Modelling mine water rebound – benefits and drawbacks of two common approaches

Modeling of mine water rebound in abandoned underground mines is an ongoing challenge for post-mining management, for example in the former Ruhr hard-coal mining area in Germany (Kessler et al. 2020). Apart from the estimation of flooding periods, public and political decision-makers want to know future mine water levels and flow rates at specific locations, e.g. surface water bodies, towns or industrial areas in the vicinity of the mine, and hydrogeological reactions in the neighbored groundwater bodies. Different numerical approaches have been developed and adapted to estimate the dynamics of mine water rebound. The most common approach are so called pond-and-pipe models that are based on volume balances of spatially delimited mining volumes and special hydraulic connections between the volumes, e. g. roadways or shafts (e.g. DMT Gmbh & Co. KG 2011, Kortas and Younger 2007). The methodological approach of these numerical models is often according to the finite volume method. The advantage of this method is a variable delineation of mining volumes depending on water transfer points. This facilitates a minimum number of mesh elements and the inclusion of many collieries within one large mining area. Mine water rebound arises out of known water in- and outflows and the estimated void volume of each model cell, typically calculated for the entire mine building. If water levels at specific points within the mine building or in the adjacent rock matrix are of interest, models require a fine-scale meshing algorithm. Finite-element models proved to be a suitable approach to discretize mining structures, voids and surrounding rock matrix at a reasonable scale. Such models need to be parameterized at every model cell and thus, are built on a broader data basis of hydraulic parameters. This contribution compares the application of the two described modeling approaches by means of a closed mining site in the Ruhr area, where mine water rebound has been observed for some 15 years. Focus was laid on the required parameter sets, the calibration, uncertainty and the interpretation of the model outcomes. First results show a high complexity regarding the parameterization of spatially refined finite-element models. It seems important to identify a reasonable level of abstraction in order to keep the model functional, but at the same time reduce the model uncertainties to an acceptable standard. However, if properly setup, such models have greater capabilities and can reduce the uncertainties, if continuously optimized with monitoring data collected during the ongoing groundwater rebound. And they also proof the results of the coarsely discretized mine water models according the finite volume method.