Faculty of Civil Engineering
Department of Geotechnics and Underground Engineering
Assessment of factors affecting groundwater ingress into tunnel Čebrať using numerical modelling
Vyhodnocení faktorů ovlivňujících přítok podzemní vody do tunelu Čebrať pomocí numerického modelování
Student: Bc. Felipe A. Gutiérrez S.
Supervisor: Prof. Ing. Nad’a Rapantová, CSc.
Ostrava 2020
2
3
4
Declaration of the student
“I hereby declare that this Master thesis was written by myself. I have quoted all the references I have drawn upon.”
In Ostrava ...
...
Student´s signature
5
I declare that
● I am informed that Act No. 121/2000 Coll. – the Copyright Act, in particular, § 35 – Utilization of the Work as a Part of Civil and Religious Ceremonies, as a Part of School Performances and the Utilization of a School Work – and § 60 – School Work, fully applies to my Master thesis;
● I take account of the VŠB – Technical University of Ostrava (hereinafter as VŠBTUO) having the right to utilize the Master thesis (under § 35(3)) unprofitably and for own use; I agree that the Master thesis shall be archived in the VŠB-TUO’s information system;
● It was agreed that, in case of VŠB-TUO’s interest, I shall enter into a license agreement with VŠB-TUO, granting the authorization to utilize the work in the scope of § 12(4) of the Copyright Act;
● It was agreed that I may utilize my work, the Master thesis or provide a license to utilize it only with the consent of VŠB-TUO, which is entitled, in such a case, to claim an adequate contribution from me to cover the cost expended by VŠB-TUO for producing the work (up to its real amount);
● Upon final submission of the Master thesis, I agree with its publishing in accordance with Act no. 111/1998 Coll. on Higher Education Institutions and on Amendments and Supplements to Some Other Acts (the Higher Education Act) without regard to the defense result.
In Ostrava ... ...
Student´s signature
6
Declaration
The topic of the thesis was requested by the company Groundwater Consulting Services s.r.o. In addition, the submitted master thesis was elaborated under the supervision of the representative of this company in Czech Republic, Ing. Jiří Beránek.
………
Assoc. Professor Eva Hrubešová, Ph.D. – supervisor of the field Geotechnics at the Faculty of Civil Engineering, VSB-Technical University Ostrava
………..
Ing. Jiří Beránek., GCS (Pty) Ltd.
7
Acknowledgements
The time here when I started studying this master, has passed so fast that it seems hard to believe. Mostly very good things have happened since then, and if it weren’t because of the difficulties caused by this pandemic-virus, the stay so far in Czech Republic without doubt, would have been absolutely perfect. That’s why I’d like to start by saying thanks to the whole Czech Republic, in particular to VŠB – Technická Univerzita Ostrava, for having given me the opportunity to study here.
I’m also infinitely thankful to Professor Eva Hrubešová, for always having very good attitude towards me, for the meaningful and interesting technical discussions. Of course, in this line I need to also thank my supervisor professor Nad’a Rapantová for her help and support, endorsement, and availability to fruitfully discuss fascinating technical topics.
I want to also thank Jiří Beránek from GCS-Water & Environmental Consultants, for his always welcoming approach and availability to help, for providing me so much information and technical advice regarding numerical hydrogeological modeling. I need to also make special mention to Marek Michna from MIDAS software Czech-Republic & Slovakia. Thanks to you both so much.
And of course, I can’t not mention Mr. professor Rosmanit, for being such a good professor and encouraging person.
Thanks so much to my girlfriend and her family, my friends, and in general to all good people I met here from all over the world. With your support and friendship, being here feels like home.
I dedicate this present work to my beloved family in Chile, specially my mother Violeta and my older brother Rodrigo, for being always there, despite of the distance and for their infinite love and support.
Felipe Andrés Gutiérrez Sepúlveda
8
Anotace
Diplomová práce se zabývá hodnocením mechanismů přítoků vody do tunelu Čebrať, který se nachází v okrese Ružomberok na Slovensku. Pro odhad přítoku vody do tunelu bylo provedeno plně 3D geo-hydromechanické numerické modelování. Tento projekt lze rozdělit do dvou fází modelování; první je geomechanická simulace na základě konstitučních modelů CWFS, Hoek-Brown a Ubiquitous-joint. Jakmile byly napěťodeformační změny kvantifikovány, byly získány hranice deformačních zón okolo tunelu - EDZ, které byly importovány do FEFLOW jako tzv. “supermesh”. Na tomto základě byl v druhé fázi sestaven hydrogeologický model ustáleného proudění, který zohlednil základní hydrogeologické struktury oblasti a antropogenní vlivy ražby tunelu.
Poté bylo provedeno několik případových studií za účelem kvantifikace dopadů tunelování a tektonického porušení na vodní bilance v povrchových tocích i v samotném tunelu. Doplňování podzemní vody infiltrací ze srážek bylo hodnoceno lineární interpolací s údaji o nadmořských výškách na základě srážek z let 2001 až 2016. Přestože se výsledky zdají být v dobré shodě s předpoklady a pozorováním, kalibrace parametrů modelu je pro budoucí analýzu nezbytná.
Klíčová slova: Numerické modelování, FEFLOW, supermesh, EDZ, zaplavení vodou.
Abstract
The thesis deals with the assessment of water ingress mechanisms in the tunnel Čebrať, located in the Ružomberok district, Slovakia. To estimate the water inflow in tunnels, fully 3D geo-hydromechanical numerical modeling has been performed. This project can be divided in two stages of modeling; the first is a geomechanical simulation by considering CWFS, Hoek- Brown and Ubiquitous-joint models. Once the strain and deformation changes around tunnels were quantified, boundaries given by EDZs, were obtained to be imported into FEFLOW as supermeshes, to perform the second stage of modeling, where main hydrogeological features of studied area and impact of tunnel construction were considered under a steady-state water flow regime.
Several cases of analysis were then studied in order to quantify the impacts of tunneling and fractures on the water budget in surface creeks, as well as on the tunnel itself. The groundwater recharge was assessed by interpolating linearly with elevation, precipitation data from years 2001 to 2016. Although results seem to be sound with observed data, parameter calibration is still a required task for future analysis.
Keywords: Numerical modeling, FEFLOW, supermesh, EDZ, water inrush.
