Evaluation of two conceptual approaches for groundwater flow simulation for a rock domain at the block-scale for the Olkiluoto site, Finland

doi:10.1016/j.enggeo.2015.05.003

Engineering Geology

Volume 193, 2 July 2015, Pages 297-304

https://doi.org/10.1016/j.enggeo.2015.05.003 Get rights and content

Highlight

•
Groundwater flow in a fractured rock with background fractures was simulated.
•
Background fractures were considered in the simulation by EPM and hybrid approaches.
•
Hybrid approach showed reasonable groundwater pathways in the study site.
•
Multiple field data were required to calibrate appropriate groundwater flow model.

Abstract

To evaluate conceptual approaches for simulating a rock domain, which is a rock unit outside a fracture zone, the block-scale groundwater flow system of the radioactive waste disposal site in Olkiluoto, Finland, was simulated using two approaches: the equivalent porous medium (EPM) approach and the hybrid approach. For the EPM approach, the rock domain was considered to be a homogeneous continuum with an equivalent hydraulic conductivity, while in the hybrid approach, it was conceptualized as a heterogeneous continuum from the stochastically generated background fracture network. In both approaches, several large fracture zones were included in the simulation models, which were calibrated by comparing the simulated hydraulic heads to the observed ones. The results showed that the simulated heads were in good agreement with the observed ones in both approaches, although the root mean square errors of the EPM approach were smaller than those of the hybrid approach. The estimated groundwater flow pathways in the hybrid approach generally had more complex geometries and longer travel times than those from the EPM approach. The simulations were evaluated by comparing the simulated flow rates at the boreholes in the site to the observed ones from the flow loggings, and the results show that the hybrid approach reproduced the observations more similarly than did the EPM approach.

Graphical abstract

Introduction

When an underground radioactive waste repository is constructed, the impact of a potential leakage of the disposed wastes on the underground and surface environments should be assessed. The majority of the nuclides released from the repository can be transported by groundwater flow. Therefore, to demonstrate the safety of the disposal of radioactive wastes for a very long time period, as well as the fate of nuclides potentially released from a repository, the groundwater flow system of a disposal site should be characterized and conceptualized mathematically for numerical simulation (SKB, 2011, Posiva, 2009). SKB (2011) proposed a stepwise conceptualization of a groundwater flow system according to the scales of concern in order to prepare groundwater pathway data for a safety assessment of deep geological disposal. A regional scale groundwater flow system was conceptualized to identify the overall hydrogeological environment, including major hydrogeological units and large fracture zones. Then, a block-scale groundwater flow was conceptualized to obtain more detailed groundwater flow information.

Because Olkiluoto Island has been considered a deep geological disposal site for spent fuels in Finland, many geological and hydrogeological investigations have been carried out at this site (Posiva, 2009). Several large fracture zones were identified at the surface, and boreholes and various hydraulic tests were conducted to estimate the hydraulic properties of the fracture zones and rock domains (Posiva, 2009). Hydraulic interference tests were performed in the study area, and the hydraulic heads and local groundwater flow rates were monitored at multiple points in the boreholes (Rouhiainen and Pöllänen, 2003). In our previous study (i.e., Ko et al., 2012), the regional groundwater flow system of the site was simulated and calibrated using data from the hydraulic tests performed at the deep boreholes. In this study, as a next step for the hydrogeological analysis of the site, a block-scale groundwater flow model was designed for the detailed evaluation of the groundwater flow pathways from the expected disposal depth, which could be required in a safety analysis of the repository. As the size of the modeling domain decreases from regional- to block-scale, some minor water conductive features, such as background fractures, might have a significant influence on the groundwater flow pathways. In this study, approaches for considering the background fractures were evaluated using the hydraulic interference test results from Rouhiainen and Pöllänen (2003).

