Deformation and transport processes in salt rocks: An ...
Transcript of Deformation and transport processes in salt rocks: An ...
I
Deformation and transport processes in salt rocks:
An experimental study exploring effects of pressure and
stress relaxation
Nawaz Muhammad
Utrecht University
No. 084
II
Members of the dissertation committee:
Prof. dr. Janos L. Urai
RWTH Aachen University, Germany
Prof. dr. Hongwei Zhou
Xi'an Technological University, China
Prof. dr. Georg Dresen
GFZ German Research Centre for Geosciences, Germany
Prof. dr. M.R. Drury
Utrecht University, The Netherlands
Dr. Chloé Arson
Georgia Institute of Technology, USA
The research was carried out at:
High Pressure and Temperature Laboratory, Faculty of Geoscience, Utrecht University
(Budapestlaan 4, 3584 CD Utrecht, The Netherlands)
Printed by: Gildeprint
Copyright © Nawaz Muhammad
All right reserved. No parts of this publication may be reproduced in any form, by print or photo print, microfilm or
other means, without written permission by the publisher.
ISBN/EAN: 978-90-6266-396-5
III
Deformation and transport processes in salt rocks:
An experimental study exploring effects of pressure and
stress relaxation
Deformatie- en transportprocessen in zoutgesteentes:een experimentele studie naar de
effecten van druk en spanningsrelaxatie
(met een samenvatting in het Nederlands)
Proefschrift
ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector
magnificus, prof. dr. G.J. van der Zwaan, ingevolge het besluit van het college voor
promoties in het openbaar te verdedigen op dinsdag 23 juni 2015
des ochtends te 10.30 uur
door
Nawaz Muhammad
geboren op 15 december 1975
te Lahore, Pakistan
IV
Promotor: Prof. dr. C.J. Spiers
Copromotoren: Dr. J.H.P. de Bresser
Dr. C.J. Peach
Dit proefschrift werd (mede) mogelijk gemaakt door een beurs toegekend door de Higher
Education Commission (HEC) van Pakistan aan Nawaz Muhammad, en door aanvullende
financiële ondersteuning onafhankelijk ter beschikking gesteld door AkzoNobel Industrial
Chemicals B.V., Nedmag Industries Mining & Manufacturing B.V., en de Nuclear Research
and Consultancy Group NRG.
V
To My Parents
VI
He created the heavens without any pillars that ye can see; He set on the
Earth Mountains standing firm, lest it should shake with you;
Surah Luqman, Chapter 31, Ayah 10
By Him in Whose Hand is my soul, the Hour (Qiyamah) will not come until
wild creatures talk to men, and a man speaks to the end of his whip and the
straps of his sandals (shoes), and his thigh will tell him about what happened
to his family after he left.
Prophet Muhammad (SAW)
Ahmad Musnad (3:84 and 85),
Hakim Mustadrak (4:467).
VII
Contents
Synopsis 1
Chapter 1: Introduction 7
1.1 Scope of the present work 8
1.2 Rocksalt 9
1.3 Bischofite and carnallite salt rocks 11
1.4 Transport properties and permeability of rocksalt 11
1.5 The main aims of this study 12
1.6 Plan of thesis 14
Chapter 2: The transition from dislocation glide to dislocation climb
controlled creep in dry rock salt in the temperature range 22-
350 oC (0.27-0.58T/Tm): using the pressure sensitivity of stress
to evaluate microphysical models
2.1 Introduction 18
2.2 Microphysical models for dislocation creep 23
2.2.1 Dislocation climb 23
2.2.2 Dislocation cross-slip 24
2.2.3 Dislocation glide 27
2.3 Method 29
2.3.1 Sample preparation 29
2.3.2 Deformation apparatus, testing and data processing 30
2.3.3 Preparation for microstructural study 33
2.4 Results 34
2.4.1 Mechanical data 34
(i) Experiments at room temperature (22 oC) 34
(ii) Experiments at 125-350 oC 34
(iii) Pressure sensitivity of normalized flow stress at different strain
values 42
(iv) General inferences from the observed pressure dependence of
creep 43
(v) n-value (stress exponent) 49
VIII
2.4.2 Microstructures 50
2.5 Discussion 55
2.5.1 Results of non-linear regression analysis 56
(i) Climb controlled creep 56
(ii) Cross-Slip controlled creep 59
(iii) Glide controlled creep 60
2.5.2 Comparison with previous studies involving tests at various
pressures 64
2.5.3 Deformation mechanism map 64
Conclusions 66
Chapter 3: Stress relaxation of synthetic and natural polycrystalline halite
3.1 Introduction 70
3.2 Method 72
3.2.1 Sample preparation 72
3.2.2 Deformation apparatus 73
3.2.3 Experiments 76
3.2.4 Data acquisition and processing 76
3.2.5 Microstructural preparations 79
3.3 Results 80
3.3.1 Mechanical data 80
(i) Stress vs. natural strain and time 80
(ii) n-value for the natural halite 86
(iii) Stress relaxation 86
3.3.2 Microstructures 90
3.4 Discussion 92
3.4.1 n-value 92
3.4.2 Composite flow law 94
Conclusions 104
IX
Chapter4: Creep behaviour of bischofite, carnallite and mixed bischofite-
carnallite-halite salt rock at in situ conditions
4.1 Introduction 107
4.2 Method 108
4.2.1 Sample preparation 111
4.2.2 Deformation apparatus 112
4.2.3 Experiments 115
4.2.4 Data acquisition and processing 115
4.2.5 Young’s modulus measurement 118
4.2.6 Microstructures (only carnallite) 118
4.3 Results 118
4.3.1 Bischofite 119
(i) Stress vs. strain curves 119
(ii) Effect of confining pressure 123
(iii) Flow behaviour 128
(iv) Stress relaxation 131
4.3.2 Carnallite 134
(i) Stress vs. strain curves 134
(ii) Flow behaviour 134
(iii) Stress relaxation 141
(iv) Microstructures 144
4.3.3 Mixture samples of bischofite, carnallite and halite 146
(i) Stress vs. strain curves 146
(ii) Flow behaviour 147
(iii) Stress relaxation 151
4.3.4 Elemental analysis using micro X-Ray Fluoroscopy (μ-XRF) 153
4.4 Discussion 160
4.4.1 Mechanical behaviour 160
(i) Creep law for bishofite using stresses at the end of constant strain
rate steps 161
(ii) Creep law for carnallite using stresses at the end of constant strain
rate steps 163
X
(iii) Comparison of steady state values 165
(iv) Relating the relaxation behaviour to the creep at near steady
state 165
(v) Composite flow laws 176
4.4.2 Effect of composition 180
Summary and conclusions 182
Chapter 5: Permeability of interfaces in layered rock salt under different
stresses and geometries
5.1 Introduction 186
5.2 Method 187
5.2.1 Samples source, composition and preparation for experiments 187
5.2.2 Apparatus and testing conditions 190
5.2.3 Calibrations 192
5.2.4 Experimental procedure and data processing 192
5.2.5 Preparations for Microstructural study 194
5.3 Results 195
5.3.1 Mechanical data 195
5.3.2 Permeability 200
5.3.3 Results Batch-I (20 MPa confinement) 201
(i) NP1 (vertical interface) 201
(ii) NP2 (oblique interface) 202
(iii) NP3 (mixed) 202
(iv) NP4 (oblique interface) 203
(v) NP5 (interlayer with salt) 204
5.3.4 Results Batch-II (10 MPa confinement) 207
(i) NP6 (vertical interface) 207
(ii) NP7 (horizontal interface) 207
(iii) NP8 (oblique interface) 207
(iv) NP4-Re-test (oblique interface) 208
5.4 Microstructures 217
(i) NP3 (mixed) 218
(ii) NP4 (oblique interface) 218
(iii) NP6 (vertical interface) 218
(iv) NP7 (horizontal interface) 219
XI
(v) NP8 (oblique interface) 220
5.5 Discussion 221
5.5.1 Summary of results 221
(i) Comparison of Batch-I results 221
(ii) Comparison of Batch-II results 223
5.5.2 Effect of confining pressure 223
5.5.3 Elastic response of existing cracks to the hydrostatic pressure 224
5.5.4 Compaction and dilatancy 226
5.5.5 Localised dilatancy at interface 226
5.5.6 Microstructures 227
5.5.7 Consequences of drilling the interfaces 229
5.5.8 Permeability: bulk and interface 231
5.5.9 Comparison with previous work and implications 232
Conclusions 233
Chapter 6: Conclusions and suggestions for further refinement 239
References 251
Samenvatting 261
Acknowledgements 269
Curriculum vitae 272
XII
1
Synopsis
2
The presence of evaporitic formations in sedimentary basins, often
dominated by the salt mineral halite, is of great influence on the structural
style developed during tectonic events. On a somewhat smaller scale, salt
rocks often host a variety of deep solution mined caverns, which are
increasingly finding use for strategic storage of energy resources in the form
of gaseous or liquid fuels and as vessels for off-peak energy storage in
compressed gas. This is in addition to the use of conventionally mined
galleries for the long term storage of hazardous waste materials. The low
permeability of most salt rocks, with weak rheological behaviour and
tendency to self-seal, has attracted engineers to use salt formations to host
such projects. The Zechstein salt deposits in the Netherlands exist in the
form of original beds and as migrated salt domes/pillows that may reach the
kilometre scale. Some contain magnesium rich salts which are solution
mined as an economic resource. Caverns constructed within these more
soluble and weaker materials pose additional challenges for long term
storage management. Careful management of fluid filled caverns requires a
full knowledge of the rheological and transport properties of the host salt
formations for a safe operation into the future.
This thesis addresses a number of deficiencies in the current knowledge of
salt mechanical behaviour regarding the mechanical creep of halite and its
mixtures with magnesium salts. In particular the pressure sensitivity of
creep is investigated, using synthetic and natural halite rock, to identify the
constituent mechanisms which contribute to the overall deformation
process, with a view to better understand the physics of salt flow. The
rheology of the magnesium bearing salts; carnallite and bischofite, is also
investigated, with special attention to the mechanical behaviour of mixtures
of these salts with halite. In addition to rheological issues, the effects of
compositional layering on permeable transport in the excavation damage
3
zones around mined cavities are investigated, using natural layered material
from mines in China, to better understand effects of bedding orientation
relative to deformational stresses expected in cavity and gallery walls at
depth.
Chapter 1 covers the literature survey on salt rheology, main aims of this
study and planning of the thesis.
Chapter 2 is about the rate controlling mechanism in the creep of
polycrystalline dry salt. The temperature range covered in this work is from
22 to 350 oC. The deformation/strain rates used lie in the range 4×10
-7 s
-1 to
10-4
s-1
, including constant strain rate and strain rate stepping tests. The main
idea behind this part of the study is to discriminate between several
dislocation creep mechanisms, such as dislocation climb, dislocation cross-
slip and dislocation glide, using the pressure sensitivity of flow stress of dry
salt. Theory suggests that there is an atomistic activation energy and
activation volume associated with solid state flow under dislocation creep.
Following theoretical models then, the salt is expected to show higher
strength under higher confining pressures for positive activation volumes
and lower strength under lower confining pressure for negative activation
volume, depending upon the rate controlling mechanism. For this
investigation, the confining pressure was varied in the range 50-600 MPa,
which was never systematically done before for fully confined salt. The salt
was found stronger at higher confining pressures, and the rate controlling
mechanism was found to be transitional from glide to climb.
Chapter 3 is about the flow law at in situ conditions of wet halite, using
synthetic and natural salt samples. The temperature and confining pressure
applied were 125 oC and 50 MPa, respectively. The samples were tested in
multi-step experiments with constant strain rate parts (5×10-5
– 5×10-8
s-1
)
4
followed by stress relaxation after selected steps. The relaxation data
showed that the rate controlling mechanism does not remain the same, but
changes from dislocation creep (at higher stresses and faster strain rates) to
grain size dependent (likely pressure solution) creep at the end of relaxation
(i.e. at lower stress and strain rates).
Chapter 4 is a compilation of tests on bischofite, carnallite and their
mixture with halite, performed at real in situ conditions of confining
pressure 40 MPa and temperature 70 oC. These tests were conducted on
polycrystalline samples obtained from natural cores. The samples were
tested employing deliquescence conditions around the sample in the same
way as reported in Chapter 3 for halite, namely in multi-step experiments
with constant strain rate parts (10-5
– 10-8
s-1
) followed by stress relaxation
after selected steps. The results revealed that the carnallite was stronger than
bischofite and the mixture was in turn stronger than carnallite. The strength
of the mixture was found to have a direct relation to the halite weight
percentage i.e. higher halite wt. % makes the mixture stronger. The
compositions of various samples were tested using a micro XRF technique.
Stress relaxation showed a change of flow mechanism from dislocation
creep at the higher stresses and strain rates (at the start of relaxation) to
grain size dependent creep towards lower stress and strain rates (at the end
of relaxation). Composite flow laws resulted for bischofite and carnallite, as
a combination of grain size insensitive creep and grain size sensitive creep
for higher and slower stresses, respectively.
Chapter 5 is about the transport properties of layered bedded salt rocks
from salt mines in China. The samples were tested for their permeability,
under increasing differential stresses, at the interface of two layers namely
halite and glauberite, using transient step argon gas permeametry method.
5
The experiments are classified into two batches, batch-I for comparatively
higher depth (20 MPa confinements) and batch-II for shallower depths (10
MPa confinements). Various geometries, with interface orientation (vertical,
oblique and horizontal) to deformation direction were tested by adding
stress in steps of 10 MPa. The axial and volumetric strains, during
deformation, were measured, which revealed that all samples were
compacted despite the increase or decrease in permeability. The
microstructural investigation revealed that local dilatancy occurred at
interfaces, which must give a local increase in permeability but was masked
by the bulk compaction of the samples.
6
7
Chapter 1
Introduction
8
1.1 SCOPE OF THE PRESENT WORK
The presence of evaporitic formations in sedimentary basins, often
dominated by the salt mineral halite, is of great influence on the structural
style developed during tectonic events (e.g. Hudec & Jackson 2007). Owing
to its low viscosity, causing Halokinesis, a whole range of complex
structures may develop involving salt rock, varying from salt pillows and
diapirs to detached salt sheets and even glaciers (e.g. Talbot 1979, Jackson
& Talbot 1991, Vendeville & Jackson 1991, Rouby et al. 2002, Debois et al.
2010). These complex structures are of great interest for the oil industry,
since many hydrocarbon findings are associated with salt structures (e.g.
Davison et al. 2000, Tang et al. 2004, Huvaz et al. 2007, see also Hudec &
Jackson 2007). On a somewhat smaller scale, salt rocks host a variety of
deep caverns and mined galleries which are increasingly finding use for
strategic storage of energy resources in the form gaseous or liquid fuels and
also as off peak buffered storage in the form of compressed air (Stormont
1997, Istavan et al. 1997) This type of strategic storage comes in addition to
the long term storage of hazardous waste materials (Hunsche & Hampel
1999, Tsang et al. 2005). The low permeability of most salt rocks, with
weak rheological behaviour and tendency to self-seal has attracted engineers
to use salt formations to contain such projects. Management of caverns
requires a full knowledge of the rheological and transport properties of the
host salt formations for a safe operation into the future. Note that in addition
to halite, this also concerns salt rocks consisting of minerals such as
bischofite and carnallite. Expanding our understanding of the rheological
and transport properties of halite, bischofite and carnallite forms the broad
aim of this study. Most importantly, such enhanced understanding will help
reliable extrapolation of laboratory results to in situ conditions, and is of
9
importance for modelling the flow of salt in settings varying from salt
diapirs to solution–mined cavities and the stability and integrity of bore
holes through sediment packages containing salt horizons.
1.2 ROCKSALT
Despite the large body of data that already exists on the mechanical
behaviour of rocksalt and resulting empirical flow laws (Heard 1972, Heard
& Ryerson 1986, Wawersik & Zeuch 1986, Carter 1993, Ter Heege et al.
2005a), the underlying question regarding what microphysical mechanisms
govern plastic flow of natural dense halite rock at in situ conditions has not
yet been answered to full satisfaction. In particular, the exact mechanism
controlling dislocation motion at relatively low temperature is still
insufficiently understood. As a result, uncertainties exist regarding the
appropriate mechanism-based flow-law for low temperature, hampering
reliable extrapolation of lab creep data to in situ strain rates. Several
dislocation models have been proposed to control plastic flow of rocksalt at
in situ conditions, such as dislocation climb (Senseny et al. 1992, Carter et
al. 1993, Franssen 1994), and dislocation cross-slip or glide (Auten et al.
1973, Skrotzki et al. 1981, Wawersik & Zeuch 1986, Conrad & Yang 1999).
Since the various dislocation mechanisms are characterized by quite
different constitutive equations, the results of extrapolation of laboratory
data may vary substantially depending on the inferred mechanism, thus
hampering reliable modelling. Many of the flow laws presented in the
literature for rocksalt are based on short term experiments with little or no
basis regarding the exact microphysical mechanism controlling creep. One
way to test which model is appropriate is by investigating the pressure
dependence of strength of rocksalt, which works against the atomistic scale
activation volume associated with the different mechanisms governing the
10
rate of dislocation creep. Dislocation glide is expected to be hardly affected
by pressure, notably at low temperature, dislocation cross slip, in which
dislocations overcome obstacles by temporary gliding onto a plane oriented
oblique to the initial glide plane, may either become easier with increasing
pressure or more difficult, depending on the details of the cross slip process,
and dislocation climb is expected to become more difficult with pressure. In
this study, new data on the rheology of rocksalt, under dry conditions, have
been tested against various models taking the effect of confining pressure
into account. The rock salt was taken dry to avoid water-involved processes
like solution-precipitation to play a role.
Salt rocks in nature are wet rather than dry (Urai et al. 1986, Spiers et al.
1990). The creep of these salt rocks usually occurs at strain rates in the
range of 10-8
to 10-15
s-1
(Heard 1972, Van Eekelen et al. 1981, Jackson &
Talbot 1986, Carter et al. 1993). In order to fully understand the creep
behaviour of halite under natural conditions, it thus is of importance to not
only know, in general, which mechanisms may control dislocation creep of
halite, i.e. dislocation glide, cross slip or climb, but also to establish to what
extent solution-precipitation mechanisms play a role and at what conditions
a given mechanism prevails. The strain rates relevant for in situ
deformation, 10-8
to 10-15
s-1
, are difficult to achieve in laboratory
experiments. However, stress-relaxation experiments (Rutter & Mainprice,
1978) allow strain rates as slow as 10-9
s-1
to be achieved, by allowing the
stress on a sample to relax through plastic deformation. In the current study,
such experiments have been carried out on a limited number of synthetic
and natural wet polycrystalline halite samples.
11
1.3 BISCHOFITE AND CARNALLITE SALT ROCKS
Many salt deposits are mainly composed of halite, but substantial amounts
of the evaporites bischofite and carnallite may be present. In order to give
practical answers to questions asked regarding salt rock behaviour during
solution mining, it is important to know the rheology of the different salts at
real in situ conditions, so that the rate of inflow into the caverns as well as
surface subsidence can be predicted. Not many studies have been performed
on these materials (Van Eekelen et al. 1981, Urai 1983, 1985), hence, the
creep behaviour of bischofite, carnallite and their mixtures, is still relatively
poorly known. As was indicated above, the slow strain rates that are
relevant, in the range of 10-8
to 10-15
s-1
, may be approached in laboratory
scale experiments by applying the stress relaxation technique. In this case
study, the mechanical properties of bischofite, carnallite and their mixtures
are studied on the basis of standard constant strain rate deformation
experiments as well as stress relaxation tests. Main aim is to produce
constitutive flow laws than can be applied at real in situ conditions.
1.4 TRANSPORT PROPERTIES AND PERMEABILITY OF
ROCKSALT
Solution mined salt caverns are of great interest for fluid storage e.g.
compressed air, natural gas etc., and are expected to show excellent healing
and sealing capacity. However, it is of utmost importance to be able to
reliably quantify the permeability of salt cavern walls, so that potential loss
of the stored asset may be assessed. In contrast to recrystallized, relatively
homogeneous domal salt, bedded salt is characterized by original
sedimentary/compositional layering which forms interfaces between various
evaporitic minerals (halite, glauberite, dolomite, clays etc.). In order to
12
reliably use caverns in layered salt for gas storage, it is essential to have
insight into the sealing capacity at interfaces between such layers. Data for
the dilatancy, permeability and damage for pure salt are readily available
(Peach & Spiers 1996, Hatzor & Heyman 1997, Stormont 1997, Popp et al.
2012), but very little is known about the permeability of layered salt under
various differential stresses. In particular, little is known regarding the effect
of differential stress on the permeability through the interfaces between salt
layers (Liang et al. 2007 and 2012). These interfaces can have different
orientations with respect to the direction of maximum differential stress, so
their response to differential stress can be expected to differ between
geometries. Consequently, the permeability can change depending on
various factors, such as bonding strength at the interface, flexure due to
different strength at the two sides of the interface, the anisotropic ductility
of the various layers under shear and orientations of bedding (vertical,
horizontal and oblique) to the axial loading direction. Management of
caverns in such layered salt, is a problem for engineers in the Hubei
province of China and this prompted experiments on layered salt from that
region within the framework of this thesis
1.5 THE MAIN AIMS OF THIS STUDY
The extrapolation of laboratory data obtained from different studies to real
in situ conditions gives a broad band of uncertainty in predicted creep rate.
Rigorous experimentation is needed to ascertain the rate controlling
mechanism in salt rock. One problem in this is the difficulty to reach in the
laboratory the slow strain rates relevant for the natural systems, from
geotechnical settings such as caverns and bore holes, to the geological
development of salt pillows and diapirs. This study has the following main
aims:
13
1. To determine the microphysical mechanism controlling dislocation
creep of halite at 20-350 C, and to develop a mechanism-based
flow law providing a solid basis for extrapolation of lab data to
long time scales. Current debate regarding which dislocation
mechanism explains the rheology of halite best, under relatively
low temperature conditions, is entered by systematically
investigating the pressure dependence of creep across a range of
pressures not attempted before (50-600 MPa). Focus is on dry,
synthetic polycrystalline halite.
2. To establish if a transition can be observed from creep behaviour
governed by dislocation mechanisms to creep behaviour controlled
by a solution-precipitation mechanism, in wet polycrystalline
halite, and if so, what the conditions of this transition are in terms
of strain rate. As an important criterion, we will the stress exponent
n of a conventional power law creep equation, expected to be above
3 in case of dislocation creep, and being ~1 for solution-
precipitation creep.
3. A new and improved flow law for bischofite and carnallite to be
used at real in situ conditions by achieving the deformation rates as
low as ~10-9
s-1
by means of stress relaxation technique.
4. To test the permeability of layered rocksalt (from mines in Hubei
province China) as response to differential stress using various
geometries/orientation to deformation direction. By studying the
bulk and local compaction/dilation (at the interface) with the help
of dilatometry in conjunction with the deformation apparatus and
microstructural studies of interface of pre tested samples.
Permeability is determined using argon gas transient step
permeametry, throughout the deformation.
14
1.6 PLAN OF THESIS
The current study is based upon experimental work.
Chapter 2 reports on constant and strain-rate stepping experiments at fixed
pressures, also constant and pressure-stepping experiments at fixed strain
rates, and focuses on the pressure sensitivity of flow stress of dry salt at
temperatures 22 to 350 oC and confining pressures 50 to 600 MPa. The
results are evaluated against various microphysical models for dislocation
creep.
Chapter 3 reports on the results of multi-step experiments on wet synthetic
and natural rock salt. The experiments involved steps of constant strain rate
in the range 5×10-5
to 5×10-8
s-1
, and relaxation steps. The experiments were
carried out at a temperature of 125 oC and a confining pressure of 50 MPa
and. The strain rates achieved during stress relaxation were as low as 10-9
s-
1.
Chapter 4 is about the study of flow behaviour of wet salt rocks containing
bischofite, carnallite and their mixtures of these two salt minerals and halite.
Multi- step strain rate stepping experiments of the same type as applied to
the wet synthetic and natural halite (chapter 3) were performed and the
results obtained are discussed with respect to the role of grain size
insensitive (dislocation) mechanism and grain size sensitive mechanism
such as solution-precipitation. A first-order impression was obtained of the
effect on the strength of the salts of the wt. % of the various salts in the
mixture.
Chapter 5 is about the transport properties of layered rocksalt with special
emphasis on the response of interface permeability to differential stress
using different geometries. Permeability change and compaction/dilation of
15
bulk and interface is discussed. An important finding about the local
dilatancy vs. bulk compaction and change of permeability revealed by
microstructural analysis is special feature of this work.
Chapter 6 contains the main conclusions of this work and gives suggestions
for future directions.
16
17
Chapter 2
The transition from dislocation glide to
dislocation climb controlled creep in
dry rock salt in the temperature range
22-350 oC (0.27-0.58T/Tm): using the
pressure sensitivity of stress to
evaluate microphysical models
Muhammad, N., C.J. Spiers, C.J. Peach & J.H.P. de Bresser, 2012
Mechanical behaviour of salt VII.
18
2.1 INTRODUCTION
It has been a long standing dispute as to what dislocation mechanism
controls the creep of dry rock salt in the temperature range of 25-350 oC.
Solving this dispute is of importance in order to come to a reliable
extrapolation of laboratory derived creep data to the in situ strain rates
relevant for the long term geomechanical behaviour of salt, for example
around solution-mined caverns. Several authors suggest that dislocation
creep at these temperatures is climb controlled (Senseny et al. 1992, Carter
et al. 1993, Franssen 1994), while others conclude control by dislocation
cross-slip or glide (Auten et al. 1973, Skrotzki et al. 1981, Wawersik &
Zeuch 1986, Conrad & Yang 1999). Since the various dislocation
mechanisms are characterized by quite different constitutive equations, the
results of extrapolation of laboratory data may vary substantially depending
on the inferred mechanism, hampering reliable modelling.
Previous experimental studies on the rheology of rock salt have explored
a wide range of temperature, strain rate and confining pressure. Aladag et al.
(1970) and Auten et al. (1973) deformed polycrystalline salt at room
temperature (RT), a strain rate of 10-4
s-1
and at various pressures ranging
0.1 to 1000 MPa. They reported a decrease in the plastic flow strength with
increasing confining pressure, via cross-slip of screw dislocations. The tests
were performed on unjacketed samples, i.e. the samples were not sealed but
open to penetration by the confining medium (n-pentane), along grain
boundaries, helping the progressive opening of fractures through lowering
of effective stress.
Heard (1972) conducted a series of confined tests on salt using a constant
confining pressure of 200 MPa, temperatures ranging from RT to 400 oC
and strain rates 10-4
to 10-7
s-1
. He concluded that at low temperature and
19
relatively fast strain rate, the rate controlling mechanism is stress-assisted
glide, whereas for slower strain rates and higher temperatures, the results
can best be explained by Weertman’s dislocation climb model (Weertman,
1955) of the type 𝜀̇ ∝ 𝜎𝑛, with stress exponent n = 5.5. The samples used
were jacketed and the water content in the samples was in the range of 20 to
45 ppm.
Carter et al. (1993) reported rheological behaviour of natural aggregates
with water content of < 100 ppm (considering it dry) while experiments
were conducted in the temperatures ranging from 50 to 200 oC. The rate
controlling mechanism was reported as transitional cross-slip, at faster strain
rates and high stresses, to dislocation climb at slower strain rates and low
stresses, as far as the power law equation 𝜀̇ ∝ 𝜎𝑛 was devised. The two
regimes, dislocation cross-slip and dislocation climb were differentiated
with n-values given as 5.3 and 3.4 respectively, along with the other
parameters obtained using best fit.
Franssen (1994) presented the results of uniaxial deformation tests on
synthetic rocksalt in the temperature range of 250-780 oC, performing
constant and strain rate stepping tests in the range of 10-3
to 10-7
s-1
. The
samples were considered dry at such high temperature conditions. The flow
law reported was given by a power law of the form 𝜀̇ ∝ 𝜎𝑛, whereas two
regimes were defined. In the temperature range of 250 to 450 oC, dislocation
climb with an n-value of 5.7, and in the temperature range of 500-780 oC
lattice diffusion with an n-value of 4.4 were found as rate controlling creep
mechanisms. The tests were conducted in unconfined state, and the
mechanical data were supported by extensive microstructural analyses.
Conrad & Yang (1999) reported unconfined creep behaviour of rock salt
(unknown water content) consistent with cross slip control, up to a
20
temperature of 264 oC, and a transition to dislocation climb above 264
oC.
However, unconfined samples are highly susceptible to ductile -brittle
transitional behaviour, due to expected micro-fracture and dilatancy of
samples under stress. This casts doubt on the reliability of the conclusions of
Conrad and Yang (1999) regarding the controlling dislocation mechanism.
Later studies have shown that salt can only be considered dry if its water
content is below ~ 5ppm (Watanabe & Peach 2002, Ter Heege et al. 2005b).
For higher water content, pressure solution mechanisms are likely to play a
role during creep (also see Urai et al. 1986, Spiers et al. 1990) complicating
the description of the mechanical behaviour in terms of a flow equation.
Ter Heege et al. (2005a) performed triaxial experiments on dense
synthetic wet rocksalt in the temperature range of 75 to 240 oC at fixed
confining pressure of 50 MPa, using constant deformation strain rates in the
range of 5×10-7
to 10-4
s-1
. The authors confirmed that the salt can be
considered dry if its water content is ≤ 5 ppm, whereas for higher water
content, 9-46 ppm, the strength of salt was found to decrease to half of the
value tested otherwise at same conditions. This higher water content was
inferred to assist pressure solution creep and recrystallization by grain
boundary migration, both enhanced by fluid assisted diffusional transport in
the wet grain boundaries. The steady state flow law reported was a
conventional power law with an n-value of 5.6.
It is clear from the above that uncertainties still exist regarding the creep
mechanism controlling rock salt, notably at relatively low temperature. In
this study, considering the important points mentioned above, jacketed and
dry polycrystalline samples were tested in the temperature range of 25 to
350 oC. The rate controlling mechanism was evaluated applying non-linear
regression to the data. The best fitting parameters thus obtained revealed
21
that the mechanism is transitional, from glide (Peierls resistance controlled),
for the temperature range of 25-125 oC, to dislocation climb with an n-value
of 4.7 in the temperature range of 250 to 350 oC.
While deforming a crystalline material, several intra-crystalline defects
are usually produced, mostly in the form of edge and screw dislocations.
These dislocations glide under stress, unless affected by some obstacle on
the glide plane. At such obstacles, piling up of dislocations may cause strain
hardening of the material, except if the obstacles can be overcome by some
recovery process. The rate of motion of the dislocations is then determined
by the rate of the recovery process. One common process for overcoming
obstacles is dislocation climb, in which edge dislocations can climb out of
their glide plane due to vacancy-ion diffusion at the obstacle.
In dislocation cross-slip, a screw dislocation temporarily changes its
glide motion by moving onto a glide plane oblique to the primary glide
plane, in order to surmount the obstacle, after which it continues to glide on
the primary glide plane.
When these recovery processes are not rate controlling, for example
because they are too slow to be effective, the dislocations are constrained on
their glide plane. The rate of dislocation motion is then limited by the rate of
overcoming barriers in the glide plane, which itself is determined by the
nature of the barriers and of the dislocation core scale mechanisms by which
they are overcome (Frost and Ashby 1982, Cannon & Langdon 1983).
Glide-controlled models are usually not capable of recovering complex
dislocation substructures, hampering steady state creep. However, plastic
deformation prevails due to the movement of atomic layers overcoming
lattice friction often referred as Peierls resistance, and it depends upon size
and nature of dislocations.
22
During thermally activated dislocation processes like climb and cross
slip, there will be a small volume difference between the initial and final
state of the activated process, expressed by the activation volume (ΔV)
(Stocker & Ashby 1973, Poirier 1976, Karato 2008, De Bresser 2002). The
activation volume determines the pressure dependence of the strain rate
(Karato 2008). For dislocation climb, ΔV is related to vacancy diffusion
(Poirier 1976). The volume change associated to cross-slip is related to the
modification of stacking faults by the activity of dislocation partials, as
explained by different models (Escaig 1968a, Escaig 1968b, Skrotzki & Liu
1982). The dislocation partials need to be constricted before cross-slip, and
subsequently get dissociated again after cross slipping. The activation
volume ΔV for the constriction part is negative, whereas for the subsequent
serial step of dissociation, it is positive. Dislocation glide does not usually
involve vacancy diffusion or cross-slip as dominant mechanisms, but it can
be expected to show a minor activation volume while overcoming the
energy barriers to glide (Peierls mechanism, Karato 2008).
Confining pressure works against the activation volume and can promote
or hinder the activated processes depending upon whether it is negative or
positive. In other words, in case of a process in which the activation volume
ΔV is positive, the confining pressure will suppress the dilation and creep
will slow down, but the process with negative ΔV, the creep will speed up
by assistance of confinement.
Following the suggestion that the effect of pressure on the mechanical
behaviour of rock salt can be used to evaluate which microphysical model
best explains creep; tests have been conducted on dry polycrystalline salt
with average grain size of 300 μm under a wide range of confining
pressures. We have used jacketed dry synthetic polycrystalline salt samples
23
in axi-symmetric testing using argon gas as confining medium at pressure
values in the range from 50 to 600 MPa. At these pressures, micro-crack
dilatancy is strongly suppressed (Peach and Spiers 1996). The temperature
range studied is from room temperature (22oC) to 350
oC (i.e. T/Tm of 0.27-
0.58), i.e. outside the field of bulk diffusional flow. In general, the
polycrystalline salt was found to be stronger at higher confining pressures,
(in particular) at the higher temperatures 125-350 oC. This is in contrast to
what was found by Aladag et al. (1970) and Auten et al. (1973)
2.2 MICROPHYSICAL MODELS FOR DISLOCATION CREEP
2.2.1 Dislocation climb
Dislocation climb is a mode of recovery in which the edge dislocations of
opposite signs annihilate one another by solid state diffusion at an obstacle,
because the obstacle is too large to be overcome by natural thermal
vibrations alone. It is a process in which strain hardening may be balanced
by recovery, and steady state creep can occur. Favourable conditions for this
process to occur are slow strain rates and high temperatures (Heard 1972). A
generalized form of Weertman’s dislocation climb model (Weertman 1955)
is given as (after Frost & Ashby 1982)
𝜀̇ =𝐴𝜇
𝑇(
𝜎
𝜇)
𝑛𝑒𝑥𝑝 [−
𝛥𝑈+𝑃𝛥𝑉
𝑘𝑇] (2.1)
where A and n are constants, μ is the shear modulus, T is temperature, σ is
the differential stress, ΔU is the activation energy for self-diffusion of
slowest ion and vacancy, P is the hydrostatic pressure, k is the Boltzmann’s
constant and ΔV is the activation volume which represents the atomistic
24
scale expansion at the lattice site when an ion jumps into a vacancy and the
crystal restores its bond lengths. Notably, the activation energy ΔU for such
diffusion process is independent of applied stress. Depending upon the
diffusion mode, i.e. bulk diffusion or through the dislocation cores, the n-
value has the range from 3 to 6.5 (Poirier 1985). The PΔV term indicates
that creep will be slower or the material will be stronger at higher
hydrostatic pressures, provided that the activation volume ΔV is positive.