9
C ONTENT
Content ... 9
1 Introduction ... 19
1.1 Background ... 19
1.2 Objective and scope of the study... 20
1.3 Structure of the thesis ... 20
1.3.1 Introduction ... 20
1.3.2 Literature review ... 20
1.3.3 Description of the area of study ... 20
1.3.4 Methodology ... 21
1.3.5 Results and discussion ... 21
1.3.6 Conclusions ... 21
2 Literature review: Natural and anthropogenic factors affecting groundwater ingress into tunnels ... 22
2.1 Damage In the rock mass due to excavation ... 22
2.1.1 Definition of the Excavation Damage Zones (EDZ) ... 22
2.1.2 Excavation influence zone (EIZ) ... 24
2.1.3 Excavation damage zone (EDZ) ... 24
2.1.4 Highly Damaged Zone (HDZ) ... 24
2.2 Types of rocks and measured EDZs in practice ... 27
2.2.1 Crystalline rocks ... 27
2.2.2 Salt rocks ... 28
2.2.3 Sedimentary rocks ... 29
2.3 Mechanical properties of a rock mass ... 31
2.3.1 Brittle behavior ... 32
2.3.2 Cohesion weakening, friction strengthening (CWFS) constitutive model ... 33
10
2.3.3 The non-linear CWFS model ... 34
2.3.4 EDZ dimension assessment: case studies ... 35
2.3.5 EDZ dimension assessment through numerical modeling ... 39
2.4 Groundwater flow ... 41
2.4.1 Equivalent porous media concept ... 42
2.4.2 Flow in fractures ... 43
2.5 Boundary conditions in hydrogeological models ... 44
2.6 Pre-grouting ... 44
2.6.1 Pregrouting execution ... 45
2.6.2 Water inflow acceptability criteria ... 46
2.6.3 Grouting requirements ... 48
2.6.4 Rock improvement with pre-grouting ... 49
2.7 Water inflow into tunnels ... 50
2.7.1 Effects of geological features... 51
2.7.2 Brief description of geological features and its impact in water inflow into tunnels 52 2.7.3 Hydraulic characterization of the EDZ ... 54
2.7.4 Groundwater prediction and modeling approaches ... 57
3 Description: Hydrogeological conditions of the case study ... 60
3.1 Geological conditions of the zone ... 61
3.2 Hydrogeological conditions of Čebrať hill ... 63
4 Methodology: Assumptions, methods and available data ... 66
4.1 Main assumptions ... 67
4.2 Software used for this study ... 68
4.2.1 MIDAS GTS-NX ... 68
4.2.2 ITASCA FLAC3D ... 69
4.2.3 DHI FEFLOW ... 70
11
4.3 Available Geological and geomechanical data ... 71
4.4 Geomechanical assessment ... 76
4.4.1 Mesh generation and construction stages ... 77
4.4.2 Constitutive models ... 77
4.5 Hydrogeological assessment ... 85
4.5.1 Geometry and meshing. ... 85
4.5.2 Hydrogeological boundary conditions ... 91
4.5.3 Numerical model setting ... 92
5 Results and discussion ... 100
5.1 Geomechanical Modeling ... 100
5.1.1 Cohesion-weakening, friction-strengthening model, MIDAS GTS-NX ... 100
5.1.2 Generalized Hoek-Brown, MIDAS GTS-NX ... 105
5.1.3 Ubiquitous-Joint, FLAC3D... 109
5.1.4 Final comments and assessment of damage and EDZ dimensions ... 114
5.2 Hydrogeological modeling ... 117
5.2.1 Prior-tunneling cases with and without fractures... 117
5.2.2 Tunneling cases with and without fractures ... 120
5.2.3 Final comments and analysis of hydrogeological modeling ... 128
6 Conclusions... 133
6.1 Recommendations for further research ... 136
7 References ... 137
12
List of figures
Figure 1 Illustration of the excavation damage zones for massive ground around a circular excavation. (ANDRA,2005). ... 23 Figure 2 Processes that induct changes in rock properties due to excavation. (Suhihara, 2008).
... 23 Figure 3 Overview of the different damage zones (Siren,2015). Up: drill and blast method.
Down: TBM mechanized method. ... 26 Figure 4 Summary of the main findings of the ZEDEX project. Modified from Emsleyl (1997).
... 28 Figure 5 Gas permeabilities, measured at various locations at the WIPP, project normalized to excavation radius dimensions and EDZ dimensions (modified from Stormont, 1997) ... 29 Figure 6 Excavation-induced brittle features in Opalinus Clay (Mont Terri rock laboratory).
(a): extensional fractures created during excavation. (Nagra, 2002). (b): bedding related spalling (Marschall et al 2006). (c): Buckling of the bedding planes around small borehole (Blümling et al 2007). ... 30 Figure 7 Conceptual models of EDZ, derived from Mont Terri Project. a) Radial distribution of extensile fracture B) Combined buckling failure and extensile fracture. C) Schematic representation of damage, showing main mechanisms that affect borehole stability. (Marschall et al 2016). ... 31 Figure 8 Stress-strain diagram of a rock showing the stages of crack development (Martin, 1993). ... 32 Figure 9 Damage zones mapped to the conceptual DISL approach of Diederichs (2003,2007), based on the concept of cohesion loss and friction mobilization as discussed by Martin (1997), Kaiser et al. (2000), Hajiabdolmajid et al. (2002), Diederichs (2003), Diederichs et al. (2004).
... 33 Figure 10 Schematic illustrating the CWFS strength model showing parameter evolution as a function of plastic strain. (Walton, 2019). ... 34 Figure 11 General properties of various URL in sedimentary rock. (Fracture systems, 2011).
... 36 Figure 12 Observed HD and EDZ extents from different URLs as dimensionless radii. (Fracture systems, 2011)... 37
13 Figure 13 Observed HD and EDZ extents from different URLs as dimensionless radii.
(continued). (Fracture systems, 2011) ... 38
Figure 14 Example of model output using arbitrary loading and material parameters. (Perras, Diederichs, 2010). ... 40
Figure 15 Rock mass and its classification according modeling approaches (Sharifzadeh, 2017). ... 42
Figure 16 Type of analysis and scale of the modeling. Sharifzadeh, 2017. ... 43
Figure 17 Grouting methodology as described in Bahadur, 2007 ... 46
Figure 18 Pre-grouting design concept in the Nygard project. (Butron et al, 2010) ... 49
Figure 19 Comparison of a conservative model with a more realistic model of possible improvements in individual Q-parameters, and how these might impact on rock mass properties and support needs. ... 50
Figure 20 Water inrush situations during tunneling. a) low flow water inrush, b) large flow water inrush. c) water with silt. d) water inflow from geological structure. (Hou, et al, 2016). ... 51
Figure 21 Geological features and its impact in numerical modeling. (Sharifzadeh, 2012). .. 51
Figure 22 Hydraulic characterization of the EDZ features around excavations in the Mont Terri rock lab. In-situ permeability measurements conducted in experimental drift. (Bossart et al. 2002). ... 57
Figure 23 a) Mont Terri site descriptive model. b) Abstraction of the EDZ to perform hydrogeological simulation, according Alcolea (2016) approach. Marschall et al 2016. ... 59
Figure 24 Streetmap of Čebrať tunnel, with its final location. Provided by: Jiří Beránek. ... 60
Figure 25 Geological map of Čebrať hill. Purple tonalities reveal predominance of dolomites while green colors, marlstone. Source: http://apl.geology.sk/gm50js/ ... 62
Figure 26 A view of the face of the tunnel in an environment of fused limestones with claystone slabs. CADECO, 2016 ... 63
Figure 27 Main hydrological features of the Čebrať region -River pattern. The interlined orange pattern was the original location of tunnel Čebrať before the change in its route. ... 64
Figure 28 Average annual rainfall in Ružomberok district (Adapted from CADECO,2016) . 65 Figure 29 Flowsheet of the work scheme followed in this thesis study. ... 66
Figure 30 Marlstones, Alicante, Spain. ... 67
Figure 31 Types of geotechnical analysis available in MIDAS GTS-NX. ... 69
Figure 32 FLAC3D basic explicit calculation cycle (Itasca, 2005) ... 70
14 Figure 33 Geometrical supermeshes and finite element meshes in FEFLOW. (DHI, 2016) .. 71 Figure 34 Empirical estimation of the UCS from rock samples tested by load point method. 72 Figure 35 Distribution of UCS estimated samples. ... 72 Figure 36 Histogram of mapped RMR values from the tunnel and empirical estimation of GSI, according to Sánchez (2018). ... 73 Figure 37 Probability plot of estimated GSI values. ... 74 Figure 38 Rock mass modulus distribution from samples. ... 74 Figure 39 Discontinuities measured in the tunnels. (Plotted from vectors measured in the field, provided by: Jiří Beránek). ... 75 Figure 40 Regional model and sub-regional model considered in geomechanical analysis. ... 76 Figure 41 Extent of the geomechanical region in MIDAS GTS-NX with hexa-dominant mesh.