Several studies have evaluated the effect of the background fractures on block-scale groundwater flow simulations for fractured rock aquifers, especially at the sites where the underground research laboratories (URLs) and nuclear-related facilities are located. Gómez-Hernández et al. (2001) conceptualized the rock domain using the stochastic continuum approach to simulate a pumping test at the Sella-field site. The rock domain was discretized into cubic cells, and the hydraulic conductivity of each cell was assigned geostatistically by considering the fractures crossing it. To simulate the block-scale groundwater flow at the Äspö site, where the URL of Sweden is located, Svensson (2001) considered the background fractures using the hybrid approach, where a heterogeneous hydraulic conductivity field of a rock domain for groundwater flow modeling was introduced from a discrete fracture network of the background fractures. In this approach, the background fractures were randomly generated using the fractal property from the distribution of fractures observed, and the hydraulic conductivity field of the simulation domain was directly determined by calculating the discharges of each grid cell. Svensson showed that his approach could reproduce the observed hydraulic head and salinity using the background fracture data. Hendricks Franssen and Gómez-Hernández (2002) applied the stochastic continuum approach to generate the hydraulic conductivity field for a block-scale rock domain of the Äspö site using an assumed variogram based on the observed hydraulic conductivities from the hydraulic tests. Selroos et al. (2002) evaluated the uncertainty of block-scale groundwater flow models for the Äspö site for different conceptual models: the stochastic continuum approach using geostatistical simulation; the discrete fracture network approach, which included randomly generated fractures using the site data; and the channel network approach that considers an uneven fracture surface, a preferential path and channeling in a single fracture. The simulated particle travel time and specific discharge were different among these models, although the minimum and median values of travel time were similar. The results of these different models indicated that hydraulic constraints, such as boundary conditions, were dominantly significant, and the influence of conceptual models for a rock domain was limited at the Äspö site.

The objective of this study is to evaluate the uncertainty in a block-scale groundwater flow model for the Olkiluoto site related to the conceptual approach for a rock domain. The equivalent porous media (EPM) and hybrid approaches were compared and evaluated. The large fracture zones were deterministically considered to be permeable zones in both approaches. For the EPM approach, the rock domain was considered to be a continuum with a representative hydraulic conductivity. In the hybrid approach, however, a discrete fracture network (DFN) for the background fractures was constructed stochastically using the statistical properties of the fractures, and a spatially heterogeneous hydraulic conductivity field was introduced from the DFN. The groundwater flow models conceptualized with the two approaches were calibrated using the hydraulic head data observed at the study site. Using the calibrated models, particle tracking simulations were performed to estimate the groundwater flow pathways. Finally, the EPM and hybrid approaches were evaluated by comparing the simulated flow rates at the boreholes to the observed flow rates from flow loggings at boreholes.

Section snippets

Study site

Olkiluoto Island, the study area, is located at the southwestern part of Finland and separated from the mainland by a narrow strait connected to the Baltic Sea (Fig. 1). Composite gneiss and granite with tonalite veins are mainly distributed as bedrocks (Anttila et al., 1999). In the central part of Olkiluoto Island, covering the modeling domain, the subsurface environment, including the groundwater flow system in the bedrocks, was investigated using several boreholes used as an underground

Approach

A three-dimensional mesh having 862,272 nodes and 838,660 elements was prepared to simulate the groundwater flow in the modeling site, which was the central part of Olkiluoto Island (Fig. 2, Fig. 3). The size of the modeling domain was 500 m × 500 m × 600 m, and the bottom of the domain was located 600 m below sea level. The observation boreholes were located at the central part of the domain (Fig. 1). In our previous study, the regional groundwater flow for Olkiluoto Island was analyzed (Ko et al.,

Results

The groundwater flow model applying to the EPM approach was calibrated by minimizing the total SSE shown in Eq. (1) with 191 measurement data at the observation boreholes with open and packed-off conditions during the no-pumping period and interference tests (Rouhiainen and Pöllänen, 2003, Klockars et al., 2006). The recharge rate was calibrated to 15.33 mm/yr. Fig. 4a shows the comparisons between the observed and simulated hydraulic heads at the boreholes, and the root-mean-squared errors

Discussion and concluding remarks

For the simulation of the block-scale groundwater flow in a fractured rock with large fracture zones and background fractures, the rock domain was conceptualized using the EPM approach and the hybrid approach, and the applicability of each approach was evaluated using the observed hydraulic heads and flow rate in the boreholes. In the comparison between the simulated and observed hydraulic heads, the residuals in the no-pumping condition were less than those in the pumping condition, and the

Acknowledgement

This work was supported by the Nuclear Research and Development Program of the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (MSIP) (No. 2012M2A8A5025589). This work is related to Äspö Task Force 7. We appreciate SKB and Posiva Oy for their help to provide the data used in this study.

