03-05-2019

Stress field assessment for the design

of underground caverns
LNEC’s recent experience

José Muralha and Luís Lamas

LNEC – Laboratório Nacional de Engenharia Civil, Lisboa, Portugal

Permission from EDP – Energies of Portugal, S.A. and Iberdrola S.A.

to publish information regarding the testing programmes is acknowledged

Stress

The Euler–Cauchy stress principle states that upon any surface (real or imaginary) that divides the
body, the action of one part of the body on the other is equivalent (equipollent) to the system of
distributed forces and couples on the surface dividing the body.

LNEC | 2

1
 03-05-2019

Stresses in rock masses

• Main factors that influence the in situ stresses in rock masses

Geology
• Tectonic stresses
• Regional and local geology (rock lythologies, jointing, faults, folds, …)
• Mechanical characteristics (deformability, heterogeneity and anisotropy)
Gravity
• Depth
• Ground topography (valleys, mountains, ridges, …)
Geometry
• Ground topography (valleys, mountains, ridges, …)
• Nearby excavations and underground openings

LNEC | 3

Stresses in rock masses

Depth and ground topography

Folding
Rock mass heterogeneity and anisotropy
LNEC | 4

2
 03-05-2019

Stresses in rock masses

Nearby excavations and underground openings

LNEC | 5

Why do we need to know the in situ stresses?

• Relief of the in situ stresses is often the most relevant action in

underground projects
• May affect the location and orientation of caverns and tunnels
• Determines the design of the support
• Influences the construction methods and excavation stages (TBM, NATM, …)

• Confinement provided by in situ stresses is conditional for the

analysis of the flow
• Industrial and radioactive waste disposal
• Fluids storage (water, hydrocarbons, liquid gas)
• Hydraulic pressure tunnels (lined and unlined)
• Fracking (shale gas and shale oil)
• Contaminants transport

LNEC | 6

3
 03-05-2019

Why do we need to know the in situ stresses?

LNEC | 7

Why do we need to know the in situ stresses?

Numerical modelling
Analytical solutions

LNEC | 8

4
 03-05-2019

Why do we need to measure the in situ stresses?

• Stresses show high variability

• Reliable estimates are difficult, demanding, tricky

LNEC | 9

Why do we need to measure the in situ stresses?

• Stresses show high variability

• Reliable estimates are difficult, demanding, tricky

LNEC | 10

5
 03-05-2019

Testing methods for in situ stress measurements

• Stress relief methods

• Complete – determination of the six component of the stress tensor (overcoring)
• Partial – determination of some components of the stress tensor
• Single – determination of just one component of the stress tensor (flat jack)
• Rock fracturing
• Hydraulic methods (HF and HTPF)
• Mechanical methods
• Observation of rock or rock mass behaviour
• Kaiser effect and strain recovery
• Borehole breakouts and core disking
• Back analysis using displacements around excavations
• Geologic interpretation of seismic-tectonic events

LNEC | 11

Testing methods
• Flat jack tests – LNEC’s SFJ equipment

25
45º 135º
Slot at 45º 20
Hor Vert
Pressure (MPa)

Horizontal 15
Slot at 135º slot
10

5
Vertical
0
slot
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Displacement (mm)

LNEC | 12

6
 03-05-2019

Testing methods
• Overcoring tests – LNEC’s STT 3D cell

LNEC | 13

Testing methods
• Hydraulic fracturing tests (HF, HTPF) – UniStra’s equipment

LNEC | 14

7
 03-05-2019

• LNEC’s SFJ equipment

LNEC | 15

• LNEC’s SFJ equipment

LNEC | 16

8
 03-05-2019

• LNEC’s SFJ equipment

TAC Pk 646 6,0

Rasgo 90º

5,0

4,0
Pressure (MPa)

3,0

2,0

1,0

0,0
-0,20 -0,15 -0,10 -0,05 0,00 0,05

Displacement (mm)

Disp. point 1 Disp. point 2 Average

LNEC | 17

• LNEC’s SFJ equipment

TAC Pk 646 6,0

Rasgo 90º

5,0

4,0
Pressure (MPa)

3,0

2,0

1,0

0,0
-0,20 -0,15 -0,10 -0,05 0,00 0,05

Displacement (mm)