... 77 Figure 42 Cohesion weakening and friction strengthening curves adopted in this work ... 79 Figure 43 Recopilation of studies where CWFS strength model was used to match brittle rock behavior observed in excavations. Walton et al (2019). ... 80 Figure 44 Mohr-Coulomb fit for Generalized Hoek-Brown envelope by considering maximum confinement pressure equal to 4 [MPa]. ... 83 Figure 45 Comparison of the volumetric regions for FLAC3D and MIDAS GTS-NX models.
In red: FLAC3D modeling region. In blue: MIDAS GTS-NX volume. ... 84 Figure 46 Geological domain considered for Ubiquitous-joint constitutive relation in FLAC3D.
... 85 Figure 47 Regional hydrogeological conceptual model. In red: 9 sets of faults, in blue: Čebrať Tunnel. ... 86 Figure 48 Fault system in the numerical model performed by CADECO (2016). In order to get a similar simulation set-up, these faults were imported into FEFLOW model. ... 87 Figure 49 Supermesh import from maps in FEFLOW. ... 88 Figure 50 Tet-gen meshing panel in FEFLOW. ... 88 Figure 51 Left: Regional supermesh imported in FEFLOW, the red dots indicate nodes that were imported under the 'Point add-in' logic, to be considered as 'hard nodes' in the upcoming mesh process. Right: Geological map of the Čebrať hill with main lithology (dolomites in green, marlstones in purple). ... 89
15 Figure 52 Geological domains considered in FEFLOW modeling. The highest elevations correspond to dolomites, while the rest of the domain is marlstones. Tunnels and EDZi were
not shown since they look very similar to the extents of EDZ. ... 90
Figure 53 Physical representation of seepage face boundary condition ... 91
Figure 54 Hydrogeological domain and boundary conditions assumed in FEFLOW. No flow boundary condition is assumed within the software automatically when no B.C. is set. ... 92
Figure 55 Distribution of hydraulic parameters in FEFLOW modeling. Since the model is isotropic, all the directions of hydraulic conductivities are the same. ... 94
Figure 56 Vertical cross-section showing infiltration, drainage, aquifer recharge, and inter aquifer flow (from Healy, 2010). ... 95
Figure 57 Probability plot of precipitation of Čebrať zone in the 16-year span from 2001 to 2016. (Adapted from CADECO,2016). ... 96
Figure 58 Water recharge linear interpolation as a function of elevation adopted in FEFLOW model... 97
Figure 59 Elevation map of the hydrogeological domain. ... 98
Figure 60 Net-recharge spatial distribution as a function of elevation. ... 98
Figure 61 Cases of analysis in hydrogeological modeling. ... 99
Figure 62 In-situ stresses in the geomechanical domain. k0=0.5. ... 100
Figure 63 Top view of the modeled yielded element extent. The zoom is focused to the region under highest overburden, and the roof of the excavation under the assumed stress-state is not under plastic state. ... 101
Figure 64 Top view of vertical cuts which show the distribution of volumetric strain, through elevations z=80 to z=110 [m]. CWFS. ... 102
Figure 65Top view of vertical cuts which show the distribution of volumetric strain, through elevations z=120 to z=150 [m]. CWFS. ... 103
Figure 66 Top view of the modeled yielded element extent. The zoomed-in zone corresponds to the same area previously analyzed under CWFS constitutive model. Noticeable number of elements in the Čebrať hill were assumed to be under plastic state. ... 105
Figure 67 Top view of vertical cuts which show the distribution of volumetric strain, through elevations z=80 to z=110 [m]. Generalized Hoek-Brown. ... 107
Figure 68 Top view of vertical cuts which show the distribution of volumetric strain, through elevations z=120 to z=150 [m]. Generalized Hoek-Brown. ... 108
Figure 69 Vertical in-situ stress in FLAC3D modeling. ... 109
16 Figure 70 Isometric view of yielded elements plot. Most damage takes place in the walls of the tunnels. ... 110 Figure 71 Section of yielded elements in the tunnel. There are approximately 20 [m] offset between these contours. ... 111 Figure 72 Section of yielded elements in the tunnels. Failure due to shear in rock mass and joints is observed... 112 Figure 73 Volumetric strain plot in FLAC3D... 113 Figure 74 Contours of deviatoric strain (s1-s3) divided by UCS (27 [MPa]). Left: CWFS model.
Right: Generalized Hoek-Brown model. ... 114 Figure 75 Contour of deviatoric stress divided by UCS. Ubiquitous-joint model. ... 115 Figure 76 Up: Procedure for EDZ assessment in hydrogeological groundwater inflow followed by Alcolea (2016). Down: description of adopted distances for EDZs boundaries in FEFLOW modeling. The distances are measured radially from the adjacent frontier. ... 116 Figure 77 Comparison between cases prior-tunneling with and without fractures. From top to bottom: Pore-pressure distribution with water table. Darcy fluxes distribution. Water budgets.
... 118 Figure 78 Water budgets measured in the numbered creeks prior tunneling considering scenarios with and with no fractures. ... 119 Figure 79 Percentual differences for creeks considering scenarios with and without fractures.
... 120 Figure 80 Comparison between tunneling cases with and without fractures. From top to bottom:
Pore-pressure distribution with water table. Darcy fluxes distribution. Water budgets. ... 121 Figure 81 Water budgets in numbered creeks in the tunneling scenarios with and without fractures... 122 Figure 82 Percentual difference histograms of water budgets in each creek. ... 123 Figure 83 Water budgets on numbered creeks in all the scenarios studied. ... 124 Figure 84 Water budget distributions inside tunnels. Top: case without fractures. Bottom: case with fractures. ... 126 Figure 85 Probability plot and basic statistics of water inflow in tunneling scenarios. In blue:
case with no fractures. In red: Case with fractures. ... 127 Figure 86 Darcy fluxes around the tunnel and fractures. ... 128 Figure 87 Water budget distributions inside tunnels. Up: case with fractures and EDZs. Bottom:
Case with fractures and no EDZs. ... 130
17 Figure 88 Probability plot and basic statistics on the tunneling scenarios. ... 131
List of tables
Table 1 Condition of tunnels according to inflow volume. Adapted from Palmström & Stille (2010). ... 47 Table 2 Classification of water inflow rates for a six (6) [m] diameter tunnel. Adapted from Sharifzadeh (2012). ... 47 Table 3 Grouting matrix according difficulty and goal hydraulic conductivity. Adapted from Stille (2012). ... 48 Table 4 Main results from the Stripa EDZ tests (Gray 1993) ... 54 Table 5 Conclusions from hydraulic parameter estimations. From (Börgesson et al. 1992) .. 56 Table 6 Adopted values for initial and residual parameters of CWFS non-linear constitutive model. The UCS value is relevant to estimate the dilation parameter. ... 80 Table 7 Main parameters used for generalized Hoek-Brown model in MIDAS GTS-NX. .... 81 Table 8 Parameters for Ubiquitous-joint constitutive model in FLAC3D. ... 84 Table 9 Main hydrogeological parameters assumed in FEFLOW 3D model. ... 93 Table 10 Approximated measurements of the extent of EDZs. Ranges depend on several zones where distances were quantified in the numerical modeling codes employed. ... 115
18
List of abbreviations
a: Parameter of Hoek-Brown envelope relation [-]
D: Damage factor (Hoek-Brown envelope) [-]
DFN: Discrete fracture network
E: Young modulus [MPa]
EDZ: Excavation damaged zone [m]
EDZ: Inner excavation damage zone [m]
EDZo: Outer excavation damage zone [m]
EdZ: Excavation disturbed zone also named as EIZ [m]
EIZ: Excavation influence zone [m]
GSI: Geological strength index [-]
HDZ: Highly disturbed zone [m]
h: Hydraulic head [m.a.s.l]
K: Hydraulic conductivity [m/s]
mb: Parameter of broken rock (Hoek-Brown envelope) [-]
mi: Parameter of intact rock (Hoek-Brown envelope) [-]
Q: Volumetric flow [m3/s]
REV: Representative elementary volume
s: Parameter of Hoek-Brown envelope relation [-]
UCS: Uniaxial compression strength [MPa]
γ: Unit weight [kN/m3]
εp: Plastic strain [-]
εv:Volumetric strain [-]
ν: Poisson number [-]
σ1: Maximum principal stress [MPa]
σ3: Minimum principal stress [MPa]
ψ Angle of dilatancy [°]
19
1 I NTRODUCTION
1.1 B
ACKGROUNDGroundwater inflow into underground constructions is an issue which potentially could bring severe consequences both economical, environmental, and as life-risk situations. Due to this, it is crucial to have control on groundwater influence in geotechnical constructions. A proper estimation of inflow rates then, becomes critical to quantify or assess things like, the need for grouting, support elements, drainage systems required and even the tunnel alignment.