References (20)

J.J. Gómez-Hernández et al.
Stochastic analysis of flow response in a three-dimensional fractured rock mass block
Int. J. Rock Mech. Min. Sci.
(2001)
J.-O. Selroos et al.
Comparison of alternative modelling approaches for groundwater flow in fractured rock
J. Hydrol.
(2002)
U. Svensson
A continuum representation of fracture networks. Part II: application to the Äspö Hard Rock laboratory
J. Hydrol.
(2001)
H. Ahokas et al.
Modelling of hydro-zones for the layout planning and numerical flow model in 2006
P. Anttila et al.
Final disposal of spent nuclear fuel in Finnish bedrock — Olkiluoto site report
H.-J.G. Diersch
FEFLOW: finite element subsurface flow and transport simulation system
H.-J.G. Diersch
Discrete feature modeling of flow, mass and heat transport processes by using FEFLOW
H.J.W.M. Hendricks Franssen et al.
3D inverse modelling of groundwater flow at a fractured site using a stochastic continuum model with multiple statistical population
Stoch. Env. Res. Risk A.
(2002)
J. Holmen et al.
Olkiluoto Geo-DFN Model, Draft Version 2
(2008)
S.-H. Ji et al.
A hybrid modeling approach to evaluate the groundwater flow system at the low- and intermediate-level radioactive waste disposal site in Gyeong-Ju, Korea
Hydrogeol. J.
(2012)

There are more references available in the full text version of this article.

Cited by (10)

