Title: Multi-scale aquifer characterization: From outcrop analogue, direct-push and borehole investigations towards improved groundwater flow models
Other Titles: Multischaal aquifer karakterisatie: van analoge ontsluitingen, direct-push technieken en boringen naar verbeterde grondwaterstromingsmodellen
Authors: Rogiers, Bart; S0160956
Issue Date: 27-Nov-2013
Abstract: Groundwater flow and solute transport modelling are affected by different kinds of uncertainty including spatial variability in aquifer properties such as hydraulic conductivity (K). This spatial variability is often present at different scales, and consequently, effective K values are very much scale-dependent. While regional-scale models often use effective K values obtained by inverse modelling, the small-scale variability has to be accounted for as well, as it has been shown that even submeter-scale heterogeneity can have important consequences on solute transport in aquifers. To support decision making related to environmental impact assessment for waste disposal sites or sites contaminated by point sources, appropriate subsurface characterization and modelling tools are thus required for accounting for subsurface heterogeneity, possibly observed at multiple spatial scales. These tools can further be used to quantify the uncertainty associated with groundwater flow and solute transport, to underpin strategies for long-term groundwater quality monitoring at disposal or contaminated sites, or for developing groundwater remediation schemes.In this work, a methodology is developed for efficient multi-scale subsurface characterization and integration of the gathered data in a stochastic regional groundwater flow and solute transport modelling approach. Different kinds of aquifer characterization technologies have to be combined to cover the centimetre- to the kilometre-scale, and to make optimal use of common or easily gathered secondary data. Additionally, for accounting for secondary data, the development of a set of tools for data calibration and interpretation is required. This is achieved by using different kinds of measurements from outcrop analogues, borehole and direct-push investigations, and by using innovative methods and techniques to obtain a sound framework for integrating all data.The case study that we use throughout the thesis is an area of ~60 km² in Mol/Dessel, Belgium, of which the subsurface consists of a succession of dipping lithostratigraphical units with varying degrees of heterogeneity, all part of the Neogene aquifer. We make use of different previous hydrogeological studies that were performed in the framework of the ONDRAF/NIRAS radioactive waste disposal programmes, and perform additional site characterization.The assessment of outcrops as analogues for the subsurface sediments shows that very useful quantitative and qualitative information can be retrieved from outcrops, but a systematic bias seems to exist between surface and subsurface data (lower K for the latter). The relative differences however seem to be valid, as is the amount of spatial heterogeneity. For including small-scale heterogeneity in the developed large-scale transport model, we make use of dispersivities estimated from the outcrop characterization.The three types of secondary K data used in this work are air permeability, grain size and cone penetration test data. For each of these we use a data-driven modelling approach to obtain estimates of K. The complexity of these approaches ranges from a simple linear model to the combination of artificial neural networks with general likelihood estimation. Each time, the site-specific estimates prove to be superior to existing models from literature.The conditioning of a regional groundwater flow model on borehole and geotechnical and hydraulic direct push data improves the model performance considerably. We use a data-driven approach for the hydrostratigraphy, by invoking a non-stationary multivariate geostatistical framework for conditioning the model. A combination of different McMC algorithms is used to estimate the uncertainty of the flow field, for the quantification of uncertainty on the corresponding solute transport. The combined algorithm is tested both in McMC sampling and optimization mode. The latter is clearly more efficient for CPU-intensive models, but only provides an approximation of the posterior.Solute transport simulations in the framework of surface disposal of radioactive waste, based on the obtained flow solutions, suggest that the reference model that considers homogeneous lithostratigraphical units produces conservative results in terms of maximum concentrations within the solute plume. For more detailed results, transport simulations using another advection solution scheme and finer numerical discretization are however recommended.
Publication status: published
KU Leuven publication type: TH
Appears in Collections:Division of Geology

Files in This Item:
File Status SizeFormat
PhD thesis - Bart Rogiers.pdf Published 26694KbAdobe PDFView/Open


All items in Lirias are protected by copyright, with all rights reserved.