However, there seems to be some degree of inefficiency of the current engineering practice to successfully predict water inflows (Fernández & Moon, 2010). Historically, the usage of empirical relations and analytical methods have been the trend, which are based on basic assumptions such as constant hydraulic potentials, homogenous and isotropic rock masses, and so on. Nonetheless, with the rise of computer power, numerical modeling is becoming a powerful and accessible tool to evaluate and decide on more complex situations, and, by examining several hypothetical scenarios.
Originally (late 90’s), in Slovakia, the development of motorways was planned to be in plains, however it was shifted to mountainous areas of the northern part of the country, where the excavations of tunnels are mandatory. These areas comprise, very difficult terrains with complex geologies and where landslides occur on a regular basis. Even the Čebrať tunnel was delayed, and re-localized due to landslide events (Frankovský, 2007).
The main purpose of this thesis project is to provide an estimation of steady-state water inflow in tunnel Čebrať, by means of fully 3D numerical modeling. The first stage of this work consists in estimating the damage area due to stress-relaxation around the tunnels, whose area exhibits an induced increase in hydraulic conductivities of surrounding rock, and the second stage, about hydrogeological modeling, where main features are modeled. Furthermore, by performing this numerical analysis it is possible to analyze the interaction between groundwater and surface water in creeks which are affected by the tunneling process, and thus, the environmental impact of the excavation can be diminished.
20
1.2 O
BJECTIVE AND SCOPE OF THE STUDY The main objective of this study is: To study the water inflow mechanisms into tunnel Čebrať and its impacts on the surrounding environment by performing 3D geo-hydromechanical numerical modeling.
This thesis also covers the following aspects:
1. Literature review on the geomechanical aspects of damage around excavations.
2. Literature review on the mechanisms of water inflow into tunnels considering hydrogeological interactions.
3. Description of the existing features of Čebrať tunnel project and highlight the geological and hydrogeological conditions of the site.
4. Carry out 3D numerical modeling for assessment of damage and water infiltration in the tunnels and its implications for the surrounding area.
5. Comparison of several water recharge scenarios, and also the impact of faults into the hydrogeological model.
6. Discussion and conclusions of the work.
1.3 S
TRUCTURE OF THE THESISIn this present work the following structure has been followed:
1.3.1 Introduction
In this section the motivations and main challenges associated to the studied case are introduced.
1.3.2 Literature review
Extensive description of the state of the art is presented here, by considering the theoretical geomechanical and hydrogeological factors that play significant role in groundwater inrush into tunnels.
1.3.3 Description of the area of study
All the technical and descriptive details about the project and the geological setting of the region analyzed are presented here.
21 1.3.4 Methodology
This section can be separated in two groups: geomechanical and hydrogeological modeling. In this section of the thesis all the considerations, assumptions, data used and procedures are detailly explained.
1.3.5 Results and discussion
All relevant results gotten by means of numerical modeling are presented here, and the results are commented and discussed according criteria shown in literature review chapter.
1.3.6 Conclusions
In this chapter, by considering all the results, and theoretical and practical background, main conclusions and challenges for future research are written. The conclusions structure is also subdivided in geomechanical and hydrogeological parts.
22
2 L ITERATURE REVIEW : N ATURAL AND ANTHROPOGENIC FACTORS AFFECTING GROUNDWATER INGRESS INTO TUNNELS
2.1 D
AMAGEI
N THE ROCK MASS DUE TO EXCAVATIONDuring the excavation of underground constructions, the surrounding rock mass experiment damage due to the induced stresses and joint interaction, triggered by the excavation process. The importance of studying the damage around tunnels is crucial, since it is well known that the induced damage, causes big changes in the permeability of the surrounding rock, which makes the tunnel or excavation, more susceptible to water ingress, to become a zone of potential contaminant transport or being a pathway to gas or radionuclide escape.
2.1.1 Definition of the Excavation Damage Zones (EDZ)
The study of the different damage zones has been analyzed since the 80’s in relation to nuclear waste disposal (Kelsall et al, 1984). There are various acronyms are used in literature to describe damage zones in underground excavations (Siren, 2015), even when several boundaries or layers can be identified and grouped according to its different geo-hydro characteristics, as shown in Figure 1. These zones are named: Highly damaged zone (HDZ), Excavation damaged zone (EDZ), and Excavation disturbed zone (EdZ).
The radius or thickness of this so-called excavation damage zone (EDZ), depends on several factors, such as, the excavation method used (TBM, Drill & Blast, etc.), dimension and size of the opening, the fracture potential and properties of the rock material and the stress field (Perras, Diederichs, 2010). Suhihara (2008) describes (Figure 2) all the processes involved in the change of rock properties induced by excavation.
23 Figure 1 Illustration of the excavation damage zones for massive ground around a circular excavation.
(ANDRA,2005).
Figure 2 Processes that induct changes in rock properties due to excavation. (Suhihara, 2008).
24 2.1.2 Excavation influence zone (EIZ)
The EIZ, or generally termed EdZ (Tsang et al. 2005), is a zone of stress change and/or elastic strain influence. The geomaterial in this zone is under elastic regime and in consequence, there is no damage induced. In this sense, this boundary represents the transition from elastic behavior, to damaged zones nearby the underground opening.
Perras & Diederichs et al 2010, state that in porous rocks the stress/strain regime could lead to hydromechanical and geochemical changes in this zone, without presenting changes in flow and transport properties. In low porosity rocks however, large stress gradients might alter joints apertures, but the authors express that the influence is minimal.
2.1.3 Excavation damage zone (EDZ)
The so-called EDZ is a zone with hydromechanical and geochemical modifications inducing changes in flow and transport properties (Perras & Diederichs. 2010). In literature, it is possible to find that there have been many experiments showing results of reduced transmissivity or permeability with distance from openings (Shao et al. 2008 and Bossart et al.
2004). These measurements in rock permeability show that the changes can be as high as 106 times that of the undisturbed rock mass.
In general terms, EDZ can be sub-divided into the outer zone, (EDZo) and the inner zone (EDZi) to make easier the hydrogeological modeling. (Perras & Diederichs, 2010).
NAGRA (2002) treated these two zones by assuming an average factor of 10x permeability increase in the inner zone (EDZi) and 5x in the outer zone (EDZo) in their safety assessment for the Opalinus Clay.
2.1.4 Highly Damaged Zone (HDZ)
The HDZ is defined as a zone where macro-scale fracturing take place. The effective HDZ permeability is dominated by interconnected fractures and is significantly greater than the undisturbed rock mass. The average permeability in this zone could be in the order of 1000 times (ANDRA, 2005) larger than the rock mass permeability. New fractures and bedding slip develop continuously at the excavation boundary within this zone, slowing and stopping soon after the excavation reaches a state of equilibrium.