Optimal selection of estimator for obtaining an accurate three-dimensional rock fracture orientation distribution
2020, Engineering Geology
Citation Excerpt :
The geometric model and the hydraulic model are fundamental models of rock structure, which have wide applications across multiple rock related engineering areas. Examples include stability evaluations of slopes, tunnels, foundations, reservoirs, and wells, permeability assessments of water conservancy projects, fluid flow conductivity analyses of CO2 storage, and production predictions of gas and oil reserves (Elmo et al., 2018; Fereshtenejad et al., 2016; Fisher et al., 2014; Ko et al., 2015; Ma et al., 2019; Reeves et al., 2013; Zheng et al., 2018). Thus, the accurate estimation of the 3D orientation distribution will benefit these engineering practices.
Advances in estimating the three-dimensional (3D) orientation distribution include i) the stereology-based Fouché estimator (FE); and ii) its recent modification (MFE), which has one less meshing step and has three modified components compared with the FE. One way to achieve an accurate estimate is by selecting the optimum estimator. Here, we developed a recognizer for the optimum estimator to help automatically select the most likely accurate estimator for use in the FE and MFE. Before this, we conducted a systematic comparison of the likelihood of obtaining an accurate estimation to clarify for which patterns FE and MFE are likely to perform accurately. The statistical comparison results showed that the MFE was more likely to perform accurately than the FE for patterns of moderately to strongly aligned fracture sets and for patterns of poorly aligned fracture sets with medium to wide intersection angles to the 1D probe, and thus, the MFE was the optimum estimator. For patterns of poorly aligned fracture sets with narrow intersection angles to the 1D probe, however, the FE was more likely to perform accurately than the MFE, and thus, the FE was the optimum estimator. On the basis of these comparison results, we developed a recognizer for the optimum estimator. This developed recognizer provided a simple, reliable tool for selecting the optimum estimator to accurately produce a 3D orientation distribution estimation, which is beneficial to realistic fracture network modelling. Weakness of the developed recognizer was also disclosed.
The equivalent discrete fracture networks based on the correlation index in highly fractured rock masses
2019, Engineering Geology
Citation Excerpt :
Deriving accurate and efficient simulations of fluid flow is the prerequisite of multi-physics coupling applications, such as hydraulic fracturing and pollutant migration (Ma et al., 2019a; Zhou et al., 2018). The most widely accepted numerical approaches to model fracture networks are conceptually categorized into continuum and discrete models (Ko et al., 2015; Larsson et al., 2013; Lee et al., 1995; Maryška et al., 2005; Zhou et al., 2018). Continuum models is preferred in handling large-scale problems, but their assumption of homogenization overlooks the geometric and hydraulic properties of individual fractures (Liu et al., 2018; Reeves et al., 2013).
In the numerical simulations of highly fractured geological formations, discrete approaches are considerably promising and adequate to describe fluid flow in detail. However, the computational complexity increases dramatically with a greater number of fractures. This becomes the primary limitation for field-scale applications. In this study, a correlation index is for the first time introduced to evaluate the significance of individual fractures, and an equivalent model is proposed to mimic the original domain with a density-reduced one. By an equivalent permeability factor, the suggested model simplifies computational complexity, but compromises result precision to minor extent. This approach is validated in typical discrete fracture networks generated with stochastic fractal models. Effects of fracture geometry are discussed based on various distribution patterns. This method improves mesh quality when dealing with a fracture-matrix domain. It is also capable of optimizing reservoir design through fast and accurate estimations of gas productivity under different boundary conditions.
Variations of hydraulic conductivity of fracture sets and fractured rock mass with test scale: Case study at Heshe well site in Central Taiwan
2016, Engineering Geology
Citation Excerpt :
While quantitative analyses are needed to elaborate design tasks in rock engineering projects, such as underground excavation (Cha et al., 2006; Huang et al., 2013; Chen et al., 2015b; Huang et al., 2016), geothermal energy development (Stober and Bucher, 2015), and the construction of underground nuclear waste repositories (Hartley and Joyce, 2013; Chen et al., 2015a; Tsang et al., 2015), difficulties often arise in the investigation of the engineering characteristics of a fractured rock mass, and the associated hydraulic characterization is quite challenging. The Discrete Fracture Network (DFN) model describes the hydraulic characteristics of a fractured rock mass, based on the spatial distribution of its fractures, and takes into account the geometric features of dominant fracture sets and their apertures; it is regarded as a theoretical method for characterizing the heterogeneity and anisotropy of its hydraulic characteristics (Reeves et al., 2013; Yang et al., 2014b; Ko et al., 2015; Matonti et al., 2015). Therefore, the investigation of the spatial distribution and geometric features of fractures in a fractured rock mass and the characterization of their associated parameters, are critical to the utilization of a DFN model in determining hydraulic characteristics (Ji et al., 2004; Elmo et al., 2014; Guo et al. 2015).