2.2.2 Dislocation cross-slip
De Bresser (2002) explains in detail the general flow law for cross-slip and
the details for the models controlled by constriction or dissociation. The
most important elements will be repeated below.
In dislocation cross slip, screw dislocations may surmount an obstacle by
temporarily gliding onto a plane oblique to the primary glide plane. Screw
dislocations of opposite sign cross slip towards each other and mutually
annihilate. Cross slip is a thermally activated process in which the activation
barrier is reduced by stress. Its general creep equation is given as (after
Poirier 1976)
𝜀̇ = 𝐾 (𝜎
𝜇)
2𝑒𝑥𝑝 [−
𝛥𝑈𝑐𝑠(𝜎,𝑃)
𝑘𝑇] (2.2)
where K is a constant and ΔUcs is the activation energy which is dependent
on stress and the hydrostatic pressure.
Screw dislocations often get dissociated into partials to form stacking
faults (e.g. Skrotzki & Liu 1982). These partials need to be constricted to be
able to move onto the oblique glide plane, in order to overcome obstacles.
25
Re-dissociation on the cross slip plane follows as start of further glide, and
the same process of constriction and dissociation takes place if the partials
move back to the primary glide plane. The width of a stacking fault is
dependent on the stacking fault energy γ (Hull and Bacon 1983) which
determines the force required to constrict two partials together
𝛾 = 𝛾𝑜 + [𝜀𝑜𝑏
2] 𝑃 (2.3)
where b is the Burgers vector length, γo is the stacking fault energy at
atmospheric pressure and εo is the maximum relative dilatation for a fault
with Burgers vector b/2. So the stacking fault energy will be higher for high
confining pressures, assisting the dislocation partials to constrict before
cross-slip and also resisting the dissociation after cross-slip. Depending
upon the slower of these serial processes, constriction or dissociation
controls cross-slip.
Considering constriction of partials as rate controlling mechanism, Wolf
(1960) and Skrotzki & Liu (1982) expressed the activation energy as
follows
𝛥𝑈𝑐𝑠(𝜎, 𝑃) = 𝛥𝑈𝑐𝑠 [𝑙𝑛𝜎𝑜
𝜇𝑜− 𝑙𝑛
𝜎
𝜇] (2.4)
where (σo/μo) is the normalized shear stress at 0 K, and ΔUcs is a constant
related to the stacking fault energy. This equation shows that if the applied
stress σ is zero, then the activation energy ΔUcs(σ,P) approaches infinity,
means that in the absence of applied stress, recovery is not possible. Or in
26
other words, this is a stress assisted recovery phenomenon. For FCC
materials, ΔUcs for large γ is described by Wolf (1960) as
𝛥𝑈𝑐𝑠 = [𝜇𝑏3
157(𝛾/𝜇𝑏)] (𝛽 − (𝛾/𝜇𝑏))
0.5 (2.5)
where β depends on the elastic constants (anisotropy) of the material. As
described by Seeger (1956), β = α/2π√3 with α = 2C44/(C11-C12), where Cij
are the values of elastic constants of the material.
Now considering dissociation as the rate controlling mechanism in cross
slip, Escaig (1968a) (see also Poirier 1976) approximates the activation
energy as
𝛥𝑈𝑐𝑠(𝜎, 𝑃) = 𝛥𝑈𝑐𝑠 [1 −𝛼𝑏𝜎
𝛾] (2.6)
where α is a geometrical constant of the order of 3, ΔUcs is constant and
represents the activation energy at zero applied stress, which means that
dissociation can occur without applying stress. For FCC materials, Escaig
(1968a) estimated the value for ΔUcs as
𝛥𝑈𝑐𝑠 = [𝜇𝑏3
1859 (𝛾/𝜇𝑏)] (𝑙𝑛
2√3
16𝜋(𝛾/𝜇𝑏))
0.5
(2.7)
The pressure dependence of stacking fault energy (Eq. 2.3) shows that,
its value increases with pressure. Using this value in Equation 2.5, will
make the ΔUcs term of Equation 2.4 decrease and consequently, the flow
rate given by Equation 2.2 will go up after using Equation 2.4 in Equation
27
2.2. In other words the material will be weaker for higher confining
pressure. On the other hand, for the dissociation control, using the pressure
dependent stacking fault energy in Equation 2.7 will decrease the constant
term ΔUcs of Equation 2.6. The 𝛥𝑈𝑐𝑠(𝜎, 𝑃) term depends on relative values
of α, σ and γ, where γ is relatively lower than α and σ. The increase in
𝛥𝑈𝑐𝑠(𝜎, 𝑃) will decrease the flow rate given by Equation 2.2, which means
that the material will be stronger at higher confining pressure.
2.2.3 Dislocation glide
Dislocations are forced to stay into their glide plane under the conditions
that climb and cross slip are unlikely to be of importance, usually at
relatively low temperature (Frost and Ashby 1982). While gliding, the
dislocations face resistance due to several factors which can determine the
deformation rate. These resistances are usually classified into two categories
(Verrall et al. 1977, Poirier 1985, De Bresser 1991, Karato 2008) namely;
discrete obstacles and intrinsic lattice friction (Peierls) resistance. Discrete
obstacles are, for example, point defects like foreign atom or dislocations
oriented oblique to the glide plane. These objects impede the dislocation
movement and consequently dislocations start piling up around the
obstacles. These obstacles can be surmounted by dislocations with the help
of external stress. The second type of resistance is offered by lattice friction.
For a comparatively pure material, the existing bonds need to be broken and
reformed in order to allow a dislocation to pass. The energy required to
overcome this periodic rise and fall in the energy of barrier, often referred as
Peierls potential hill, can be overcome by external stress or temperature.
This lattice friction phenomenon may involve a serial recovery process like
dislocation climb, but rate controlling mechanism is still Peierls stress being
slower (see for more details, Weertman 1957, De Bresser 1991). The
28
respective model equations after (Verrall et al. 1977, De Bresser 1991 and
Haseeb 2006) are given as:
Barrier controlled
𝜀̇ = 𝐾𝑒𝑥𝑝 [−(𝛥𝑈+𝑃𝛥𝑉)
𝑘𝑇(1 − (
𝜎
𝜎𝑜)
𝑝)
𝑞
] (2.8)
𝜀̇ = 𝐾 (𝜎
𝜇)
2𝑒𝑥𝑝 [−
(𝛥𝑈+𝑃𝛥𝑉)
𝑘𝑇(1 − (
𝜎
𝜎𝑜)
3/4)
4/3
] (2.9)
Peierls stress controlled
𝜀̇ = 𝐾 (𝜎
𝜇)
2.5𝑒𝑥𝑝 [−
(𝛥𝑈+𝑃𝛥𝑉)
𝑘𝑇(1 −
𝜎
𝜎𝑜
𝜋
2)] (2.10)
where 𝜀̇ is the strain rate of deformation, K is mechanism constant, P is the
confining pressure, ΔU and ΔV are the activation energy and activation
volume for a certain process to occur, k is the Boltzmann constant, T is
temperature, σ is applied stress, σo is the stress at 0 K and p and q are
parameters depending upon the nature of obstacle. For regularly arranged
obstacles (Eq. 2.8) and for Peierls resistance controlled glide (Eq. 2.10)
these are taken as 1, and for lattice resistance controlled glide these have got
the best values as 3/4 and 4/3 respectively (Verrall et al. 1977).The 0 K term
σo is in fact a pressure dependent term (Karato 2008), but since this value
can be expected to be high compared with the differential stress values
applied, this effect of pressure is neglected here.
29
These Equations (2.8-2.10) suggest that for positive activation volume,
the pressure will reduce the strain rate, in other words, the material will be
stronger at higher confining pressure. According to Karato (2008), the
pressure dependence of glide can be expected to be less than that of
conventional power law (Eq. 2.1).
In order to actually use the effect of pressure on flow stress to evaluate
the creep controlling mechanism, the pressure (and temperature)
dependence of the shear modulus has to be taken into account. The
following relationship is applied (based on Clark 1966, Simmons & Wang
1971, Frost & Ashby 1982):
𝜇 = 18199 + 1.342𝑃 − 9.94𝑇 (2.11)
where μ is the shear modulus in MPa, P is the confining pressure in MPa
and T is the temperature in Kelvin.
2.3 METHOD
2.3.1 Sample preparation
The salt samples were prepared artificially in the laboratory. Analytical
quality sodium chloride, powder salt was obtained from Merck salt
company, with an average grain size in the range 200-400 m. An amount
of 50 gram was weighed and placed in vacuum chamber with a ~ 1.5 gm of
water for 12 hours, in order to moisturize it to less than 1.0 wt. % water
content. The moisturized powder was sealed in rubber sleeves after
evacuation, followed by cold pressing at 100 MPa confining pressure for
one hour. The cold pressed rough cylindrical shaped salt samples were
30
machined down to the required dimension of 25 mm length and 10 mm
diameter. The cylindrical samples were subsequently sealed in FEP
(Fluorinated Ethylene Propylene) sleeves, enclosed by two specially
designed steel end pieces provided with a valve which allowed evacuation
of the sleeve. The evacuated and sealed assembly was placed inside a high
pressure vessel at 100 MPa confining pressure and 150 oC for annealing for
one week, to allow ‘Hot Isostatic Pressing’ (HIPing). After annealing, the
samples were measured for their volume and mass, to calculate density and
porosity. This procedure of sample preparation resulted in dense samples
(99.3 % of the theoretical density. The annealed samples were then dried in
a continuous, dry argon gas flow atmosphere, at 520 oC for 12 hours,
applying a very slow heating and cooling rate of 0.1 oC min
-1. The samples
obtained were ultra-dry with a water content varying between 1 to 10 ppm
(measured using FTIR spectroscopy). Each dried sample was finally
encapsulated between hard steel end pieces in soft metal jackets (Indium for
temperature ≤ 125 oC, Lead for temperature up to 250
oC and zinc for 350
oC) to avoid contamination and penetration of the confining medium into the
sample during the triaxial testing.
2.3.2 Deformation apparatus, testing and data processing
The experimental testing of the salt samples was done using an axial
symmetric, horizontally mounted Instron 1362 frame loading samples
confined within a 1.0 GPa capacity argon gas medium, volume
compensated, vessel (so called Gas Apparatus). It is equipped with two load
cells with 100 kN capacity; one external for control purposes and the other
internal to allow sample load to be measured without the uncertainty of seal
friction. Both cells had 20 N resolutions and are accurate to within ± 0.1%.
The internal load cell is pressure sensitive, and its signal was carefully
31
calibrated at various pressures, so that it could be corrected for the small gas
leakage if found. The dynamically sealed deformation piston was driven by
servo controlled electromechanical actuator of the Instron loading frame,
using software to set the deformation speed according to the desired
equivalent strain rate value for a particular experiment. The pressure is
measured using a strain gauge type pressure transducer with a resolution of
.01 MPa. The deformation/shortening of the sample was measured using an
externally mounted standard linear variable differential transformer
(LVDT), +/- 50 mm range, 0.1% linearity and 0.05 m resolution (also used
by the Instron position controller). The Gas Apparatus is equipped with an
internal three zone furnace placed around the sample and the temperature on
the sample was controlled with a three term (PID) controller using a type-S
thermocouple close to the furnace kanthal windings. Measurement of
sample temperatures and gradient was done by using three type-S
thermocouples around the sample jackets. Maximum gradient along the
sample was recorded as to be within maximum of 3 oC. The cooling of the
pressure vessel was done using continuous flow of cooling water maintained
at 21 oC. For more detailed description of the deformation apparatus, see
McDonnell et al. (1999) and De Bresser (2002).
The tests were performed in the temperature range of 22-350 oC (0.27 -
0.58 T/Tm), using argon gas as a confining medium in the pressure range of
50-600 MPa. Strain rates ranged 5×10-7
to 10-4
s-1
). Maximum natural strain
achieved during deformation was in the order of 0.25. For all individual
tests, the temperature was kept constant. The main focus of this study was to
investigate the pressure sensitivity of the strength of salt. For that, three
types of experiments were carried out:
32
a) Constant displacement rate approaching constant strain rate tests at
different but constant confining pressures (50, 300 and 600 MPa).
b) Strain rate stepping tests using (4×10-7
- 4×10-6
- 4×10-5
- 4×10-6
-
4×10-7
s-1
) at different but constant confining pressures (50, 300 and
600 MPa). In these tests, the strain rate was first instantaneously
increased from 4×10-7
to 4×10-6
to 4×10-7
and then decreased again
to 4×10-7
and 4×10-7
. The subsequent strain rate was employed by
stopping the piston and switching to the new strain rate (by adjusting
the equivalent speed of deformation piston). This was done quickly
(~ 5 s) to avoid the relaxation of the stress already given to the
sample from the last step.
c) Pressure stepping test (stepping down) at fixed strain rate. For each
confining pressure step, the sample was deformed through natural
strain of 0.02 - 0.03, subsequently the piston was pulled back and
pressure was reduced to next lower value and deformation was
started once again, after stability of the signals were achieved.
The data during the test, including confining pressure, load on the
sample, temperature and position of deforming piston were logged every 10
seconds at 16-bit precision. At the end of test, the sample was unloaded
immediately to avoid relaxation of the sample under load. The data were
processed to calculate the stress, strain and strain rate of the sample, taking
into account the change of area during shortening of sample (assuming
homogeneous deformation at constant volume). For the jacket correction,
the cylinders of the same metals as that of jackets, i.e. indium (In) for the
temperature range of 25-125 oC, lead (Pb) for the temperatures of 250
oC
and zinc (Zn) for 350 oC were deformed at similar conditions of
temperatures and strain rates. The stress data were used to calculate the
33
equivalent force on jacket according to its thickness, and subsequently the
force on jacket was removed to get the real force on the sample.
The experimental data on salt samples were corrected for the stiffness of
the machine, the pressure dependence of the load cell signal (in case of
pressure stepping), the jacket strength, the effect on the load cell output
signal in case of occasional small gas leaks, and drift in the internal load
signal if observed by correcting its base signal using slope from calibration
tests. From the reduced data and application of all the corrections, the errors
in the final calculated stress are about 2.5 % (De Bresser 2002).
2.3.3 Preparation for microstructural study
The deformed samples were carefully removed from their jackets and
selected samples were prepared for microstructural study. These samples
were cut along their length using a diamond tipped saw and using so-called
’evaporating oil’ (Shell light oil (organic), S4919) as lubricant. The sample
halves thus obtained were bonded onto glass slides and were polished, first
by SiC papers and then finalized to an optical finish of 1.0 micron using
diamond-oil suspension (Metadi, Buehler). To reveal the microstructure,
samples were treated with a chemical etchant (95% saturated NaCl solution
+ 5% de-ionized water + 8.0 gm FeCl3 per litre) followed by rinsing with n-
hexane spray and drying using hot air. The images were made using a Leica
optical polarization microscope equipped with high resolution digital image
capturing and analysis system.
34
2.4 RESULTS
2.4.1 Mechanical data
The results of all experiments performed on the dry salt samples are given in
Table 2.1. Below, the results obtained for the tests at different temperatures,
pressures and strain rates are presented in detail.
(i) Experiments at room temperature (22 oC)
Two experiments (N108 and N119) were performed to determine the
strength of the material at room temperature, at 50 and 600 MPa confining
pressures and constant strain rate (10-4
s-1
). The differential stresses thus
obtained are plotted against natural strain in Figure 2.1. The Figure shows
that the samples behaved elastically at start, at natural strains up to 0.01,
followed by plastic deformation pre-dominated by strain hardening until the
end of the test at a natural strain of about 0.17. Overall, the sample
deformed at a confining pressure of 600 MPa is slightly stronger than the
one at 50 MPa, but the difference decreases towards higher strain.
(ii) Experiments at 125-350 oC
At 125 oC, 12 experiments were performed (see Table 2.1). The pressure
sensitivity of the flow stress was tested using two sets of constant strain rate
tests (1×10-6
and 10-4
s-1
) at constant pressures (50 to 600 MPa), and one
pressure stepping test (stepping down: 600 – 450 – 300 – 150 – 50 MPa) at
constant strain rate 1×10-6
s-1
(Figure 2.2a & 2.2b). All stress-strain curves
show strain hardening, though the hardening rate appears to decrease
towards higher strain (up to 0.17), in particular at the slower strain rate. The
salt samples were consistently found stronger at higher confining pressures.
35
Figure 2.1. Stress strain graph of room temperature (22 oC) experiments, showing
very little difference in strength of salt as a function of pressure
For the deformation at 1×10-6
s-1
, the increase in the maximum flow stress
was observed to be 63% (i.e. from an average of 19 to 31 MPa) for
confining pressures of 50 and 600 MPa respectively. In the pressure
stepping test at constant strain rate 10-6
s-1
(Fig. 2.2b), the flow stress of salt,
measured at the end of each step, was found to decrease from ~ 24.2 MPa to
20.9 MPa for a stepwise pressure drop from 600 to 50 MPa. In other words,
the results of the pressure stepping tests are consistent with those of the tests
at constant pressure (Fig. 2.2a), namely a higher strength at higher confining
pressure.
At faster strain rate of (10-4
s-1
), samples were tested using 50, 300 and
600 MPa confining pressures, and the salt presented positive pressure
sensitivity, i.e. salt appeared stronger at higher
0
10
20
30
40
50
60
70
0.00 0.05 0.10 0.15 0.20 0.25
Dif
fere
nti
al S
tres
s [M
Pa]
Natural strain
Room temperature
𝜀̇ = 10-4 s-1
N108_50 MPa
N119_600 MPa
36
(a)
(b)
0
5
10
15
20
25
30
35
0.00 0.05 0.10 0.15 0.20 0.25
Dif
fere
nti
al S
tres
s [M
Pa]
Natural strain
N74_600 MPa
N81_600 MPa
N78_300MPa
N98_150 MPa
N80_50 MPa
N75_50 MPa
Temp = 125 oC
𝜀̇ = 10-6 s-1
0
5
10
15
20
25
30
35
0.00 0.05 0.10 0.15 0.20 0.25
Dif
fere
nti
al S
tres
s [M
Pa]
Natural strain
N83_Pressure stepping
Temp = 125 oC
𝜀̇ = 10-6 s-1
600 MPa 450 300 150 50
37
(c)
(d)
Figure 2.2. Differential stress vs. natural strain at temperature 125 oC, a) pressure
sensitivity at strain rate 10-6
s-1
, b) pressure stepping test at 10-6
s-1
, c) pressure
sensitivity at strain rate 10-4
s-1
, d) strain rate stepping tests (step up followed by
step down) at fixed confining pressures of 50, 300 and 600 MPa.
0
5
10
15
20
25
30
35
40
0.00 0.05 0.10 0.15 0.20 0.25
Dif
fere
nti
al S
tres
s [M
Pa]
Natural strain
N104_600 MPa
N106_300 MPa
N101_50 MPa
N102_50 MPaTemp = 125 oC
𝜀̇ = 10-4 s-1
0
5
10
15
20
25
30
35
40
0.00 0.05 0.10 0.15
Dif
fere
nti
al S
tres
s [M
Pa]
Natural strain
N109_600 MPa
N105_300 MPa
N103_50 MPaTemp = 125 oC
4×10-7
4×10-7 4×10-6
4×10-6 4×10-5
38
confining pressure (Fig. 2.2c). For the two extreme pressures of 50 and 600
MPa, the flow stress appeared to be 30.2 and 37.9 MPa respectively, i.e. salt
was found 27% stronger at 600 MPa than at 50 MPa.
At similar temperature of 125 oC, three strain rate stepping experiments
(4×10-7
- 4×10-6
- 4×10-5
- 4×10-6
- 4×10-7
s-1
) were performed at confining
pressures of 50, 300 and 600 MPa (Fig. 2.2d). The samples showed positive
strain rate sensitivity, i.e., were stronger at higher strain rates.
Within each step, the samples showed limited strain hardening and
appeared stronger at repeat strain rate steps at higher strains (compare stress
strain segments of same strain rate at low and higher natural strain, Fig.
2.2d).
At 250 oC, three samples were tested at constant confining pressures of
50, 300 and 600 MPa, at constant strain rate of 10-6
s-1
(Table 2.1). All
stress-strain curves show near steady state flow behaviour. As was found at
125 oC, the strength of the samples was higher at higher confining pressures
(Fig. 2.3a). Maximum natural strain imparted to individual sample was
~0.16. The final value of the flow stress at 50, 300 and 600 MPa
confinement was found to be 8, 12 and 15 MPa, respectively (Table 2.1), i.e.
a total of 88% increase in strength of the salt samples if pressure is increased
from 50 to 600 MPa, at strain rate of 10-6
s-1
. The strain rate sensitivity was
tested at a constant pressure of 50 MPa by performing a 5-steps (4×10-7
-
4×10-6
- 4×10-5
- 4×10-6
- 4×10-7
s-1
) strain rate stepping test. The sample
was deformed through 0.02-0.03 natural strain for each step (Fig. 2.3b).
Figure 2.3b shows that the strength of the salt was higher for faster
deformation steps. The slow deformation at 4×10-7
s-1
shows behaviour
approaching steady state, at the faster deformations steady state could not be
reached within the limited strain imposed.
39
At 350 oC, one pressure stepping experiment (600 – 300 – 50 MPa) was
performed at a constant strain rate of 4×10-5
s-1
(Fig. 2.4a). For each step,
the sample was deformed through a natural strain of 0.06-0.07. The
maximum strength of the salt at 600 MPa was found to be 10 MPa, whereas
the strength at both 300 and 50 MPa was found to be 8 MPa, i.e. in this
experiment, no significant difference in strength was resolved for 300 and
50 MPa confinements. Three strain rate stepping tests were performed at
confining pressures of 50, 300 and 600 MPa. Each test consisted of five
strain rate-steps (4×10-7
- 4×10-6
- 4×10-5
- 4×10-6
- 4×10-7
s-1
) (Fig. 2.4b).
The sample showed positive strain rate dependence, i.e. higher strength at
faster deformation steps. Whereas no distinct difference could be resolved
for the tests performed at 50 and 300
(a)
0
2
4
6
8
10
12
14
16
18
0.00 0.05 0.10 0.15 0.20
Dif
fere
nti
al S
tres
s [M
Pa]
Natural strain
N145_600 MPa
N146_300 MPa
N142_50 MPa
Temp = 250 oC
𝜀̇ = 10-6 s-1
40
(b)
Figure 2.3. Differential stress vs. natural strain at temperature 250 oC, a) strain rate
10-6
s-1
, b) strain rate stepping at confining pressure of 50 MPa.
(a)
0
2
4
6
8
10
12
14
16
18
0.00 0.05 0.10 0.15 0.20
Dif
fere
nti
al S
tres
s [M
Pa]
Natural strain
𝜀̇ _stepping
Temp = 250 oC
𝑃 = 50 MPa
4×10-7
4×10-6
4×10-7
4×10-5
4×10-6
0
2
4
6
8
10
12
14
16
18
0.00 0.05 0.10 0.15 0.20
Dif
fere
nti
al S
tres
s [M
Pa]
Natural strain
N164_P_stepping
Temp = 350 oC
𝜀̇ = 4×10-5 s-1
600 MPa 300 50
41
(b)
Figure 2.4. Differential stress vs. natural strain at temperature 350 oC, a) pressure
stepping at strain rate of 4×10-5
s-1
, b) strain rate stepping tests (stepping up
followed by down stepping).
(a)
0
2
4
6
8
10
12
14
16
18
0.00 0.05 0.10 0.15 0.20
Dif
fere
nti
al S
tres
s [M
Pa]
Natural strain
N161_50 MPa
N162_300 MPa
N163_600 MPa
Temp = 350 oC
𝜀̇ stepping
-3.8
-3.6
-3.4
-3.2
-3.0
-2.8
-2.6
-2.4
-2.2
0 100 200 300 400 500 600
Lo
g(σ
/µ)
Confining Pressure P [MPa]
22 ºC, 1E-4 s^-1 125 ºC, 1E-4 s^-1
125 ºC, 1E-6 s^-1 250 ºC, 1E-6 s^-1
350 ºC, 4E-5 s^-1 350 ºC, 4E-6 s^-1
350 ºC, 4E-7 s^-1
10-4 s-1
10-6 s-1
4×10-5 s-1
4×10-7 s-1
10-4 s-1
10-6 s-1
4×10-6 s-1
42
(b)
Figure 2.5. Pressure sensitivity of flow stress, normalized to pressure and
temperature corrected shear modulus, at natural strains of, a) 0.12, b) 0.16.
MPa confinements, perhaps it was beyond the limits of the measurements.
The flow stress presented steady state behaviour for all steps.
(iii) Pressure sensitivity of normalized flow stress at different
strain values
In Figures 2.5a and 2.5b, the normalized flow stress is plotted against the
confining pressure by taking the data at two values of natural strain, i.e. 0.12
and 0.16. The Figure forms a summary of the observations described above,
namely that strength of the synthetic dry rock samples is higher at higher
confining pressure, in particular at slow strain rate and high temperature.
-3.8
-3.6
-3.4
-3.2
-3.0
-2.8
-2.6
-2.4
-2.2
0 100 200 300 400 500 600
Log(σ
/µ)
Confining Pressure P [MPa]
22 ºC, 1E-4 s^-1 125 ºC, 1E-4 s^-1
125 ºC, 1E-6 s^-1 250 ºC, 1E-6 s^-1
350 ºC, 4E-5 s^-1 350 ºC, 4E-7 s^-1
350 ºC, 4E-6 s^-1
10-4 s-1
10-6 s-1
4×10-5 s-1
4×10-7 s-1
10-4 s-1
10-6 s-1
4×10-6 s-1
43
(iv) General inferences from the observed pressure dependence
of creep
The observed pressure dependence of the flow behaviour of the dry salt is
illustrated by the stress–strain curves of Figures 2.1-2.4, for room
temperature, 125, 250 and 350 oC, respectively. The stress-strain curves
demonstrate limited pressure sensitivity at room temperature, but a distinct
effect of pressure at all higher temperatures. The effect of pressure is also
visualized in Figures 2.5a-b and 2.6a-f plotting the maximum stress values
obtained per test, normalized to the pressure and temperature corrected
shear modulus (see Eq. 2.11), against confining pressure, strain rate and
reciprocal temperature (1/T). The sensitivity of the differential stress to
strain rate clearly changes with increasing temperature, illustrated by a
change in slope in the graph, while the effect of pressure on this sensitivity
is rather limited (compare Figs. 2.6a, b & c). Figures 2.6d-f show that the
logarithmic normalized stress is not linearly dependent on the reciprocal
temperature. This non-linear dependence is visible at all pressures tested.
These observations on the sensitivity of the differential stress, as a function
of strain rate (Figs. 2.6a-c) and temperature (Figs. 2.6d-f), suggest that a
change in rate controlling mechanism occurs going from room temperature
22 oC to 350
oC. Note also in this respect that all stress-strain curves at RT-
125 o
C show strain hardening, while near steady state behaviour was
observed at 250-350 oC.
44
(a) Log-log plots of flow stress, normalized to shear modulus, vs deformation rates
at fixed pressure (50 MPa) and at: 125, 250 and 350 oC. Higher temperature tests
are more sensitive to strain rates.
(b) Log-log plots of flow stress, normalized to shear modulus, vs deformation rates
at fixed pressure (300 MPa) and at: 125 and 350 oC. Higher temperature tests are
more sensitive to strain rates. (Only one point at 250 oC for 300 MPa confining
pressure, no trend).
-3.8
-3.6
-3.4
-3.2
-3.0
-2.8
-2.6
-2.4
-2.2
3 4 5 6 7
Log (
σ/μ
)
-Log (Strain rate
125 ºC 50 MPa
250 ºC 50 MPa
350 ºC 50 MPa
[s-1])
-3.8
-3.6
-3.4
-3.2
-3.0
-2.8
-2.6
-2.4
-2.2
3 4 5 6 7
Lo
g (
σ/μ
)
-Log (strain rate [s-1])
125 ºC 300 MPa
250 ºC 300 MPa
350 ºC 300 MPa
45
(c) Log-log plots of flow stress, normalized to shear modulus, vs deformation rates
at fixed pressure (600 MPa) and at: 125 and 350 oC. Higher temperature tests are
more sensitive to strain rates. (Only one point at 250 oC for 600 MPa confining
pressure, no trend).
(d) Log of flow stress, normalized to shear modulus, vs reciprocal temperature at
fixed confining pressure 50 MPa. The slope of data varies with temperature.
-3.8
-3.6
-3.4
-3.2
-3.0
-2.8
-2.6
-2.4
-2.2
3 4 5 6 7
Log (
σ/μ
)
-Log (strain rate [s-1])
125 ºC 600 MPa
250 ºC 600 MPa
350 ºC 600 MPa
-3.8
-3.4
-3.0
-2.6
-2.2
0.0015 0.0025 0.0035
Lo
g (
σ/μ
)
1/T [K-1]
1.00E-06
1.00E-04
4.00E-07
4.00E-06
4.00E-05
10-6 s-1
10-4 s-1
4×10-7 s-1
4×10-6 s-1
4×10-5 s-1
46
(e) Log of flow stress, normalized to shear modulus, vs reciprocal temperature at
fixed confining pressure 300 MPa.
(f) Log of flow stress, normalized to shear modulus, vs reciprocal temperature at
fixed confining pressure 600 MPa. The slope of data varies with temperature
Figure 2.6. Sensitivity of differential stress (normalized by shear modulus, μ), a, b,
c) strain rate sensitivity at confining pressures of 50, 300 and 600 MPa at various
temperatures, d, e, f) 1/T plots at 50, 300 and 600 MPa confining pressures at
various strain rates.
-3.8
-3.4
-3.0
-2.6
-2.2
0.0015 0.0025 0.0035
Log (
σ/μ
)
1/T [K-1]
1.00E-06
1.00E-04
4.00E-07
4.00E-06
4.00E-05
10-6 s-1
10-4 s-1
4×10-7 s-1
4×10-6 s-1
4×10-5 s-1
-3.8
-3.4
-3.0
-2.6
-2.2
0.0015 0.0025 0.0035
Log (
σ/μ
)
1/T [K-1]
1.00E-06
1.00E-04
4.00E-07
4.00E-06
4.00E-05
10-6 s-1
10-4 s-1
4×10-7 s-1
4×10-6 s-1
4×10-5 s-1
47
Table 2.1: Mechanical data with conditions of tests
Test T
[oC]
P
[MPa] 𝜀̇
[s-1
]
σ
[MPa] ε
H2O
[ppm]
N108 25 50 10-4
64 0.18 0.9
N119 25 600 10-4
63.6 0.18 13.7
N75 125 50 10-6
18.7 0.16 20
N80 125 50 10-6
19.6 0.16 4.3
N94 125 150 10-6
24.8 0.15 2.2
N98 125 150 10-6
22.3 0.16 3.8
N78 125 300 10-6
25.9 0.17 5.3
N99 125 450 10-6
24.4 0.16 ---
N74 125 600 10-6
32.7 0.17 2
N81 125 600 10-6
31.7 0.17 21
N83* 125 600 10-6
24.2 0.05 10
N83* 125 450 10-6
22.7 0.08 10
N83* 125 300 10-6
23.6 0.11 10
N83* 125 150 10-6
21.9 0.14 10
N83* 125 50 10-6
20.9 0.17 10
N101 125 50 10-4
30.2 0.16 9
N102 125 50 10-4
28.8 0.16 18.3
N106 125 300 10-4
33.1 0.16 8.9
N104 125 600 10-4
37.9 0.16 1.1
N103** 125 50 4×10-7
18.2 0.16 8.4
N103** 125 50 4×10-6
23.5 0.16 8.4
N103** 125 50 4×10-5
29.4 0.16 8.4
N105** 125 300 4×10-7
21 0.16 8.4
N105** 125 300 4×10-6
27.5 0.16 11.1
N105** 125 300 4×10-5
34.2 0.16 11.1
N109** 125 600 4×10-7
22 0.16 13.7
N109** 125 600 4×10-6
29.4 0.16 13.7
N109** 125 600 4×10-5
35 0.16 13.7
48
Table 2.1: contd.
Test T
[oC]
P
[MPa] 𝜀̇
[s-1
]
σ
[MPa] ε
H2O
[ppm]
N147** 250 50 4×10-7
8.3 0.16 5.1
N147** 250 50 4×10-6
10.5 0.16 5.1
N147** 250 50 4×10-5
14.8 0.16 5.1
N142 250 50 10-6
7.7 0.16 8.6
N146 250 300 10-6
12 0.16 6.4
N145 250 600 10-6
14.8 0.16 6.6
N164** 350 600 4×10-5
10.9 0.16 2
N164** 350 300 4×10-5
8.2 0.16 2
N164** 350 50 4×10-5
7.8 0.16 2
N161** 350 50 4×10-7
2.8 0.16 2
N161** 350 50 4×10-6
4.7 0.16 2
N161** 350 50 4×10-5
6.8 0.16 2
N162** 350 300 4×10-7
2.8 0.16 4
N162** 350 300 4×10-6
4.2 0.16 4
N162** 350 300 4×10-5
7.9 0.16 4
N163** 350 600 4×10-7
4 0.16 5
N163** 350 600 4×10-6
5.2 0.16 5
N163** 350 600 4×10-5
11 0.16 5
T: test temperature
P: confining pressure
𝜀̇: strain rate
σ: differential stress measured at the end of individual step or test
ε: maximum value of natural strain in a test
*P-stepping experiment
**strain-rate-stepping experiment
Note: Water content was measured at the end of test
49
Figure 2.7. Differential stress vs strain rate data taken from Table 2.1. Note
the change in n-value with pressure and temperature.
(v) n-value (stress exponent)
In order to obtain a first order impression of the sensitivity of the flow stress
on strain rate, the flow stress obtained at the end of each experiment is
plotted against strain rate in log-log space in Figure 2.7. A conventional
power law of the type 𝜀̇ = 𝜎𝑛 was used to determine the stress exponent n
for the individual isotherms at confining pressures of 50, 300 and 600 MPa.