25 This zone can be linked with notorious plastic strain, yielded elements, tensile failure, dilatation of fractures and significant stress relaxation. The transition between the HDZ and EDZ will be sharp in brittle rocks and more gradational in ductile rocks. (Perras & Diederichs, 2010).
The HDZ is a highly damaged zone that is generated by the chosen excavation process, this way appropriate blasting techniques are necessary to minimize the extent of the HDZ.
However according to Jonsson (2009), in certain kind of rocks the HDZ will be present regardless of the excavation method employed. A blasting experiment on granite rock masses concluded that normal blasting could induce up to 1.7 [m] thick EDZ, whereas a controlled blasting can reduce the latter thickness to 1 [m]. (Jonsson et al. 2009). In Figure 1, it is possible to see the different boundaries generated due to tunneling. Siren (2015) also states that the extents of EDZs is highly influenced by the excavation method used.
26 Figure 3 Overview of the different damage zones (Siren,2015). Up: drill and blast method. Down: TBM mechanized method.
27
2.2 T
YPES OF ROCKS AND MEASUREDEDZ
S IN PRACTICE 2.2.1 Crystalline rocksThese rocks in general terms present very low degree of permeability. Major excavation response experiments in crystalline rocks have been finalized including Stripa and Äspö (Sweden), AECL URL1 (Canada), Grimsel (Switzerland), and KURT (South Korea).
The URL project was developed in the Canadian Shield for the purpose of carrying out large scale in situ tests in the granite of Lac Du Bonnet. The main working levels of the URL are at depths of 240 and 420 m. Souley (2001) proved that the in-situ measurements and forecast of permeability, were in agreement, and they estimated the extend of the EDZ to be around 50 to 70 [cm]. The cross-section of the tunnel was elliptical with axis of 4.4 m horizontally and 3.5 m vertically approximately. Rutqvist et al. (2009) performed numerical modeling for the excavation-induced damage, changes in permeability, and fluid-pressure variations. As a result, it was obtained that the increase in permeability above the tunnel was due to fracturing under high deviatoric stress, while the changes in permeability were caused by the opening of pre-existed fractures because of the drop in confining mean stress.
In South Korea, a small-scale underground research laboratory (KURT) was developed in granite to investigate the behavior of barriers (Cho, 2008). Borehole drilling, geological survey, and in situ and laboratory tests were performed to validate the numerical modeling. It was obtained that the size of the EDZ ranged from 1.1 m to 1.5 m around a 6 x 6 m horseshoe shaped tunnel. The elastic modulus and rock strength within the EDZ envelope, were around 50% and 15%, respectively.
In Äspö, at an overburden of 500 [m] approximately, several studies were carried out.
The average depth of the damaged zone ranged from 5 to 20 [mm] from the excavation wall around the 5 [m] diameter tunnel. The extent of the EDZ and the hydraulic conductivity of the EDZ adjacent to the floors of the tunnels were generally larger than in the roof and wall sections (Autio et al,2005).
1 Underground research laboratory
28 From acoustic emission measurements, EDZ was measured in Äspö (ZEDEX project).
As seen in figure 3 the extent of the EDZ is larger around the Drill and blast tunnel compared to the TBM drilled tunnel. The extent of the damage zone is about 0.3 m in the wall and about 0.8 m in the floor in the drill and blast drift. Particularly, in this study the method of excavation did not affect the extent of the disturbance zone.
Figure 4 Summary of the main findings of the ZEDEX project. Modified from Emsleyl (1997).
2.2.2 Salt rocks
The behavior of salt rock under the action of stress is very complex in nature, since this type of material presents a time-related behavior (creeping) as well as plastic deformation. Due to this, the long-term behavior of these rocks is the vital importance, and it is a challenging problem to assess.
In the U.S, in the WIPP site, using gas permeabilities Stormont (1997) showed that damage within a salt formation could be detected and that an EDZ could be delineated, as shown in Figure 4. The results of the permeability tests show a rapid increase in gas permeability within one radius of the excavation, with the measurements coming from a variety of cavern shapes and time durations after excavation (Stormont 1997).
The authors found out that permeability measurements can provide a useful insight at detecting the EDZ. However, the creep phenomenon, may expand the extent of the EIZ and EDZ, so the long-term stability in salt rocks is more concerning for the safety in nuclear waste disposals than in crystalline rocks.
29 Figure 5 Gas permeabilities, measured at various locations at the WIPP, project normalized to excavation radius dimensions and EDZ dimensions (modified from Stormont, 1997)
2.2.3 Sedimentary rocks
The research done in sedimentary rocks is mostly focused on the argillaceous formations, which are thought to be a very favorable medium for disposal of radioactive waste, due to their low permeability and diffusivity. These rocks also present an apparent self-healing property to close induced fractures (Bock ,2010). This process of healing may reduce the water flow through the EDZ with time.
The French Institute for Protection and Nuclear Safety has selected the Tournemire site for its own research program on deep geological disposal (Bonin, 1998). The Tournemire experiment shows that the excavation generates hydro-mechanical disturbances which can extend up to 6 times the mean radius of the gallery (2.5 m).
In central Japan, at Tono mine, it was found that the size of the EDZ in the Neogene sedimentary rock depends on the excavation method used. The mechanical way is more effective than blasting in limiting the thickness of the EDZ (Sato et al. 2000).
30 In Switzerland, the Opalinus clay at Mont Terri has been chosen as a potential zone for radioactive waste disposal thanks to favorable tectonic condition (Martin et al. 2003, Bossart et al. 2008). Extensive research and experiments were carried out to measure EDZ. The principal conclusions were that the EDZ is dependent on stress-induced fractures and variable moisture conditions. The inner zone was characterized as a highly interconnected network of unsaturated joints connected with the free face of the opening. On the contrary, in the outer zone, it was observed that the damage induced fractures were not noticeable connected. The EDZ was found to be strongly linked to the existing in-situ stress and bedding plane anisotropy (Popp et al., 2008), which leads to non-symmetrical EDZs generation.
Figure 6 Excavation-induced brittle features in Opalinus Clay (Mont Terri rock laboratory). (a):
extensional fractures created during excavation. (Nagra, 2002). (b): bedding related spalling (Marschall et al 2006). (c): Buckling of the bedding planes around small borehole (Blümling et al 2007).
Other of the most important finding in Opalinus clay response (Marschall, 2016), is that the mechanical behavior is brittle, the characteristics of the EDZ show extensional fracturing, bedding parallel slip, kink failures and tectonic structure reactivation.
In the short term, the findings prove that the rock mass is stress and structurally controlled, or combination of both. The stress damage is highly dependent on the excavation method whereas the structural-controlled behavior is a product of sedimentary and tectonic structures. This regime is very hard to predict, due to the irregularity of the pre-existing fracture systems. Moreover, the long-term behavior is dominated by ductile failure mechanisms which are attributed to creep.
31 Figure 7 Conceptual models of EDZ, derived from Mont Terri Project. a) Radial distribution of extensile fracture B) Combined buckling failure and extensile fracture. C) Schematic representation of damage, showing main mechanisms that affect borehole stability. (Marschall et al 2016).
2.3 M
ECHANICAL PROPERTIES OF A ROCK MASSThe strength of the rock mass is a difficult problem to assess due to the fact that in-situ rock is heterogeneous, non-continuous, and an anisotropic medium. From laboratory testing, parameters can be obtained from intact rock samples, which need to be escalated later on under certain criterion to represent the nature of the rock mass. Thus, the geological characterization of the rock mass is fundamental to understand its behavior.
Underground openings cause a redistribution of the in-situ stress field, and therefore zones of high stress and damage develop from the excavation damaged zone. In this zone, the material experiences a degradation of strength properties, as a consequence of the interaction of pre-existing joints with the damage-induced ones.