Disp. point 1 Disp. point 2 Average

LNEC | 18

9
 03-05-2019

• LNEC’s SFJ equipment

LNEC | 19

• LNEC’s STT SFJ equipment. Issues

• Direct measurement of a single stress component

• 3D stress tensor measurement requires several tests
• Reliable results in good quality rock
adequate choice of the test locations and surface finishing
avoid rock surfaces close to the floor of adits

LNEC | 20

10
 03-05-2019

• LNEC’s STT 3D cell

Strains (10-6)

Time (min)

LNEC | 21

• LNEC’s STT 3D cell

LNEC | 22

11
 03-05-2019

• LNEC’s STT 3D cell. Calculations

LNEC | 23

• LNEC’s STT 3D cell. Calculations

LNEC | 24

12
 03-05-2019

• LNEC’s STT 3D cell. Issues

• Point wise measurement of the 3D stress tensor

• Performed down to 100 m depth in large borehole (120 – 140 mm i diameter)
• Needs around 50 cm of intact rock, which is difficult to tell in advance
• Recovery of the core with the cell to perform bi-axial test to calculate
the elastic constants
• Requires a specialized drilling team

LNEC | 25

Testing methods
• Verification

LNEC | 26

13
 03-05-2019

Testing methods
• Hydraulic fracturing tests (HF, HTPF) – UniStra’s equipment

LNEC | 27

Why is a global stress interpretation model needed?

• Number of stress measurement tests is usually small.

• The tests only characterize the stress state in the locations where
they were executed.
• Allows to integrate all results from tests performed:
• in different locations
• using different measurement techniques
• The main factors that affect the stress field are taken into account:
• significant topographic changes,
• the existence of underground excavations in the zone of interest,
• the variability of the rock mass mechanical properties, etc.

LNEC | 28

14
 03-05-2019

Stress state global analysis

• Methodology for the stress field evaluation

Main assumption: The in situ state of stress, prior to the disturbance caused by
significant topographic changes or existing excavations, varies linearly with depth:

σ j 0 = kj γ h (j = 1,...,6)

where:
kj - parameters that characterise the variation with depth of the in situ state of stress
γ - rock mass unit weight
h - depth
j = 1, … , 6 - 6 loadings σj0 (corresponding to the 6 initial stress components),
with unit kj values, are applied separately in a numerical model of the rock mass.

LNEC | 29

Stress state global analysis

• Methodology for the stress field evaluation
?
∑ A ijk j = Mi (i = 1, …, N) (j = 1, …, 6)

stress values in the N

in situ measurements
Component of the state of stress
obtained for each stress measurement
(i = 1, ..., N), for the loading j kj parameters obtained by the
least squares method

Estimated stress field

σm = ∑ k j σmj (m = 1,…,6) (j = 1,…,6)

LNEC | 30

15
 03-05-2019

Stress state global analysis

• Loading j = 1 loading σ10 with unit k1 value

y h

z x
γh

LNEC | 31

Stress state global analysis

• Loading j = 2 loading σ2 with unit k2 value

LNEC | 32

16
 03-05-2019

Stress state global analysis

• Loading j = 3 loading σ3 with unit k3 value

LNEC | 33

Stress state global analysis

• Castelo de Bode Tunnel (Sousa, Martins & Lamas, 1986)

LNEC | 34

17
 03-05-2019

Application examples
Hydropower
Date Rock type Type of tests
scheme
Venda Nova II 2003 Granite STT
Baixo Sabor 2005 Granite STT
Picote II 2006 Granite STT
Bemposta II 2008 Gneiss and micaschist STT+SFF
Salamonde II 2009/2012 Granite SFJ/STT
Foz Tua 2010 Granite and metagreywacke STT
Alvito 2010 Schist and greywacke STT
Diamer Bhasha* 2010 Gabronorite STT
Neelum Jhelum* 2010 Sandstones, mudstones STT
Paradela II 2011 Granite STT+SFJ+HTPF
Gouvães 2016 Granite and micaschist SFJ+STT
* LNEC provided the equipment, trained the technicians and performed the calculations