The n-value corresponds to -1/slope of the best fit lines. As only one strain
rate was applied at room temperature, no n-value could be determined at this
condition. The data of 125 oC shows that the n-value increases with
pressure, giving 7.4 (± 2.2), 11.4 (± 1.5) and 15.3 (± 7.8) for 50, 300 and
600 MPa confining pressures respectively. For 250 oC, the data were only
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
3 4 5 6 7
Log (
flow
str
ess[
MP
a])
-Log (strain rate[s-1])
125 ºC 50 MPa 250 ºC 50 MPa 350 ºC 50 MPa
125 ºC 300 MPa 250 ºC 300 MPa 350 ºC 300 MPa
125 ºC 600 MPa 250 ºC 600 MPa 350 ºC 600 MPa
7.4
7.1
4.8
11.4
15.3
t
4.3
4.4
50
available for 50 MPa confinements, and n-value came out to be 7.1 (± 1.6).
The n-value for 350 oC came out to be in 4.8 (± 0.4), 4.3 (± 0.4) and 4.4 (±
0.8) at confining pressure of 50, 300 and 600 MPa. So overall, the n-value
has a decreasing trend with temperature, and the error in the value is higher
for higher confining pressure.
2.4.2 Microstructures
The microstructures of selected deformed salt samples are shown in Figure
2.8. The undeformed (Fig. 2.8a) microstructure shows a dense aggregate of
polygonal and cubic grains with an average size of ~300 μm. The straight
grain boundaries are often intersecting at high angle ~120o, with no
evidence of intra-crystalline deformation inherited from the sample
preparation. A few holes are visible in and at boundaries of grains, probably
representing limited porosity.
The microstructures of the samples tested at different confining pressures
and otherwise similar conditions of temperature, strain and strain rates, did
not show any difference, e.g. Figure 2.8b and 2.8c are microstructures of
samples deformed at 600 and 50 MPa respectively.
Figures 2.8c-f show samples tested at 50 MPa confining pressure at
different temperature. The microstructure of the sample deformed at room
temperature (Fig. 2.8c) shows grains full of small scale elongated sub-grains
(< 10 μm) with following somewhat wavy trajectories through the grains.
No evidence has been found of grain boundary bulging or more pervasive
recrystallization.
The microstructures of the samples at tested at 125 oC are shown in
Figure 2.8d-e, deformed at 10-6
and 10-4
s-1
respectively. The microstructures
51
show flattened grains with slightly elongated sub-grains that appear to be
aligned in wavy arrays or deformation bands. In some grains, equi-axed sub-
grains are visible. No signs were found of recrystallized grains, neither at
grain boundaries nor internally in grains. A few grains do not show much
sub-grain activity (e.g. Fig 2.8b, c & e), which can be associated to
heterogeneous etching of the sections.
The micrographs of the samples deformed at 250 and 350 oC (Figs. 2.8f-
g) show a well-defined substructure of close-to-equidimensional sub-grains,
with average size of ~50 μm, in grains with straight grain boundaries. The
sub-grains appearing near the grain boundary appear to have a somewhat
smaller size (~20 μm) than the ones on the centre of the grains,
demonstrating heterogeneous straining.
(a)
52
(b)
(c)
53
(d)
(e)
54
(f)
(g)
55
Figure 2.8. Microstructures of the dry salt samples. Deformation/shortening
direction is horizontal, all deformed microstructures have similar confinement
condition of 50 MPa and the scale bar is also similar 50 μm a) undeformed or
reference sample, showing annealed structure with well-defined boundaries
intersecting at high angle ~120o with average grain size of 300 μm, b-c) N108 (Pc =
50 MPa), N119 (Pc = 600 MPa), T = 22 oC, 𝜀̇ = 1×10
-4 s
-1, 𝜀𝑚𝑎𝑥 = 0.18, flattened
grains with elongate sub-grains often aligned in wavy arrays or deformation bands,
d) N75, T = 125 oC, 𝜀̇ = 1×10
-6 s
-1, 𝜀𝑚𝑎𝑥 = 0.16, flattened grains, with a few equi-
axed sub-grains, no signs of recrystallized grains, e) N101, T = 125 oC, 𝜀̇ = 1×10
-4 s
-
1, 𝜀𝑚𝑎𝑥 = 0.16, grains with different orientation as revealed by light reflection, no
sign of grain bulging or recrystallization at grain boundaries or inside the grains, f)
N142, T = 250 oC, 𝜀̇ = 1×10
-6 s
-1, 𝜀𝑚𝑎𝑥 = 0.16, polygonised with clear grain
boundaries, each bigger grain has developed sub-grain of an average size of 50 μm,
no wavy boundaries as seen in lower temperature deformed samples, the
development of a new grain formation is also obvious at the middle of
microstructure, g) N161, T = 350 oC, 𝜀̇ = 4×10
-5 s
-1, 𝜀𝑚𝑎𝑥 = 0.16, comparatively less
dense in sub-grain formation than 250 oC and inferred (Franssen 1994) to show
dislocation climb features, no signs of wavy lines.
2.5 DISCUSSION
The change in characteristic microstructures when going from room
temperature to 350 oC, as described in section 2.4.2, supports the
interpretation that the rate controlling mechanism changes. The ill-defined
sub-grain structure observed at the low temperatures appears to rule out
climb-controlled creep at those conditions, while that mechanism may
become more prominent at higher temperature. The wavy features seen at
temperature 22-125 oC may correspond to structures seen in other deformed
salts (e.g. Skrotzki et al. 1981, Skrotzki & Welch 1983, Franssen 1994)
where they have been interpreted as indicating cross slip of screw and jogs
of edge dislocations. Whereas the well-defined boundaries of polygonal
grains and the newly formed strain free sub-grains at higher temperature
(250 to 350 oC), are representative of cross slip and dislocation climb
process (Skrotzki et al. 1981, Franssen 1994).
56
In the next part of this Discussion, different models for dislocation creep
will be tested against the data in order to investigate which are the
mechanisms that most likely explain the behaviour seen at the lower and
higher temperature ranges tested. In particular, the observed effect of
pressure will be used to discriminate between different models, focussing on
dislocation glide, cross-slip and climb. The experimental data will be fit to
the model equations applying non-linear regression best fitting. The
resulting values for the various parameters of the models will be put against
constraints from the literature.
2.5.1 Results of non-linear regression analysis
(i) Climb controlled creep
Non-linear regression of the data was carried out applying Equation (2.1),
testing three temperature ranges, 125 to 350 oC, 125 to 250
oC and 250 to
350 oC. The best fitting parameters are listed in Table 2.2. The best fit for
125-350 oC gives poor result in terms of quality of fit. This was as expected,
given the gradual change in stress exponents n with temperature (Fig. 2.5)
and confirms that a single climb controlled creep mechanism cannot explain
the behaviour observed in the dry rock salt. For temperatures 125 to 250 oC,
the quality of fit is better, but the n-value obtained (8.9 ± 0.6) is too high to
be supportive of climb controlled creep, in which values of 3-4.5 are to be
expected in case of lattice diffusion control of climb, or 6.5 if dislocation
core (pipe) diffusion plays a role (Frost & Ashby 1982, Poirier 1985, De
Bresser 1991). The most reasonable result was found for the temperature
range of 250 to 350 oC, both in terms of quality of fit and considering the
fitting values for the model parameters. The stress exponent of power law
came out to be 4.7 ± 0.3, which is in agreement with the theory of climb
57
controlled creep (Weertman 1968, Sherby & Weertman 1979, Poirier 1985)
and also corresponds well to values obtained previously in other studies
(Franssen 1994, dry salt in the T-range of 250-780 oC, Ter Heege et al.
2005a on wet rock salt in the T-range of 75 to 250 oC). The value obtained
for the activation energy, 126 ± 7 kJmol-1
is also reasonably close to values
reported in literature for short circuit diffusion through dislocation cores,
pipe diffusion (Wawersik & Zeuch 1986, Franssen 1994). The n-value
obtained is slightly higher than the value fit with standard dislocation climb
flow law (i.e. 4.5). The activation energy for halite is normally associated
with diffusion of the slowest moving ionic species (Cl-), and the value
reported is 146 kJmol-1
for dislocation core diffusion (Barr et al. 1960), or in
the range of 103 to 155 kJmol-1
(Wawersik and Zeuch 1986, Franssen
1994). This energy also looks in good agreement with the predicted value
that, activation energy for core diffusion lies in the range of 0.5 to 0.7 of
activation energy of lattice diffusion of the slowest moving Cl- ion which is
Table 2.2. Power law, dislocation climb model best fit parameters
Temp.
[oC]
n LOGA
ΔV
(×10-29
)
[m3]
ΔV/Vm
ΔU
(×10-19
)
[J]
ΔU
[kJmol-1
]
R2
125-
350
5.4
(±0.5)
10.9
(±2.3)
1.3
(±0.4)
0.30
(±0.09)
1.1
(±0.1)
66.9
(±7.0) 0.55
ONLY
125
and
250
8.9
(±0.6)
22.2
(±2.7)
2.4
(±0.4)
0.54
(±0.08)
1.3
(±0.1)
76.7
(±7.7) 0.77
ONLY
250
and
350
4.7
(±0.3)
14.4
(±1.4)
2.9
(±0.4)
0.66
(±0.08)
2.1
(±0.1)
125.9
(±7.2) 0.90
n:power law stress exponent
LOGA: constant
ΔV: activation volume
ΔU: activation energy
R2: correlation coefficient
58
Table 2.3. Cross-slip dissociation control model best fit parameters
Temp.
[oC]
LOGK γo
[Jm-2
] εo α R
2
22-125- 11.4
(± 1.2 )
0.026
(± 0.002)
0.006
(±0.004)
0.58
(± 0.06) 0.80
125-350 4.8
(± 0.3)
0.040
(± 0.002)
-0.017
(± 0.010)
1.74
(± 0.09) 0.77
ONLY 125
and 250
4.2
(± 0.5)
0.050
(± 0.005)
-0.044
(± 0.020)
1.95
(± 0.20) 0.77
ONLY 125
and 350
4.8
(± 0.3)
0.040
(±0.0.003)
-0.018
(± 0.020)
1.73
(± 0.09) 0.79
LOGK: constant
γo: stacking fault energy at atmospheric pressure
εo: maximum relative dilatation for a fault with Burgers vector b/2
α: geometrical constant
R2: correlation coefficient
Table 2.4.Dislocation glide model best fit parameters after non-linear regression
analyses
Glide
mode
Temp.
[oC]
LOG
K
ΔV
×10-29
[m3]
σo
[MPa]
ΔV/Vm
ΔU
×10-19
[J]
ΔU
[kJmol-1
]
R2
Lat
tice
resi
stan
ce
con
tro
lled
22
-12
5
14.1
(±1.5)
3.0
(±0.5) 219
(±11.4) 0.66
(±0.11)
1.5
(±0.1)
138.2
(±14.1) 0.83
Ob
stac
le
con
tro
lled
22
-12
5
13.1
(±1.8)
2.8
(±0.5)
139.0
(±4.4)
0.63
(±0.10)
2.8
(±0.3)
168.6
(±15.5) 0.81
Pei
erls
resi
stan
ce
con
tro
lled
22
-12
5
17.1
(±1.7)
2.6
(±0.4)
238.1
(±8.3)
0.58
(±0.10)
2.3
(±0.2)
138.5
(±15.1) 0.82
LOGK: constant
ΔU: activation energy
ΔV: activation volume
σo: stress at 0 K
R2: correlation coefficient
59
205-221 kJmol-1
(Franssen 1994, Ter Heege et al. 2005a). The activation
volume ΔV parameter for halite is not well constrained in literature, but as
theory of ion-vacancy diffusion suggests that annihilation of vacancy and
ion can cause atomic scale expansion to restore the lattice, so the value
0.66Vm looks reasonable for the associated dislocation climb recovery under
discussion. This obtained activation volume value (0.66Vm) lies in close
approximation to the value reported in literature for metals (e.g. Werner &
Mehrer 1983, Mehrer 2011), where authors have reported that for fcc
metals, the activation volume may lie in the range of 0.6Ω to 1Ω (Ω is the
atomic volume of the metal).
On the basis now of the mechanical behaviour observed (near steady
state stress-strain and stress-strain rate trends), the microstructural
observations, and the best fitting analysis, it is concluded that dislocation
climb is the rate controlling mechanism in the temperature range of 250-350
oC. The flow law describing the behaviour is:
𝜀̇ = 2.51 × 1014 (𝜇
𝑇) (
𝜎
𝜇)
4.7𝑒𝑥𝑝 [−
2.1×10−19+𝑃(0.66𝑉𝑚)
𝑘𝑇] (2.12)
Please note that in Equation 2.12, pressure (P), differential stress (σ) and
shear modulus (μ), are in MPa.
(ii) Cross-Slip controlled creep
Cross-slip is a serial process of constriction and dissociation of partials
allowing dislocations to glide on planes oblique to the initial glide plane, in
order to overcome obstacle hindering glide (see section 2.2.2). Confining
pressure promotes constriction and hinders dissociation of partial
dislocations, causing faster or slower creep of the material, respectively,
60
depending on which of these two, constriction of dissociation, is the slowest
step. In the current study, the dry polycrystalline salt showed higher strength
at higher confining pressures, so the constriction part does not appear to be
rate controlling (section 2.2.2). Non-linear regression was done by using the
data with the dissociation controlled cross slip model testing four
temperature ranges, 22 to 125 oC, 125 to 350
oC, 125 to 250
oC and 250 to
350 oC. The best fit parameters are listed in Table 2.3. The values for the
stacking fault energy γo lie in the range 0.03 - 0.05 Jm-2
, which is
substantially lower than the values reported for rock salt in literature; 0.195
- 0.288 Jm-2
(Fontaine 1968, Fontaine & Haasen 1969, Mohammed &
Langdon 1974, Tasker & Bullough 1981, Skrotzki & Liu 1982). The best fit
values of the dilatation constant εo are rather small and negative, ranging -
0.044 to -0.017, whereas in previous studies the reported value of this
dilatational constant was ~0.3 (Fontaine & Haasen 1969). Given these
results, it appears unlikely that dissociation controlled cross-slip is the rate
controlling mechanism in the lower temperature range of 25-125 oC.
In conclusion, thus, neither constriction controlled cross slip, nor
dissociation controlled cross slip is likely to describe the creep behaviour of
dry rock salt at 25-350 oC.
(iii) Glide controlled creep
The best fitting results using non-linear regression for glide controlled creep
are given in Table 2.4 for the three models of glide; discrete obstacle
control, intrinsic lattice resistance including kink nucleation process, and
lattice resistance including dislocation climb (see section 2.2.3). Analysis is
confined to only one temperature range, RT-125 o
C, given that climb
controlled creep appears well established for the higher temperature range
61
250-350 oC, and cross slip is ruled out as likely controlling mechanism
overall. The quality of fit is rather good for all three models.
The fitting parameters for the obstacle controlled glide model include an
activation energy of 169 kJmol-1
and a value of 139 MPa for σo , the stress at
0 K. Verrall et al. (1977) and Frost & Ashby (1982) quote 289 kJmol-1
for
the activation energy and only 38 MPa for σo. These values deviate from the
fitting values, casting doubt on the applicability of the obstacle controlled
glide model.
The fitting parameters for the two lattice resistance models are very
much alike, with almost identical activation energies of ~138 kJmol-1
and
values of 219-238 MPa for σo. This agreement is of course related to the fact
that the model equations (section 2.2.3) are very much alike. Verrall et al.
(1977) and Frost & Ashby (1982) quote 69 kJmol-1
for the activation energy
and 272 MPa for σo. for the double kink model. No constraints are known to
the author regarding the climb-related glide model. However, the best fitting
value for the activation volume, ~0.6-0.7 of the molecular volume of halite
(Table 2.4), is of the same order of magnitude as that for climb controlled
creep described by Equation (2.1). For that reason, the glide model
involving climb is preferred above the lattice resistance kink nucleation
model. The flow law describing the behaviour of dry rocks salt in the
temperature range RT-125 oC accordingly is:
𝜀̇ = 1.26 × 1017 (𝜎
𝜇)
2.5𝑒𝑥𝑝 [−
(2.310−19+𝑃(0.58𝑉𝑚))
𝑘𝑇(1 −
𝜎
238
𝜋
2)] (2.13)
Please note that in Equation 2.12, pressure (P), differential stress (σ) and
shear modulus (μ), are in MPa.
62
The activation energy parameter fit value in the current data lies in the
range of 138-169 kJmol-1
. Verrall et al. 1977 and Frost & Ashby 1982
proposed the activation energy in the range of 69 to 289 kJmol-1
, where the
lower 69 kJmol-1
value is proposed for Peierls/lattice resistance controlled
glide and 289 kJmol-1
is for obstacle controlled dislocation glide
mechanism. The 0 K stress values obtained by the best fit of the current data
found in the range of 139 to 238 MPa. Whereas the values given by Verrall
et al. are 38 MPa for obstacle and 272 MPa for of Peierls/lattice resistance
controlled glide. So, the closest values to the published work are for Peierls
resistance with activation energy slightly higher and 0 K flow stress slightly
lower. So we propose Peierls resistance glide model with the best fitted
parameter as in this study to consider as the rate controlling mechanism for
the temperature range of 22-125 oC.
The complete data set of flow stress and strain rate is plotted in log-log
space in Figure 2.9. The obtained stress exponents n, i.e. the reciprocal
values of the slopes of the isotherms, change from rather high values at 125
oC (>7), to ~4-5 at 350
oC (Fig. 2.7). In fact, the distribution of data points at
125 oC fits better to a slightly curved isotherm, flattening off at the faster
strain rates (i.e. an exponential relation), than to a linear trend (i.e. at
constant n).
The number of data points at 250 oC is too limited to conclude with
confidence whether these data correlate better with the non-linear stress-
strain rate relation of 125 oC or with the linear trends at 250
oC. However,
the near steady state behaviour observed at 250 oC (Fig. 2.3a) matches the
350 oC data rather than the 125
oC and RT data (compare Fig. 2.3a with Fig
2.1, 2.2 and 2.4).
63
Figure 2.9. Complete data with best fit lines in log space. Shows the changing trend/slope for 22- 125 oC and constant trend at 350
oC
in log-space. The data of previous study by Franssen 1994 on dry salt is also plotted along, which shows slightly higher strength at
350 oC as compared with current study, Colour lines are the glide model predictions at room temperature and 125
oC, at 50, 300 and
600 MPa, pointing higher strength at higher confining pressures. The green (double arrow line at 1.2 (=LOG (16 [MPa])) presents the
transition between glide and climb control creep.
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
-7.0-6.0-5.0-4.0-3.0
Log (
stre
ss [
MP
a])
Log (strain rate [s-1] )
22 ºC 50 MPa
22 ºC 600 MPa
125 ºC 50 MPa
125 ºC 300 MPa
125 ºC 600 MPa
250 ºC 50 MPa
250 ºC 300 MPa
250 ºC 600 MPa
350 50 MPa
350 ºC 300 MPa
350 ºC 600 MPa
Franssen 1994 350 ºC
Unconfined
Peierls
resistance
glide
models at
50, 300 and
600 MPa.
Dislocation
s climb
models at
50, 300 and
600 MPa.
64
On the basis of the above, it is inferred that within the range of
conditions tested, the dislocation mechanism controlling creep of dry rock
salt changes when going from low to high temperature. The transition is
likely to occur in between 125 and 250 oC.
2.5.2 Comparison with previous studies involving tests at various
pressures
Limited previous studies have been performed exploring effects of pressure
on the creep of halite. Aladag et al. (1970) and Auten et al. (1973) reported
that salt (single and polycrystalline) got weaker at higher confining pressure
while tested at room temperature. This is inconsistent with the results of the
current study. In the studies of Aladag et al. (1970) and Auten et al. (1973),
the samples were not jacketed to protect them from infiltration of the
confining fluid. In addition, the sample preparation was done using abrasive
papers followed by polishing with water and rinsing with methanol and
ether. So the water content, as well as, the impurity content of the samples
was not known. In current work, the salt samples were prepared from
reagent grade pure salt powder and special care was taken to dry and jacket
the samples. The testing of the samples, deformed at similar conditions to
(Aladag et al. 1970, Auten et al. 1973), i.e. room temperature and using
strain rate of 10-4
s-1
, did not show decrease in strength, but increase with
confining pressure. It is thought that, given the special care in sample
preparation and testing procedure, these current results are more reliable and
should form the basis for extrapolation to in situ conditions.
2.5.3 Deformation mechanism map
Figure 2.9 can be regarded as a deformation mechanism map show for dry
rock salt, in logarithmic stress-strain rate space, with the predictions of the
65
lattice resistance glide model (Eq. 2.13) and of the climb controlled model
(Eq. 2.12). These predications can of course be extrapolated to strain rates
10-11
– 10-15
s-1
, representing realistic strain rates in salt mines (maximum)
and salt domes (minimum) according to Carter et al. (1982). The question
now is where exactly the boundary between the two creep mechanisms is
located. In the experiments presented in the current Chapter, the transition
from glide control to climb control occurs in between 125 and 250 oC. The
character of the creep Equations, exponential vs. power law for glide and
climb control, respectively, is illustrated in Figure 2.9 by the straight and
curved lines in log-log space of stress and strain rate. The location of the
boundary is very sensitive to the values for the fitting parameters (Equations
2.12 and 2.13) and in fact, the current data set is not sufficient to define this
boundary with full confidence. We thus, for now, follow the approach by
Frost & Ashby (1982) in which the boundary between glide and other
mechanisms is at a stress value independent of strain rate or temperature.
For the transition from glide to climb for the dry polycrystalline halite
investigated here, the corresponding stress value is 16 MPa. For common in
situ conditions, for both geotechnical applications like at caverns and mines,
and at natural salt halokinetic settings such as at salt domes, the climb-
controlled creep Equation (Eq. 2.12) thus appears a meaningful description
of flow behaviour of (dry) rock salt.
It is noted that from the analysis given above, it was concluded that the
rate controlling mechanism for creep of dry rock salt is dislocation glide or
climb, depending on extrinsic conditions, rather than cross-slip. However, it
cannot be ruled out that cross slip does play some role, as microstructures
suggested that there exists waviness at sub-grain level, pointing towards the
cross-slip. However, the mechanism is not rate controlling at tested
66
conditions and accordingly, no cross slip regime is included in the
deformation mechanism map of Figure 2.9.
CONCLUSIONS
In this chapter, the results are presented of new experiments on jacketed
samples of dry synthetic rock salt. Aim was to determine the microphysical
mechanism controlling dislocation creep of halite at 20-350 C, and to
present a mechanism-based flow law providing a solid basis for
extrapolation of lab data to long time scales. One way of distinguishing
between the various dislocation mechanisms that may control creep in dry
rock salt is by the effect of confining pressure. For that reason, systematic
pressure stepping tests were carried out across a range of pressures not
attempted before (50-600 MPa). The following was concluded:
In the temperature range of 22 to 350 oC, the rate controlling
dislocation mechanism changes from Peierls resistance controlled
glide at room temperature to climb controlled creep governed by
dislocation diffusion at 350 oC. At all conditions tested, the dry
rock salt is stronger if the confining pressure is higher. The
activation volume for creep was found to be about 0.58Vm for the
Peierls resistance controlled glide, and about 0.66Vm for climb
controlled creep, where Vm is the molecular volume of halite.
The data allowed to obtain values for the fitting parameters for the
Peierls resistance controlled glide model and for the climb
controlled creep model, compiled in two creep laws (equations 2.12
and 2.13 above).
The temperature at which the transition from glide to climb control
takes place was found to lie in between 125 and 250 oC, at a stress
67
of about 16 MPa. Given the slow strain rates and low stresses
normally relevant for in situ conditions, dislocation creep of
rocksalt in nature will likely be controlled by dislocation climb.
68
69
Chapter 3
Stress relaxation of synthetic and
natural polycrystalline halite
70
3.1 INTRODUCTION
The mechanical behaviour of halite has been studied extensively before in
the laboratory; both for dry and wet conditions, at different temperatures and
pressures (see e.g. Heard 1972, Heard & Ryerson 1986, Wawersik & Zeuch
1986, Spiers et al. 1990, Senseny et al. 1992, Peach et al. 2001, Ter Heege
et al. 2005a). For dry conditions, Carter & Hansen (1982) suggested
dislocation climb process as rate controlling mechanism, Franssen (1994)
concluded that at laboratory strain rates and relatively low temperatures,
250-450 oC, the behaviour is best explained by climb-controlled dislocation
creep with climb itself being governed by diffusion through dislocation
cores, while at higher temperatures, 500-780 oC, lattice diffusion controlled
creep is dominant. In chapter 2 of this thesis, pressure stepping tests have
been used to investigate the creep behaviour of dry halite in the temperature
range 22-350 oC, where a transition was observed from glide to dislocation
climb. For wet conditions, several authors have suggested that under
laboratory conditions of temperature 23 to 500 oC, dislocation creep is the
main mechanism controlling creep (Heard 1972, Heard & Ryerson 1986,
Carter et al. 1993), while other authors concluded that if grain size is small
and temperature is low, solution-precipitation creep plays an important role
(Urai et al. 1986, Spiers et al. 1990). Ter Heege et al. 2005b) argued on the
basis of an experimental study that in wet polycrystalline halite, both
dislocation and solution-precipitation mechanisms might contribute to the
overall deformation, provided that microstructural modification processes
such as dynamic recrystallization and grain growth can take place freely.
Salt rocks in nature are wet rather than dry (Roeder & Bassett 1981, Urai
1983). The creep of these salt rocks usually occurs at strain rates in the
range of 10-8
to 10-15
s-1
(Heard 1972, Van Eekelen et al. 1981, Jackson &
71
Talbot 1986). In order to fully understand the creep behaviour of halite
under natural conditions, it thus is of importance to not only know, in
general, which mechanisms may control creep of halite, i.e. dislocation vs.
solution-precipitation mechanisms, but also to be able to constrain the strain
rates under which a given mechanism prevails. The strain rates relevant for
in situ deformation, 10-8
to 10-15
s-1
, are difficult to achieve in laboratory
experiments. However, one way of approaching such slow strain rates is to
perform stress-relaxation experiments (Rutter & Mainprice, 1978). In this
technique, strain rates as slow as 10-9
s-1
can be achieved by allowing the
stress on a sample to relax through plastic deformation. The experiments are
time consuming, but result in valuable data regarding creep behaviour at low
stresses and slow strain rate.
In the current study we have used both synthetic and natural wet
polycrystalline halite samples to investigate what controls the rate of
deformation at real in situ conditions, by applying the stress relaxation
technique. The main research question to be answered was if a transition can
be observed, from creep behaviour governed by dislocation mechanisms to
creep behaviour controlled by a solution-precipitation mechanism, and if so,
what the conditions of this transition are in terms of strain rate. As an
important criterion, we used the stress exponent n of a conventional power
law creep equation, expected to be above 3 in case of dislocation creep, and
being ~1 for solution-precipitation creep.
The experiments were performed as multi-step strain rate stepping while
keeping the temperature and confining pressure as constant, at 125 oC and
50 MPa, respectively. Indeed, for higher stresses and strain rates, we found a
high stress exponent n of the standard power law, in the range 10 to 13,
while towards lower stress and slower strain rate, the n-value showed a
72
decreasing trend and reduced to ~1 at the end of the relaxation step. This
transition took place over the strain rate interval 10-8
to 10-9
s-1
, at stresses 10
to 3 MPa, at 125 oC.
3.2 METHOD
3.2.1 Sample preparation
In this study, synthetic as well as natural salt samples were used. All
samples were manufactured such that they became cylindrical samples with
length in the range of 80-85 mm and diameter 35-36 mm.
The synthetic rock salt sample (halite1) was prepared in the laboratory
starting from analytical grade NaCl powder from Merck, with an average
particle size of 200-400 μm. The powder salt was cold pressed in a hardened
steel, piston-cylinder assembly. The pistons used on both sides of the
cylinder were perforated at the centre to provide evacuation of air molecules
from the powder prior to cold pressing, using a vacuum pump. The cylinder
was polished and lubricated with Teflon spray before pouring the salt
powder in. The powder was axially pressed at 200 MPa for 20 minutes The
resulting cylindrical sample was sealed in a Viton rubber sleeve and was put
in a silicone oil pressure vessel for annealing under 100 MPa confining
pressure and a temperature of 150 oC for one week. The sample thus
obtained had a theoretical density (mass to volume ratio) of 99.5%. For
more details about the sample preparation technique, see Peach (1991).
Two natural halite samples (halite2 and 3) were prepared from
‘Speisesalz’ cores. These cores come from the Asse mine, Germany, and
were taken from the ~800m gallery level and from a depth of >3m inside the
horizontal gallery wall (Spiers et al. 1986, Urai et al. 1987, Peach 1991).
73
The samples had a grain size in the range of 3-10 mm. The constituents of
the cores were mainly halite (> 98%), a small amount of poly-halite
(K2SO4.MgSO4.2CaSO4.2H2O; ~1%), and some minor quantity of
anhydrite. The microstructure shows high angle ~120o boundaries, with a
few cracks and very little internal dislocation structure and corresponding
low dislocation densities.
In order to create a deliquescence condition in our samples, comparable
to that at in-situ, these were moisturized with water < 0.5 wt. % using
atomizer in a chamber and carefully measured for its mass increase. This
was followed by wrapping up the samples in a double layer of perforated
glass fibre sheet (0.3 mm starting thickness) creating an equilibrated humid
environment around the sample. These wrapped samples were further sealed
in 1.0 mm thick polymer “ethylene propylene diene monomer (EPDM)”
jackets to avoid contamination of the samples by the confining medium
(silicone oil) used in the deformation apparatus. The deformation pistons
were inserted in the open ends of the jacket. To seal the ends, stainless steel
wires were tightly wound in grooves of steel pistons from over the jacket.
3.2.2 Deformation apparatus
The apparatus used for this study was the so-called “Shuttle Vessel” (Fig.
3.1) of the experimental rock deformation (HPT) laboratory at the
department of Earth Sciences at Utrecht University. The Shuttle Vessel
machine is an internally heated 100 MPa confining pressure vessel mounted
on a standard 100kN Instron 1362 loading frame with an electro-mechanical
servo controlled positioning system. This machine can be used to deform the
sample at a constant piston speed, approaching constant strain rate if total
strains are kept relatively low. The machine is provided with a (Instron
standard, +/- 50 mm) linear variable differential transformer (LVDT), but to
74
Figure 3.1. Shuttle vessel triaxial deformation apparatus
come to an accurate measurement of the sample deformation, another
(LVDT_2, 0- 25 mm range, H.F. Jensen, Denmark) was installed at the top
of the vessel and near the sample, to reduce the effect of the elastic
distortion of the apparatus and measure accurately the shortening of the
sample, especially during stress relaxation, where very limited natural strain
in the order of 0.001 is to be monitored. The temperature was measured by
thermocouples at two locations inside the vessel; one was positioned close
to the middle of sample and, the other was at the top of the sample. K-type
thermocouples were used, which are accurate within ± 1 oC. The axial load
on the sample was measured using a 100 kN load cell accurate within 0.1 %
of the full scale. The confining pressure was created by using silicone oil
which is kept at constant pressure within ± 0.1 MPa using a servo pump.
Measurement of the pressure was done using a diaphragm pressure
LVDT_2
Deformation piston
Thermocouples
connections
Pressure vessel
containing sample
assembly
75
Figure 3.2. Sample with piston assembly and steel discs
transducer (Teledyne 2403, 100 MPa range). To avoid rusting of the
deformation pistons and related contamination of the samples, two grooved
stainless steel discs having 2.0 mm thickness and 35.0 mm diameter were
used as separators, along with 50 μm thin teflon sheets to reduce the friction
between sample and the deforming pistons (Fig. 3.2). The triaxial apparatus
used is very sensitive to the environmental conditions. For example, a little
variation in laboratory temperature not only causes a drift in the load cell
signal, but also may cause a small change in the pressure, which would
influence the load cell signal. Such (small) change in pressure will result in
some elastic expansion/contraction of the pressure vessel, hampering
straightforward processing of the data from the LVDT_2. Therefore the
apparatus was carefully calibrated for the effect of pressure and temperature
on the load cell. The elastic distortion of machine was corrected on the basis
of tests using a steel dummy of known Young’s modulus. This was done by
applying gradually increasing axial load to the dummy and measuring the
LVDT_2 reading in parallel, then using a 10 order polynomial to relate the
LVDT_2 value to the internal load.
35 mm
76
3.2.3 Experiments
In this study, total three multistep experiments were performed; one on
synthetic (halite1) and two on natural (halite2 and 3). A typical experiment
consisted of a few steps at constant strain rate, in the range 5×10-5
to 5×10-8
s-1
. During the constant strain rate part of the test, the sample was deformed
until a steady (or near steady) state of stress was reached. Then the piston
was arrested and the stress on the sample was allowed to relax until the
diminishing force on the sample reached the limits of the load cell
resolution. The duration of each relaxation step was a few days.
The synthetic sample, halite1, was deformed by a 3-step repetitive strain
rate of 5×10-7
s-1
, for a natural strain of 0.01-0.03 per step and each step was
followed by stress relaxation. The natural samples, halite2 and 3, were
tested in a seven strain rate steps (5×10-5
, 5×10-6
, 5×10-7
, 5×10-8
, 5×10-7
,
5×10-6
, 5×10-5
s-1
), and the relaxation was followed after the constant strain
rate parts of 5×10-6
, 5×10-8
and 5×10-6
s-1
. The experiments were performed
at 125 oC sample temperature and 50 MPa confining pressure.
3.2.4 Data acquisition and processing
The data, containing pressure, temperature, load and position (LVDT_1 and
2) were logged throughout the test. The stress on the sample was calculated
from the load values by assuming constant volume deformation, correcting
the instantaneous area value for progressive change in length of the sample.
The cumulative strain (shortening) of the sample was calculated as the
natural strain, i.e. the natural logarithm of the ratio of the final length and
the initial length. The zero of the internal load signal was tested before the
deformation piston touched the sample and after unloading at the end of
experiment. If found necessary, correction was applied for zero-level shift.
77
The data acquired during stress relaxation was analysed with a dedicated
code to produce the plastic strain rate of the sample.
The stress relaxation technique is based on the following. During
deformation at constant rate (imposed by a moving piston), some energy
gets stored inside the material as elastically stored energy. During
relaxation, when the loading piston is arrested, this energy is dissipated
through plastic deformation of the sample. Ideally, this requires an infinitely
stiff machine which does not participate by means of the elastically stored
energy in its own frame. In practice, such a machine does not exist without
exceptional servo control to keep constant sample length, so the elastic
modulus of machine should be taken into account. In other words, the
elastically stored energy inside the active part (under force) of the machine
also dissipates through the plastic strain of the sample, so the data also need
to be corrected for the stiffness of machine. In order to calculate the sample
straining, we take the simple assumption that the stress is proportional to
strain, provided that other physical conditions (confining pressure,
temperature, microstructure) are constant (Rutter & Mainprice, 1978). The
strain rate at any instant will then be proportional to the stress relaxation
rate, with the elastic (Young’s) modulus of sample as a constant of
proportionality.