Martin and Chandler (1993, 1994) and thereafter Eberhardt et al. (1999) state that the stress-strain relationship for an intact brittle rock, can be subdivided in five regions, as shown in figure 5: in region I, the rock experiments the closure of pre-existing microcracks. In the region II, the rock exhibit an approximately linear-elastic behavior. Region III however, the rock starts to experience dilation, which is a stable crack growth. Region IV represents the onset of unstable crack growth. The uniaxial compression strength of the intact rock depicts
32 the beginning of the region V. Martin and Chandler (1994), showed that the initiation of new cracks and propagation of cracks occur approximately at 40% of UCS.
Figure 8 Stress-strain diagram of a rock showing the stages of crack development (Martin, 1993).
A typical stress-strain relation obtained from a uniaxial compression test is presented in Figure 5, where σcc is the crack closure stress, σci is the crack initiation stress, σcd is the crack damage stress, and σc is the peak stress at failure. The three stress thresholds, i.e., σci, σcd, and σc, represent important stages in the development of the macroscopic failure process of intact rocks.
2.3.1 Brittle behavior
For brittle rocks that have a dominant tendency to split or spall near the excavation boundary, the conceptual model of Diederichs (2007), using crack initiation and long-term strength, can be adopted to delineate the three zones around the excavation. This proposed model, named in literature as DISL, is and improvement based on Hoek-Brown failure
33 criterion, and considers values of the intact rock thresholds as well as a stress path-based approach to assess damage in brittle rocks.
Figure 9 Damage zones mapped to the conceptual DISL approach of Diederichs (2003,2007), based on the concept of cohesion loss and friction mobilization as discussed by Martin (1997), Kaiser et al.
(2000), Hajiabdolmajid et al. (2002), Diederichs (2003), Diederichs et al. (2004).
2.3.2 Cohesion weakening, friction strengthening (CWFS) constitutive model
Novel research on brittle behavior of rocks around excavations, have proved that the use of a cohesion-weakening, friction-strengthening based constitutive relation in continuum modeling, has provided excellent match with observed characteristics such as fracture zone size and shape, and ground displacements. (Walton, 2019).
The origin of the CWFS model it dates back to decade of the 60’s, with the study of Schmertmann & Osterberg (1960) who proved that the frictional behavior in granular materials only shows up if there is movement between the particles involved. Thus, when thinking about rock, prior to the onset of yielding, there should be only one component of strength, given by cohesion. Once the damage appears, and cracks start forming, the initial cohesion is degraded, as shearing between individual particles begin to occur and the rock dilates, the friction component becomes mobilized. This mechanism was extensively tested in laboratory, during compression observations of Lac Du Bonnet granite (Martin, 1997; Diederichs, 1999, 2004, 2010).
34 To reproduce the observed mechanical behavior, Hajiabdolmajid et al (2002 & 2003) proposed the so called CWFS, where the cohesion and friction angle are a function of plastic strain. Using the presented model, the authors were able to reproduce the progressive failure observed in tunnels.
Figure 10 Schematic illustrating the CWFS strength model showing parameter evolution as a function of plastic strain. (Walton, 2019).
Although the CWFS model was first thought to be applicable to brittle rocks (crystalline rocks), the methodology proposed by Walton & Diederichs (2015c), have also given sound results for sedimentary rocks, which present a strain-softening behavior rather than brittle.
2.3.3 The non-linear CWFS model
The non-linear CWFS model proposed by Renani & Martin (2018) is a modification of the original CWFS model proposed by Hajiabdolmajid (2002). Cohesion and friction are mobilized as a function of the equivalent plastic strain (εp):
𝜏 = 𝜎 𝑡𝑎𝑛 𝜙 𝜀 + 𝑐 𝜀 (2.1)
Where:
𝜀 = 2𝜀 𝜀
3 𝑑𝑡 (2.2)
Renani and Martin (2018), recommended the following relationships for mobilized cohesion and friction values for in-situ rocks:
35
𝑐 𝜀 = 𝑐 + (𝑐 − 𝑐 ) 2 − 2
1 + 𝑒 ,
(2.3)
𝜙 𝜀 = 𝜙 + (𝜙 − 𝜙 ) 2
1 + 𝑒 ,
− 1 (2.4)
Where the initial cohesion and friction are the values when there is no plastic strain, whereas the residual values correspond to the ultimate mobilized values for cohesion and friction angle. The parameter 𝜀 , is the equivalent plastic strain where 99% of the residual values have been reached. The values according literature for 𝜀 , are in the range of 0.001 and 0.003, with higher values for soft rocks and lower values for strong rocks.
(Walton, 2019).
Back-analyzed data of the study of AECL mine, showed that 𝜀 , > 𝜀 , (Martin, 1997; Hajiabdolmajid et al.,2002; Zhao et al., 2010). According to Walton (2019), it is possible to generally conclude that 𝜀 , ≥ 𝜀 , , where the differences between these parameters are due to the brittleness of the rock mass.
2.3.4 EDZ dimension assessment: case studies
As discussed in section 2.2 numerous experiments have conducted EDZ dimension assessment. Among the experiments on crystalline rock, it is possible to enumerate: (Fracture systems, 2011).
The Colorado School of Mines Test Site in Colorado, USA.
The BWIP near surface test facility, in Washington, USA.
The Grimsel Test Site in Switzerland.
The Kamaishi Mine, Japan.
The AECL Underground Research Laboratory in Manitoba; Canada.
The Äspö Hard Rock laboratory in Sweden;
The Kurt facility in Korea.
The Onkalo facility at Olkiluoto in Finland.
The MIU URL at Mizumani in Japan.
36 The first underground research facilities in argillaceous rock was established at Mol, Belgium. The following URL have been built since then:
Mont Terri URL in Switzerland
Tournemire URL in France
Meuse/Haute Marne URL in France
Tono Mine in Japan
Honorobe URL in Japan
Some general characteristics of URL in sedimentary rocks are summarized in Figure 11.
Figure 11 General properties of various URL in sedimentary rock. (Fracture systems, 2011).
The observed EDZ in the previous URLs can be seen in Figures 12 and 13.
37 Figure 12 Observed HD and EDZ extents from different URLs as dimensionless radii. (Fracture
systems, 2011).
38 Figure 13 Observed HD and EDZ extents from different URLs as dimensionless radii. (continued).
(Fracture systems, 2011)
39 2.3.5 EDZ dimension assessment through numerical modeling
The assessment of EDZ extent with numerical modeling can be performed by considering short- and long-term behavior as well as continuum and discontinuum approaches.
Naturally, the long-term behavior will be ruled by the creep effect, whereas the short-term regime can be forecasted by means of elasto-plastic modeling.
Su (2007) states that the extension of the very limited fracture zone and of the micro- fractured zone (EDZ) was accurately predicted by elasto-plastic modeling in the MODEX-REP project (Lebon et al 2002).
There are outputs in numerical modeling which can provide and insight of the limits of the EDZs. Yielded elements are one of them. A yielded zone shows that region is no longer under elastic regime, and therefore cohesion has decayed whereas friction has increased according to plastic strain. Practically speaking, the rock mass in plastic yielding is damaged due to connected and unconnected macro and micro fractures (Diederichs, 2003, 2007). The zone beyond the plastic yield which is in elastic state, can be categorized as EIZ.
The yielded elements give and idea of EDZi and EDZo combined together. The maximum extent of the plastic zone is the outer limit of EDZo (Sharma, 2019).
Perras & Diederichs (2016) suggested that the isoline given by volumetric strain equal to zero, represents the boundary between EDZi and EDZo. Likewise, the beginning of the EDZ, can be interpreted as the start of volumetric strain change (Perras & Diederichs, 2010).
According to Sharma (2019), with the CWFS model, plastic yield can still occur under contractive volumetric strain. Physically this means that cracking is taking place under high confining stress. This situation prevents the interaction of fractures and therefore its growth.