LNEC | 35

Application examples
Hydropower
Date Rock type Type of tests
scheme
Venda Nova II 2003 Granite STT
Baixo Sabor 2005 Granite STT
Picote II 2006 Granite STT
Bemposta II 2008 Gneiss and micaschist STT+SFF
Salamonde II 2009/2012 Granite SFJ/STT
Foz Tua 2010 Granite and metagreywacke STT
Alvito 2010 Schist and greywacke STT
Diamer Bhasha* 2010 Gabronorite STT
Neelum Jhelum* 2010 Sandstones, mudstones STT
Paradela II 2011 Granite STT+SFJ+HTPF
Gouvães 2016 Granite and micaschist SFJ+STT

LNEC | 36

18
 03-05-2019

Picote II
• New underground cavern with 22 m width

LNEC | 37

Picote II
• New underground cavern with 22 m width

LNEC | 38

19
 03-05-2019

Picote II
• Six STT overcoring tests in two boreholes inclined at 70º
• Test depths between 40 and 80 m

LNEC | 39

Picote II
• Assumptions:
• The main action is the excavation of the valley by the river
• Plane strain condition
• σv, (vertical stress) σh (horizontal stress in the model 2D plane) and
σn (horizontal stress perpendiculr to the model 2D plane) are principal stresses
• σv prior to the valley excavation equal to the overburden
• σh and σn prior to the valley excavation increase linearly with depth
• σh = k1 γ h, σv = γ h and σn = k3 γ h

LNEC | 40

20
 03-05-2019

Picote II
Borehole STT1 STT2 Values considered in the
78.35
• FLAC2D model Depth (m)
E (GPa)
66.10
38.4 41.5
60.60
41.3
77.45
39.3
numerical model
40
• Length – 1000 m ν(-) 0.16 0.16 0.26 0.35 0.2

• Elements: 200 x 300

• Elements in STT region: 2.5 x 1.75 m

LNEC | 41

Picote II
• Calculated stress field k1 = 1.70 and k3 =1.75,
corresponding to σh ≈ 4 in the powerhouse cavern area

LNEC | 42

21
 03-05-2019

Picote II
• The high horizontal stresses were confirmed by small flat jack tests
performed during construction in chambers excavated in access tunnels and
incipient spalling observed along particular orientations of the access tunnels

FLAC3D 3.00
Step 6315 Model Perspective
18:05:56 Mon Dec 22 2008

Center: Rotation:
X: -1.733e+000 X: 0.000
Y: 5.000e-001 Y: 0.000
Z: 2.889e+000 Z: 0.000
Dist: 2.770e+002 Mag.: 3.81
Ang.: 22.500

Plane Origin: Plane Orientation:

X: 0.000e+000 Dip: 90.000
Y: 5.000e-001 DD: 0.000
Z: 0.000e+000

Contour of SZZ
Plane: on
Magfac = 0.000e+000
Gradient Calculation
-2.4445e+007 to -2.0000e+007
-2.0000e+007 to -1.5000e+007
-1.5000e+007 to -1.0000e+007
-1.0000e+007 to -5.0000e+006
-5.0000e+006 to 0.0000e+000
0.0000e+000 to 1.6104e+006
Interval = 5.0e+006
Principal Stresses
Plane: on
Magfac = 0.000e+000
Local face system
EDP Energias de Portugal
PIEC/Fundações e Estruturas

LNEC | 43

Bemposta II
• Repowering of a hydraulic scheme
with a new powerhouse in a shaft
(80 m in height and 30 m in diameter)

LNEC | 44

22
 03-05-2019

Bemposta II
• Tests were performed in two locations
• In each, three small flack jack and three
overcoring tests were performed

LNEC | 45

Bemposta II
• Assumptions:
• The main action is the excavation of the valley by the river
• Plane strain condition
• σv, (vertical stress) σh (horizontal stress in the model 2D plane) and
σn (horizontal stress perpendiculr to the model 2D plane) are principal stresses
• σv, σh and σn prior to the valley excavation increase linearly with depth
• σh = k1 γ h, σv = k2 γ h and σn = k3 γ h

STT2

B’

LNEC | 46

23
 03-05-2019

Bemposta II
• FLAC3D model to take into account the excavation of the adit where
tests were performed