The measured total displacement” xtotal” (from LVDT_2) can be written
as the sum of the sample deformation (elastic + plastic) and elastic distortion
of apparatus
𝑥𝑡𝑜𝑡𝑎𝑙 = 𝑥𝑠𝑎𝑚𝑝𝑙𝑒 + 𝑥𝑎𝑝𝑝𝑎𝑟𝑎𝑡𝑢𝑠 (3.1)
78
After differentiating w.r.t. time and normalizing this equation using the
instantaneous length of the sample, the following relation of plastic strain
rate of the sample is obtained:
𝜀�̇�𝑙𝑎𝑠𝑡𝑖𝑐 = 𝜀�̇�𝑜𝑡𝑎𝑙 −1
𝐸𝑠𝑎𝑚𝑝𝑙𝑒(�̇�𝑠𝑎𝑚𝑝𝑙𝑒) −
1
𝐿𝑡𝑆 (
𝜕𝐹
𝜕𝑡) (3.2)
where,
𝜀�̇�𝑙𝑎𝑠𝑡𝑖𝑐 = plastic strain rate of sample [s-1
]
𝜀�̇�𝑜𝑡𝑎𝑙 = total strain rate measured by LVDT_2 [s-1
]
𝐸𝑠𝑎𝑚𝑝𝑙𝑒 = Young’s modulus of sample [MPa]
�̇�𝑠𝑎𝑚𝑝𝑙𝑒 = sample stress relaxation rate [MPas-1
]
𝐿𝑡 = instantaneous length of sample [m]
𝑆 = machine stiffness correction constant [mN-1
]
(𝜕𝐹
𝜕𝑡) = rate of change of force [Ns
-1]
We then have to determine the total strain rate, stress relaxation rate and
the rate of change of force, as on right hand side of Equation (3.2), to
calculate the plastic strain rate of the sample. The first term, the total strain
rate (𝜀�̇�𝑜𝑡𝑎𝑙), was directly measured using the external LVDT_2 attached at
the top of the vessel. The second and third terms contain the calculation of
the change of the force and stress on the sample as a function of the time,
which was done by defining a variable window size on each data point with
minimum 10 % of its value on high and low side of the data series of force
and stress.
79
During the experiments on halite reported here, at 50 MPa confining
pressure, the friction between the deformation piston and seal was rather
high, and comparatively, it was more pronounced during relaxation than
during deformation at constant strain rate (compare Figs. 3.3 and 3.5). As a
result, the irregularity in the relaxation curves was found to be substantial,
which obscured the actual trend/path of stress and strain rate during
relaxation by giving a number of different strain rate data points for a single
stress value (see high scatter in log strain rate axis). In order to resolve the
true value of strain rate during stress relaxation, a statistical analysis was
performed, in which the moving average value of strain rate at particular
(decreasing) stress values, during relaxation, was calculated (using a
variable time window to ensure a fixed level of error in the derivative of the
noisy displacement/time curve). The obtained average values of strain rates
gave a better picture of data and to understand the trend. These average
values were used for all relaxation steps (see comparison of real relaxation
data and its mode in Fig. 3.5a).
3.2.5 Microstructural preparations
After deformation, the samples were prepared for microstructural study. The
samples were cut along their length using a diamond tipped saw lubricated
by so-called ’evaporating oil’ (Shell light oil (organic), S4919). The sample
halves were bonded onto glass slides and were polished, first by using SiC
papers and then finalized to an optical finish of 1.0 micron using diamond-
oil suspension (Metadi, Buehler). To reveal the microstructure, samples
were undergone chemical etching (95% saturated NaCl solution + 5% de-
ionized water + 8.0 gm FeCl3 per litre) followed by rinsing with n-hexane
spray and drying using hot air. Photographic images were made in reflected
80
light using a Leica optical polarization microscope equipped with a high
resolution digital image capturing and analysis system.
3.3 RESULTS
3.3.1 Mechanical data
The results of the experiments performed on the synthetic and natural halite
poly-crystals are given in Table 3.1. The stress vs.; strain and time curves
are shown in Figures 3.3a-f.
(i) Stress vs. natural strain and time
The synthetic salt sample, halite1, was deformed in three steps using a
similar strain rate of 5×10-7
s-1
. In each step, the sample was deformed
through a very limited amount of strain of 0.015-0.02. The stress-strain
curve of Figure 3.3a shows that steady state was not reached in any of the
three steps, but the strain hardening rate appears to decrease with increasing
strain.
The natural salt samples, halite2 and 3 were deformed in 7-steps (see
Table 3.1 for details), adding 0.025-0.03 strain at each step. The strength of
the natural halite samples is almost twice that of the synthetic salt, at strain
rate 5×10-7
s-1
. Steady state was not reached in any of the strain rate steps.
The rate of strain hardening differed depending on the strain rate and
appeared higher at the higher rates. For the slowest deformation step, 5×10-7
s-1
, a few stick slip events, due to
81
(a)
(b)
0
5
10
15
20
25
0.00 0.05 0.10 0.15 0.20 0.25
Dif
fere
nti
al s
tres
s [M
Pa]
Natural strain
Halite1
Temp = 125 oC
Pc = 50 MPa
𝜀̇ = 5×10-7 s-1
0
5
10
15
20
25
0 10 20 30
Dif
fere
nti
al s
tres
s [M
Pa]
Time [Days]
Halite1
Temp = 125 oC
Pc = 50 MPa
𝜀̇ = 5×10-7 s-1
82
(c)
(d)
0
5
10
15
20
25
30
35
0.00 0.05 0.10 0.15 0.20 0.25
Dif
fere
nti
al s
tres
s [M
Pa]
Natural strain
Halite2
Temp = 125 oC
Pc = 50 MPa
5×10-5 5×10-6
5×10-7
5×10-7
5×10-8
5×10-6 5×10-5
0
5
10
15
20
25
30
35
0 10 20 30 40
Dif
fere
nti
al s
tres
s [M
Pa]
Time [Days]
Halite2
Temp = 125 oC
Pc = 50 MPa
5×10-5
5×10-6
5×10-7
5×10-7
5×10-8
5×10-6
5×10-5
83
(e)
(f)
Figure 3.3. Differential stress curves against natural strain and time, a-b) halite1, c-
d) halite2, e-f) halite3
0
5
10
15
20
25
0.00 0.05 0.10 0.15 0.20 0.25
Dif
fere
nti
al s
tres
s [M
Pa]
Natural strain
Halite3
Temp = 125 oC
Pc = 50 MPa
5×10-5
5×10-6
5×10-7
5×10-7
5×10-8
5×10-6
5×10-5
0
5
10
15
20
25
0 5 10 15 20 25
Dif
fere
nti
al s
tres
s [M
Pa]
Time [Days]
Halite3
Temp = 125 oC
Pc = 50 MPa 5×10-5
5×10-6
5×10-5
5×10-7
5×10-8
5×10-7
5×10-6
84
Table 3.1. Mechanical data
Sample 𝜀̇
[s-1
] εa
σ
[MPa]
Halite1
5×10-7
0.01 7.7
5×10-7
0.03 8.7
5×10-7
0.05 9.2
Halite2
5×10-5
0.04 22.6
5×10-6
0.07 21.9
5×10-7
0.11 16.0
5×10-8
0.13 13.1
5×10-7
0.15 15.7
5×10-6
0.17 20.3
5×10-5
0.21 21.1
Halite3
5×10-5
0.04 21.8
5×10-6
0.07 20.9
5×10-7
0.10 15.3
5×10-8
0.13 11.9
5×10-7
0.16 15.2
5×10-6
0.18 19.5
5×10-5
0.22 21.9
𝜀̇ is the strain rate
εa is the natural axial strain at the end of the particular step
σ is the differential stress value at the end of the deformation step
high friction between deformation piston and seal, were also observed, as
seen by the irregular nature of the stress-strain curve.
Although the hardening rate per individual step in strain rate could be
established, for both the synthetic and the natural samples, the data as a
whole were insufficient to uncover a robust trend in changing hardening rate
when going from one strain rate to the other. This hampered reliable
extrapolation to higher strain, preventing a comparison of differential stress
85
(a)
(b)
Figure 3.4. Log-log plot of strain rate vs. differential stress using the maximum
differential stress values at the end of each step. Best fit lines to the data of halite2
and 3 represent power law fits, with corresponding n-values indicated.
-8.0
-7.5
-7.0
-6.5
-6.0
-5.5
-5.0
-4.5
-4.0
1.0 1.1 1.2 1.3 1.4
Log (
stra
in r
ate
[s-1
])
Log (stress [MPa])
step1 to step4
step4 to step7
Halite2
Temp = 125 oC
Pc = 50 MPa
11.8 ± 2.4
10.8 ± 2.1
-8.0
-7.5
-7.0
-6.5
-6.0
-5.5
-5.0
-4.5
-4.0
1.0 1.1 1.2 1.3 1.4
Lo
g (
stra
in r
ate[
s-1])
Log (stress [MPa])
step1 to step4
step4 to step7
11.1 ± 1.2
13.6 ± 2.6
Halite 3
Temp = 125 oC
Pc = 50 MPa
86
as a function of strain rate at the same strain. Thus, in Table 3.1, only the
values of the differential stress at the end of each step are given.
(ii) n-value for the natural halite
The dependence of the differential stress on the strain rate for the natural
halite samples (Fig. 3.4) was tested by applying a conventional power law
creep of the type 𝜀̇ = 𝐴𝜎𝑛, where 𝜀̇ is the strain rate, σ is the flow stress, and
A and n are constants. Since both halite2 and 3 presented substantial strain
hardening, any estimate of the constants A and n can only give a first-order
impression of the creep behaviour of the material the data were divided into
two parts, for both halite2 and 3; step1 to step4 (decreasing strain rate) and
step4 to step7 (increasing strain rate). Linear regression analysis in log-log
space resulted in high n-values, ranging 10.8 to 13.6 (Figs. 3.4a-b). Note
that the strain rate is the independent variable, but it is shown as dependent
in the Figure 3.4 (i.e. on Y-axis) to compare with the stress relaxation curves
(Figs. 3.5a-d).
(iii) Stress relaxation
The stress relaxation behaviour of the tested halite samples are plotted in
Figures 3.5a-d. The steady state/maximum stress values obtained at the end
of each strain rate deformation step are marked in the graphs along with the
projected slopes representing the stress exponent n-values (n = 1, 5 and 10)
according to scale. As explained earlier, the stick slip events between
deformation piston and seal were more conspicuous during stress relaxation
than during deformation. Such events caused great noise in the signal of the
measured displacement and, hence, resulted in substantial scatter in
calculated strain rate, so the mode values were used instead (see section
3.2.4).
87
Generally, the graphs show that the calculated plastic strain rate at the
start of each relaxation period fits to a trend with high n-value (>10, i.e. a
steep slope), implying that the stress is not very sensitive to strain rate.
Progressively, the sensitivity increases and the corresponding n-value
appears to approach n = 1 for lower stress and strain rate values near end of
each relaxation step. A free hand dashed-line drawing, showing trend of
relaxation is also plotted along to show the expected trend during relaxation.
(a) Noise in strain rate data and the corresponding mode values
-10.0
-9.5
-9.0
-8.5
-8.0
-7.5
-7.0
-6.5
-6.0
-5.5
-5.0
0.0 0.5 1.0 1.5
Log (
stra
in r
ate
[s-1
])
Log (stress [MPa])
Step 2 5E-7 s^-1
Mode step 2
5×10-7 s-1
88
(b)
(c)
89
(d)
Figure 3.5. Log-log plot of strain rate vs. differential stress including the results of stress relaxation steps. Closed triangles are the
steady state/maximum differential stress values obtained at the end of each step, closed rectangles are the mode values of the
calculated strain rate data. Three slopes representing the n-values of 1, 5 and 10 are included to be compared with the trend during
relaxation, represented by the free-hand drawn dashed line. a) selected example of the collected data (close diamonds), showing the
scatter due to sudden change of stress during stick slip events, and open square symbols are the mode corresponding mode values, b)
halite1, c) halite2, d) halite3
90
3.3.2 Microstructures
Two samples were prepared for microstructural study. Halite1 (fine grained,
reagent grade pure synthetic polycrystalline sample) shows a microstructure
of a dense aggregate of flattened grains. The grains show lobate and locally
bulged grain boundaries indicating dynamic recrystallization. The grain
boundaries often intersect at triple points with angles deviating from 120o.
Sub-grain walls and sub-grains are recognizable close to at least some of the
grain boundaries, but overall, the internal parts of the grains do not show
well-developed substructures. This sample was allowed to relax under stress
for one week, under 50 MPa confinement at 125 oC, after deformation at a
strain rate of 5×10-7
s-1
. Halite2 (coarse grained, natural polycrystalline)
sample with a grain size of 2-5 mm. The sample was deformed at 5×10-5
s-1
as last step of its stepping test, and was not relaxed but was taken out of the
machine for microstructural steady. Microstructure shows slightly open
grain boundaries but no sign of dislocation substructures, but recrystallized
and shows flattened grains. The open grain boundaries are due to handling
effects, as, at the end of experiment, the confining pressure (50 MPa) was
taken off before the sample was properly cooled down (to below the boiling
point of water).
91
Figure 3.6. Microstructures after tests, deformation direction is horizontal, wet
samples (H2O ≈30 ppm) A) synthetic salt halite 1, recrystallized grains with no
internal structure, average grain size of ~300 μm, B) natural salt halite 2, grain size
2-5 mm, slightly open grain boundaries, no dislocation structures.
100 μm A
10 mm B
92
3.4 DISCUSSION
The aim of this study was to investigate if a transition can be observed from
creep behaviour of halite governed by dislocation mechanisms to creep
behaviour controlled by a solution-precipitation mechanism, and if so, what
the conditions of this transition are in terms of strain rate. Below, we will
discuss our observations and compare the strength of the halite and its
change with decreasing strain rate with the mechanical behaviour
established in other studies.
3.4.1 n-value
Fitting the stress-strain rate data obtained during the constant strain rate
parts of the multi-step experiments to a conventional power law creep of the
type 𝜀̇ = 𝐴𝜎𝑛 resulted in a value for n larger than 10. Microphysical models
for creep controlled by dislocation climb generally result in power law creep
equations (e.g. see Chapter 2), but the values for n usually range 3-4.5 for
climb controlled by lattice diffusion, or up to 6.5 in case of climb controlled
by dislocation core (pipe) diffusion (Senseny et al. 1992, Carter et al. 1993,
Franssen 1994) . The value of n > 10 appears to rule out climb control for
the creep behaviour of wet polycrystalline halite for the conditions tested.
Rather, a glide or cross slip controlled creep model may apply (cf. De
Bresser et al. 2002). The results presented in Chapter 2 on the dry rock salt
showed a low sensitivity of the stress on the strain rate similar to the wet
halite reported here, at comparable conditions of temperature and pressure,
at 125 oC and 50 MPa. For the dry halite, it was concluded that a glide
mechanism (Weertman 1957, Poirier 1985) rather than a cross slip
mechanism (cf. Auten et al. 1973, Skrotzki et al. 1981) controls flow at the
given conditions. Though we have not tested the effect of pressure on the
93
Figure 3.7. Log-log plot of strain rate vs. differential stress, comparing the current
results with the results of previous studies on wet and dry halite at 125 oC. (Heard
1972, Wawersik & Zeuch 1986, Carter et al. 1993, Ter Heege et al. 2005a), (see
Table 3.2 for details).
strength of wet halite, which may serve as a way of discriminating between
mechanisms (Chapter 2), we infer that the conclusion regarding glide
control also holds for the current wet halite.
The stress relaxation behaviour of the wet halite (Figs. 3.5a-d) shows that
the n-value gradually decreases with decreasing stress and strain rate,
reaching ~1 at strain rates below 10-8
s-1
. This trend is observed in both
synthetic and natural halite samples. This strongly suggests that a transition
takes place towards grain size sensitive (GSS) creep (Spiers et al. 1990, Ter
Heege et al. 2005a).
If we compare our work with previous studies on halite by using their
flow laws, parameters given in Table 3.4, (Heard 1972, Wawersik & Zeuch
-12
-10
-8
-6
-4
-2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Log (
stra
in r
ate
[s-1
])
Log (differential stress [MPa])
Halite1
Halite2
Halite3
Heard 1972
Wawersik & Zeuch 1986
Carter et al. 1993_high
Carter et al. 1993_Low
Ter Heege et al 2005a
94
1986, Carter et al. 1993 and Ter Heege et al. 2005a), for 125 oC, we see in
Figure 3.7 that the current data on natural halite2 and 3 have got the upper
limit by Wawersik & Zeuch (1986) (unknown water content) and Ter Heege
et al. 2005a (wet), whereas the lower bounds are given by Carter et al. 1993
(dry). The data of synthetic sample halite1 show weaker behaviour and lie in
approximation of Wawersik & Zeuch (1986) and Ter Heege et al. (2005a).
3.4.2 Composite flow law
Provided physical conditions (like pressure and temperature) are fixed, the
microstructure of the sample may be expected to remain constant during
stress relaxation with limited straining (see Rutter & Mainprice, 1978),
while processes such as dynamic (syn-deformational) recrystallization and
grain growth are likely to affect the microstructure during deformation that
results in substantial increase in strain. The relaxation data show that the
dependence of strain rate on stress approaches linearity at low stress and
slow strain rate, which observation has been used above to suggest that
grain size sensitive behaviour might play a role.
We thus should consider if the flow behaviour of wet halite should be
described by a composite flow equation of grain size insensitive (GSI) and
grain size sensitive (GSS) behaviour as follows
𝜀̇ = 𝐴∗𝜎𝑛 + 𝐵∗𝜎𝑑−𝑝 (3.3)
Where A* and B* are constants at a given temperature, n is the usual stress
exponent and p is the grain size exponent.
95
Table 3.2. The updated Table of flow law parameters after Ter Heege et al. 2005a.
Material/
composition
H2O
[ppm]
P
[MPa] 𝜀̇
[s-1]
T
[oC]
σ
[MPa]
LOGA
[MPa-n s-1]
Stress
exponent n
ΔU
[kJmol-1]
Source/
comments Synthetic pure rocksalt
5-10 50-600 4×10-7 to 10-4 250-350 2.8-15.2 -14.4 ±1.4 4.7 ±0.3 126 This study (Chapter 2)
Synthetic pure
rocksalt 20–45 200 10-1 to 10-8 23-400 1.6-47 5.58 ±0.8 5.5 ±0.4 98 ±8 Heard (1972)
Synthetic pure
rocksalt 20–45 200 10-1 to 10-8 23-400 1.6-47 0.7 ±0.4 5.8 ±0.2 96 ±3
Heard & Ryerson
(1986)
Natural (>95% rocksalt)a
? 14, 21 10-6 to 10-11 23-160 8.3-24 3.36-6.03 4.1 ±6.3 50-83 Wawersik & Zeuch (1986)
Natural
(>99% rocksalt)b <100 2.5-20.7 10-6 to 10-9 50-200 6.9-20.7 3.8 5.3 ±0.4 68 ±4
Carter et al.
(1993)
high 𝜀̇, σ
Natural
(>99% rocksalt)b <100 2.5-20.7 10-7 to 10-9 100-200 2.5-10.3 4.09 3.4 ±0.1 52 ±1
Carter et al.
(1993)
low 𝜀̇, σ
Synthetic pure
rocksalt Dry Uncof. 10-3 to 10-7 250-780 0.4-14.8 -0.76 ±0.2 5.7 ±0.3 129 ±8
Franssen (1994)
Low T
Natural (rocksalt)c ? Unconf.
+ 15-20 10-3 to 10-11 30-250 1.7-40 -- 7 110
Hunsche & Hampel
(1999)d
Natural
(>98% rocksalt)c 500 3-30 3.5×10-7 150 11-13 -- -- --
Peach et al.
2001
Synthetic pure
rocksalt 9–46 50 10-4 to 10-7 75-200 7.2-22.4 1.56 ±0.54 5.6 ±0.5 80 ±6
Ter Heege et
al 2005
a: Range of parameters for natural rocksalt from five different locations: Salado (New Mexico), West Hackberry and Bayou Choctaw (Louisiana), Bryan Mound
(Texas) and Asse (Germany).
b: Avery Island (Louisiana). c : Asse Speisesalz (Germany).
d: In this study, mechanical data were fitted to a composite law. Stress exponent and activation energy quoted here are from their best fit to a power law equation.
96
In order to evaluate Equation (3.3), we need to estimate what the grain
size was during deformation of our halite samples. We do this by using a
conventional Piezometric relation, allowing predicting what the dynamically
recrystallized grain size was at a certain stress value (e.g. De Bresser et al.
2001)
𝑑 = 𝐾𝜎 −𝑚 (3.4)
where K and m are material and mechanism specific constants. This
relationship between stress and recrystallized grain size is generally
assumed to be independent of temperature, although that there is evidence
that this assumption is not generally valid.
Ter Heege et al. (2005b) established a Piezometric relation between the
grain size d and differential stress σ by measuring the grain size of
experimentally deformed synthetic halite samples. The data by the authors
are shown in the Figure 3.8 along with the best fit linear trend. The halite
samples tested here (halite2 and 3) are natural salt samples, but we assume
that the same piezometer relation holds at current conditions of temperature
and pressure; 125 oC and 50 MPa respectively.
Using the slope and intercept values of the best fit line in Figure 3.8, we
obtained values for K and m as 27.1 and 1.75 respectively, so the Equation
(3.4) reduces to
𝑑 = 27.1𝜎 −1.76 (3.5)
Using this Equation, the grain size at different stress values were
calculated and are given in Table 3.3. These values represent the grain sizes
97
at the end of the constant strain rate parts of the experiments, and hence, the
grain sizes at the start of the relaxation parts. Assuming now that the
microstructure remains constant during relaxation, these are also the grain
sizes applicable at the behaviour at n ≈ 1. This allows us to estimate the
value for p of Equation (3.3), in which stress is linearly dependent on strain
rate. Taking logarithms and simplifying Equation (3.3), we get
𝑙𝑜𝑔𝜀 ̇ = (𝑙𝑜𝑔𝐵∗ + 𝑙𝑜𝑔𝜎) − 𝑝𝑙𝑜𝑔𝑑 (3.6)
Or
𝑙𝑜𝑔𝜀 ̇ = 𝑖𝑛𝑡𝑒𝑟𝑐𝑒𝑝𝑡 − 𝑝𝑙𝑜𝑔𝑑 (3.7)
To keep the intercept part of this equation as a true constant, a fixed value of
log stress (0.8) was selected (i.e. 6.3 MPa). For this fixed value of stress, log
strain rate values were picked from the stress relaxation curves (Figs. 3.5c-
d) (see relaxation data in Table 3.4). Note that halite1 was not used in this
analysis being different. The selected strain rate values are thus plotted
against the calculated grain size d values in Figure 3.9.
Regression analysis revealed a p-value of -1.1 (±0.3). This p-value
supports the hypothesis that during stress relaxation, grain size sensitive
(GSS) creep might play a role. It is well known that (GSS) solution-
precipitation mechanisms may play an important role in the creep of fine
grained halite, creep (e.g. Raj, 1982, Spiers et al. 1990, Schutjens 1991,
Cox & Paterson 1991, Schutjens & Spiers 1999). According to these
authors, a p-value of about 1 suggests, that the rate controlling step in the
process will be dissolution or precipitation rather than diffusion.
Accordingly, we infer that dissolution/precipitation controlled pressure
98
Figure 3.8. Log-log plot of differential stress vs. grain size for wet synthetic halite
(Ter Heege 2005b).
Figure 3.9. Strain rates picked from relaxation curves at fixed stress value (σ
=6.3[MPa]) against the recrystallized grain size calculated using the piezometer
relation of Eq. 3.5.
y = -0.5688x + 0.8151
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
Log (
stre
ss [
MP
a])
Log (grain size [mm]
-8.6
-8.5
-8.4
-8.3
-8.2
-8.1
-8.0
-7.9
-7.8
-1.0 -0.8 -0.6 -0.4
Lo
g (
stra
in r
ate
[s-1
])
Log (grain size [mm])
Halite2 and 3
99
Table 3.3. Calculated grain size using piezometer using maximum stress at the end
of deformation step
Sample Step 𝜀̇
[s-1
]
σ
[MPa]
d
[mm]
Halite 1
1 5×10-7
7.7 0.750
2 5×10-7
8.7 0.603
3 5×10-7
9.2 0.542
Halite2
1 5×10-5
22.6 0.113
2 5×10-6
21.9 0.120
3 5×10-7
16.0 0.206
4 5×10-8
13.1 0.294
5 5×10-7
15.7 0.213
6 5×10-6
20.3 0.136
7 5×10-5
21.1 0.127
Halite3
1 5×10-5
21.8 0.120
2 5×10-6
20.9 0.129
3 5×10-7
15.3 0.224
4 5×10-8
11.9 0.347
5 5×10-7
15.2 0.228
6 5×10-6
19.5 0.146
7 5×10-5
21.9 0.119
𝜀̇ is the strain rate
σ is the differential stress value at the end of the deformation step
d is the calculated recrystallized grain size, applying Eq. 3.5
solution processes dominates the creep of wet halite at low stress and strain
rate, at 125 oC.
Now using the intercept value of the best fit line in Figure 3.9, the
unknown constant B* gets the value of 1.68x10-10
(MPa-1
mm1.1
s-1
), so the
GSS creep equation becomes
100
Table 3.4: Data picked from relaxation curves at similar differential stress values
Sample Deformation
strain rate [s-1
] 𝜀̇
[s-1
]
LOG
(𝜀̇[s-1])
σ
[MPa]
LOG
(σ[MPa])
Halite1
5×10-7
4×10-8
-7.4 6.3 0.8
5×10-7
4.3×10-8
-7.4 6.3 0.8
5×10-7
3.4×10-8
-7.5 6.3 0.8
Halite2
5×10-6
8.6×10-9
-8.1 6.3 0.8
5×10-8
4.7×10-9
-8.3 6.3 0.8
5×10-6
7.0×10-9
-8.2 6.3 0.8
Halite3
5×10-6
1.4×10-8
-7.9 6.3 0.8
5×10-8
2.9×10-9
-8.5 6.3 0.8
5×10-7
9.7×10-9
-8.0 6.3 0.8
𝜀̇ is the strain rate values picked from stress relaxation curves at fixed differential
stress value (6.3 MPa)
Deformation strain rates are the values used to deform the sample and set the stress
value for relaxation.
σ is the fixed differential stress selected to pick the data from relaxation curves
𝜀�̇�𝑆𝑆 = 1.68 × 10−10𝜎𝑑−1.1 (3.8)
For the GSI part of Equation (3.3), we assume that conventional power law
is applicable; plotting the steady points of halite2 and 3, against the
deformation strain rates (Figure 3.10), the values of n and A* come out to be
~11 and 4.365×10-20
s-1
MPa-11
, so the GSI part can be written as
𝜀�̇�𝑆𝐼̇ = 4.365 × 10−20𝜎11 (3.9)
The GSS flow law (Eq. 3.8) is based on the calculated grain size - strain rate
data at σ = 6.3 MPa under the assumption that there is no influence of GSI
flow at that condition. So, the GSS flow law (Eq. 3.8) is the right-handed
101
Figure 3.10. Log-log plot for conventional power law (𝜀̇ = σn). The slope of the
curve represents the stress exponent n-value. Three slopes lines corresponding to n
= 1, 5 and 10 are also projected according to scale. The slope of halite2 and 3
combined data very high (i.e. n ~ 11).
𝜀�̇�𝑆𝐼̇ = 4.365 × 10−20𝜎11 (3.9)
end member of the composite description of Equation (3.3). Its trends for
four different grain sizes are plotted in Figure 3.11, where the trend for the
steady state data (halite1 and 2) is included, as described by Equation (3.9).
Equation (3.9), however, cannot simply be regarded as the left-hand (GSI)
end member of the composite flow law (Eq. 3.3), since grain size sensitive
behaviour might have influenced the steady state creep without realizing it.
We thus took the data of halite2 and 3 from Table 3.1 and fitted them to
Equation 3.3, using the established values for n, p and B* and applied non-
linear regression. This resulted in a value of 6.42×10-20
(MPa-11
s-1
) for A*.
The composite flow law (Eq. 3.3) now can be written, as
y = 10.97x - 19.36
-8.0
-7.5
-7.0
-6.5
-6.0
-5.5
-5.0
-4.5
-4.0
1.0 1.1 1.2 1.3 1.4
Log (
stra
in r
ate
[s-1
])
Log ( flow stress [MPa])
Halite 2 and 3
n = 1
n = 5
n = 10
102
Figure 3.11. Log-log plot of strain rate and differential stress showing the predicted trends for GSS creep applying Eq. 3.8 for
different grain sizes (GSS1, 2, 3 and 4 for d = 0.1, 0.2, 0.3 and 0.4 mm respectively), satisfying the lower stress and strain rate data.
Note that higher grain sizes are lower in strain rates. The trend for GSI creep applying Eq. 3.9, satisfies the steady state values of
constant deformation data.
-9.0
-8.5
-8.0
-7.5
-7.0
-6.5
-6.0
-5.5
-5.0
-4.5
-4.0
0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
Log (
stra
in r
ate
[s-1
])
Log (differential stress [MPa])
Halite 2 and 3 steady state data
GSI
GSS1
GSS2
GSS3
GSS4
Halite2 and 3 data from relaxation curves
103
Figure 3.12. Log-log plot of strain rate and differential stress showing the predicted trends applying the composite creep Eq. 3.10, for
a grain size of 0.1 mm (composite1), 0.2 mm (composite 2),0.3 mm (composite 3) and 0.4 mm (composite 4).
-9.0
-8.5
-8.0
-7.5
-7.0
-6.5
-6.0
-5.5
-5.0
-4.5
-4.0
0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
Log (
stra
in r
ate
[s-1
])
Log (differential stress [MPa])
Halite2 and 3 steady state data
composite1
composite2
composite3
composite4
Halite2 and 3 data from relaxation curves
104
𝜀̇ = 6.42 × 10−20𝜎11 + 1.68 × 10−10𝜎𝑑−1.1 (3.10)
The trend lines in Figure 3.12 show that the influence of grain size is
effective at lower stresses and strain rates. For higher stresses, these curves
satisfy the steady state points, whereas on lower stress/strain rates, these
trends satisfy the data picked from relaxation curves. So the composite flow
law gives a complete picture of the creep characteristics of natural wet halite
samples, in two regimes of GSI (n ~ 11) and GSS (n ~ 1) that gradually pass
into each other on higher stress and strain rate side.
CONCLUSIONS
The main research question addressed in this chapter was if a transition can
be observed, in wet polycrystalline halite, from creep behaviour governed
by a (grain size insensitive) dislocation mechanisms to creep behaviour
controlled by a (grain size sensitive) solution-precipitation mechanism, and
if so, what the conditions of this transition are in terms of strain rate.
The experiments on the synthetic and natural halite samples at a confining
pressure of 50 MPa and temperature of 125 oC revealed that at faster strain
rates and higher stresses, dislocation creep (GSI) plays its role, which can be
described by a conventional power law creep equation with n = 11. With
such a high value, dislocation climb is unlikely to be rate controlling, but a
dislocation glide mechanism should be considered. During the stress
relaxation parts of the experiments, the n-value decreased to ~1 and a grain
size sensitive creep mechanism, most probably pressure solution, becomes
operative with the dissolution/precipitation step as rate controlling, as
105
revealed by an analysis of the grain size exponent of creep (p ~ 1), using a
flow equation of the type 𝜀̇ = 𝐵∗𝜎𝑑−𝑝.
The transition from (presumably) glide controlled dislocation creep with
high power law n-value to grain size sensitive creep at n ~ 1, at 125 oC,
occurs at a strain rate of about 10-9
s-1
. So, at slow strain rates ~10
-9 s
-1, at a
depth equivalent to an overburden/hydrostatic pressure of 50 MPa, grain
size sensitive creep mechanism plays a more important role than dislocation
mechanisms.
The number of experiments performed at the given conditions are
limited, i.e. one experiment on reagent grade synthetic fine grain
polycrystalline salt (halite1) and two on natural coarse grained salt samples
(halite2 and 3), whereas the composite flow law is proposed for natural salt
for at in situ conditions. Doing more experimentation on different grain
sized halite samples, both on synthetic and natural would help to make more
general flow law. However, the story (i.e. high n-value for faster strain rate
and higher stress and low n-value for slower strain rate and lower stress)
does not seem to change much as we have seen in the current work.
106
107
Chapter 4
Creep behaviour of bischofite,
carnallite and mixed bischofite-
carnallite-halite salt rock at in situ
conditions
108
4.1 INTRODUCTION
The salt deposits at Veendam (northern part of the Netherlands) are mainly
composed of the evaporites bischofite, carnallite and halite in the form of
layers and mixtures, with sulphates in minor quantities. In order to solution
mine the caverns, it is important to know the rheology of the different salts
at real in situ conditions, so that the rate of inflow into the caverns as well as
surface subsidence can be predicted. The strain rate of such salts in
underground mines is normally in the range of 10-8
to 10-15
s-1
(Heard 1972,
Van Eekelen et al. 1981, Jackson & Talbot 1986) which is a rate that cannot
be achieved easily in laboratory scale experiments. However, laboratory
experiments can be used to define a flow law that allows extrapolation to
real in situ conditions. In order to perform such extrapolation in a reliable
manner, good understanding of the deformation mechanism of the material
is needed, so that the characteristics of the flow law can be related to the
microphysical mechanism controlling creep. In this case study, the
mechanical properties of rock salts bearing bischofite, carnallite and their
mixtures, are studied with the main aim of producing constitutive flow laws
than can be applied at real in situ conditions.
Van Eekelen et al. (1981) and Urai (1983) have tested the creep
behaviour of bischofite at a fixed confining pressure of 28 MPa, in the
temperature range of 40-80 °C, with varying water content including dry
samples. The authors suggested that a conventional power law of the type
𝜀̇~𝜎𝑛, relating strain rate 𝜀̇ to stress , can be used to describe the flow
behaviour of bischofite, as long as two regimes were defined, one at low and
one at relatively high differential stress. The flow laws of the two regimes
show different values of the power law stress exponent n, namely 1.5 and 4
for the low and high stress regime, respectively. However, the nature of the
109
two regimes was not fully understood and accordingly, a good basis for
establishing the rate of inflow relevant in the case of cavern evolution
associated with solution mining is still missing.
The creep properties of dry and wet carnallite have been studied by Urai
(1985), by performing triaxial deformation experiments at a temperature of
60 oC, using a range of strain rates and confining pressures. Urai (1985)
proposed a conventional power law for steady state creep of wet carnallite
with stress exponent n = 4.8 ± 0.1. The strength of carnallite was found to be
substantially higher than that of bishofite. It is yet unknown what the
rheology of mixtures of bishofite and carnallite is like, and which of these
salts dominates creep in a mixture.