40 Figure 14 Example of model output using arbitrary loading and material parameters. (Perras, Diederichs, 2010).
Regarding damage, or loss of material around excavations, it was obtained through back analyzed cases over the last decades states that the maximum long-term strength of the rock mass is given by (Diederichs, 2007):
𝜎 , = 𝐶𝐼 + (1 𝑡𝑜 2)𝜎 (2.5)
Where CI is the crack initiation threshold which typically is around 35% to 50% the value of unconfined compressive strength of the intact rock. The latter expression is valid under low confined stress, where spalling damage occurs. Thus, a quick indicator of spalling damage is given by the envelope:
𝜎 − 𝜎 = 0.35 × 𝑈𝐶𝑆 (2.6)
41
2.4 G
ROUNDWATER FLOWThe general equation for 3D groundwater flow is given by Equation 2.7, which is a derivation from Darcy’s law and conservation of mass.
𝜕
𝜕𝑥 𝐾 𝜕ℎ
𝜕𝑥 + 𝜕
𝜕𝑦 𝐾 𝜕ℎ
𝜕𝑦 + 𝜕
𝜕𝑧 𝐾 𝜕ℎ
𝜕𝑧 = 𝑆 𝜕ℎ
𝜕𝑡 − 𝑊 ∗ (2.7) Where Ss is the specific storage coefficient of media. And K are the hydraulic coefficients in the x, y and z directions. W corresponds to the water recharge/discharge.
For the purpose of hydrogeological modeling, the rock mass can be considered as an assemblage of intact blocks that are separated by discontinuities. Figures 15 and 16 give and insight of the different possible abstraction on how to model the hydro-behavior of the rock mass depending if the medium is considered as a continuum or discontinuum.
It is usual to adopt the concept of fractured porous medium to refer to a geological material that can be discretized in porous blocks divided by fractures. For such media, the void space consists of two parts including a network of fractures, and blocks of porous medium (Sagar & Runchal, 1982; Bear et al., 1993) thus, the water flow through rock masses can be analyzed by means of continuum and discontinuum tools.
The simplest difference between continuum and discontinuum tools is that for the latter, the inclusion of fractures is explicitly done. In continuum tools, the jointed rock mass is thought as an equivalent porous medium with an equivalently averaged property.
42 Figure 15 Rock mass and its classification according modeling approaches (Sharifzadeh, 2017).
2.4.1 Equivalent porous media concept
The advantage of continuum approaches, especially for large scale problems, is that the behavior of the rock mass can be simplified to consider averaged-equivalent properties. Thus, the modeling stage is much simpler since inherent complexities are seen from a further point of observation.
According to Sharifzadeh (2017) two basic condition must be met to consider a rock mass as an equivalent porous medium:
The equivalent permeability of a fractured rock shows insignificant variations in the value with a small addition or subtraction to the sampling volume.
An equivalent hydraulic conductivity tensor should exist for rock mass that appropriately predicts the flow vector when the direction of a constant gradient is changed.
43 The first criterion implies that a REV2 should exist for a given rock mass. The second criterion means that the hydraulic tensor, is symmetric. In other words, the measured or calculated directional hydraulic conductivities in different directions (for samples larger than REV) should be plotted as an ellipsoid. If these criteria are not fulfilled, the rock mass cannot be represented as a continuum equivalent model, and therefore, the geological structures should be considered.
Figure 16 Type of analysis and scale of the modeling. Sharifzadeh, 2017.
2.4.2 Flow in fractures
The groundwater flow through fractured media is highly dependent on the structure of void spaces and its interconnections. (Javadi et al., 2009, 2010; Sharifzadeh et al., 2009; Javadi
& Sharifzadeh, 2013). Attributes such as roughness, contact area, aperture, are of the fundamental importance in groundwater flow (Javadi et al., 2014b; Sharifzadeh et al., 2006).
The fluid through fractures can be estimated by the cubic law, which is only valid for laminar flow, and it is expressed as (Whitherspoon et al, 1980):
2 REV: Representative Elementary volume: a volume which is statistically meaningful to describe the properties of the rock mass as a whole.
44 𝑄 = −𝑤𝑎
12𝜇 = −𝑘𝐴
𝜇 ∇𝑃 (2.8)
Where Q is the volumetric flow rate, w is the fracture width perpendicular to pressure gradient, ah is the hydraulic aperture, k is the permeability and Af is the cross-section area of fracture. Cubic law is only exact for smooth-wall fractures with homogenous aperture.
Considering the previously defined concepts, the appropriate modeling approach depends on the scale of the situation to simulate. Thus, the most suitable method will be function of the size of the excavation relative to geometrical characteristics of joints and the scope of the study.
2.5 B
OUNDARY CONDITIONS IN HYDROGEOLOGICAL MODELSThere are basically three kinds of boundary conditions in hydrogeological modeling:
Type 1. Specified head boundary (Dirichlet conditions) where head along the boundary is set at a known value. Heads along a specified head boundary may vary with space, or as a function of elevation. The direct implication of this boundary condition is that the water supply is inexhaustible.
Type 2. Specified flow boundary (Neumann conditions) where the derivative of head at the boundary is specified. Flow is calculated from Darcy’s law. A no flow boundary is a special case Type 2 boundary where the flow across the boundary is zero.
Type 3. Head-dependent boundary (Cauchy conditions) where flow across the boundary is calculated from Darcy’s law using a gradient calculated as the difference between a specified head outside the boundary and the head computed by the model at the node located on or near the boundary. This type of boundary condition is sometimes called a mixed boundary condition because it relates a boundary head to a boundary flow.
2.6 P
RE-
GROUTINGIn tunneling projects, pre-grouting is carried out with the goal of setting an impervious zone around the tunnel periphery by reducing the penetrability of the highest conductive fractures in the rock mass. (Grov, 2001). Barton et al (2019) states that pre-grouting has three main functions:
45
To control water inflow into tunnel
To limit groundwater drawdown above the tunnel
To make tunneling progress more predictable since rock mass quality is effectively improved
Pre-grouting is defined as the process of injecting grout materials into selected boreholes drilled in the rock mass, with the purpose of sealing surrounding fractures. This way, the rock mass around the excavation is less permeable and geologically more competent. The pattern of grouting depends on the local hydrogeological conditions, but as a general rule, the grout hole orientations are chosen to cross as many joints as possible, and at angles which are normal to the strike and dip of the structures.
2.6.1 Pregrouting execution
Pre-grouting can be performed either from the surface or from inside the tunnel. The method of drilling and injection of grout relies on local conditions. A generic grouting methodology is summarized below (Bahadur, 2007) as shown in Figure 17.
1) Drilling of 40-75mm diameter hole to required length and inclination as per the site condition. The pattern and spacing of the grout holes will be based on groutability test.
2) Installation of a pipe with internal diameter to fit the expandable packers.
3) Placement of packer at the very end of the pipe and injection of a cement-based grout which fills the annular space between the rock and the pipe. Hardening for about 12 hours.
4) When the grout is hardened, drilling through the pipe to feasible length.
5) Placement of packer and pressure injection with appropriate cement grout for penetration into the rock mass in the drilled length of the hole. Termination criteria as per grout mix design.
6) After hardening of the injected grout, re-drilling through the pipe and injected area to design length beyond last drilled length.
7) Placement of the packer in the pipe and inject (repetition of step 5).
46 Figure 17 Grouting methodology as described in Bahadur, 2007
2.6.2 Water inflow acceptability criteria
Depending on the project needs, an acceptability criteria on admissible groundwater flow must be set. To examine if the pre-grouting operation fulfills the design inflow goals, measurements of water inflow into the tunnel must be done before and after grouting takes place. By considering these measurements (pore pressure, settlements, water table level), the sealing efficiency of the grouting can be assessed.
