Numerical Assessment of Steady State and Transient Flow Effects on Groundwater Contaminant Transport

Sammanfattning

Climate change and contamination sources are putting groundwater as drinking water resource under threat. An important tool to predict alterations in groundwater systems is groundwater contaminant transport modelling. In groundwater modelling, steady state and transient flow approaches can be used to balance accuracy and computational requirements for solving the model. This thesis investigates how different flow solutions affect predicted contaminant concentrations for Swedish climatological and hydrogeological conditions. The objective is to formulate guidelines for selecting the optimal modelling approach. In my thesis, I implemented three modelling strategies using the software FEFLOW: i) steady state flow and transport, ii) transient flow and transport with different groundwater recharge averaging periods, iii) steady state flow and transient transport considering model parameter uncertainty in a Monte Carlo simulation. I calculated groundwater recharge with an adapted Rodhe method. A simplified 3D box model with observation points was used to simulate contaminant transport over 10 or 100 years. I evaluated the simulation results with random forest regression to determine feature importance. Predicted contaminant concentrations varied under parameter uncertainty. Highest observed ranges in concentrations after 40 years at one observation point were between 0-60 mg/L. Most of the transient flow and transport predictions were within the interquartile uncertainty bands of the steady state flow and transient transport realisations. Observed steady state flow and transport and transient flow and transport concentrations converged after 10-40 years simulation time. Computational runtimes for different modelling approaches varied substantially. Transient flow and transport simulations considering daily groundwater recharge for a 10-year period versus steady state flow and transport calculations had a run time ratio of 4680:1. I found that model parameter uncertainty plays a more important role in groundwater contaminant transport modelling than temporal groundwater recharge resolution. Considering long term predictions (> 10-40 years), steady state flow and transport simulations can provide good results under minimal computational effort. Transient flow and transport simulations are necessary for short-term investigations or scenarios considering variable flow directions. A compromise can be made with a steady state flow and transient transport approach, to keep computational requirements low while the simulation accuracy stays high. These findings provide valuable guidelines for groundwater modelling practitioners.