Also halite, both as natural single and poly-crystals and as artificially
prepared dense aggregates, has been studied before for its creep properties
under dry and wet conditions (Heard 1972, Heard & Ryerson 1986,
Wawersik & Zeuch 1986, Urai et al. 1986, Spiers et al, 1990, Senseny et al.
1992, Carter et al. 1993, Spiers & Carter 1998, Ter Heege et al. 2005b,
Muhammad et al. 2012). Also for this material, a conventional power law
appears to describe the behaviour well. For wet salt, the proposed stress
exponent n lies in the range of 4.1 to 5.7, in the temperature range of 23 to
400 oC.
In the current work, we have performed triaxial deformation experiments
on polycrystalline natural samples of bischofite, carnallite and their mixture
with halite, at real in situ conditions of confining pressure 40 MPa and at a
fixed temperature of 70 oC. All deformation tests were done in strain rate
stepping mode, with several steps being followed by stress relaxation
(Rutter & Mainprice 1978), with the aim to achieve strain rates as low as ~
10-9
s-1
, approaching the natural strain rates in salt caverns. Aims of the
110
work were: 1) to determine the creep behaviour of relatively pure bishofite
and carnallite under in situ conditions, 2) to establish the role of dislocation
creep, recrystallization and, possibly, pressure solution creep in the
behaviour of bishofite and carnallite, and construct creep laws that allow
reliable extrapolation, and 3) obtain a first order impression of the creep
behaviour of mixtures of bishofite, carnallite and halite, and compare this
with the behaviour of the end members.
The mechanical data of the samples tested confirmed that bischofite is
much weaker than carnallite, and revealed that carnallite is weaker than the
bishofite-carnallite-halite mixture. During stress relaxation, we observed
that, at 70 oC, the n-value for bischofite and carnallite gradually changed
with changing conditions, from n = 5 at higher stress (and higher strain rate)
to n = 1 for lower stress (and corresponding lower strain rate). For the
mixtures, the n-value is observed to be rather high at relatively high stress (n
> 10), but also reduced to n = 1 during relaxation to slow strain arte and low
stress. A higher n-value (n > 3) is usually related to grain size insensitive
(GSI) dislocation creep processes and a lower n-value is consistent with
grain size sensitive (GSS) mechanisms such as pressure solution. We have
evidence that the rate controlling mechanism might not be governed solely
by dislocation motion, but may also be grain size dependent, depending on
deformation conditions. Accordingly, the flow behaviour of bischofite and
carnallite can be explained by flow laws combining GSI and GSS creep. We
therefore suggest that the established flow laws for bischofite and carnallite
(Urai 1983, 1985) require modifications. These results help putting
constraints on the behaviour of salt mixtures, as for example in the case of
the salt deposits at Veendam.
111
4.2 METHOD
4.2.1 Sample preparation
The natural cores of bischofite, carnallite and their mixture with halite,
extracted by Nedmag Industries Mining & Manufacturing B.V. during the
so-called TR9 drilling project, were provided with 100 mm diameter and 1
m length, along with the description. The selection of the cores was initially
based on the description, later on; compositional analysis was done locally
using micro XRF technique.
The cores were first cut down to rectangular rods of about 100 mm
length and 50 mm diameter using a hand saw. These rods were then shaped
down to samples with the required dimensions of 35 mm diameter and 85
mm (average) length, using Silicone Carbide papers. Since the salts under
investigation are hygroscopic, the samples were prepared in a low humidity
room with relative humidity (R.H.) < 15% to control the water content.
Natural salts in general are wet (Roedder & Bassett 1981, Urai 1983). In
order to create a deliquescence condition in our samples, comparable to that
at in-situ conditions, the samples were first equilibrated with air with R.H. >
30% (Urai 1985, Christov 2009) and, in parallel, carefully measured for
increase in weight by water absorption. This was followed by wrapping up
the samples in a double layer of perforated glass fibre sheet (0.3 mm starting
thickness) creating an equilibrated humid environment around the sample.
These wrapped samples were further sealed in 1.0 mm thick polymer
“ethylene propylene diene monomer (EPDM)” jackets to avoid
contamination of the samples by the confining medium (silicone oil) used in
the deformation apparatus. The deformation pistons were inserted in the
open ends of the jacket. To seal the ends, stainless steel wires were tightly
112
wound in grooves of the steel pistons from over the jacket. After the
experiments, the samples were taken out immediately, sealed in plastic
wrapper and stored at -20 oC to freeze the microstructures by preventing
fluid-assisted recrystallization processes.
In order to obtain a first-order insight into the composition of the mixed
salt rock samples used, the micro XRF (X-ray Fluorescence) technique was
applied. In this technique, flat surfaced samples were prepared by cutting
and polishing, and these surfaces were analysed for the number of counts
being received as per characteristic X-Rays of the elements. The output data
obtained from the XRF analysis contains a list of elements and their atomic
and weight percentage present at the surface of the selected area. These
elements can be combined according to their contribution in the chemical
formulae of the salts involved, and the results then show in wt. % the
composition. Unfortunately, the XRF machine used is not calibrated for
elements with atomic number smaller than 10. So the lighter elements
hydrogen and oxygen could not be detected. As a consequence, the water
content of our mixtures could not be studied.
4.2.2 Deformation apparatus
The apparatus used for this study was the so-called “Shuttle vessel” (Fig.
4.1) of the experimental rock deformation (HPT) laboratory at the
department of Earth Sciences at Utrecht University. The Shuttle vessel
machine consists of an internally heated 100 MPa confining pressure vessel
mounted on a standard Instron 1362 loading frame with a servo controlled
positioning system. This machine can be used to deform the sample at a
constant piston speed, approaching constant strain rate if total strains are
kept relatively low. The machine is provided with a (Instron standard) +/- 50
113
Figure 4.1. Shuttle vessel triaxial deformation apparatus.
Figure 4.2. Sample with piston assembly and steel discs
35 mm
LVDT_2
Deformation piston
Thermocouples
connections
Pressure vessel
containing sample
assembly
114
mm linear variable differential transformer (LVDT_1), but to come to an
accurate measurement of the sample deformation, another LVDT (LVDT_2)
with a 25 mm stroke from Jensen was installed at the top of the vessel and
near the sample, to reduce the effect of the elastic distortion of the apparatus
and measure accurately the shortening of the sample, especially during
stress relaxation, where very limited axial strain in the order of ~0.1% is to
be monitored. The temperature was measured by thermocouples at two
locations inside the vessel; one was positioned close to the middle of sample
and, the other was at the top of the sample. K-type thermocouples were
used, which are accurate within ± 1 oC. The axial load on the sample was
measured using a 100 kN load cell accurate within 0.1% of the full scale.
The confining pressure was created by using silicone oil which is kept at
constant pressure within ± 0.1 MPa using a servo pump. Measurement of the
pressure was done using a diaphragm pressure transducer (Teledyne 2403,
100 MPa range). To avoid rusting of the deformation pistons and related
contamination of the samples, two grooved stainless steel discs having 2.0
mm thickness and 35.0 mm diameter were used as separators, along with 50
μm thin PTFE sheets to reduce the friction between sample and the
deforming pistons (Fig. 4.2). The triaxial apparatus used is very sensitive to
the environmental conditions. For example, a little variation in laboratory
temperature not only causes a drift in the load cell signal, but also may
cause a small change in the pressure, which would influence the load cell
signal. Such (small) change in pressure will result in some elastic
expansion/contraction of the pressure vessel, hampering straightforward
processing of the data from the LVDT_2. Therefore the apparatus was
carefully calibrated for the effect of pressure and temperature on the load
cell. The elastic distortion of machine was corrected on the basis of tests
using a steel dummy sample of known Young’s modulus. This was done by
115
applying a gradually increasing axial load to the dummy and measuring the
LVDT_2 reading in parallel, then using a 10th order polynomial to relate the
LVDT_2 value to the internal load.
4.2.3 Experiments
In this study, multistep experiments were performed. A typical experiment
consisted of a few steps at constant strain rate, in the range 10-5
to 10-8
s-1
,
interrupted by periods of stress relaxation. During the constant strain rate
part of the test, the sample was deformed until a steady (or near steady) state
of stress was reached. This usually required about 2-4% of shortening. Then
the piston was arrested and the stress on the sample was allowed to relax
until the diminishing force on the sample reached the limits of the load cell
resolution. The duration of each relaxation step was a few days.
In detail, each experiment included a maximum of seven strain rate steps
(10-5
, 10-6
, 10-7
, 10-8
, 10-7
, 10-6
, 10-5
s-1
), followed by stress relaxation; after
each step in case of bischofite and after three steps (10-6
, 10-8
, 10-6
see Table
4.1) in case of carnallite and mixture samples. The experiments were
performed at 70 oC sample temperature and 40 MPa confining pressure for
real in situ conditions, whereas the bischofite was additionally tested at 70
MPa confining pressure to obtain a first-order impression of the pressure
sensitivity of creep.
4.2.4 Data acquisition and processing
Confining pressure, temperature, load and position (LVDT_1 and 2) were
logged throughout the test. The stress on the sample was calculated from the
load values by assuming constant volume deformation, correcting the
instantaneous area value for progressive change in length of the sample. The
116
cumulative strain (shortening) of the sample was calculated as the ratio of
the change in length of sample and the initial length of sample. The zero of
the internal load signal was tested before the deformation piston touched the
sample and after unloading at the end of experiment. If found necessary, a
correction was applied for zero-level shift. The data acquired during stress
relaxation was analysed with a dedicated code to produce the plastic strain
rate of the sample.
The stress relaxation technique is based on the following. During
deformation at constant rate (imposed by a moving piston), some energy
gets stored inside the material as elastically stored energy. During
relaxation, when the loading piston is arrested, this energy is dissipated
through plastic deformation of the sample. Ideally, this requires an infinitely
stiff machine which does not participate by means of the elastically stored
energy in its own frame. In practice, such a machine does not exist, so the
elastic modulus of machine should be taken into account. In other words, the
elastically stored energy inside the active part (under force) of the machine
also dissipates through the plastic strain of the sample, so the data also need
to be corrected for the stiffness of machine. In order to calculate the sample
straining, we take the simple assumption that the stress is proportional to
strain, provided that other physical conditions (confining pressure,
temperature, microstructure) are constant (Rutter & Mainprice 1978). The
strain rate at any instant will then be proportional to the stress relaxation
rate, with the elastic (Young’s) modulus of sample as a constant of
proportionality.
The measured total displacement” xtotal” (from LVDT_2) can be written
as the sum of the sample deformation (elastic + plastic) and elastic distortion
of apparatus
117
𝑥𝑡𝑜𝑡𝑎𝑙 = 𝑥𝑠𝑎𝑚𝑝𝑙𝑒 + 𝑥𝑎𝑝𝑝𝑎𝑟𝑎𝑡𝑢𝑠 (4.1)
After differentiating w.r.t. time and normalizing this Equation using the
instantaneous length of the sample, the following relation of plastic strain
rate of the sample is obtained:
𝜀�̇�𝑙𝑎𝑠𝑡𝑖𝑐 = 𝜀�̇�𝑜𝑡𝑎𝑙 −1
𝐸𝑠𝑎𝑚𝑝𝑙𝑒(�̇�𝑠𝑎𝑚𝑝𝑙𝑒) −
1
𝐿𝑡𝑆 (
𝜕𝐹
𝜕𝑡) (4.2)
where,
𝜀�̇�𝑙𝑎𝑠𝑡𝑖𝑐 = plastic strain rate of sample [s-1
]
𝜀�̇�𝑜𝑡𝑎𝑙 = total strain rate measured by LVDT_2 [s-1
]
𝐸𝑠𝑎𝑚𝑝𝑙𝑒 = Young’s modulus of sample [MPa]
�̇�𝑠𝑎𝑚𝑝𝑙𝑒 = sample stress relaxation rate [MPa s-1
]
𝐿𝑡 = instantaneous length of sample [m]
𝑆 = machine stiffness correction constant [m N-1
]
(𝜕𝐹
𝜕𝑡) = rate of change of force [N s
-1]
We then have to determine the total strain rate, stress relaxation rate and
the rate of change of force, as on right hand side of Equation (4.2), to
calculate the plastic strain rate of the sample. The first term, the total strain
rate (𝜀�̇�𝑜𝑡𝑎𝑙), was directly measured using the external LVDT_2 attached at
the top of the vessel. The second and third terms contain the calculation of
the change of the force and stress on the sample as a function of time, which
was done by defining a variable window size on each data point with
118
minimum 10% of its value on high and low side of the data series of force
and stress.
4.2.5 Young’s modulus measurement
The second term of Equation (4.2) contains the Young’s modulus E of the
sample material. Unfortunately, its value for bischofite and carnallite is not
reported in literature. We thus determined these values for the material
tested, by performing ultra-sonic time of flight sound velocity measurements
on unconfined samples of bischofite, carnallite and the mixture under study,
using the facilities at Technical University Delft. The results thus obtained
are given in Table (4.2).
4.2.6 Microstructures (only carnallite)
Three carnallite samples were selected for microstructural study, in
particular with the aim to determine the grain size. The samples were cut
along the cylindrical axis at the middle. Subsequently, the halves were
ground using silicon carbide papers, followed by polishing and etching to
reveal the grain boundaries in reflected light microscopy. The thick sections
of the samples were photographed with a high resolution photographic
camera. The line-intercept method was used to determine values for the
average grain size of the samples.
4.3 RESULTS
The mechanical results of bischofite, carnallite and mixture samples, along
with the experimental conditions have been tabulated in Table (4.1), where
the differential stress values obtained at the end of each step are given. Since
the main aim of this study is to produce a flow law to be applied at in situ
119
conditions, we will assess how the obtained mechanical data fit to
conventional power laws (Van Eekelen et al, 1981, Urai 1983, Ter Heege et
al. 2005b), describing grain size insensitive (GSI – dislocation) creep and/or
grain size sensitive (GSS – diffusion/pressure solution) creep, as given
below
GSI: 𝜀̇ = 𝐴∗𝜎𝑛 (4.3)
GSS: 𝜀 ̇ = 𝐵∗𝜎𝑑−𝑝 (4.4)
where 𝜀̇ is the strain rate, A* is a substituted constant term for Aexp(-Q/RT),
where A is constant (for constant temperature), Q is the activation energy, R
is the gas constant, T is the temperature, B* is also constant at a given
temperature T, σ is the flow stress of the sample, n is the stress exponent, d
is the average grain diameter, p is the grain size exponent.
4.3.1 Bischofite
(i) Stress vs. strain curves
In total, six experiments were performed on bischofite, all at a temperature
of 70 oC. Samples bischofite 5, 6 and 7 were tested at in situ condition of 40
MPa confining pressure, while bischofite 2, 3 and 4 were tested at 70 MPa
confining pressure (see Table 4.1). Figures 4.3a-c show the differential
stress vs. natural strain, d-f show the differential stress vs. time. Bischofite5
was a 7-step strain-rate stepping stress-relaxation experiment with relaxation
periods between every constant strain rate part. Bischofite6 was a 5-step
stepping experiments with stress relaxation followed by each step. The first
two steps of bischofite6 were performed at the same strain rate (10-6
s-1
) but
the stress obtained at the end of the first step was slightly lower than that at
120
(a)
(b)
121
(c)
(d)
122
(e)
(f)
Figure 4.3. Mechanical data bischofite5, 6 and 7, (a-c) differential stress vs. natural
strain (d-f), differential stress vs. time
123
the end of the second step (see Table 4.1), demonstrating that steady state
was not yet reached during the first step. Bischofite7 was a repeat of the
bischofite5 experiment, with the only difference being that the sample was
given comparatively more time for stress relaxation (see Table 4.1). This
test gave reasonably reproducible/overlapping differential stress values at
similar strain rates.
(ii) Effect of confining pressure
Bischofite 2, 3 and 4 were tested at 70 MPa confining pressure. Figures
4.4a-c show the stress-strain curves of these experiments. The stress values
obtained were very similar to the results at 40 MPa confining pressure
(Table 4.1).
(a)
124
(b)
(c)
Figure 4.4. Mechanical data bischofite2, 3 and 4, (a-c) differential stress vs. natural
strain
125
The current experimental data are shown in Figures 4.5a-b. Data from
Urai (1983), at temperatures 60 and 80 oC, at similar strain rate and fixed
confining pressure 28 MPa, are included in the Figure. Comparison of the
results revealed that there is no measurable effect of confining pressure on
the flow stress of bischofite in the range 40-70 MPa.
(a)
(b)
Figure 4.5. Pressure sensitivity of differential stress as compared with Urai (1983),
a) this study (70 oC) vs. Urai’s (60
oC), b) this study (70
oC) vs. Urai’s (80
oC).
126
Table 4.1: Test conditions and results of constant deformation parts, bischofite,
carnallite and mixture
Test
P
[MPa]
𝜀̇
[s-1
]
σ
[MPa]
Experiment
duration
[hrs.]
Bischofite5
40 10-5
5.3
576
40 10-6
3.8
40 10-7
2.3
40 10-8
1.2
40 10-7
2.2
40 10-6
3.4
40 10-5
5.0
Bischofite6
40 10-6
3.2
520
40 10-6
3.8
40 10-5
5.7
40 10-5
5.5
40 10-7
2.3
Bischofite7
40 10-5
5.2
1338
40 10-6
3.4
40 10-7
2.1
40 10-8
1.2
40 10-7
2.0
40 10-6
3.5
40 10-5
5.3
Bischofite2
70 10-6
3.7
45.3 70 10-5
5.9
70 10-4
9.5
Bischofite3
70 10-6
3.4
66.3
70 10-6
3.5
70 10-6
3.5
70 10-5
5.5
70 10-5
5.3
127
Table 4.1: contd.
Test
P
[MPa]
𝜀̇
[s-1
]
σ
[MPa]
Experiment
duration
[hrs.]
Bischofite4
70 10-5
5.5
106
70 10-6
3.0
70 10-7
2.0
70 10-8
---
70 10-7
2.2
70 10-6
3.2
70 10-5
4.9
Carnallite1
40 10-5
20.8
1050
40 10-6
13.2
40 10-7
9.0
40 10-8
5.5
40 10-7
9.0
40 10-6
14.0
40 10-5
24
Carnallite2
40 10-5
21.5
953
40 10-6
13.4
40 10-7
9.0
40 10-8
5.2
40 10-7
9.0
40 10-6
14.5
40 10-5
24.3
Carnallite3
40 10-5
28.1
73 40 10-6
18.8
40 10-7
11.1
Carnallite4 40 10-5
17.4 143
128
Table 4.1: contd.
Test
P
[MPa]
𝜀̇
[s-1
]
σ
[MPa]
Experiment
duration
[hrs.]
Carnallite5
40 10-5
21.9
579
40 10-6
13.8
40 10-7
9.4
40 10-8
6.1
40 10-7
10.2
40 10-6
15.2
40 10-5
23.8
Mixture1
40 10-5
25.7
1014
40 10-6
26.1
40 10-7
23.2
40 10-8
20.6
40 10-7
23.3
40 10-6
27.7
40 10-5
33.3
Mixture2
40 10-5
24.9
611
40 10-6
21.6
40 10-7
16.8
40 10-8
10.8
40 10-7
16.7
40 10-6
21.1
40 10-5
25.2
P is the confining pressure
σ is flow/steady state stress (at the end of each deformation step)
𝜀̇ is strain rate of deformation
(iii) Flow behaviour
All stress values obtained at the ends of the constant strain rate steps (Table
4.1), approaching steady state, are plotted against corresponding
deformation strain rates in Figure 4.6. Note that the dependent variable
(stress) is plotted along the x-axis in this figure, to allow easier comparison
129
with the relaxation data presented below. Best fit linear regression (in log-
log-space) resulted in n = 4.8 ±0.2 when applying the conventional power
law of Equation (4.3).
The steady state stress values have also been used to calculate n-values (cf.
Eq. 4.3) for every individual step in strain rate of experiments bischofite 5-
6-7. The resulting values are plotted in Figure (4.7). The n-value changes
from about 6 at the higher stress to about 4 at the lower end of the stress
range. This shows that a simple power law of the type of Equation (4.3)
might not be applicable to the full data set for bischofite. This will be
explored further in the Discussion part of this Chapter.
Table 4.2. Elastic modulus of the specimens
Sample ρ
[kg m-3
]
Vp
[m s-1
]
Vs
[m s-1
] ν
G
[GPa]
E
[GPa]
Carnallite 1600* 3938 1988 0.33 6.3 16.8
Bischofite 1600* 4312 2037 0.36 6.6 18.0
Mixture 1600* 3970 2207 0.28 7.8 19.8
Halite 2100* -- -- -- -- 39*
ρ: density of the material
Vp: longitudinal component of velocity (measured)
Vs: shear component of velocity (measured)
ν: Poisson’s ratio (calculated)
G: shear modulus (calculated)
E: Young’s modulus (calculated)
*From literature
130
Figure 4.6. Complete data of bischofite tests at 40 MPa confining pressure.
Figure 4.7. n-value at different steady state stress values in bischofite5, 6 and 7.
-8.5
-8.0
-7.5
-7.0
-6.5
-6.0
-5.5
-5.0
-4.5
-4.0
0.0 0.2 0.4 0.6 0.8
Log (
stra
in r
ate[
s-1])
Log (flow stress [MPa])
Bischofite data
0
1
2
3
4
5
6
7
8
-1.0 -0.5 0.0 0.5 1.0
n-v
alue
Log (flow stress [MPa])
n = 4.8 ±0.2
131
(iv) Stress relaxation
The stress relaxation results obtained for the individual bischofite
experiments at 40 MPa confining pressure are plotted in Figures 4.8a-j. The
steady state stress values obtained at the end of each strain rate step are
included in the graphs and form the starting points for relaxation. Results at
similar strain rate are combined. Generally, the graphs show that the
evaluated plastic strain rate at the start of each relaxation fit to a trend with
relatively high n-value (n ~ 5), implying that stress is relatively insensitive
to strain rate. During relaxation then, the stress-strain rate sensitivity
gradually increases and the n-value appears to approach n = 1 near the end
of each relaxation step. Strikingly, the individual relaxation curves, starting
at a particular strain rate and gradually decreasing in strain rate, do not pass
through the steady state values obtained during the constant strain rate parts
at lower rates. If compared at the same strain rate, the strength of the
material during relaxation is always less than that during the constant strain
rate part. It is noted that occasionally, there were stick slip events during the
steps, occurring due to piston/seal friction (see irregularities in the curves).
This resulted in apparent sudden rise in calculated plastic strain rate, but it
did not change the trend as it regained its progression toward low n-value
after every event.
132
(a) (b) (c)
(d) (e) (f)
133
(g) (h) (i)
(j)
Figure 4.8. Stress relaxation curves bischofite tests at P = 40 MPa, for
comparison, the relaxation curves of similar strain rate deformation steps
are combined, a) Bischofite5, step1 and step7 𝜀̇ = 10-5
s-1
, b) Bischofite5,
step2 and step6 𝜀̇ = 10-6
s-1
, c) Bischofite5, step3 and step5 𝜀̇ = 10-7
s-1
, d)
Bischofite5, step4 𝜀̇ = 10-8
s-1
, e) Bischofite6, step1 and step2 𝜀̇ = 10-6
s-1
, f)
Bischofite6, step3 and step4 𝜀̇ = 10-5
s-1
, g) Bischofite7, step1 and step7 𝜀̇ =
10-5
s-1
h) Bischofite7, step2 and step6 𝜀̇ = 10-6
s-1
, i) Bischofite7, step3 and
step5 𝜀̇ = 10-7
s-1
, j) Bischofite7, step4 with strain rate 10-8
s-1
134
4.3.2 Carnallite
(i) Stress vs. strain curves
Five carnallite samples were tested and their stresses vs. natural strain data
(Table 4.1) are shown in Figures 4.9a-e. Carnallite1 and 2 were both 7-step
strain-rate stepping tests with three relaxation periods after constant strain
rate deformation steps at 10-6
, 10-8
and 10-6
s-1
, respectively. Carnallite3 was
a 3-step stepping test without relaxation, carnallite4 was a single step
deformation test followed by stress relaxation, and carnallite5 was a repeat
experiment of carnallite 1 and 2. The samples showed strain rate sensitivity
of stress, and strengths at the end of similar (repeat) strain rates were
slightly higher for steps at higher total strains (see Table 4.1). The strengths
of carnallite1, 2 and 5 are in good agreement with each other, demonstrating
reproducibility. Carnallite3 was comparatively too strong when compared
with carnallite 1, 2 and 5 at similar strain rates, while carnallite4 was found
to be slightly weaker than the other samples.
(ii) Flow behaviour
All stress values obtained at the ends of the constant strain rate steps (Table
4.1), approaching steady state, are plotted against strain rate in Figure 4.10.
As was the case in Figure 4.6, the dependent variable (stress) is plotted
along the x-axis in this figure, to allow easier comparison with the
relaxation data presented below. Best fit linear regression (in log-log-space,
Fig. 4.11) resulted in n = 5.1 ±0.3 when applying the conventional power
law of Equation (4.3).
135
(a)
(b)
136
(c)
(d)
137
(e)
(f)
138
(g)
(h)
139
(i)
(j)
Figure 4.9. Mechanical data carnallite1-5, (a-e) differential stress vs. natural strain,
(f-j) differential stress vs. time.
140
Figure 4.10. Steady state stress values of carnallite samples against the strain rate on
log space. Note: the strain rate is independent in this work, but it is plotted on
dependent axis (Y-axis) to compare with the relaxation data.
Figure 4.11. n-values at different steady state values of carnallite experiments
-8.5
-8.0
-7.5
-7.0
-6.5
-6.0
-5.5
-5.0
-4.5
-4.0
0.5 0.7 0.9 1.1 1.3 1.5
Log (
stra
in r
ate[
s-1])
Log (stress [MPa])
Carnallite all data
0
1
2
3
4
5
6
7
0.5 0.7 0.9 1.1 1.3 1.5
n-v
alu
e
Log Stress [MPa]
n = 5.1 ±0.3
141
(iii) Stress relaxation
The stress relaxation results obtained for carnallite tests are plotted in Figure
4.12. The steady state values obtained at the end of each deformation step
are included in the graphs and form the starting points of the relaxation. For
comparison, similar strain rate deformation steps are combined. Generally,
the graphs show that the evaluated plastic strain rate at the start of each
relaxation fit to a trend with relatively high n-value (n ~ 5), implying that
stress is relatively insensitive to strain rate. During relaxation then, the
stress-strain rate sensitivity gradually increases and the n-value appears to
approach n = 1 near the end of each relaxation step. The relaxation curves
found not passing through the steady state values obtained during the
constant deformation steps, but stay on lower stress side. i.e. the strength of
the material is low during relaxation if compared at similar strain rates. It is
noted that occasionally, there were stick slip events during the steps,
occurring due to piston/seal friction (see irregularities in the curves). This
resulted in apparent sudden rise in calculated plastic strain rate, but it did not
change the trend as it regained its progression toward low n-value after
every event.
142
(a) (b) (c)
(d) (e) (f)
143
(g)
Figure 4.12. Stress relaxation curves bischofite tests at P = 40 MPa, for comparison, the relaxation curves of similar strain rate
deformation steps are combined, a) Carnallite1 step2 and step6 𝜀̇ = 10-6
s-1
, b) Carnallite1 step4 𝜀̇ = 10-8
s-1
, c) Carnallite2 step2 and
step6 𝜀̇ = 10-6
s-1
,d) Carnallite2 step4 𝜀̇ = 10-8
s-1
, e) Carnallite4; The only relaxation step 𝜀̇ = 10-5
s-1
, f) Carnallite5 step2 and step6 𝜀̇ =
10-6
s-1
, g) Carnallite5 step4 𝜀̇ = 10-8
s-1
,
144
(iv) Microstructures
Samples carnallite1, 3 and 4 and one undeformed sample have been used for
microstructural analysis. The undeformed microstructure (Fig. 4.13a) shows
a dense aggregate of grains with an average grain size of 4.9 mm. The grain
size resulting at the end of the multi-step (strain rate stepping and
relaxation) experiment carnallite1 was measured to be 2.3 mm. Carnallite3
only included strain rate steps, no relaxation, and resulted in an average
grain size of 3.5 mm. Carnallite4 was one step (10-5
s-1
) deformation
experiment followed by relaxation. The grain size measured is 5.4 mm. The
measurements are summarized in Table (4.3).
Table 4.3 Measured grain sizes of different carnallite samples
Sample 𝜀̇
[s-1
]
σ
[MPa]
d
[mm]
Stress
Relaxation
ε
Undeformed -- -- 4.865 -- --
Carnallite1 10-5
24 2.335 No 0.16
Carnallite3 10-7
11.1 3.535 No 0.96
Carnallite4 10-5
17.4 5.425 Yes 0.54
𝜀̇: deformation strain rate
σ: maximum/steady stress at the end of the corresponding strain rate step
d: measured grain size from microstructure of sample, using line intercept method
ε: maximum natural strain at the end of the constant strain rate deformation
(a)
10 mm
145
(b)
(c)
10 mm
10 mm
146
(d)
Figure 4.13. The microstructures of carnallite samples, a) carnallite undeformed;
showing dense aggregate of grains in grey scale, with average grain size 4.9 mm, b)
Carnallite1, σ = 24 MPa (at the end of experiment), deformation strain rate 𝜀̇= 10-5
s-1
, without relaxation, showing deformed (slightly flattened) grains by deformation,
compression direction is horizontal, c) carnallite3, σ = 11.1 MPa (at the end of
experiment), deformation strain rate 𝜀̇= 10-7
s-1
, without relaxation, compression
direction horizontal. The grain boundaries look slightly opened up, which might be
due to handling of sample while preparing for microstructural analysis, d)
Carnallite4 (after stress relaxation), deformation strain rate 𝜀̇= 10-5
s-1
, compression
direction horizontal. The grains appear slightly grown during relaxation. The
average grain size showing is 5.4 mm, which is bigger than other three measured
samples (i.e. undeformed, carnallite1 and 3).
4.3.3 Mixture samples of bischofite, carnallite and halite
(i) Stress vs. strain curves
Two multistep experiments were performed on mixed bischofite-carnallite-
halite samples. mixture1 was a 7-step strain-rate stepping test with three
relaxation periods after constant strain rate steps at 10-6
, 10-8
and 10-6
s-1
,
respectively (see Table 4.1). The sample showed strain hardening in all
10 mm Fracture line during sample preparation
147
constants strain rate steps (Fig. 4.14a), except for the one at the lowest rate.
Mixture 2 was also a 7-step strain-rate stepping test with three relaxation
periods (Table 4.1), following the same sequence of steps as applied in
mixture1 (Fig. 4.14a). The sample mixture2 contained a distinct carnallite
band of ~14 mm width, appearing as a ~30o tilted disc in the middle of the
sample (see Figure 4.15b). Mixture 2 was found to be weaker than mixture
1, and did not show the strain hardening as observed in mixture 1, but rather
approached steady state stress behaviour in individual steps (Figure 4.15a).
At the end of the experiment, the sample was taken out of the testing
machine and localized strain along the pre-existing carnallite band (Figure
4.15b) was observed. This localized shear strain at the carnallite band most
probably caused some leakage of the rubber jacket in the final stage of the
experiment, which allowed some silicon oil to effuse through jacket and
contaminate the sample.
(ii) Flow behaviour
The sensitivity of stress to strain rate of the two mixtures is illustrated in
Figures 4.16a-b, plotting the stress values at the end of each step as a
function of strain rate, again with strain rate along the y-axis. Figure 4.16a
quite clearly illustrates the hardening behaviour of mixture1 and shows that
the sensitivity of the differential stress to strain rate is rather low (high n-
value). Figure 4.16b shows that the flow behaviour of mixture2 does not
follow a linear trend in log-log space. Expressed using the power law n-
value (Eq. 4.3), the trend is from n ~ 10 at the higher stress and faster strain
rate, to n ~ 5 at lower stress and slower strain rate.
148
(a)
(b)
Figure 4.14. a) Stress-strain curve for mixture1, b) mixture1 sample after
deformation, loading direction vertical.
10 mm
149
(a)
(b)
Figure 4.15. a) Stress strain curve for mixture2, b) mixture2 sample before and after
deformation, loading direction vertical.
10 mm
150
(a) (b)
Figure 4.16 Steady state (or maximum) stress values at each step, showing the strain
history. a) Mixture1 has higher stress values towards higher strain; b) mixture2 has
reproducible stress values for similar strain rates. Note: Scale along stress axis is
not same and mixture1 is stronger at the end, if compared at similar strain rates.
151
(iii) Stress relaxation
The stress relaxation data of mixtures are shown in Figures 4.17(a-d). The
steady state values obtained at the end of each deformation step are included
in the graphs and from the starting points of relaxation. For comparison,
similar strain rate steps are combined. At start of relaxation, the n-value is
high (n > 10) showing steep slope, implying that the stress is relatively
insensitive to strain rate. During relaxation then, the n-value decreases and
appears to approach n = 1 near the end of each relaxation step. Distinctly,
mixture1 relaxation curves progressively, pass through the steady state
points obtained during deformation steps (compare Figs. 4.17a and c),
pointing that the mixture2 shows weaker behaviour during relaxation than
during deformation. A similar trend was observed during stress relaxation of
bischofite and carnallite (compare with Figs. 4.8 and 4.12). Note that the
relaxation curves showed crest-troughs due to day and night temperature
rhythm, which is quite substantial.
152
(a) (b)
(c) (d)
Figure 4.17. Stress relaxation behaviour of mixture samples. a-b) mixture1, c-d) mixture2. Note the x-axis (Log stress) scale of
mixture 2 is different from that of mixture1.
153
4.3.4 Elemental analysis using micro X-Ray Fluoroscopy (μ-XRF)
In order to obtain some first-order insight into the composition of the
samples used in the experiments, the micro XRF (X-ray Fluorescence)
technique was applied. In total 5 number of samples were analysed;
carnallite1, 3 and 5, mixture1 and 2.
As a first step, a relatively large cross-section of the sample was mapped
for different elements. This gave a qualitative overview. After this step, one
smaller rectangular area per sample (20 × 5 mm2), taken as representative of
the whole sample (e.g. mixture1 and 2, Figs. 4.18a-b), was selected for
detailed elemental analysis. The results for the selected areas are visualized
in Figure 4.19. In the figures, darker colour reflects higher wt. % of the
contributing salts. The figures show that carnallite samples (1, 3 and 5)
contain; higher percentage of carnallite (KMgCl3), i.e. 64 to 66 wt. %,
bischofite 17 to 19 wt. % and halite as 6 to 8 wt. % and other trace elements.
The mixture1 has higher halite wt. % (65%), the next abundant mineral is
bischofite (14.15%) followed by carnallite (4.13%) and traces are 14.18%.