LNEC | 47

Bemposta II
• Results: k1 = 0.604 k2 = 0.906 k3 = 0.748

LNEC | 48

24
 03-05-2019

Bemposta II
• FLAC3D model validation

Model: 1m Model: 11m Model: 63m Model: 141m

k1 0.604 0.604 0.608 0.615
k2 0.906 0.906 0.907 0.911
k3 0.748 0.746 0.756 0.759

LNEC | 49

Salamonde II

Cavern overburden: 150 m

Rock mass: granite

LNEC | 50

25
 03-05-2019

Salamonde II
• 6 overcoring tests during the
design stage
In 2 boreholes, S8 and S13, before
construction started
Cávado river

Mau river
• 10 flat jack tests during the S13
construction stage Existing powerhouse
Salamonde I

At 3 locations, SFJ1, SFJ2 and SFJ3, SFJ3 New powerhouse

during early construction stages Salamonde II

SFJ2
SFJ1
0 20 40 60 m
S8

LNEC | 51

Salamonde II
• Integration of OC+FJ results
3D numerical model (FLAC3D)
350m

350m

193m

0 20 40 60 m
210m

Ground topography
Rock mass: continuous, linearly elastic, homogeneous and isotropic,
(E = 35 GPa ; ν = 0,20 ; γ = 27 kN/m3)

LNEC | 52

26
 03-05-2019

Salamonde II
• Assumptions:
• The ground surface geometry and the excavations at the time the flat jack
tests were performed are included in the model
• Plane strain condition
• σv, (vertical stress) is a principal stress and equal to the overburden
• Initial vertical (normal) stress σ03 = k3 γ h
• Initial horizontal (normal) stresses, σ01 = k1 γ h and σ02 = k2 γ h
• Initial shear stress in the horizontal plane σ04 = k4 γ h

LNEC | 53

Salamonde II
• The ground surface geometry and the excavations at the time the flat jack
tests were performed are included in the model

SFJ2
SFJ1

SFJ3

LNEC | 54

27
 03-05-2019

Salamonde II
• Calculated initial stress field

k1 = 0.88 (x)
k2 = 0.98 (y)
k3 = 1.00 (z)
k4 = 0.31 (xy)

LNEC | 55

Salamonde II
• Stresses in the roof (side-drifts) after excavation

σxx Pillar 3 σyy

Pillar 2
Access Pillar 1
tunnel

(Pa)
σzz σxy

LNEC | 56

28
 03-05-2019

Paradela II
• Time dependent effects

Cavern overburden: 500 m

LNEC | 57

Paradela II
HF/HTPF

• 6 overcoring tests OC/SFJ

In boreholes PD1 and PD2

• 12 flat jack tests

At locations SFJ1, SFJ2 and SFJ3

• HF/HTPF tests
Adit
At depths between 200 m and
500 m in boreholes PD19 and
PD23

LNEC | 58

29
 03-05-2019

Paradela II
• HF/HTPF (lines) and OC (dots) test results

LNEC | 59

Paradela II
• Interpretation of HF/HTPF results
3D numerical model (FLAC3D) Rock mass: homogeneous,
continuous and isotropic

Consideration of time dependent

effects:
- Maxwell viscoelastic model
- “Relaxed” equivalent elastic
properties in the 3D model:
ν E

ν = 0.47 (due to gravity loading)

LNEC | 60

30
 03-05-2019

Paradela II
• Interpretation of HF/HTPF results
Measured versus calculated stresses

Borehole PD19 Borehole PD23

LNEC | 61

Paradela II
• Interpretation of HF/HTPF results
Measured and calculated principal stresses

FM – results from the viscoelastic model

LNEC | 62

31
 03-05-2019

Gouvães

Access adit Cavern overburden - 500 m

Mica-schist (lower E)
Tâmega overlying granite (higher E)
River
Transformers Powerhouse
cavern cavern

mica-schist

contact
surface

granite Mica-schist
Granite

LNEC | 63

Gouvães
• Stress measurements
Were performed when the exploratory adits were completed.
For each rock type:
12,24 m
- 2 x 4 flat jacks in SFJ
13,10 m
STT
perpendicular rock SFJ
4 flat jacks 30,59 m
31,55 m 6 tests
faces 12,09 m
4 flat jacks
46,25 m

STT
- 6 overcoring tests in 6 tests
28,38 m
30,35 m Perspective from West
vertical boreholes 31,31 m