The sealing effect of the grouting operation, can be evaluated as (Dalmalm, 2004):
𝑆𝑒𝑎𝑙𝑖𝑛𝑔𝑒𝑓𝑓𝑒𝑐𝑡 =𝐼𝑛𝑓𝑙𝑜𝑤 𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑔𝑟𝑜𝑢𝑡𝑖𝑛𝑔 − 𝐼𝑛𝑓𝑙𝑜𝑤 𝑤𝑖𝑡ℎ 𝑔𝑟𝑜𝑢𝑡𝑖𝑛𝑔
𝐼𝑛𝑓𝑙𝑜𝑤 𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑔𝑟𝑜𝑢𝑡𝑖𝑛𝑔 (2.9) Research done by Palmström and Stille (2010) categorized the water inflow into underground excavation Table 1.
47 Table 1 Condition of tunnels according to inflow volume. Adapted from Palmström & Stille (2010).
Condition Inflow Volume (per 100 [m] of tunnel)
Dry <0.1 L/min
Seepage 0.1-1 L/min
Dripping 1-10 L/min
Flowing 10-500 L/min
Heavily flowing 30-300 m3/h
Water in-burst >300 m3/h
The thickness of the grouted zone is also an important parameter to consider on the sealing efficiency. Going beyond the optimal thickness could lead to an uneconomical project, and if the thickness is lower than needed, the inflows goals cannot be achieved (Dalmalm, 2004).
Another classification of water inflow can be seen from Table 2, which specifies inflow rate per meter of excavation, consequences and dominant geological structure (Sharifzadeh, 2012):
Table 2 Classification of water inflow rates for a six (6) [m] diameter tunnel. Adapted from Sharifzadeh (2012).
Class Inflow rate
[L/min/m] Description Flow mechanism
Dominating geological
feature
Consequences
I <12.5 Very low Dripping Porosity Insignificant
II 12.5-35 Low Leakage Porosity,
Bedding
Decreasing construction rate.
III 35-150 Medium Inflow Fracture, Dyke 1-7 days’ work delay
IV 150-350 High High inflow Fold, Fault
1-30 days’ work delay, with simultaneous ground improvement.
V 350-1000 Very high Inrush Fault, Karst
More than 1-month delay, required ground
grouting before constructing.
48 VI >1000 Extremely
high Water burst Fault, Karst
Construction stopped.
Required complex ground improvements.
Norwegian and Swedish experience allow as a maximum water flow of 2-4 L/min per 100 meters of tunnel for the former, and 2-5 L/min per 100 meters of tunnel for the latter. In Finland the accepted volume of water is less than 2 L/min per 100 meters. These are values set for areas where the consequences of water drainage can be severe. (Norwegian Tunneling Society, 2011; Stille, 2015; Sievänen, 2012). It is understood that depending on the tunnel’s usage, a higher water ingress can be tolerated. The acceptable flow of water ingress for traffic tunnels in Norway, are often set below 20 L/min per 100 [m]. (Holmøy & Nilsen, 2014).
2.6.3 Grouting requirements
Several geological parameters need to be considered in pre-grouting, to be able to reach the project goals. Among the most important factors are fracture systems of the rock mass, its mechanical properties and the in-situ stress. These geological characteristics affect the gout flow as well as the applicable grout pressure.
Stille (2012) categorized the difficulty in grouting, as a function of desired sealing effect and hydraulic conductivity of the grouted area:
Table 3 Grouting matrix according difficulty and goal hydraulic conductivity. Adapted from Stille (2012).
Required sealing efficiency Required conductivity
<90% 90-99% >99%
>10-7 Uncomplicated
grouting Fair Grouting Difficult grouting 10-7 to 10-8 Fair grouting Difficult grouting Very difficult grouting
<10-8 Difficult grouting Very difficult grouting Very difficult grouting
49 2.6.4 Rock improvement with pre-grouting
Not only changes in permeability are expected after pre-grouting. Barton (2019) reports an 8-fold increase in bulk modulus of the rock mass after grouting, a rotation in the three principal permeability tensors and more than one order of magnitude in reduction of hydraulic conductivity.
Furthermore, Barton et al (2019) also adds that if micro-cements and micro-silica-based additives rather than the common industrial cement and bentonite, dramatic changes in the effective rock mass properties are to be expected, as well as compressibility waves velocities around the tunnel.
Figure 18 Pre-grouting design concept in the Nygard project. (Butron et al, 2010)
Barton (2011) shows how the Q-system can be used to quantify improvements in the rock mass parameters after grouting. In his model, he assumes the rock mass to have a UCS equal to 100 [Ma]. (Figure 19).
50 Figure 19 Comparison of a conservative model with a more realistic model of possible improvements in individual Q-parameters, and how these might impact on rock mass properties and support needs.
As expected, a rock mass more damaged, has potentially more chances to be greatly improved after grouting (Barton, 2019). The improvements in Q index, have also positive impact in the needs of rock support.
2.7 W
ATER INFLOW INTO TUNNELSGroundwater inflow ingress into excavations, is a complex issue due to the convolutedness of mechanisms that take place in the phenomenon, and these are also governed by several other factors. For this said reason, the development of accurate and robust predictive models of inflow mechanisms requires necessarily an extense knowledge of the physical phenomena that govern the groundwater flow and its interaction with geological features.
Despite of the previously mentioned, the governing physical processes could become more or less complicated depending on the scale of the model, which could or could not take into account the micro and mega scales interconnections of the phenomena. While, for example the geological and regional hydrology govern the behavior in a macro-scale, features like shape, construction method, etc. control directly, the water inflow mechanisms triggered.
51 The main flow paths are interconnections between void pores, which are represented by primary porosity, and fractures. The latter greatly influences the hydraulic behavior of the host rock
Figure 20 Water inrush situations during tunneling. a) low flow water inrush, b) large flow water inrush. c) water with silt. d) water inflow from geological structure. (Hou, et al, 2016).
2.7.1 Effects of geological features
A diagram with the main geological interactions and its importance in influencing the groundwater flow is shown in Figure 21 (Sharifzadeh, 2012): by moving clockwise through the geological features, aspects like prediction capability and modeling capability are increased, whereas by moving in the opposite direction, inflow hazards and water control problems are incremented.
Figure 21 Geological features and its impact in numerical modeling. (Sharifzadeh, 2012).
52 An important fact is that by considering discrete features like faulting and karst formations, the matrix permeability becomes negligible in comparison, and the hydraulic behavior of the rock mass becomes heavily influenced by the characteristics of fractures. For this reason, fractures are the most important geological feature in regard to water inflow into tunnels (Sharifzadeh, 2012).
2.7.2 Brief description of geological features and its impact in water inflow into tunnels
Porosity
Effective porosity, which reveals the degree of interconnection between pores, where the water goes through, controls the permeability in soils and undisturbed rock masses. This parameter is function of shape size, arrangement and distribution of sizes of the grains, etc.
Under the assumptions of isotropic and homogenous materials, the water inflow into tunnels can be estimated through simple closed-form equations. (Goodman,1965).
Fracture
In geological media with low matrix permeability, fluid flow takes place in an important manner, through fractures and its networks. Fracture-dominated flow is key to understand, since the assumption of equivalent porous media is not always valid, depending on the scale of observation. A fracture or joint system behave more or less like a continuum when (Long, 1982):
1. Fracture density is increased.
2. Apertures are constant rather than following a distribution.
3. Orientations are more distributed instead of being homogenous.
4. The scale of the sample is large.
Bedding
Bedding planes can behave as a boundary of discontinuity between porous layers, as well as they can determine the fault-pattern in geological formations. In a layer between brittle and ductile rock, fractures tend to initiate within the stiffer beds and ends at the contact with