Mixture2 has got bischofite (18.6%), halite (29.8%) and carnallite (36.8 wt.
%). In short, all samples contain the neighbour minerals and not very pure in
composition. In mixture1, the wt. % of halite is rather high, 65.4%, while
bischofite is 14.2%, carnallite is 4.1% and the remaining elements are
considered as trace elements as their contribution is very limited. In
mixture2, the halite composition is less than in mixture1, and is ~30%,
carnallite is slightly higher than in mixture1, ~37%, and bischofite is 18.6%.
154
(a)
(b)
Figure 4.18. Selection of area for detailed inspection of elements for micro XRF. a)
mixture1,b) mixture2
10 mm
10 mm
155
Carnallite1
156
Carnallite3
157
Carnallite5
158
Mixture1
159
Miixture2
Figure 4.19. Micro XRF analyses of the samples showing composition in wt. %.
Bischofite (MgCl2), Carnallite (KMgCl3), Halite (NaCl)
160
4.4 DISCUSSION
4.4.1 Mechanical behaviour
The stress strain plots of the three salts tested; bischofite (Figures 4.3 and
4.4), carnallite (Figure 4.9) and their mixture including halite (Figures 4.14
and 4.15) showed that these materials have different strengths for similar
strain rate and temperature. Bischofite was found the weakest in the series.
Carnallite appeared stronger than bischofite and the mixture was found even
stronger than carnallite. Note that mixture1 showed strain hardening
throughout the test and was found the strongest in the series, while mixture2
approached steady state behaviour. The stress values reached at the end of
the constant strain rate tests were related to the strain rates via the standard
power law of Equation (4.3), resulting in n = 4.8 ±0.2 for bischofite, n = 5.1
±0.3 for carnallite, and n ~ 5-10 for mixtures. Commonly, n-values ranging
3 to 4.5 are related to dislocation climb as the rate controlling mechanism,
involving bulk vacancy diffusion though the lattice (Weertman 1968,
Sherby & Weertman 1979, Poirier 1985), where the n-value may rise to 6.5
if pipe diffusion along dislocation cores plays a controlling role (Frost &
Ashby 1982, De Bresser 1991). Climb controlled dislocation creep is a grain
size insensitive (GSI) mechanism. The stress relaxation trend of all three
materials showed rather low sensitivity of stress to strain rate at the start (n
~ 5 for bischofite and carnallite and n > 5 for mixtures), but this sensitivity
changed towards the end of relaxation, when the n-value approached 1.
Such low n-values are usually interpreted as indicating grain size sensitive
(GSS) behaviour.
We will now first consider the flow law for bischofite and carnallite
taking only the data resulting from the constant strain rate tests into account,
161
i.e. representing the GSI behaviour of the materials, and then discuss how to
use the relaxation data to come to a more complete description of the salts
and their mixture. Two important elements need to be taken into account in
this discussion: i) the gradual change to n ~ 1 observed towards slow strain
rate during relaxation, and ii) the observations that the relaxation curves of
all three salts tested, gradually covering slower strain rates, did not pass
through the steady state values determined at slower rates than where
started.
(i) Creep law for bishofite using stresses at the end of constant
strain rate steps
The stress exponent n determined for every individual step in strain rate of
experiments Bischofite 5-6-7 (Fig. 4.8) changed from about 6 at the higher
stress to about 4 at the lower end of the stress range, showing that a simple
power law of the type of Equation (4.3) might not be applicable. Rather, a
general flow equation may be used (De Bresser et al. 2002, Renner et al.
2002), of the type
𝜀̇ = 𝐴𝜎𝑛𝑒𝑥𝑝 (𝜎
𝐵) 𝑒𝑥𝑝 (
−𝑄
𝑅𝑇) 𝑑−𝑝 (4.5)
where A and B are constants, Q represents the activation energy of the rate
controlling process, and R is the gas constant. The exp(σ/B) term in this
equation makes that stress-strain rate data plotted in log-space do not follow
linear trends. Comparing the current bischofite data at temperature 70 oC (as
N70) with Urai’s 1983 at temperatures 60 and 80 oC (as U60 and U80),
Figure 4.20, we see that the trend of N70 is slightly different, as it shows
slightly weaker behaviour for faster strain rates (10-6
s-1
and higher) and
stronger behaviour for slower strain rates (10-7.5
s-1
and beyond).
162
Figure 4.20. Bischofite current steady state stress values (at 70 oC) comparison
with Urai's 1983 (at 60 and 80 oC).
For constant temperature conditions, the second exponential term drops out
and assuming, for now, grain size independent behaviour, Equation (4.5) is
reduced to
𝜀̇ = 𝐴′𝜎𝑛𝑒𝑥𝑝 (𝜎
𝐵) (4.6)
Taking the stress-strain rate data at steady state (Table 4.1) and applying
non-linear regression analysis, the following best fit values of the
parameters included in Equation (4.6) were found: n = 3.4, 𝐴′= 10(-8.519)
, and
B = 2.261 (correlation coefficient, R2 = 0.99). According to this best fitting,
the n-value approaches 3.4 at low stress of 0.1 MPa and increases to ~ 6 at
higher stress (~0.6 MPa, cf. Fig. 4.7).
-10
-9
-8
-7
-6
-5
-4
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Log (
stra
in r
ate
[s-1
])
Log (flow stress [MPa])
U60 U80 N70
163
Of course, in this analysis, it is assumed that a single GSI creep law
applies to the data. Later this discussion, we will discuss the possibility that
a composite creep law combining GSI and GSS might suit the observations
better.
(ii) Creep law for carnallite using stresses at the end of constant
strain rate steps
The conventional power law of the type of Equation (4.3) can be written
in full as:
𝜀̇ = 𝐴𝑜𝑒𝑥𝑝 (−𝑄
𝑅𝑇) 𝜎𝑛 (4.7)
where Ao is a constant, Q represents the activation energy of the rate
controlling process, and R is the gas constant. Since all our experiments
were carried out at the same temperature (70 oC), our data do not allow
determining a value for the activation energy Q. However, we can compare
our data with the results of Urai (1985) on carnallite. Using the steady state
(or the maximum) stress values for each step and plotting these against the
strain rate in the corresponding step, all data are plotted in Figure 4.21. The
graph shows that the current results are in good agreement with Urai’s wet
sample results. Urai (1985) obtained n = 4.5 ±1.0 for steady state flow of
wet carnallite. The value obtained for the stress exponent n in the current
work is 5.1 ± 0.3, hence is in agreement with the result of Urai. More
importantly, it is a far more accurate determination of n than previously
established. Given the agreement of our n-value with that of Urai (1985),
taking the errors into account, it is justified to apply the Q-value determined
by Urai to our data, which is 83700 (± 20000) J/mole.
164
Figure 4.21. Current study carnallite data vs. Urai (1985) dry and wet carnallite
sample data.
Taking the steady state data of all carnallite samples, the temperature
used T = 70 oC (343 K), n = 5.1 and Q = 83700 Jmol
-1, a value of 6.4 (MPa
-
5.1 s
-1) was thus determined for Ao. With these values for Ao, n and Q,
Equation (4.7) can now be used to describe the steady state flow of
carnallite at various temperatures:
𝜀̇ = 6.4 𝑒𝑥𝑝 (−83700
𝑅𝑇) 𝜎5.1 (4.8)
For the in situ temperature of 70 oC (343 K), Equation (4.8) reduces to:
𝜀̇ = 1.13 × 10−12𝜎5.1 (4.9)
-10
-9
-8
-7
-6
-5
-4
0.0 0.5 1.0 1.5 2.0
Log (
stra
in r
ate
[s-1
])
Log (flow stress [MPa])
Carnallite
Urai_Dry
Urai_B1.15%
This study
165
(iii) Comparison of steady state values
For a comparison of the strengths measured as steady state values of
bischofite, carnallite and mixture, averaged values at similar strain rates
were used to calculate strength ratios and plotted in Figure 4.22. The
comparison shows that the strength ratio is higher at slower strain rates.
Carnallite is 4.3 times and mixture is 5.1 times stronger than bischofite at a
strain rate of 10-5
s-1
; this ratio increases towards slower deformation strain
rates, resulting in values 4.7 times and 13 times at strain rate of 10-8
s-1
,
respectively The strength of the mixture is 1.2 times larger than that of
carnallite for a strain rate of 10-5
s-1
and 2.8 times at 10-8
s-1
. If we compare
our work with previous studies on halite by Heard (1972), Wawersik &
Zeuch (1986), Carter (1993) and Ter Heege et al. (2005a), we see (Figure
4.23a) that the upper limit (highest stress at a given strain rate) is
constrained by the flow law by Heard (1972 - confined tests on
polycrystalline natural halite aggregates) at the temperature of 70 oC.
Carnallite is weaker and lies around the flow law given by Wawersik &
Zeuch (1986), whereas the bischofite is clearly the weaker end in the family.
(iv) Relating the relaxation behaviour to the creep at near steady
state
We observed that the behaviour of bischofite, carnallite and (notably)
mixture2 during relaxation is different from that during the constant strain
rate parts of the testing; in all three multi-step experiments, the material was
weaker during relaxation than during deformation at constant strain rate, if
compared at the same strain rate or stress. In addition, if the dependence of
strain rate on stress is considered by means of the power law n-value (Eq.
166
Figure 4.22. Strength ratios of mixture by taking bischofite and carnallite as
reference. B/B: bischofite to bischofite, C/B: carnallite to bischofite, M/B: mixture
to bischofite and M/C is mixture to bischofite strength ratio at similar strain rates.
(a)
0
2
4
6
8
10
12
14
1.E-09 1.E-07 1.E-05
Str
ength
rat
io
Strain rate [s-1]
B/B
C/B
M/B
M/C
-14
-12
-10
-8
-6
-4
-2
0.0 0.5 1.0 1.5 2.0
Lo
g (
stra
in r
ate
[s-1
])
Log(flow stress [MPa])
Bischofite Carnallite
Mixture1 Mixture2
Ter Heege 2005a Heard 1972
Carter 1993 Wawersik&Zeuch1986
167
(b)
Figure 4.23 a) Projected curves for flow laws of wet halite from previous studies, b)
steady state values comparison of bischofite, carnallite and mixture (with halite%
composition)
4.3), the n-value during relaxation is lower than at constant strain rate, again
if compared at the same strain rate or stress.
During relaxation, when changes in strain are very limited, the
microstructure may be expected to remain constant (see Rutter & Mainprice
1978), while processes such as dynamic (syn-deformational)
recrystallization and grain growth are likely to affect the microstructure
during deformation covering substantial changes in strain (De Bresser et al.
2001, Ter Heege et al. 2005a). The relaxation data suggest that dependence
of strain rate on stress may approach linearity (i.e. n-value approaching 1),
which suggests that grain size sensitive behaviour might play a role at low
stress (Spiers et al. 1990, Ter Heege et al. 2005a). We thus speculate that
during relaxation, the creep behaviour of bischofite, carnallite and mixtures
-14
-12
-10
-8
-6
-4
-2
0.0 0.5 1.0 1.5 2.0
Log (
stra
in r
ate[
s-1])
Log ( flow stress[MPa])
Bischofite Carnallite Mixture1 Mixture2
168
might go through a transition from GSI creep (Eq. 4.3) at the faster strain
rates (at the start of relaxation) to GSS creep (Eq. 4.4) at relatively slow
rates. This transition then would occur at constant structure, notably at
constant grain size, while the transition from GSI to GSS creep might be
prevented during steady state creep when microstructural modification is
effective. Since our study did not systematically involve experiments on
materials with different grain size, we explore the possible role of GSS
creep using predictions on the basis of recrystallized grain size piezometers
for the salts under consideration. In a recrystallized grain size piezometer,
the size of the grains is directly related to the differential stress (e.g. Twiss
1977, Shimizu 2008, De Bresser et al. 2001) according to:
𝑑 = 𝐾𝜎 −𝑚 (4.10)
Or
𝜎 = 𝐾1
𝑚 𝑑−1
𝑚 (4.11)
where K and m are material and mechanism specific constants. This
relationship between stress and recrystallized grain size is generally
assumed to be independent of temperature, although that there is evidence
that this assumption is not generally valid (de Bresser et al. 2001).
Bischofite
Van Eekelen et al. (1981) measured the recrystallized grain size in a number
of bischofite samples experimentally deformed at 60 °C (Figure 4.24). We
used their data to calibrate the values for K and m in Equations (4.10) and
(4.11) by simply applying
169
Figure 4.24. Flow stress vs. mean grain size at 60oC
𝑑 = 𝐾(𝜎)−1
𝑠𝑙𝑜𝑝𝑒⁄ (4.12)
The resulting Piezometric relation for bischofite is:
𝑑 = 4.725𝜎 −1.15 (4.13)
Using the piezometer of Equation 4.13, we can now estimate the
recrystallized grain size at any stress, under the assumption the dynamic
recrystallization was fully effective at the conditions imposed. We have
done this for the average stresses relevant for our bischofite samples at the
start of the relaxation periods, after steps at constant strain rate of 10-5
, 10-6
,
10-7
and 10-8
s-1
. The results are given in Table (4.4). The role of grain size
can now be further evaluated by (1) assuming that during relaxation the
grain sizes will remain the same, and (2) that a GSS creep law of the type of
Equation 4.4, with n = 1, applies at the final stages of relaxation, i.e. at the
y = -0.8686x + 0.5858
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
-1.0 -0.5 0.0 0.5
Log (
flow
str
ess
[MP
a])
Log (grain size [mm])
Van Eekelen et al. 1986 bischofite data
170
Figure 4.25. Log strain rate values picked from relaxation curves at similar stress of
1.0 MPa
very slow strain rates, This allows calculating the p-value (Equation 4.4).
Taking logarithm and simplifying the Equation (4.4) we get
𝑙𝑜𝑔𝜀 ̇ = 𝑖𝑛𝑡𝑒𝑟𝑐𝑒𝑝𝑡 − 𝑝𝑙𝑜𝑔𝑑 (4.14)
The intercept includes the stress and is only a constant if data are
considered at single value of stress. For this, we picked σ = 1.0 MPa (i.e.
Logσ = 0) in the stress relaxation curves (Figures 4.8(a-j)), determined the
strain rate values at the various relaxation curves, and plotted the results, in
log-log space (cf. Eq. 4.14, see Fig. 4.25), against the values for the grain
size from Table (4.4). Then using Equation 4.14, the grain size exponent ‘p’
can be determined as from the slope of the best fit line (Figure 4.25).
As is clear from the Figure 4.25, the scatter is quite substantial, so the
resulting value for p, 0.8 ±0.2, must be regarded as only a crude estimate.
Nevertheless, the result confirms the suggestion that grain size sensitive
y = -0.7539x - 7.4049
-8
-7.9
-7.8
-7.7
-7.6
-7.5
-7.4
-7.3
-7.2
-7.1
-7
-0.5 0 0.5 1
Lo
g (
stra
in r
ate
[s-1
])
Log (grain size [mm])
Bischofite this study
171
Table 4.4: Estimated grain size from piezometer
Salt 𝜎𝑎𝑣𝑒𝑟𝑎𝑔𝑒
[MPa]
𝜀̇ [s
-1]
d
[mm]
Bis
cho
fite
5.3 10-5
0.69
3.5 10-6
1.10
2.2 10-7
1.91
1.2 10-8
3.83
Car
nal
lite
22.7 10-5
1.81
14.7 10-6
2.99
9.5 10-7
4.92
5.6 10-8
9.06
𝜎𝑎𝑣𝑒𝑟𝑎𝑔𝑒 : average stress value for similar strain rate of deformation,
taken from Table 4.1
𝜀̇: deformation strain rate
d: estimated grain size from piezometer
behaviour is likely to play a role in the flow behaviour of bischofite, at least
during the relaxation. A possible mechanism controlling flow in this GSS
regime is dissolution or precipitation controlled pressure solution creep (e.g.
Spiers et al. 1990, Schutjens 1991). A possible mechanism controlling flow
in this GSS regime is dissolution or precipitation controlled pressure
solution creep (e.g. Spiers et al. 1990, Schutjens 1991). According to these
studies, the inter-granular pressure solution creep (IPS) is a serial process
comprising of three steps; dissolution of material at high stress zones,
transportation via diffusion to lower stress zones and precipitation at low
stress zones. The slowest of these will determine the creep rate. The creep
rate is inversely related to the grain size (‘d-p
’), where p = 1, 2 and 3 are the
sensitivities of the rate controlling mechanism to grain size, defining
dissolution, diffusion and precipitation respectively.
It cannot be excluded that during the long time spans needed for
relaxation down to slow strain rates, some microstructural modifications has
172
taken place, notably (static) grain growth. This then undermines our
assumption of constant structure. However, the range of (predicted) grains
sizes in our samples is rather limited (Table 4.4), the durations of the
various relaxation steps are quite comparable, and all tests were done at the
same temperature. Consequently, the possible change in grain size during
relaxation due to grain growth will be almost the same for all steps, hardly
affecting estimates of the value for p.
Using the intercept value of the best fit line (Fig. 4.25), the unknown
constant B* of Equation (4.4) was calculated to be 3.94×10-8
(MPa-1
mm0.8
s-
1). Hence, the GSS flow law for bischofite at 70
oC, with grain size d in mm
is
𝜀�̇�𝑆𝑆 = 3.94 × 10−8𝜎𝑑−0.8 (4.15)
Carnallite
No recrystallized grain size Piezometric relation is available in the literature
for carnallite. We plotted our grain size data of two carnallite samples as a
function of stress together with the data for halite (Ter Heege et al. 2002b)
and bischofite (Van Eekelen et al. 1981) in Figure 4.26. Bischofite and
halite follow very comparable trends, but the grain size of carnallite at a
given stress is substantially larger than that of halite and bischofite. We
assumed that the trend observed for halite and bischofite also holds for
carnallite, coming from the same family of materials. We thus suggest that
the recrystallized grain size Piezometric relation for carnallite (Fig. 4.26) is:
𝑑 = 65.92𝜎 −1.15 (4.16)
173
Figure 4.26. Carnallite measured grain sizes in log space with halite and bischofite
Figure 4.27. Strain rate values picked from relaxation data plotted against stress
equivalent grain size from piezometer
Now doing the same exercise for carnallite as carried out for bischofite,
applying Equations 4.4 and 4.14 under the assumption of constant structure
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
-1.5 -0.5 0.5
Log (
flow
str
ess
[MP
a])
Log (grain size [mm])
Bischofite Halite
Carnallite Estimated grain size
y = -1.079x - 7.3413
-9
-8.8
-8.6
-8.4
-8.2
-8
-7.8
-7.6
-7.4
-7.2
-7
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Lo
g (
stra
in r
ate
[s-1
])
Log (grain size [mm])
Carnallite
174
(grain size) during relaxation, we can obtain a crude estimate for the p-value
for carnallite. We picked σ = 4.0 MPa (i.e. Logσ = 0.6) in the stress
relaxation curves (Figures 4.12a-g), determined the strain rate values at the
various relaxation curves, and plotted the results, in log-log space against
the values for the grain size from Table (4.4), see Figure 4.27. The resulting
value for p was: 1.0 ±0.26.
Using the intercept value of the best fit line (Figure 4.27), the unknown
constant B* of Equation (4.4) was calculated to be 1.01×10-8
(MPa-1
mm1.0
s-
1). Hence, the GSS flow law for bischofite at 70
oC, with grain size d in mm
is
𝜀�̇�𝑆𝑆 = 1.01 × 10−8𝜎𝑑−1.0 (4.17)
In literature, the value p = 1 is suggested for dissolution or precipitation
controlled pressure solution creep), while p = 3 for diffusion controlled
pressure solution creep (e.g. Raj, 1982).
Mixture
The mixture samples have shown a change in the trend of relaxation
comparable with bischofite and carnallite; starting with a relatively high n-
value at the higher stress and approaching n = 1 towards the end of the
relaxation curve. Hence, the suggestion made for bischofite and carnallite
that a transition occurs from GSI to GSS creep during relaxation may also
hold for the mixtures. The relaxation curve of mixture2 (halite 30%,
carnallite 37%, bischofite 18.6%) does not pass through the steady state
stress values obtained during constant deformation steps, at slower strain
rate, whereas the relaxation curve of mixture1 (halite 65%, carnallite 4%,
175
bischofite is 14%) does seem to pass through the slower strain rate steady
state points. We infer that the mixture2 has experienced grain size evolution
during relaxation. The data set on the mixtures is not extensive enough to
allow a meaningful definition of the creep behaviour in terms creep GSS
equation cf. Equation 4.4.
Table 4.5: Data picked from relaxation curves at similar differential stress values
Sample Deformation
strain rate [s-1
] 𝜀̇
[s-1
]
LOG
(𝜀̇[s-1])
σ
[MPa]
LOG
(σ[MPa])
Bischofite5
10-5
8.6×10-8
-7.1 1.0 0
10-6
3.0×10-8
-7.5 1.0 0
10-7
1.6×10-8
-7.8 1.0 0
10-8
1.4×10-8
-7.9 1.0 0
Bischofite6 10
-5 5.3×10
-8 -7.3 1.0 0
10-6
2.1×10-8
-7.7 1.0 0
Bischofite7
10-5
5.1×10-8
-7.3 1.0 0
10-6
4.2×10-8
-7.4 1.0 0
10-7
3.1×10-8
-7.5 1.0 0
10-8
2.0×10-8
-7.7 1.0 0
Carnallite1
10-6
9.3×10-9
-8.0 4.0 0.6
10-8
4.7×10-9
-8.3 4.0 0.6
10-6
1.2×10-8
-7.9 4.0 0.6
Carnallite2
10-6
1.9×10-8
-7.7 4.0 0.6
10-8
4.1×10-9
-8.4 4.0 0.6
10-6
2.4×10-8
-7.6 4.0 0.6
Carnallite4 10-5
3.0×10-8
-7.5 4.0 0.6
Carnallite5
10-6
1.2×10-8
-7.9 4.0 0.6
10-8
4.0×10-9
-8.4 4.0 0.6
10-6
5.1×10-9
-8.3 4.0 0.6
𝜀̇ : the strain rate values picked from stress relaxation curves at fixed differential
stress value (6.3 MPa)
Deformation strain rates are the values used to deform the sample and set the stress
value for relaxation.
σ : the fixed differential stress selected to pick the data from relaxation curves
176
(v) Composite flow laws
In the above, we have established creep laws for the GSI deformation
behaviour of bischofite and carnallite, Equations (4.6) and (4.9)
respectively, and for the GSS behaviour, (4.15) and (4.17) respectively. The
basis for determining these flow laws were the data obtained during the
deformation at constants strain rate, providing near steady state stresses, and
the analysis of the relaxation curves under the assumption of constant
structure. We will now evaluate if combining GSS and GSS creep in the
form of a composite creep law is of added value. We start from:
𝜀̇ = 𝐴∗𝜎𝑛 + 𝐵∗𝜎𝑑−𝑝 (4.18)
Equation (4.18), however, cannot simply be regarded as the sum of the GSI
and GSS creep laws defined separately, since grain size sensitive behaviour
might have influenced the steady state creep while this was not taken into
account in performing the best fitting exercise. In other words, the
established GSI creep law might not hold for the complete stress – strain
rate range covered. We assume now that at the highest stress, GSI creep is
robust. Regression analysis taking only the data into account at strain rate
10-6
and 10-5
s-1
, i.e. at the higher stresses, then results in n = 5.4 ±0.4 for
bischofite and n = 5.3 ±0.7 for carnallite. Using these new values for the
stress exponent n, holding for the GSI part of the composite Equation 4.18,
and taking the established GSS flow equations, non-linear regression best
fitting resulted in values for A* of Equation 4.18.
For bischofite:
𝜀̇ = 1.1 × 10−9𝜎5.4 + 3.94 × 10−8𝜎𝑑−0.8 (4.19)
177
For carnallite:
𝜀̇ = 3.70 × 10−13𝜎5.3 + 1.01 × 10−8𝜎𝑑−1 (4.20)
The trends representing composite flow for bischofite and carnallite
following Equations 4.19 and 4.20 are shown in Figures 4.28 and 4.29.
The trend lines in Figures (4.28-4.29) show that the influence of grain
size is effective at lower stresses and strain rates. For higher stresses, these
curves satisfy the steady state points, whereas on lower stress/strain rates,
these trends satisfy the data picked from relaxation curves. In Figure 4.28b,
the trend lines to Urai’s (1983) 60 and 80 oC data are also plotted along
using best fit polynomial. The two isotherms seem to present lower and
upper limits of the composite flow, whereas the grain size (Eq. 4.19) looks
to play role to suggest the strain rate at lower stress region.
(a)
-10
-9
-8
-7
-6
-5
-4
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Lo
g (
stra
in r
ate
[s-1
])
Log (flow stress [MPa])
Bischofite steady state data
Data picked at 0 from relaxation curves
composite 1
composite 2
composite 3
composite 4
178
(b)
Figure 4.28. a) Trends resulting from the composite GSI + GSS flow law (19) for
bischofite, covering the steady state behaviour (diamond data points) as well as the
gradual decrease in n-value during relaxation (from n ~ 5 to n ~ 1 when going
towards low stress and strain rate), b) comparison with Urai’s data and
prediction on lower stresses and strain rates
So the composite flow laws give a complete picture of the creep
characteristics of bischofite and carnallite, in two regimes of GSI (n ~ 5) and
GSS (n ~ 1) that gradually pass into each other.
The final question to be addressed now is what flow law to use, the laws
determined on the basis of the steady state stress-strain rate values (Eqs. 4.6
and 4.9), or the composite laws (4.19 and 4.20). It is recalled here that the
GSS creep laws were determined using the relaxation data under the
assumption of constant structure. During deformation in nature, ongoing
-10
-9
-8
-7
-6
-5
-4
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Log (
stra
in r
ate
[s-1
])
Log (flow stress [MPa])
Bischofite steady state data
Data picked at 0 from relaxation curves
composite 1
composite 2
composite 3
composite 4
U80 oC
U60 oC
U80 oC
U60 oC
179
Figure 4.29. Trends resulting from the composite GSI + GSS flow law (20) for
carnallite, covering the steady state behaviour (diamond data points) as well as the
gradual decrease in n-value during relaxation (from n ~ 5 to n ~ 1 when going
towards low stress and strain rate)
microstructural modification is likely to occur. De Bresser et al. (2001)
proposed that in the GSI creep field, dynamic recrystallization will reduce
the grain size, moving the material to the boundary with the GSS creep
field, so increasing the contribution of a grain size sensitive mechanism to
the overall creep rate. On the other hand, fine grained starting material
deforming by a GSS mechanism might show grain growth, also moving the
material towards the GSS-GSI boundary, now increasing the contribution of
a grain size insensitive mechanism.
Flow behaviour within the boundary zone between GSI and GSS creep
should be described by a composite flow law of the type: 𝜀�̇�𝑜𝑡𝑎𝑙 = 𝜀�̇�𝑆𝐼 +
𝜀�̇�𝑆𝑆, cf. Equations (4.19 and 4.20, bischofite and carnallite respectively). In
-10
-9
-8
-7
-6
-5
-4
0.0 0.4 0.8 1.2 1.6 2.0
Log (
stra
in r
ate
[s-1
])
Log (flow stress [MPa])
Carnallite steady stress
Data picked at 0.6 from relaxation curves
composite 1
composite 2
composite 3
composite 4
180
case, however, that the grain size reduction related to the GSI part of
composite creep is balanced by grain growth associated with GSS flow of
fine grained material, both mechanisms may contribute about equally to the
overall flow rate so, 𝜀�̇�𝑆𝐼 ≈ 𝜀�̇�𝑆𝑆. The composite Equation then boils down
to a flow law without grain size sensitivity:
𝜀�̇�𝑜𝑡𝑎𝑙 = 2𝜀�̇�𝑆𝐼 (4.21)
The factor of 2 included in Equation (4.21) holds if GSI and GSS creep both
contribute half to the total creep rate. This not necessary is the case, while a
balance between the two mechanisms might still be maintained (see De
Bresser et al. 2001). We suggest that our experimental data indicate that
during straining at constant strain rate, the microstructure is continuously
being reworked; allowing some balance to develop at the boundary between
the GSI and GSS mechanisms, cf. Equation (4.21). The GSI flow laws as
established for bischofite and carnallite, Equations (4.6) and (4.9), then form
solid descriptions of the creep behaviour of these materials.
However, in cases that effective microstructural modification cannot be
assumed, as for example during transient creep in the walls of salt caverns,
the composite creep laws established form better descriptions.
4.4.2 Effect of composition
The mixture1 sample was found to be stronger than mixture2. The XRF
analysis of the samples (Fig. 4.19) has shown a higher halite wt. % in
mixture1 compared to mixture2, ~65% and 30%, respectively, while the
difference in bischofite wt. % is limited, namely 14 and 19%, respectively.
Samples carnallite 1, 3 and 5 (Fig. 4.19) were obtained from a different part
181
Figure 4.30. Strengths of the mixtures 1 and 2 and samples carnallite 1 and 5 (that
have limited amounts of halite and hence can also be considered as mixtures),
against wt. % halite, carnallite and bischofite. The plot shows that the higher the
halite % in the mixture, the higher the strength of the material.
182
of the TR-9 core that formed the source of our samples. The micro XRF
analysis (carnallite 1, 3 and 5) revealed that these samples were not pure
carnallite samples, as they contain up to 6-8% halite and 17-19% bischofite.
So these samples can, to a first-order, be considered as mixtures as well, and
their mechanical behaviour can be compared with that of the “true” mixtures
1 and 2.
Accordingly, the strengths of mixtures 1 and 2 and carnallite samples 1
and are plotted against the wt. % of halite, carnallite and bischofite in Figure
4.30. The Figure shows that there is a systematic dependence of strength of
materials tested, notably on the halite content. With increasing halite
content, the mixed salt becomes 2 to 4 times stronger (at 10-6
, 10-8
s-1
respectively) than the carnallite material tested. We thus infer that the
difference in strength from one mixture to the next is best explained by the
difference in halite wt. % between the samples.
SUMMARY AND CONCLUSIONS
Deformation experiments were conducted on bischofite, carnallite and
mixed bischofite-carnallite-halite samples obtained from natural cores. Main
aim was to produce constitutive flow laws that can be applied at real in situ
conditions. The experiments were carried out at a confining pressure of 40
MPa and a temperature of 70 ºC. The experiments were multi-step tests
consisting of constant strain rate parts and stress relaxation parts. The flow
laws developed are mainly on the basis of mechanical data, microstructural
work is for future.
The main findings are:
183
1. Carnallite is 4-5 times stronger than bischofite. The bischofite-
carnallite-halite mixtures, at their turn, are stronger than carnallite,
and hence also stronger than bischofite. With increasing halite
content, the mixed salt becomes 2 to 4 times stronger (at strain rates
10-6
, 10-8
s-1
respectively) than the carnallite material tested. We
infer that the difference in strength from one mixture to the next is
best explained by differences in halite wt. %.
2. The constant strain rate parts of the multistep experiments allowed
defining (dislocation/grain size insensitive GSI) creep laws for
bischofite and carnallite.
3. For bischofite as well as carnallite, we observed that during stress
relaxation, the conventional power law stress exponent in the creep
laws changes from ~5 at 10-5
to ~1 at 10-9
s-1
. This is interpreted as
reflecting a change from grain size insensitive (GSI) dislocation
creep at the faster strain rates to grain size sensitive (GSS)
behaviour at slow strain rate.
4. Assuming that recrystallization was effective during deformation at
constant strain rate, and that the recrystallized grain size remained
constant during relaxation, composite creep laws combining GSI
creep and GSS creep, holding at 70 oC were established for
bischofite and carnallite.
5. If during deformation of bischofite and carnallite the microstructure
is continuously being reworked, some balance might develop
between the GSI and GSS mechanisms, at the boundary between the
creep regimes. The established single GSI flow laws then form solid
descriptions of the creep behaviour of these materials. In cases that
effective microstructural modification cannot be assumed, as for
184
example during transient creep in the walls of salt caverns, the
composite creep laws form better descriptions.
185
186
187
Chapter 5
Permeability of interfaces in layered
rock salt under different stresses and
geometries
Muhammad, N., C.J. Spiers, C.J. Peach, J.H.P. de Bresser & W. Liu, 2015
Mechanical Behavior of Salt VIII.
L. Wei, Nawaz Muhammad, Y. Li, C.J. Spiers, C. Yang & H. Ma, 2014
Yanshilixue Yu Gongcheng Xuebao
188
1.1 INTRODUCTION
Salt formations have been of great importance for storage of gas and liquid
fuel for the last few decades due to easy solution mining and excellent
sealing capacity. In order to fully assess the potential, it is important to
know the permeability of the salt cavern walls. The permeability is mainly
an inter-crystalline phenomenon or through fractures (Gloyna & Reynolds
1961). In contrast to domal salt, layered salt has the additional feature of
interfaces between different compositional layers. Due to the multilayer
character, it is not only important to know the bulk permeability, but also
important to know the permeability through interfaces and any changes,
such as damage from material incompatibility, due to application of
differential stress. Selected and cored samples with various interface
geometries were tested for their permeability to argon gas. Five samples
(NP1 to NP5: Batch-I) were provided from the Jintan salt mine at a depth of
(873-1047 m) and three samples (NP6 to NP8: Batch-II) from the Yunying
salt mine at a shallower depth of (811-815 m) from Hubei province in
China. In order to simulate the near real in situ conditions of caverns, the
samples were tested at a confining pressure of 20 and (for shallower depth)
10 MPa respectively. Data for the dilatancy, permeability and damage for
pure salt are readily available (Peach & Spiers 1996, Hatzor & Heyman
1997, Stormont 1997, Popp et al. 2012), but very little is known about the
permeability of layered salt under various differential stresses. There is
some indication (Hatzor & Heyman 1997) from effects of bedding
anisotropy on dilation that there could be effects on permeability. In
particular, little is known regarding the effect of differential stress and
bedding orientation on permeability through interfaces between salt layers
(Liang et al. 2007 and 2012). Models exist relating damage to estimates of
189
permeability (Arson & Pereira 2013, Pereira & Arson 2013), but it is
beyond the scope of these experiments to provide a full quantification of
porous microstructure to test such models, where radial permeability
measurements and exact mapping of porosity would be necessary.
For the first time, in this research work, focus is made on the response of
interlayer bond integrity of interfaces to differential stress, by measuring the
permeability at various steps of increasing differential stress, using argon
gas as pore fluid.
5.2 METHOD
5.2.1 Samples source, composition and preparation for experiments
The sample preparation was done at the laboratory of rock and soil
mechanics and engineering, Wuhan Institute of Rock and Soil Mechanics
(IRSM), Chinese Academy of Sciences. The salt cores were extracted from
two different mines. Batch-I samples (NP1 to NP5) from Jintan mines,
formed in Cenozoic, Paleogene system,Palaeocene to Eocene series.