61,48 m
63,93 m
SFJ
3 flat jacks
STT

GEC
STT
SFJ
4 flat jacks
SFJ
3 flat jacks Perspective from East

LNEC | 64

32
 03-05-2019

Gouvães
Elevation
532.0 m • 3D numerical model
- topography
Elevation
415.2 m - existing excavations
Elevation
330.1 m - different rock lythologies
- contact surface
Elevation
350.9 m
Elevation at the
700 m bottom 20 m
570 m

Mica-schist
Mica schist
Granite

Granite

LNEC | 65

Gouvães
• Estimated and measured elastic constants
Laboratory tests Hoek-Brown Flat jack tests Biaxial tests for
Rock type (rock) (rock mass) (rock mass) overcoring (rock)
Ei (GPa) ν Em (GPa) ESFJ (GPa) ESTT (GPa) ν
Mica-schist 29.0 0.25 11.8 18.2 16.9 0.25
Granite 40.4 0.24 25.5 17.2 55.1 0.23

Very large difference between the values of the rock mass Young’s modulus
- estimated from laboratory tests and GSI (Hoek-Brown),
- and measured in the field by flat jacks.

• Two situations were studied

- homogeneous case (Egranite = Emica-schist = 25 MPa)
- heterogeneous case (Egranite = 25 MPa, Emica-schist = 12.5 MPa)

LNEC | 66

33
 03-05-2019

Gouvães
• Calculations
Homogeneous Heterogeneous
Principal situation situations
stress / A – Mica-schist B – Mica-schist
direction C – Mica-schist D – Granite
+ Granite + Granite
k Azimuth k Azimuth k Azimuth k Azimuth
V 1.00 vertical 1.00 vertical 1.00 vertical 1.00 vertical
Hmax 1.23 45º 1.22 46º 1.26 49º 1.19 42º
Hmin 0.66 135º 0.68 136º 0.54 139º 0.77 132º

A – reference situation
B – check the influence of the proximity of the test locations to the contact surface
C – tests in mica-schist only, to represent more accurately the stresses in the mica-schist
D – tests in granite only, to represent more accurately the stresses in the granite

LNEC | 67

Gouvães
N
Elevation
Calc. B Calc. C Calc. D Powerhouse
(m)
• Principal directions 233.0
- - -
longitudinal axis
(mica schist)
and stresses (MPa) 170.6
-9.16
-9.42
- - -
for calculations (granite)
-7.92 -8.18

σI
B, C and D

- sI sub-horizontal, aligned σII

-7.37
-8.62
W E
with the Tâmega River -7.36
-8.62

- sII subvertical
- sIII sub-horizontal σIII
-5.54
-4.23
-4.94
-3.38
Elevation 233.0 m (mica schist)
Elevation 170.6 m (granite)

Tâmega river S

LNEC | 68

34
 03-05-2019

Gouvães
Horizontal plane at mid Vertical plane normal to
cavern height cavern axis

Mica schist
Mica schist

Situation B
heterogeneous Granite
initial principal
Granite
stresses

LNEC | 69

Gouvães
• Direct consideration of the stress field in the complex numerical
models developed to support the design

LNEC | 70

35
 03-05-2019

Gouvães
• Direct consideration of the stress field in the complex numerical
models developed to support the design

LNEC | 71

Concluding Remarks

• Relief of the in situ state of stress is usually the most important action to take into
account for design of underground powerhouse caverns.

• Tests are expensive and their number is usually small.

• Variability of test results is usually high.

• Individual tests:
- involve small volumes of rock,
- provide the stress state only in the locations where they are performed,
- supply results of reduced interest when considered individually.

• The stress measurement programme must take into account the main rock mass
features and the method of analysis of the test results.

LNEC | 72

36
 03-05-2019

Concluding Remarks

• Integration of stress measurement results, using different testing methods, improves

the quality of the in situ stress estimates.

• A global analysis of the stress measurement results, using an inverse methodology,

provides an estimate of the most likely stress field in the rock mass in the zone of
interest for the problem under analysis.

• Numerical models allow considering ground topography, nearby excavations,

heterogeneities and other relevant geometrical or mechanical features.

• Influence of gravity, topography, tectonic forces or time dependent effects can easily
be simulated.

• The initial in situ stress field, which otherwise would have to be assumed, can be
provided to the designers of the underground structures.

LNEC | 73

Muchas Gracias

37