Variations of the equivalent hydraulic conductivity of a fractured rock mass with investigating scale typically cause difficulties in the hydraulic characterization of a site. A mechanistic model regarding water flow through fractured rock masses and associated methods of multiple regression analysis have been established to determine the transmissivities of dominant fracture sets with statistical significance using in situ hydrogeological test results. Accordingly, this manuscript decomposes the transmissivity of a fracture set into the product of its hydraulic conductivity and its aperture. The ultimate purpose is to investigate the effects of scale on the hydraulic characteristics of fractured rock masses. The results of in situ hydrogeological tests at various scales, performed at the Heshe well site, are utilized to determine the equivalent hydraulic conductivity thereof. A series of laboratory tests are planned and carried out to obtain the apertures of fracture sets, which are applied in determining the respective hydraulic conductivities of those sets. Analyzed results reveal that the hydraulic conductivities of fracture sets can be determined with statistical significance. The comparison of equivalent hydraulic conductivities that are obtained in in situ multiple borehole pumping (MBP) tests with those determined using the combination of the hydraulic conductivities and geometric parameters of fracture sets verifies the proposed method. The hydraulic conductivities of fracture sets that are determined from their apertures, which are obtained from laboratory tests, differ from those obtained by in situ MBP tests by eight orders of magnitude. The magnitudes of hydraulic conductivities of fracture sets initially increase with the test scale, subsequently approaching an asymptote, exhibiting a negative exponential distribution. The proposed method yields the hydraulic conductivities of fracture sets and the equivalent hydraulic conductivity of a fractured rock site from the spatial distribution, hydraulic conductivities, and aperture of each dominant fracture set. Such an approach can minimize the deviation of the hydraulic conductivity of a fracture set with the measured aperture, and can be used to determine hydraulic conductivity, which is affected by both test scale and in situ stress.
Determining transmissivity of fracture sets with statistical significance using single-borehole hydraulic tests: Methodology and implementation at Heshe well site in central Taiwan
2015, Engineering Geology
Citation Excerpt :
Fractures that are interlaced in rock masses provide major channels for groundwater flows and dominate the permeability of the rock masses. Among various innovative methods that have been proposed for evaluating the permeability of a fractured rock mass (Ni et al., 2013; Grenon et al., 2014; Zha et al., 2014; Ko et al., 2015), that based on the Discrete Fracture Network (DFN) model is one of the most effective for determining the highly changeable permeability of rock masses and its anisotropy thereof (Oda et al., 1987; Cacas et al., 1990; Chesnaux et al., 2009; Pan et al., 2010; Bang et al., 2012; Dorn et al., 2013; Reeves et al., 2013). When a DFN model is utilized, the spatial distribution and hydraulic conductivity of its fractures must be given.
This work develops a novel method for determining the transmissivity of fracture sets with statistical significance and validates the method by performing in situ hydrogeological tests. An ideal mechanistic model is firstly considered, in which the injection discharge in a single-borehole hydraulic test is the sum of the flow of water through the intact rock and that through the fractures. The model assumes that fractures of similar attitudes have mutually consistent hydraulic properties and that the law of mass conservation applies in hydrogeological tests, facilitating statistical analysis. To validate the proposed method and demonstrate its practical implementation, an experimental well site is designed using a discrete fracture network concept and information about the spatial distribution of fractures at the site. Ten wells within 10 m of each other are drilled to perform 59 groups of single-borehole hydraulic tests at specific locations where fractures of distinct attitudes are observed in the wells. A deviation of more than three orders of magnitude in the measured equivalent hydraulic conductivity among the wells reveals that the fractures dominate the hydraulic characteristics at the study site. Therefore, the transmissivity of each dominant fracture set is statistically determined using the proposed method, and fractures with transmissivity deviated from their mean are identified. Most test statistics are evaluated with a confidence level of greater than 95%, indicating that the transmissivity of dominant fracture sets can be statistically inferred with high confidence. The efficacy of the proposed method for determining the transmissivity of fracture sets in fractured rock masses is thus verified, so the method can be used to effectively describe the hydraulic conductivity of fractured rock masses using a DFN model.
Minimum Scanline-to-Fracture Angle and Sample Size Required to Produce a Highly Accurate Estimate of the 3-D Fracture Orientation Distribution
2019, Rock Mechanics and Rock Engineering
A double scale methodology to investigate flow in karst fractured media via numerical analysis: The Cassino plain case study (central apennine, Italy)
2018, Geofluids

View all citing articles on Scopus

View full text

Technical NoteEvaluation of two conceptual approaches for groundwater flow simulation for a rock domain at the block-scale for the Olkiluoto site, Finland

Highlight

Abstract

Graphical abstract

Introduction

Section snippets

Study site

Approach

Results

Discussion and concluding remarks

Acknowledgement

Int. J. Rock Mech. Min. Sci.

J. Hydrol.

J. Hydrol.

Modelling of hydro-zones for the layout planning and numerical flow model in 2006

Final disposal of spent nuclear fuel in Finnish bedrock — Olkiluoto site report

FEFLOW: finite element subsurface flow and transport simulation system

Discrete feature modeling of flow, mass and heat transport processes by using FEFLOW

3D inverse modelling of groundwater flow at a fractured site using a stochastic continuum model with multiple statistical population

Stoch. Env. Res. Risk A.

Olkiluoto Geo-DFN Model, Draft Version 2

A hybrid modeling approach to evaluate the groundwater flow system at the low- and intermediate-level radioactive waste disposal site in Gyeong-Ju, Korea

Hydrogeol. J.

Technical Note
Evaluation of two conceptual approaches for groundwater flow simulation for a rock domain at the block-scale for the Olkiluoto site, Finland