Batch-II samples (NP6 to NP8) from Yunying basin, which is located in the
northeast part of Jianghan Sunken. The salt mine locates in the centre
location of Yunying salt basin. This was an inland fault-bounded salt lake,
formed during the Cretaceous through to the early Tertiary period.
The salt cores (under investigation) were located in Lower Paleogene,
Gypsum and Salt Group, and are composed of halite and glauberite as bulk
minerals, whereas the interlayer itself is, fine grained, white coloured and
argillaceous (Figure 5.1). The mineralogical components in the cores of
Batch-I were determined by dissolution method, which revealed that
190
composition of this interlayer varies; NaCl: 17.50%-23.62%, Na2SO4:
11.19%-19.43%, CaSO4: 23.86%-26.35%, remainder of insoluble minerals
such as quartz, feldspar, argillaceous minerals etc. The sample bulk is
mainly salt NaCl > 95% and rest is glauberite Na2SO4·CaSO4. The fine
material in the interlayers of Batch-I is clay-rich (see Appendix III). Only
one sample from Batch-II was analysed for its bulk composition using XRD
and the results revealed the composition; glauberite: 78.42%, halite:
11.88%, dolomite: 4%, quartz: 2.53%, magnesite: 3.16%.
Cylindrical cores (of ~100 cm length and 10 cm diameter) were extracted
by drilling vertically into the horizontal salt beds. From these cores, samples
were prepared incorporating interfaces oriented, vertical, oblique and
horizontal to deformation direction, by hand sawing. The blocks were then
machined into cylinders by saw-cuts followed by Silicon Carbide paper
grinding and polishing to reach the final required diameter of 50 mm.
The eight samples obtained (denoted “NP1” to “NP8”) had an average
diameter of 50 mm and lengths in the range 85 to 100 mm. Three different
geometries have been investigated, namely interfaces that are oriented
vertical, horizontal and oblique to the sample axis or show more complex
configurations (“mixed” e.g. NP3, Figure 5.1). All permeability
measurements were taken parallel to the cylindrical axis, with gas
transmission via the end faces. Three samples (NP2, 4 and 8) were prepared
with their interface at an oblique and one sample NP7 with interface
oriented horizontal to measurement direction. Consequently, the interface
between layers was not necessarily directly accessible by argon gas from
flat end faces and interfacial effects could therefore be masked by
impermeable sample ends. To study the permeability characteristics at the
interface, it was devised to drill perpendicular holes in both flat faces till
191
Figure 5.1. Pictures of the samples with different interface geometries and their
schematic diagrams. The two main layers are marked with their abbreviations;
halite (H) and glauberite (G)
Sal
t
H
H H
H
G
G
G G
NP1 NP2 NP3 NP4
15 mm
NP6 NP7 NP8
H
H
H
G
G
G
H
G
NP5
15 mm
192
interface depth in order to provide a direct access for argon gas. A 2.5 mm
diameter drill was used for all samples, whereas the depth of each hole
purely depends upon the distance of the interface from a flat face (see Table
5.2 for more details).
The samples were placed between steel pistons, and were subsequently
jacketed with butyl rubber jackets, to allow even confinement and avoid
contamination of samples with the confining medium, silicone oil. The
jacket was stretched on both ends, rolled over the pistons and tied with
stainless steel wires embedded in circular grooves within the pistons. The
pistons were perforated at centre with a 2.5 mm hole for the argon gas
passage. For uniform spread of gas on the both flat faces of the sample, a
double layer of fine glass fibre sheeting was incorporated on both sides. To
reduce the friction between the polished pistons and sample plus glass fibre,
during deformation, a 50 µm thick perforated PTFE
(polytetrafluoroethylene) sheet was inserted in between.
5.2.2 Apparatus and testing conditions
The experiments were carried out by coupling two instruments, namely a
“Heard” triaxial deformation apparatus (named after its original designer,
the late H.C. Heard, Lawrence Livermore Lab., CA), and a transient-step
argon gas permeameter (designed and fabricated at HPT laboratory Utrecht
University) which works on the basic principle given by Sutherland and
Cave (1980), for measurement of low-permeability samples. For more
details see Peach and Spiers (1996). The tests were conducted at room
temperature and confining pressures of 20 and 10 MPa for Batch-I and
Batch-II respectively. The confining medium used was silicone oil and
pressure was maintained using a servo pump control system accurate within
193
0.01 MPa. The displacement of the confining pressure control system pump
acted as a dilatometer (with a resolution of ± 0.1l), allowing the
measurement of the change in total confining fluid volume to be interpreted
as bulk compaction/dilatancy of the sample, after correction for machine
distortion.
The samples were deformed at a strain rate ( ) of about 10-5
s-1
, reached
by employing a constant piston displacement rate of 1.086 µm/s, at 50:1
drive-gear ratio of the Heard apparatus and the differential stress on sample
was measured. Displacement was measured externally by an LVDT (linear
variable differential transformer) with 25 mm range. Axial load was
measured using a semi-internal force gauge with 400 kN capacity,
insensitive to seal friction but affected by confining pressure. This
sensitivity was corrected by calibration against confining pressure measured
using the 100 MPa pressure system control transducer (type Jensen HFJ PE
1000 LS). All signals were computer logged using a 16-channel, 16-bit
resolution, 250 kSamples/sec analogue to digital data acquisition system
(National Instruments, USB-6221). After every 10 MPa increase in
differential stress, the deformation piston was arrested and permeability was
measured using the argon gas transient-step permeametry. For permeability
measurement, the mean Argon-gas pressure used was ≥ 1.5 MPa, which is
high enough (Peach and Spiers 1996) to suppress the Klinkenberg gas
slippage effect (Klinkenberg, 1941) (in salt samples for κ > 10-19
m2). For
Argon gas pressure, two absolute pressure transducers (Keller PR33X, 2
MPa range, high precision 0.01% and accuracy of 0.05%, with 16 bit
precision data transfer and temperature compensation) monitored the
upstream pressure decay and downstream pressure rise following the
194
approach to a new equilibrium. During equilibration, the drop in pressure of
the gas was logged until equilibrium was reached.
5.2.3 Calibrations
Both triaxial deformation machine and the gas permeameter were carefully
calibrated for the testing conditions. The triaxial apparatus was calibrated
for the machine stiffness (axial elastic distortion) by using a stainless steel
dummy (length = 100 mm and diameter = 50 mm) of known Young’s
modulus, recording elastic distortion as a function of the axial load, at 10
and 20 MPa confining pressures. Volumetric distortion was determined at
constant pressure, using the volumometer control system during the axial
stiffness determinations. Since the apparatus is also volumetrically sensitive
to temperature, a small change in temperature can cause expansion or
contraction of the vessel thereby changing the signal of dilatometer, it was
also calibrated for volume versus temperature.
For transient-step permeametry, the precise attached volumes on both
sides of the sample, including volume of pipes and reservoirs on upstream
and downstream sides are required. An impermeable plastic dummy sample
was used to separate the two system halves and the connected volumes
determined by use of Boyle’s law (PV = constant). The unknown volumes
were calculated by monitoring the drop of pressure at constant temperature
relative to an evacuated pre-calibrated volume in the attached permeameter
by opening a connecting valve.
5.2.4 Experimental procedure and data processing
Before permeability measurement, the system, including gas pipes and
sample assembly, was evacuated and flushed with argon gas three times to
195
drive out any traces of air molecules. The sample was then equilibrated with
> 1.5 MPa argon mean gas pressure for a few hours, to fill the gas into the
pores of the sample. Then the pressure step (down) of 0.2 MPa was created
in the reservoir on one side of the sample and rapidly applied to the pre-
equilibrated sample, by opening a valve to start the transient test. The
pressure difference across the sample versus time data were digitally
recorded until the pressure transient dissipated to below 10% of its starting
value. To study any reversible permeability response of the samples (elastic
response of the crack networks), the permeability at zero-load conditions
was measured at two confining pressure values (5 and 20 MPa for Batch-I; 7
and 10 MPa for Batch-II) at the beginning and end of the test sequence.
Assuming negligible sample storage as compared to the reservoirs
attached, constant volume of the system plus sample at the provided
pressure values and no gas slippage at the boundaries of the walls and
keeping temperature constant, applying the following exponential pressure
decay equation with time as given by Sutherland and Cave (1980):
𝛥𝑃 = (∆𝑃𝑜)𝑒−∝𝑡 (5.1)
∝= 𝜅 (𝐴
𝑙) (
𝑉1+𝑉2
𝑉1𝑉2) (
1
𝜇𝛽) (5.2)
𝜅 =𝑑
𝑑𝑡(𝑙𝑛 (
∆𝑃
∆𝑃𝑜))
𝑙
𝐴(𝑉1𝑉2
𝑉1+𝑉2)𝜇𝛽 (5.3)
where
ΔPo = initial pressure difference [Pa]
ΔP = pressure difference at time t [Pa]
196
α = constant for given sample and experimental conditions [s-1
]
t = time [seconds]
κ = permeability [m2]
A = cross-sectional area of sample [m2]
l = length of sample [m]
V1, V2 = volumes of gas reservoirs on two sides of sample assembly [m3]
µ = dynamic viscosity of argon gas at room temperature [Pa s]
β = argon compressibility [Pa-1
]
The value of the permeability (κ) can be directly determined from the slope
of a graphical plot of ln(ΔP/ΔPo) against time (t). This analysis was
performed by a dedicated computer program (Peach, 1991) with argon gas
properties determined from a virial equation of state and interpolated
viscosity values for the mean temperature and pressure of the argon gas used
in each test.
5.2.5 Preparations for Microstructural study
The deformed samples were studied for the permeable pathways by slicing
into halves along the length. Prior to that, the samples were impregnated
with a low viscosity dark blue resin. The resin was prepared in a hollow
cylinder and the samples were submersed in it. The assembly was first
placed in a vacuum chamber for one hour to drive out the gas/air molecules
from inside the samples and then it was brought to normal atmospheric
pressure which helped drive the resin into the sample through open grain
boundaries. It was kept for 24 hours to allow the resin to set. Later the
samples were taken out and halved along the length (and perpendicular to
interface) using a hand-saw and subsequently were polished with SiC papers
197
to reveal the resin impregnation marks. The porous locations then appeared
dark in greyscale of scanned images. Five out of total eight numbers of
samples were selected for such microstructural study.
5.3 RESULTS
5.3.1 Mechanical data
The mechanical response of the tested samples is shown in Figure 5.2. The
measured stress and volumetric strain of all samples is plotted against the
axial strain. The maximum axial strain (𝜀𝑎𝑥𝑖𝑎𝑙) imparted to any individual
sample was less than 3.5%, except for NP3 (𝜀𝑎𝑥𝑖𝑎𝑙 ~ 7%). The bulk
volumetric strain was also measured during deformation and was found
negative for all samples, indicating compaction of the samples. None of the
samples macroscopically failed during deformation, neither at the interface
nor elsewhere. At the end of the experiments the samples were measured for
their final dimensions. Such measurement on the vertical interface sample
NP6 revealed that the halite half was relatively shorter and plastically
deformed, whereas the glauberite half behaved elastically.
198
(a)
(b)
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0
10
20
30
40
50
60
0 1 2 3
com
pac
tio
n ɛ
v [
%]
dil
atio
n
Dif
fere
nti
al S
tres
s [M
Pa]
Axial strain [%]
NP1
Stress vs. axial strain
Volumetric strain
vs. axial strain
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0
10
20
30
40
50
60
0 1 2 3
com
pac
tion ɛ
v [
%]
dil
atio
n
Dif
fere
nti
al S
tres
s [M
Pa]
Axial strain [%]
NP2
Stress vs. axial
strain
Volumetric strain vs.
axial strain
199
(c)
(d)
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0
10
20
30
40
50
60
0 1 2 3 4 5 6 7
com
pac
tion ɛ
v [
%]
dil
atio
n
Dif
fere
nti
al S
tres
s [M
Pa]
Axial strain [%]
NP3
Stress vs. axial
strain
Volumetric strain vs.
axial strain
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0
10
20
30
40
50
60
0 1 2 3
com
pac
tio
n ɛ
v [
%]
dil
atio
n
Dif
fere
nti
al S
tres
s [M
Pa]
Axial strain [%]
NP4
Stress vs. axial
strain
Volumetric strain vs.
axial strain
200
(e)
(f)
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0
10
20
30
40
50
60
0 1 2 3
com
pac
tion ɛ
v [
%]
dil
atio
n
Dif
fere
nti
al S
tres
s [M
Pa]
Axial strain [%]
NP5
Stress vs. axial
strain
Volumetric strain vs.
axial strain
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0
5
10
15
20
25
30
35
40
0 1 2 3
com
pac
tio
n ɛ
v [
%]
dil
atio
n
Dif
fere
nti
al S
tres
s [M
Pa]
Axial strain [%]
NP6
Stress vs. axial
strain
Volumetric strain vs.
axial strain
201
(g)
(h)
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0
5
10
15
20
25
30
35
40
0 1 2 3
com
pac
tion ɛ
v [
%]
dil
atio
n
Dif
fere
nti
al S
tres
s [M
Pa]
Axial strain [%]
NP7
Stress vs. axial
strain
Volumetric strain vs.
axial strain
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0
5
10
15
20
25
30
35
40
0 1 2 3
com
pac
tio
n ɛ
v [
%]
dil
atio
n
Dif
fere
nti
al S
tres
s [M
Pa]
Axial strain [%]
NP8
Stress vs. axial
strain
Volumetric strain vs.
axial strain
202
(i)
Figure 5.2. Mechanical data showing differential stress and volumetric strain vs.
axial strain, a-h) NP1-NP8 and i) retest NP4
5.3.2 Permeability
The interface permeability (κ) to argon gas was measured as response; to
hydrostatic confinement (at start and at the end of each experiment) and to
stepwise increase in differential stress. The change in permeability with
hydrostatic pressure was studied to see the elastic response of permeable
pathways, necessarily at zero differential loading. The response of sample
permeable pathways to differential stress was done at constant confining
pressures of (20 and 10 MPa for Batch-I and II respectively). The stepwise
obtained permeability values are plotted in Figures 5.3 and 5.4.
As a group, all samples of both batches, exhibit a wide range of initial
permeability (10-15
to < 10-20
m2). This makes comparison difficult, so it
becomes necessary to concentrate on relative change. Taking the starting
zero-load (zero differential stress) permeability (κo) as reference value (at
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0
10
20
30
40
50
60
0 1 2 3
com
pac
tion ɛ
v [
%]
dil
atio
n
Dif
fere
nti
al S
tres
s [M
Pa]
Axial strain [%]
NP4_retest
Stress vs. axial
strain Volumetric strain vs.
axial strain
203
confining pressure 20 or 10 MPa for Batch-I or II respectively) and κ as the
instantaneous permeability at each higher differential stress, relative
permeability values (κ/κo) were calculated and plotted against the axial and
volumetric strain Figures 5.5 and 5.6.
5.3.3 Results Batch-I (20MPa confinement)
(i) NP1 (vertical interface)
The initial permeability measured at zero differential stress and Pc = 20 MPa
was 1.07×10-16
m2. It progressively went down with differential stress and
decreased to 4.71×10-17
m2 at the maximum differential stress of 30 MPa.
The sample was fully unloaded before starting a second run on the same
sample. During this second run, the differential stress has stepwise increased
to values higher than reached in the first run, to a maximum value of 55
MPa, and the permeability found decreasing and reached final value of
1.61×10-17
m2.
The relative change in permeability κ/κo, with differential stress, is plotted
against; axial strain in Figure 5.5a and volumetric strain in Figure 5.5b.
Figures 5.5 (a, b) show that with (stepwise increase in differential stress and
corresponding) axial and volumetric strain of the sample, the relative
permeability is decreasing. After the first run of experiment (0-10-20-30
MPa differential stress), the sample was fully unloaded by pulling back the
piston. Thereafter, the value of permeability was measured as zero-load
value, resulting in a permeability that was nearly the same as the last under
load value (at 30 MPa differential stress). This shows that some permanent
change (notably compaction) has taken place in the sample. This was further
confirmed by analysing the strain of the sample at the initial stage of the
second run. Most of the strain at the first part of second run is elastic, from 0
204
to 20 MPa differential stress, during which part of the test the permeability
hardly changed. Only after exceeding the previous maximum differential
stress, did the permeability resume its decrease for the remaining test steps.
(ii) NP2 (oblique interface)
The permeability measured with increasing differential stress is plotted in
Figure 5.3b, which shows a decreasing trend. The zero-load permeability
value at Pc = 20 MPa was found to be 2.9×10-17
m2, which decreased with
differential stress till last step at a differential stress of 40 MPa and value
found to be 6.21×10-18
m2. The only increase in permeability at the end of
the test is due to hydrostatic pressure decrease at zero differential stress.
The relative change in permeability κ/κo is plotted in Figures 5.5(c, d). The
decrease in κ/κo is systematic and appears directly related to axial and
volumetric (compaction) strain. The deformation is steady and permeability
is decreased to about 80% of its initial value at the end of test for maximum
differential stress of 40 MPa.
(iii) NP3 (mixed)
Initially, under zero differential load and Pc = 20 MPa the sample did show
a measurable permeability (6×10-17
m2). Providing the first step differential
stress of 10 MPa, the permeability decreased by 20% (1.29×10-17
m2). For
the subsequent steps of differential loading of 20 MPa and higher, the
permeability became too low to be measured by the system (i.e. κ < 10-21
m2). While measuring the permeability of the sample at 20 MPa differential
stress, it took so long (a few days) that the stress on the sample was relaxed.
For a repeat measurement of permeability, the sample was deformed again
205
to 20 MPa differential stress, the permeability was still too low (i.e. κ < 10-21
m2) to be measured.
The relative permeability κ/κo measured showed a steep fall to a value of
0.21κo for 10 MPa differential stress, but for the following steps, no
information could be retrieved, as the absolute permeability dropped below
the apparatus limit (<10-21
m2).
(iv) NP4 (oblique interface)
Sample NP4 is an impure salt rock showing an oblique interface of salt with
glauberite. Overall, the sample shows a decreasing trend of permeability
with differential stress (Figure 5.3d). At the unloaded condition of step5,
which is a repeat of step4, a slight increase in permeability was observed.
This difference was attributed to non-equilibration of argon gas pressure
within the sample before starting the transient test in step 4 (step5 was given
a longer equilibration time before the permeability test).So, the starting
permeability at zero differential stress and Pc = 20 MPa is 1.16×10-17
m2,
which shows decreasing trend and reduced to 3.93×10-19
m2 for differential
stress of 30 MPa. For next steps its values remains almost the same for
further deformation to 40 MPa and also completely unloading the
differential stress.
The relative permeability κ/κo calculated is plotted against axial and
volumetric strain in Figures 5.5(g, h). Graph shows there is much change in
κ/κo during first three deformation steps (i.e. for differential stress of 10, 20
and 30 MPa). For the subsequent step of 40 MPa, the curve does not show
much change and shows final value as 0.04κo.
206
(v) NP5 (interlayer with salt)
This salt sample had the lowest permeability among the other samples in
Batch-I. The starting permeability value at zero differential stress and Pc =
20 MPa was found 1.07×10-20
m2, which almost remained the same for first
step deformation to 10 MPa. Subsequent deformation lowered the
permeability value of the sample, finishing with 7.26×10-21
m2 for a
maximum differential stress of 30 MPa. The corresponding relative
permeability κ/κo is plotted against axial and volumetric strain in Figures
5.5(i, j). This shows a negligible change for the first step deformation of 10
MPa followed by progressive decrease and finishing with 0.68κo for a
maximum differential stress of 30 MPa.
(a)
0
10
20
30
40
50
60
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
0 1 2 3 4 5 6 7 8 9 1011121314151617
[MP
a]
Per
mea
bil
ity [
m2]
Number of test steps
NP1 PermeabilityDifferential StressConfining Pressure
207
(b)
(c)
0
10
20
30
40
50
60
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
0 1 2 3 4 5 6 7 8 9 10
[MP
a]
Per
mea
bil
ity [
m2]
Number of test steps
NP2 Permeability
Differential Stress
Confining Pressure
0
10
20
30
40
50
60
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
0 1 2 3 4 5 6 7 8 9 10
[MP
a]
Per
mea
bil
ity [
m2]
Number of test steps
NP3 Permeability
Differential Stress
Confining Pressure
208
(d)
(e)
Figure 5.3. Permeability of Batch-I samples, (a-e) (NP1-NP5) as measured at
different steps in sequence of increasing differential stress for Batch-I samples.
Confining pressure and the differential stress are plotted on secondary y-axis.
0
10
20
30
40
50
60
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
0 1 2 3 4 5 6 7 8 9 10 11 12 13
[MP
a]
Per
mea
bil
ity [
m2]
Number of test steps
NP4 Permeability
Differential Stress
Confining Pressure
0
10
20
30
40
50
60
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
0 1 2 3 4 5 6 7 8
[MP
a]
Per
mea
bil
ity [
m2]
Number of test steps
NP5
Permeability
Differential Stress
Confining Pressure
209
5.3.4 Results Batch-II (10 MPa confinement)
(i) NP6 (vertical interface)
Distinctly, this sample has shown an increasing trend of permeability with
differential stress (Figure 5.4a). The initial value at zero load and at Pc = 10
MPa was found 4.54×10-20
m2 that increased to 6.1×10
-19 m
2 for a maximum
differential stress of 30 MPa.
The relative permeability (κ/κo) is plotted against axial and volumetric strain
in Figures 5.6(a, b). In Figures we see the gradual rise in relative value with
strain and reaching to 13κo for maximum strain with a maximum differential
stress of 30 MPa.
(ii) NP7 (horizontal interface)
The starting permeability at zero differential loading and Pc = 10 MPa was
found 6.13×10-18
m2. The permeability was found to decrease at first during
stepwise increasing differential stress to 10 MPa (Figure 5.4b), but
decreased for the subsequent steps to 20 and 30 MPa, ending at about twice
the value at start 1.26×10-17
m2.
The relative permeability (κ/κo) plotted in Figures 5.6(c, d) which shows a
small dip for differential loading of 10 MPa, recovery to original value at 20
MPa and a further increase to ~2κo for final differential stress of 30 MPa.
(iii) NP8 (oblique interface)
This sample had a high starting permeability 6.09×10-16
m2 as measured at
zero load and at Pc = 10 MPa. With stepwise increase in differential stress,
its permeability value was found to decrease (Figure 5.4c), ending up with
210
an order of magnitude lower value 7.06×10-17
m2, than at start for a
maximum differential stress of 30 MPa.
The relative permeability (κ/κo) plot is shown in Figures 5.6(e, f), which
shows an almost linear trend of decreasing permeability with axial and
volumetric strain and finishing with 0.12κo for maximum differential stress
30 MPa.
(iv) NP4-Re-test (oblique interface)
The sample NP4 was tested again at lower confining pressure Pc = 10 MPa
for a comparative study. The results appeared to be consistent as the
permeability decreased with increasing differential stress (Figure 5.4d). As
compared with the first test at 20 MPa confining pressure, this time the
sample was given more time to become fully equilibrated with argon gas
before starting experiment, and therefore no anomalies were found in the
trend of the permeability change with differential stress. Starting
permeability at zero differential stress was found almost same as was in the
first test, i.e. 1.02×10-17
m2. Deforming the sample caused the reduction of
permeability as measured in stepwise order; the final value appeared to be
4.60×10-18
m2.
The relative permeability (κ/κo) plots (Figures 5.6(g, h)) show that the
maximum decrease in permeability is 0.45κo which is comparatively less
than its first test where it reduced to 0.21κo.
211
(a)
(b)
0
10
20
30
40
50
60
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
0 1 2 3 4 5 6 7 8
[MP
a]
Per
mea
bil
ity [
m2]
Number of test steps
NP6
Permeability
Differential Stress
Confining Pressure
0
10
20
30
40
50
60
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
0 1 2 3 4 5 6 7 8
[MP
a]
Per
mea
bil
ity [
m2]
Number of test steps
NP7
Permeability
Differential Stress
Confining Pressure
212
(c)
(d)
Figure 5.4. Permeability of Batch-II samples, (a-d) (NP6-NP8 & NP4-retest) as
measured at different steps in sequence of increasing differential stress for Batch-II
samples. Confining pressure and the differential stress are plotted on secondary y-
axis.
0
10
20
30
40
50
60
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
0 1 2 3 4 5 6 7 8
[MP
a]
Per
mea
bil
ity [
m2]
Number of test steps
NP8
Permeability
Differential Stress
Confining Pressure
0
10
20
30
40
50
60
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
0 1 2 3 4 5 6 7 8
[MP
a]
Per
mea
bil
ity [
m2]
Number of test steps
NP4-retest
Permeability
Differential Stress
Confining Pressure
213
(a) (b)
(c) (d)
(e) (f)
214
(g) (h)
(i) (j)
Figure 5.5 (i-j). Relative permeability vs. axial and volumetric strains of Batch-I
(confining pressure 20 MPa) semi logarithmic plots
(a) (b)
215
(c) (d)
(e) (f)
(g) (h)
Figure 5.6. Relative permeability vs. axial and volumetric strains of Batch-
II (confining pressure 10 MPa) semi logarithmic plots (a-h)
216
Table 5.1 Step wise permeability values for Batch-I and II, and the conditions used
Sample No of
step
Confining
pressure
[MPa]
Differential
stress [MPa]
Average
Argon
pressure
[MPa]
Permeability
[m2]
NP
1
(ver
tica
l in
terf
aced
)
1 5 0 1.4 1×10-15
2 5 0 1.5 1×10-15
3 20 0 1.6 1×10-16
4 20 0 1.6 1×10-16
5 20 10 1.6 9×10-17
6 20 20 1.6 7×10-17
7 20 30 1.6 5×10-17
8 5 0 1.6 8×10-17
9 5 0 1.6 8×10-17
10 20 0 1.6 4×10-17
11 20 20 1.6 4×10-17
12 20 40 1.6 3×10-17
13 20 55 1.6 2×10-17
14 20 55 1.6 2×10-17
15 20 0 1.6 2×10-17
16 5 0 1.6 3×10-17
NP
2
(ob
liq
ue
inte
rfac
ed)
1 5 0 1.5 6×10-17
2 20 0 1.7 3×10-17
3 20 0 1.7 3×10-17
4 20 10 1.7 3×10-17
5 20 20 1.7 2×10-17
6 20 30 1.7 1×10-17
7 20 40 1.7 6×10-18
8 20 0 1.7 7×10-18
9 5 0 1.7 2×10-17
NP
3
(mix
ed)
1 5 0 1.6 1×10-16
2 5 0 1.6 1×10-16
3 20 0 1.6 6×10-17
4 20 10 1.6 1×10-17
5 20 20 --- <10-21
6 20 30 1.6 <10-21
7 20 40 1.6 <10-21
8 20 0 1.6 <10-21
9 5 0 1.6 <10-21
NP
4
(ob
liq
ue
inte
rfac
ed) 1 5 0 1.6 1×10
-15
2 5 0 1.6 1×10-15
3 5 0 1.6 1×10-15
4 20 0 1.6 *2.×10-18
5 20 0 1.6 1×10-17
217
Table 5.1: contd.
Sample No of
step
Confining
pressure
[MPa]
Differential
stress [MPa]
Average
Argon
pressure
[MPa]
Permeability
[m2]
NP
4
(ob
liq
ue
inte
rfac
ed)
6 20 10 1.6 4×10-18
7 20 10 1.7 2×10-18
8 20 20 1.7 3×10-19
9 20 30 1.6 4×10-19
10 20 40 1.6 4×10-19
11 20 0 1.6 4×10-19
12 5 0 1.7 8×10-19
NP
5
(in
terl
ayer
) 1 5 0 1.5 3×10-20
2 20 0 1.5 1×10-20
3 20 10 1.5 1×10-20
4 20 20 1.5 8×10-21
5 20 30 1.5 7×10-21
NP
6
(ver
tica
l
inte
rfac
ed)
1 7 0 1.9 5×10-20
2 10 0 1.9 5×10-20
3 10 10 1.9 5×10-20
4 10 20 1.9 1×10-19
5 10 30 1.9 6×10-19
6 10 0 1.9 8×10-19
7 7 0 1.9 8×10-19
NP
7
(ho
rizo
nta
l
inte
rfac
ed)
1 7 0 1.9 1×10-17
2 10 0 1.9 6×10-18
3 10 10 1.9 5×10-18
4 10 20 1.9 5×10-18
5 10 30 1.9 1×10-17
6 10 0 1.9 1×10-17
7 7 0 1.9 2×10-17
NP
8
(ob
liq
ue
inte
rfac
ed)
1 7 0 1.9 7×10-16
2 10 0 1.9 6×10-16
3 10 10 1.9 3×10-16
4 10 20 1.9 2×10-16
5 10 30 1.9 7×10-17
6 10 0 1.9 7×10-17
7 7 0 1.9 9×10-17
218
Table 5.1: contd.
Sample No of
step
Confining
pressure
[MPa]
Differential
stress [MPa]
Average
Argon
pressure
[MPa]
Permeability
[m2]
NP
4 (
rete
st)
(ob
liq
ue
inte
rfac
ed)
1 7 0 1.9 1×10-17
2 10 0 1.9 1×10-17
3 10 10 1.9 7×10-18
4 10 20 1.9 5×10-18
5 10 30 1.9 5×10-18
6 10 0 1.9 5×10-18
7 7 0 1.9 6×10-18
*Non-equilibrated start of test (poor value) repeated correctly in following step
Table 5.2. Sample dimensions and changes after test
Test
Initial
length
[mm]
Final
length
[mm]
Initial
dia.
[mm]
Final
dia.
[mm]
Max.
Axial
strain
[%]
Max.
Volume
strain
[%]
NP1 87.50 *87.33
**86.93 49.75 49.753 0.7 -0.32
NP2 86.02 84.37 49.87 *51.48
**49.85 2.4 -0.3
NP3 92.63 86.6 49.94 53.13 6.8 -0.82
NP4 87.64 84.93 49.82 *50.11
**50.46 3.23 -0.44
NP5 85.13 85.1 50.0 50.02 0.275 -0.41
NP6 85.25 *85.22
**84.84 50.01 50.13 0.70 -0.24
NP7 100 98.75 49.84 *49.86
**50.41 1.55 -0.08
NP8 84.47 *82.85
**82.63 49.81 49.95 2.56 -0.48
NP4
retest 84.93 -- 50.26 0.5 -0.5
* glauberite half
** salt half
Note that the initial and final lengths were measured outside the machine, whereas
strains were measured during deformation.
219
5.4 Microstructures
Five deformed samples were selected for microstructural study, considering
the different geometries tested NP3, 4, 6, 7 and 8. These were prepared for
an optical inspection by resin impregnation, cutting and polishing. Cross-
sectional views of these samples are shown in Figure 5.7. In this series of
views, the original cylindrical sample is shown on the left (A), its cross-
section at centre (B) and a magnified view of the interface of (B) on the
right (C). The dark bands, related to resin impregnation, indicate the zones
with permeable channels, which appear as micro-fractures parallel to the
interface.
(i) NP3, A) cylindrical view, upper half glauberite rich, lower half halite rich B)
cross-sectional view of upper and lower halves (at same scale as A), C) magnified
view of encircled region showing resin impregnation marks.
15 mm
A B
C
Resin
7.5 mm
220
(i) NP3 (mixed)
It is a sample without a visible interface. Figure 5.7i shows three different
views; (A) is the cylindrical view where we see the glauberite and halite as
dark and light grey areas. The upper half appears more dominated by
glauberite and the lower half is dominated by halite. The cross-sectional
view (B) shows the upper and lower half of the sample. The glauberite rich
upper half does not show any sign of impregnation, while the lower halite
rich half has a few resin-filled trans granular fractures which appear
disconnected in this two-dimensional view (magnified view C), but must be
connected in depth (to be impregnated).
(ii) NP4 (oblique interface)
The oblique interface of the layers of glauberite and halite is clearly visible
in cylindrical view Figure 5.7ii (A) where the upper half is halite rich
appearing as light and the lower half is glauberite rich and is appearing as
dark grey. The cross-sectional view of the interface is shown in (B). The
encircled area is shown in (C) as magnified image, where we see the resin
impregnated pathways (inter-granular partings) as black lines around grains.
(iii) NP6 (vertical interface)
Figure 5.7iii shows the vertical interface sample. The cylindrical view (A)
of the sample before test shows two layers; the left half is halite-rich and the
right half dominantly consists of glauberite. The cross sectional view (B)
shows the impregnated parts appearing very dark in this grey scale image.
The right half (glauberite-rich) contains only a limited amount of resin,
whereas the halite-half has far more resin impregnation, visible via dark
bands. There is a clear pathway parallel to the interface (see magnified
image (C)). Not only the interface, but also the bulk material away from the
221
(ii) NP4, A) cylindrical view, upper half glauberite rich, lower half halite rich, white
dotted lines show the drilling positions B) cross-sectional view of upper and lower
halves (at same scale as A), C) magnified view of encircled region showing
permeable pathways at interface as resin impregnated.
interface contains parted grains and permeable paths visible as resin filled
dark grey channels (dark grey in micrograph).
(iv) NP7 (horizontal interface)
The cylindrical sample view (Figure 5.7iv (A)) of the horizontal interface
sample clearly shows the two halves; the upper half which is halite-rich and
the lower half which is sulphate-rich (light and dark grey colours,
respectively). The cross-sectional view (B) shows that the halite-rich half
contains more resin that the half with glauberite, which hardly shows any
permeable pathways. At the interface, there is a network of connected paths,
appearing as blackish lines in (B and C) forming a rectangular pull apart
15 mm
A
B
Resin
C
7.5 mm
222
(iii) NP6, A) cylindrical view, left half is halite-rich and right half is glauberite-rich
B) cross-sectional view (at same scale as A) with outlined interface C) magnified
view of clearly connected pathways at interface appearing as darker line.
lattice resembling boudinage. The sample had drilled holes (2.5 mm
diameter) in both flat faces, but the walls of the holes were not sealed, so the
resin also gained access to the sample via these holes, resulting in local
darkening the halite.
(v) NP8 (oblique interface)
In the cylindrical view of the oblique interface sample, the salt-rich part
shows loosely bound grains (Figure 5.7v (A)) as well as inter-granular paths
that appear black upon impregnation (B). In this sample, flat face at top of
sample has a clearly visible interface, while the flat face at the bottom of
sample was drilled, until the interface was reached. In the magnified image
15 mm
B C
Resin 7.5 mm
A
223
(iv) NP7, A) cylindrical view, upper half halite-rich, lower half glauberite-rich,
white dotted lines show the drilling positions B) cross-sectional view (at same scale
as A) with highlighted interface, C) magnified view of highlighted region showing
marks of resin impregnation.
