Artificial Infiltration in Rock and Soil - A Comparison of Responses in Central Gothenburg using Numerical Models

Abstract

Underground constructions can cause lowering of the groundwater table, due to groundwater leakage into the structure. To minimize the risk of drawdown, artificial infiltration can be implemented which can maintain or raise the groundwater table to a desired level. Water can be infiltrated either in rock or soil, and observation data has shown that the two materials exhibit different responses. In literature, limited amount of research has been conducted comparing the responses between rock and soil infiltrations and it was therefore of interest to investigate this further. This thesis aimed to compare the responses obtained from rock and soil infiltration using numerical finite difference modelling. To enable a systematic comparison between the different approaches two types of models were created, using Excel and GMS MODFLOW. Consequently the thesis also resulted in a comparison between the models, to conclude which model approach could simulate the observed data but also to see how the characteristics of the models affected the results. In addition, a risk object, which in this case is a building sensitive to groundwater fluctuations, was included to exemplify the results obtained from the models by suggesting the most suitable infiltration approach. The comparison was applied to a stretch of the Västlänken project, which is a railway tunnel currently being constructed in Gothenburg, Sweden. The results from both models show that the response generated from rock infiltration have more of a linear response, compared to soil infiltration which demonstrated an exponential trend. The results from the Excel model were correlated to the observed data when it comes to the propagation of the responses, indicating a more diffusive response in rock compared to soil. The GMS MODFLOW model however, did not show this type of pattern. Both numerical models managed to simulate the differences in responses, but to which extent the results coincide with the observed data varied. For the risk object, it was suggested that rock infiltration might be the best option since a more linear response makes it easier to predict how much the levels will be raised. Further, the exponential response from a soil infiltration would require the infiltration facility to be located close by, which might be difficult since the area is densely built.