would be expected. It seems that on the time scale considered in the present modelling the littoral zones act as hydraulic barriers around some of the lakes. This means that the spreading through horizontal transport in the upper layers in these areas is less important than the effects of vertical flow directions. The dispersion processes transports solutes from the advective zones into more stagnant zones. The results show that this process is important when evaluating the risk of solute spreading into zones in which the transport pattern is more diffuse. Therefore, it is important to estimate the dispersion coefficients and to avoid numerical dispersion. Furthermore, since sorption leads to a delay in the peak arrival time, it is an important factor to include in simulations when the estimation of the solute peak arrival time is crucial.