(C), a set of connected and continuous dark lines along the interface are
clearly visible, which appear to anastomose.
5.5 DISCUSSION
5.5.1 Summary of results
(i) Comparison of Batch-I results
Five samples tested at room temperature and confining pressure of 20 MPa
showed a decreasing trend of permeability with differential stress. The
absolute permeability value is different depending upon the interface
Partly intersected, resin
filled, drilled hole.
15 mm
Resin
7.5 mm
C
B A
224
(v) NP8, A) cylindrical view, top half glauberite-rich, bottom half halite-rich, white
dotted lines show the drilling positions B) cross-sectional view of interface (at same
scale as A), C) magnified view of encircled region showing marks of resin
impregnation at the interface.
Figure 5.7. Three different views of each sample. Dotted white lines show the drill-
hole position and depth. The deformation direction is vertical. A) Side view of
cylindrical sample B) cross-sectional view (at same scale as A) and C) (at least) × 2
magnified view of interface.
orientation to loading, as these were similar in terms of extraction location
and depth (i.e. from the same mine and core). Glauberite is found
considerably stronger than salt, and remains elastic with limited plastic
deformation. The bonding at the interface is found strong enough to
withstand the applied differential stress and no general failure occurred. The
volumetric strain in all samples is found negative i.e. compaction.
15 mm
7.5 mm
A B C
Resin
225
(ii) Comparison of Batch-II results
Three samples (NP6, 7 and 8) extracted from a different mine and one
sample from Batch-I (NP4) were tested for shallower depth conditions using
a confining pressure of 10 MPa. The vertical interfaced sample (NP6)
showed a significant increase in permeability with differential stress and
deformation, the horizontal interfaced sample (NP7) sample did not show
any significant change in the permeability, if compared the initial and final
permeability values, though there was change in trend from decrease to
increase with differential stress. The oblique interfaced sample NP8 showed
progressive decrease of permeability with deformation, and NP4-retest
showed a further minor decrease in permeability maintaining its trend in
Batch-I (i.e. at Pc = 20 MPa). The volumetric strain for all of the samples
was negative, i.e. compaction.
5.5.2 Effect of confining pressure
The permeability is mainly an inter-crystalline phenomenon or through
fractures (Gloyna & Reynolds 1961). Under mine conditions, salt rocks are
always in compressed state due to hydrostatic confinement, that makes the
inter-crystalline pathways compacted (Walsh 1965) and reduces the pore
fluid transport and hence permeability. Further, the deformation of salt rocks
also causes dilatancy with strain (Hunsche 1998, Schulze et al. 2001), in the
form of crack opening, intra-crystalline fractures or intra-granular tensile
cracks which in turn creates more pathways by spreading and percolating
(Stormont and Daemen 1992) causing increase in permeability. The
threshold value of confining pressure to suppress the dilatancy is reported to
be Pc > 17.5 MPa (Peach and Spiers 1996). So there is a direct effect of
226
confining pressure on permeability, i.e. higher confining pressure causes
compaction that in turn reduces permeability.
5.5.3 Elastic response of existing cracks to the hydrostatic
pressure
The data obtained on the basis of the experiments show that there is a
permanent change in permeability with respect to differential stress, as the
final permeability was always different from the starting value and did not
reverse on unloading. However, a minor elastic response of the material can
be recognized. We have plotted the permeability values measured by
changing the confining pressure from 5 to 20 MPa at the start and from 20
back to 5 MPa at the end of the experiments of Batch-I (Figure 5.8), and
also the similar treatment using 7 and 10 MPa of Batch-II, under zero
differential load conditions (Figure 5.9). The graph shows that the
permeability at the higher confining pressure is less than at the lower
confining pressure. On the basis of the classical crack-closure model by
Walsh (1981), the permeability is proportional to the 3rd
power of mean
crack width (i.e. κ ∝ <ω>3), which highlights the sensitivity of permeability
to confining pressure by elastic strain, seen in each of the tests. Hence, we
infer that this change in permeability is due to elastic response of crack
width to the hydrostatic loading. This elastic response is only instantaneous,
and gives a qualitative picture of existing cracks and permeable pathways,
which can be predicted to be healed/sealed on long time scales ~1000 years
if salt is dry (Koelemeijer et al. 2012) or on short time scales ≥ 2 years for
relative humidity of 75% (Houben et al. 2013). That is also the reason why
salt rocks are the good choice for underground storage.
227
Figure 5.8. Elastic response of existing cracks to hydrostatic (confining) pressure
Batch-I
Figure 5.9. Elastic response of existing cracks to hydrostatic (confining) pressure
Batch-II
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14P
erm
eabil
ity [
m2]
NP1 Δκ start NP2 Δκ start NP3 Δκ start
NP4 Δκ start NP5 Δκ start NP1 Δκ end
NP2 Δκ end NP4 Δκ end
5 20 20 5 MPa
1.E-21
1.E-20
1.E-19
1.E-18
1.E-17
1.E-16
1.E-15
1.E-14
Per
mea
bil
ity [
m2]
NP6 κ start NP7 κ start NP8 κ start NP6 κ end NP7 κ end NP8 κ end NP4 redo Δκ start NP4 redo Δκ end
10 7 10 7 MPa
228
5.5.4 Compaction and dilatancy
According to Peach & Spiers (1996), halite can be expected to show
dilatancy at confining pressures lower than 18 MPa. However, none of our
test specimens showed bulk dilatancy. This is probably due to the fact that
the total strain imparted < 3%, (except for NP3 that contained comparatively
higher wt. %age of halite) was not high enough to actually take the sample
into the dilatant field, where Peach & Spiers reached final axial strains of
10%, and to the fact that the glauberite layers are stronger than halite and
thus load supporting. In fact, resin impregnation into permeable pathways,
after the experiments, suggests there actually is some localised dilatancy
near to the interfaces.
5.5.5 Localised dilatancy at interface
Overall the permeability of the samples has showed a decreasing trend with
differential stress, especially the Batch-I tests (NP1-NP5) which were all
performed at 20 MPa confining pressure (well above dilatancy limit, Peach
and Spiers 1996). This is interpreted as that compaction and plastic
deformation during loading stages have caused closure of already existing
pathways in the bulk volume of the samples. The vertical interface NP6 has
shown an increasing trend in permeability with differential stress and the
horizontal interface NP7 has shown an increase in permeability towards the
end of test. The mechanical data showed that both of these samples were
compacted by stepwise increase in differential stress. We speculate that the
compaction of the samples is measured as for the bulk, whereas the increase
in permeability is associated to the local dilatancy on the interface and
additionally by more connectivity of already existing pore networks. Indeed,
resin filling at interfaces shows clear evidence of local crack linkage
pathways along interfaces (Figure 5.7). On the other hand, the decrease in
229
the permeability of oblique interface (NP8) is due to bulk compaction by
choking the already existing pathways. However, the resin filling at the
interface shows a localized dilation and permeable connectivity which we
infer is masked by the bulk compaction.
5.5.6 Microstructures
The cross-sectional views of the selected samples are shown in Figures
5.7(i-v) B and C. In general the glauberite-rich parts did not show any
visible sign of impregnation, but still we found non-zero permeability of the
samples. For example, the sample NP5, which is glauberite rich, presented a
low. but measurable, permeability. So there must be some connected
pathways inside the sample inaccessible to the resin. The halite-rich parts
also appear to contain resin, indicative of dilatation, despite all samples
showing overall compaction. NP3, which is halite-rich, showed high
permeability at start of test, its microstructure shows resin filling around
grains, but in the form of discontinuous networks. This sample also
achieved comparatively higher strain, which might have partly closed the
pre-existing cracks that resulted in a dramatic loss of permeability. This
suggests pre-existing inter- and intra-granular crack networks in the halite,
possibly due to damage and present before the deformation.
NP4 was tested twice; its microstructure has prominent resin
impregnation networks connecting through the interface. Although the
sample has shown compaction at both conditions of confining pressures,
locally, dilatancy is visible on the interface as impregnation, which must
have been masked in magnitude by the bulk compaction.
NP6 (vertical interface) sample was initially low permeability and
presented a different trend, i.e. increase in permeability with differential
230
stress. Its microstructure shows that not only the interface, but also the bulk
material away from the interface contains parted grains and permeable paths
visible as resin filled sub-horizontal dark grey channels (dark grey in
micrograph). Again, the bulk showed compaction, which shows that the
permeability increase is due to connectivity of local dilatancy along the
interface, where resin highlights the pathways. The spread of dilatancy
horizontally into the halite from the interface probably reflects the
differential unloading of the halite after its plastic deformation relative to
the more elastic glauberitic half (semi-circular disking fractures confined to
the halite half).
The horizontal interfaced NP7 had a slightly tilted interface inside the
sample. The decrease in permeability at start might be due to closure the
existing connected pathways running sub-horizontally along interface, but
due to confinement within the dilatant regime, further straining of the
sample caused some shear slippage on the interface resulting in crack
opening in lateral directions and increase in permeability. These interfacial
cracks are resin filled along with some sub-horizontal inter-granular
fractures in the halite half, especially close to the resin filled drill hole.
The widespread resin impregnated cross-section of the halite rich part of
NP8 shows that the halite grains were loosely bound (even at the end of
testing). The bulk compaction of the sample with differential stress and drop
in permeability shows that the plastic straining of the samples has caused
narrowing down of the permeable pathways through the bulk, whereas the
local dilatancy which is visible at the oblique interface is masked by this
bulk compaction.
231
5.5.7 Consequences of drilling the interfaces
The permeability values given in Table 5.1 are calculated at sample scale,
using the full sample length and diameter, but the samples NP2, 4, 7 and 8
were actually drilled till the interface of the layers and we measured the
properties of interface with that of sample volume.
The exact path followed by argon is hard to describe, because:
a) the gas was spread uniformly on both faces of the sample and
b) walls of the holes were not sealed or isolated from argon gas
So, there was a potential chance that the Argon gas volume received on
downstream side is the sum of; volume passed through interface only plus
the volume through the bulk of the sample.
The basic principle implicitly used in Equation 5.3 is the empirical law
given by Darcy (1856), which is given as
𝑄 =−𝜅𝐴𝛥𝑃
𝜇𝐿 (5.4)
Where, Q is volume flow rate (m3-s
-1), κ is the permeability of the
medium (m2), A is the cross-sectional area of the sample (m
2), μ is the
dynamic viscosity of the fluid (Pa-s) and (ΔP/L) is the macroscopic fluid
pressure gradient (Pa-m-1
). Re-arranging Equation 5.4 for permeability:
𝜅 = −𝑄𝜇𝐿
𝐴𝛥𝑃 (5.5)
232
Figure 5.10. NP7 schematic diagram, A) cylindrical transparent view, B) cross-
sectional view of interface showing tips of the drilled holes and the cubical
rectangle connecting, C) dimensions of cubical rectangle with (t is the thickness of
area facing)
The permeability is inversely proportional to the pressure gradient
‘ΔP/L’ and directly proportional to the area ‘A’ open to flow for a fixed Q
and μ.
The listed permeability values (Table 5.1) are calculated for all samples
including the drilled-hole samples at the full sample scale.
If only the pathway at the drilled interface is considered, the effective
length ‘L’ is actually the tip to tip distance of the drilled holes, and the
cross-sectional area ‘A’ is the area perpendicular to the connected interfacial
path. This area is hard to estimate due to multiple pathways in the interface
from tip to tip (some of these pathways were highlighted by resin
Drilled holes
A B
28.6 mm
50.0 mm
t
C
233
impregnation). A possible connected path is represented by a rectangular
cube with its length as tip-to-tip distance, breadth as 35.35 mm calculated as
base of a right triangle with its hypotenuse as diagonal of the rectangle (i.e.
diameter of the sample, which is 50 mm for all samples) and thickness ‘t’
can be estimated from the impregnated pathways in the microstructures.
For simplicity, we consider the horizontal interfaced sample NP7, for
which the tip-to-tip inter-holes distance ‘L’ is measured to be 28.6 mm
(Figure 5.10). The thickness ‘t’, estimated from Figure 5.7(iv) is ~ 0.1 mm.
Consider the step4 permeability value (5.05×10-18
m2), calculated at sample
scale (L = 100 mm and diameter = 50 mm) measured at confining pressure
of 10 MPa and differential stress of 20 MPa. Now changing the length ‘L’ to
the new constant value of 28.6 mm and calculating the permeability value
‘κ’ for an new area of 3.535 mm2, we get the value 7.32×10
-16 m
2, which is
~102 times higher than what is listed in Table 5.1. It is important to note that
the main idea behind this work was to test the differential stress sensitivity
of permeability characteristics at the interface of layers. So the absolute
value is less important.
5.5.8 Permeability: bulk and interface
As explained above, that the drilling of a few samples was devised to
provide direct access of Argon gas to the interface, but as a consequence,
the absolute permeability of the samples becomes uncertain. For example;
two samples NP3 and NP5 are the example of relatively dominant minerals
halite and glauberite respectively. Both of these samples were tested without
drilling holes. The starting permeability for NP3 was found in the order of
10-16
m2, which dropped by an order of magnitude for differential stress of
10 MPa. However, further deformation caused the decrease of permeability
234
to become too low to be measured by the apparatus used and it was only
realized after 10 days period while sample was under permeability test with
20 MPa differential stress. NP5 was found less permeable than other
samples, from the start, κ ~10-20
m2 and the sample took several days to be
filled in by Argon gas and reach a starting equilibrium pressure before test
could be started. In short, it was indispensable to drill the oblique and
horizontal interface samples and by-pass the low permeable bulk to
investigate the effects of the interfaces.
As a matter of fact, the Argon gas could not be restricted to pass through
the interface only, since the hole surfaces were not sealed. The permeability
values thus reported have uncertainty in the absolute values, but more
importantly, response of interface permeability of the samples to differential
stress was still measurable and can give guidance about possible leakage
under the real in situ conditions for gas storage.
5.5.9 Comparison with previous work and implications
The potential of using rock salt caverns repositories has already been widely
discussed in the literature and representative cores have been tested for
transport properties (e.g. Gloyna & Reynolds 1961, Sutherland & Cave
1980, Stormont & Daemen 1992, Peach & Spiers 1996, Schulze 2001, Zhou
et al. 2009). All of these studies have reported that rock salt, in general, is a
very suitable choice for repositories, for its sealing and healing capabilities.
However, the characteristics of the salt formations vary substantially from
location to location, in particular regarding homogeneity or interlayering
with other associated minerals. Bedded salt rocks with different layers and
interfaces have never been tested before regarding permeability
characteristics with effect of differential stress included. There is some work
235
on permeability of bedded salt (Stormont & Daemen 1992, Stormont 1997),
but the cores used were without interface of layers. The current study
focusses in particular on layered halite-glauberite salt and presents some of
the interface properties of different layers in bedded salt in different
geometries, covering various possibilities in repositories.
Undisturbed domal salt (5 m deep in cavern walls) has very low
permeability ≤ 10-21
m2, but for the first 1 m depth, this permeability often
has higher values reaching 10-16
m2 (Sutherland and Cave 1980, Peach 1991,
Stormont and Daemen 1992, Stormont 1997) as a result of dilatational creep
in the excavation damage zone. For repositories, such caverns are apparently
a suitable choice.
Different geometries of interface were tested for permeability change
with increasing differential stress, and the absolute magnitude of
permeability was found to vary per sample. The permeability measured, lies
in the wide range of 10-15
to < 10-21
m2. These cores were just extracted by
vertically drilling the horizontal beds, but do show signs of damage and the
initial permeabilities were very high for cavern use. Most of the deformation
effects are masked by this initial dilatation and all samples compacted,
under stress, as a result. More samples from deeper in walls should be tested
to check this hypothesis, but the effect of interfaces is clear in that local
dilatancy will increase permeability in their vicinity, when stressed, and
fluid could therefore escape along these deformation-induced pathways.
CONCLUSIONS
The permeability of layered rock salt with different geometries has been
studied by triaxial testing, with emphasis on interface response to
236
differential stress. Two batches were made according to confining pressure
values; 20 and 10 MPa, simulating depths of ~1000 and 600 m respectively.
The samples were multi-layered (mainly halite and glauberite), often with a
fine grained layer (argillaceous Batch-I) at the major interface. The interface
orientation ranges from horizontal to vertical through oblique direction.
The main findings are follows:
1) Generally, the samples showed an instantaneous response of
permeability to straightforward hydrostatic confinement at the start/end
of the tests. This is interpreted as indicating simple elastic
closure/opening of pre-existing cracks and planar pathways;
2) The absolute values for the permeability varied per sample, due to
differences in interlayer/interface character and initial damage. The high
initial permeability of some of the samples suggests already damaged
material and masked the subsequent effect of the localised deformation
induced damage.
3) All samples showed a decrease in bulk volume with axial strain,
demonstrating progressive bulk compaction with increasing differential
stress; presumably due to crack closure in the bulk.
4) In layer parallel shortening, the applied stress is mostly supported by
harder layer of glauberite in elastic manner, whereas the halite half is
plastically deformed at the same time, as confirmed by the measurement
of the final dimension of the samples after tests.
5) On the basis of the results of our experiments, pre-existing fractures in
the halite are expected to become closed at higher stress values.
6) The interfaces with fine grained material (clay in Batch-I) do not appear
to become debonded.
237
7) The observed decrease in the permeability, with differential stress, of
vertical interface NP1 (high initial permeability) is inferred as due to
closure of grain boundaries and associated sealing of inter-granular
pathways, but the increase in permeability of vertical interface NP6 is
due to local dilatancy at the interface, as confirmed by resin
impregnation after test and a much lower initial permeability.
8) The oblique interfaced samples (NP2, 4 and 8) all showed decrease in
permeability.
9) The samples with higher volume of halite (e.g. NP3) compacted more
and became much less permeable (at the deformation conditions
employed) than that of the other samples, with permeability values
beyond the limit of measurement of our system (i.e., κ<10-21
m2).
10) Since the permeability increases, inferred from microstructural
observations (resin filled pathways), are masked by the bulk compaction
and permeability reduction of pre-existing damage, future work should
endeavour to test less damaged material better reflecting in situ
permeabilities. Without microstructural examination of the interface, the
dilatational effects of the interfaces would have been unnoticed.
238
Data supplied by Chinese Academy of Sciences (CAS) for Batch-I (Appendix I-III)
Appendix I
Petro physical information of samples
Sample Length
[mm]
Diameter
[mm]
Depth
[m] Lithotypes
Density
[g/cm3]
Porosity
[%]
NP1 87.50 49.75 972.74-973.25 horizontal
interface 2.44 4.5
NP2 86.02 49.87 897.11-897.71 inclined
interface 2.33 1.4
NP3 92.63 49.94 968.41-968.86 pure rock
salt 2.23 1.5
NP4 87.64 49.82 873.75-874.1 impure
rock salt 2.26 1.7
NP5 85.13 50.0 1047.1-1047.5 interlayer
with salt 2.59 5.2
Appendix II
Compositions solution results for each sample
Reference
No.
Cl-
[%]
SO42-
[%]
Na+
[%]
K+
[%]
Ca2+
[%]
Mg2+
[%]
Insolution
[%]
Total
[%]
NP1 52.92 1.31 34.8 0.02 0.46 0.01 10.48 100
NP2 57.28 0.84 37.36 0.01 0.29 0.02 4.2 100
NP3 56.73 0.87 37 0.03 0.34 0.01 5.02 100
NP4 53.82 0.24 34.97 0.01 0.19 0.02 10.75 100
NP5 16.47 5.73 11.51 0.02 1.83 0.04 64.4 100
Appendix III
Mineralogical determination of the (nonsaline) interlayers using XRD (Batch-I).
Minerals [%]
Quartz 34.865
Analcime-C 11.645
Montmorillonite-14A 7.635
Halite,syn 2.67
Illite 11.01
Montmorillonite-chlorite 11.155
Feldspar 8.36
Albite 7.2
Pyrite 5.46
Total 100
239
240
241
Chapter 6
Conclusions
242
The focus of this thesis was on the deformation and transport
processes in salt rocks. The study was an experimental study with
special emphasis on the effects of confining pressure and stress
relaxation. Below, the aims of the study, as outlined in Chapter 1, are
repeated and the main conclusions are given, following the chapter
organization of the thesis.
In Chapter 2, the results are presented of new experiments on jacketed
samples of dry synthetic rock salt. The aim was to determine the
microphysical mechanism controlling dislocation creep of halite at
20-350 C, and to develop a mechanism-based flow law providing a
solid basis for extrapolation of lab data to long time scales. One way
of distinguishing between the various dislocation mechanisms that
may control creep in dry rock salt is by the effect of confining
pressure. For that reason, systematic pressure stepping tests were
carried out across a range of pressures not attempted before (50-600
MPa). The following was concluded:
In the temperature range of 22 to 350 oC, the rate controlling
dislocation mechanism changes from Peierls resistance
controlled glide at room temperature to climb controlled creep
governed by dislocation core diffusion at 350 oC. At all
conditions tested, the dry rock salt is stronger if the confining
pressure is higher. The activation volume for creep was found
to be about 0.58Vm for the Peierls resistance controlled glide,
and about 0.66Vm for climb controlled creep, where Vm is the
molecular volume of halite.
243
The resulting creep equations are:
𝜀̇ = 1.26
× 1017 (𝜎
𝜇)2.5
𝑒𝑥𝑝 [−(138 + 𝑃(0.58𝑉𝑚))
𝑘𝑇(1
−𝜎
𝜎𝑜
𝜋
2)]
for Peierls resistance controlled glide
and
𝜀̇ = 2.51 × 1014𝜎4.7𝑒𝑥𝑝 [−126 + 𝑃(0.66𝑉𝑚)
𝑘𝑇]
for climb controlled creep
The temperature at which the transition from glide to climb
control takes place was found to lie in between 125 and 250
oC, at a stress of about 16 MPa. Given the slow strain rates
and low stresses normally relevant for in situ conditions,
dislocation creep of rocksalt in nature will likely be controlled
by dislocation climb.
In Chapter 3, the results are presented of experiments on dense
aggregates of both synthetic and natural wet halite. The aim was to
establish, for this wet material, if a transition can be observed from
creep behaviour governed by dislocation mechanisms to creep
244
behaviour controlled by a solution-precipitation mechanism, and if so,
what the conditions of this transition are in terms of strain rate. All
experiments were multi-step experiments at a temperature of 125 oC,
consisting of constant strain rate parts and stress relaxation parts.
Strain rates as low as 10-9
s-1
were achieved during stress relaxation.
The following was concluded:
The stress exponent n obtained by fitting the mechanical data
to a conventional power law is very high (> 10), suggesting
that dislocation glide may play a role as rate controlling
mechanism, as concluded for dry rock salt deformed at
relatively low temperature. During stress relaxation, a
progressive decrease in the n-value was observed, reaching to
a value of ~1 at the slowest strain rates attained. Such value of
n of ~1 points to a grain size sensitive mechanism controlling
creep at slow strain rate and low stress, likely to be solution-
precipitation in the case of wet halite.
Assuming i) that recrystallization was effective during the
deformation at constant strain rate, modifying the grain size in
relation to the differential stress, and ii) that the grain size
remained constant during relaxation, allowed to establish a
composite creep law combining grain size insensitive
(dislocation/GSI) creep and grain size sensitive (solution-
precipitation/GSS) creep, holding at 125 oC. This composite
creep law is as follows:
245
𝜀̇ = 6.42 × 10−20𝜎11 + 1.68 × 10−10𝜎𝑑−1.1
Natural salt appeared stronger than synthetic salt at otherwise
similar conditions, which is inferred to be due to the
impurities in natural salt.
Chapter 4 reports on the creep behaviour of bischofite, carnallite and
mixtures of bischofite-carnallite-halite. Aims were: 1) to determine
the creep behaviour of these salt rocks under in situ conditions, 2) to
establish the role of dislocation creep, recrystallization and, possibly,
pressure solution creep in their behaviour, and to construct creep laws
that allow reliable extrapolation to natural conditions, and 3) to obtain
a first order impression of the difference in strength between mixtures
of bishofite-carnallite-halite mixtures and their end members.
Experiments have been carried out at real in situ conditions of
confining pressure 40 MPa and at a fixed temperature of 70 oC. All
experiments were multi-step experiments at a temperature of 70 oC,
consisting of constant strain rate parts and stress relaxation parts. The
following was concluded:
The strength of bischofite is much less than that of carnallite
and bishofite-carnallite-halite mixtures.
The constant strain rate parts of the multistep experiments
allowed defining (dislocation/GSI) creep laws for bischofite
and carnallite:
246
𝜀̇ = 10−8.519𝜎3.4𝑒𝑥𝑝 (𝜎
2.261)
𝜀̇ = 1.13 × 10−12𝜎5.1
Making the same assumptions as done for wet halite (Chapter
3), namely i) that recrystallization was effective during the
deformation at constant strain rate, modifying the grain size in
relation to the differential stress, and ii) that the grain size
remained constant during relaxation, composite creep laws
combining grain size insensitive (dislocation/GSI) creep and
grain size sensitive (GSS) creep, holding at 70 oC were
established for bischofite and carnallite:
𝜀̇ = 1.1 × 10−9𝜎5.4 + 3.94 × 10−8𝜎𝑑−0.8
𝜀̇ = 3.70 × 10−13𝜎5.3 + 1.01 × 10−8𝜎𝑑−1
If during deformation of bischofite and carnallite the
microstructure is continuously being reworked, some balance
might develop between the GSI and GSS mechanisms, at the
boundary between the creep regimes. The established GSI
flow laws then form solid descriptions of the creep behaviour
of these materials. In cases that effective microstructural
modification cannot be assumed, as for example during
247
transient creep in the walls of salt caverns, the composite
creep laws form better descriptions.
The bischofite-carnallite-halite mixtures are stronger than
carnallite, and hence also stronger than bischofite. With
increasing halite content, the mixed salt becomes 2 to 4 times
stronger (at 10-6
, 10-8
s-1
respectively) than the carnallite
material tested. We inferred that the difference in strength
from one mixture to the next is best explained by the
difference in halite wt. % between the samples.
Chapter 5 describes the results of permeability measurements on
experimentally deformed samples of layered rocksalt from mines in
the Hubei province China. The aim was to test the permeability as
response to differential stress using various geometries/orientation to
deformation direction. By studying the bulk and local
compaction/dilation (at the interface) with the help of dilatometry in
conjunction with the deformation apparatus and microstructural
studies of interface of pre-tested samples. Permeability was
determined using argon gas transient step permeametry, throughout
the deformation. The following was concluded
Hydrostatic and differential stresses both have immediate
effect on permeability, due to initial damage of samples.
The differential stress can cause compaction and permeability
reduction at bulk level if the initial damage levels are high and
may mask dilation at the local interface level. Since both may
248
occur together, it is of utmost importance to study the specific
interlayer characteristics of the walls of potential underground
storage. Local dilatation due to differential stress will cause
local increase in permeability at interfaces in layered salt
rocks similar to those tested here, based on the observations
from resin impregnation and permeametry
Less damaged samples should be tested to better quantify the
effects of local dilation at interfaces and measure the
permeability increases revealed by microstructural analysis
following impregnation. Damaged samples already possess
permeabilities above those expected for storage cavern use.
Relatively pure salt is less permeable and has a better sealing
capacity than sulphates like glauberite.
Suggestions for further research and refinement
In this thesis project, new data have been obtained on the creep
behaviour of dry synthetic rock salt (Chapter 2), wet synthetic and
natural rock salt (Chapter 3), and on polycrystalline bischofite,
carnallite and their mixtures (Chapter 4). In addition, the permeability
of interfaces in salt samples was investigated, in various geometries
(Chapter 5). Although a substantial body of new data was obtained, a
number of uncertainties remained. The following suggestions for
future work are made:
Though a positive pressure sensitivity of flow stress was
found throughout the temperature and strain rate ranges tested,
249
the number of data obtained at low temperature is still limited.
More data for the temperature range 22-125 oC are needed to
further check if the conclusions that under common in situ
conditions, at relatively low stress, climb controlled creep is
the process governing salt flow. Additional tests are
particularly needed to constrain more exactly, than currently
established, what the temperature-stress conditions are of the
transition from glide to climb control.
The number of experiments performed on wet synthetic and
natural polycrystalline halite was relatively limited, partly
owing to the long-time duration of the type of experiments
performed, including stress relaxation. Nevertheless, a
transition was inferred from grain size insensitive (dislocation
– GSI) creep at high stress to grain size sensitive (likely
pressure solution- GSS) at low stress. Grain sizes were
estimated on the basis of application of a recrystallized grain
size piezometers for halite. Independent measurements of
grain sizes are needed to investigate to what extent the
approach that was followed is robust. This concerns both
samples deformed at constant strain rate, assumed to
microstructurally adjust themselves to the operative stress, as
to samples there were let to relax, in which constant structure
was applied. In addition, experiments were only carried out at
125 oC – additional data at higher temperature, where
recrystallization is more prominent, are needed to further
250
constrain the composite GSI-GSS creep law, notably
regarding the activation energy for the GSS-part.
An important outcome of the work on bischofite, carnallite
and their mixtures was the apparent dependence of the
strength of mixtures on the relative wt.% of the various salts
that are present, notably the wt.% halite. In order to study the
contribution of each salt (bischofite, carnallite and halite) in a
systematic way, reagent grade synthetic salt samples should be
prepared and tested for their creep properties. This then can be
followed by the preparation of a series of mixtures with
known composition, and subsequently testing these under the
same conditions as applied on the single salt samples. This
will help to quantify the participation of each member in the
composition, and make a flow law according to volume
contribution of each salt. The results need to be compared
with predictions of models of creep behaviour of
polymineralic minerals.
In investigations of layered salt permeability, the difficulty of
measuring layer parallel permeability changes at high angles
to applied differential stress could be lessened by using radial
permeability tests, rather than relying on pathway
modifications by drilling to allow axial permeability
measurement. In any case, samples with less initial damage
should be used. Impregnation with coloured low viscosity
251
resin and microstructural analysis remains a valuable tool in
elucidating permeable pathways.
Finally, the results of this work are directly relevant for the modelling
of flow of salt in settings varying from salt diapirs to solution–mined
cavities and the stability and integrity of bore holes through sediment
packages containing salt horizons. It would be most useful if the creep
equations for glide and climb controlled creep for dry halite, the
composite creep law obtained for wet halite, and the (composite)
creep laws for bischofite and carnallite are included in models, to
evaluate their full implications.
252
253
REFERENCES
Aladag, E., L.A. Davis & R.B. Gordon, 1970. Cross slip and the plastic
deformation of NaCl single and polycrystals at high pressure.
Philosophical Magazine, 21, 469-478.
Arson, C. & J.M. Pereira, 2013. Influence of damage on pore size
distribution and permeability of rocks. International Journal for
Numerical and Analytical Methods in Geomechanics, 37(8), 810-831.
Auten, T.A., L.A., Davis, & R.B. Gordon, 1973. Hydrostatic pressure and
the mechanical properties of NaCl polycrystals. Philosophical Magazine,
28, 335-341.
Barr, L.W., J.A. Morrison, & P.A. Schroeder. Anion diffusion in crystals of
NaCl. Journal of Applied Physics 36.2 1965: 624-631.
Cannon, W.R., & T.G. Langdon, 1983. Creep of ceramics. Journal of
Materials Science, 18(1), 1-50.
Carter, N.L. & F.D. Hansen, 1983. Creep of rocksalt. Tectonophysics, 92(4),
275-333.
Carter, N.L., S.T. Horseman, J.E. Russell & J. Handin, 1993. Rheology of
rocksalt. Journal of Structural Geology, 15 (9-10), 1257-1271.
Christov C., 2009. Isopiestic Determination of the Osmotic Coefficients of
an Aqueous MgCl2 + CaCl2 Mixed Solution at (25 and 50) °C. Chemical
Equilibrium Model of Solution Behavior and Solubility in the MgCl2 +
H2O and MgCl2 + CaCl2 + H2O Systems to High Concentration at (25
and 50) °C J. Chem. Eng. Data 2009, 54, 627–635.
Clark, S.P. (Ed.). 1966. Handbook of physical constants (Vol. 97).
Geological Society of America.
Conrad, H. & D. Yang, 1999. The rate-controlling mechanism(s) during
plastic deformation of polycrystalline NaCl at 0.28--0.75 Tm. Journal of
materials science, 34 (4), 821-826.
Cosenza, Ph., M. Ghoreychi, B. Bazargan-Sabet & G.De Marsily, 1999. In
situ rock salt permeability measurement for long term safety assessment
of storage. International Journal of Rock Mechanics and Mining
Sciences 36, no. 4: 509-526.
254
Cox, S.F.& M.S. Paterson, 1991. Experimental dissolution‐precipitation
creep in quartz aggregates at high temperatures. Geophysical Research
Letters 18.8: 1401-1404.
De Bresser, J.H.P.,1991. Intracrystalline deformation of calcite, Geol.
Ultraiectina, 79, Ph.D. thesis, Utrecht Univ., The Netherlands.
De Bresser, J.H.P., J.H. Ter Heege & C.J. Spiers, 2001. Grain size reduction
by dynamic recrystallization: can it result in major rheological
weakening?. International Journal of Earth Sciences, 90(1), 28-45.
De Bresser, J.H.P., B. Evans & J. Renner, 2002. On estimating the strength
of calcite rocks under natural conditions. Geological Society, London,
Special Publications 200, no. 1: 309-329.
De Bresser, J.H.P., 2002. On the mechanism of dislocation creep of calcite
at high temperature: Inferences from experimentally measured pressure
sensitivity and strain rate sensitivity of flow stress. Journal of
geophysical research, 107 (B12), 2337.
Desbois, G., P. Závada, , Z. Schléder & J.L. Urai, 2010. Deformation and
recrystallization mechanisms in actively extruding salt fountain:
Microstructural evidence for a switch in deformation mechanisms with
increased availability of meteoric water and decreased grain size (Qum
Kuh, central Iran). Journal of Structural Geology, 32(4), 580-594.
Dienes, J.K., 1982. Permeability, percolation and statistical crack
mechanics. In: R.E. Goodman and F.E. Heuze (Editors), Issues in Rock
Mechanics. American Institute of Mining, Metallurgical and Petroleum
Engineering, New York, NY.
Escaig, B., 1968a. Sur le glissement de´vie´ des dislocations dans la
structure cubique a` faces centre´es, J. Phys., 29, 225– 239.
Escaig, B., 1968b. L’activation thermique des de´viations sous faibles
contraintes dans les structures h.c. et c.c, Phys. Status Solidi, 28, 463–
474.
Fontaine, G., 1968. Dissociation des dislocations sur les plans (110) dans les
cristaux ioniques du type NaCl. Journal of Physics and Chemistry of
Solids 29.2: 209-214.
255
Fontaine, G. & P. Haasen, 1969. Hydrostatic pressure and plastic
deformation of the alkali halides. physica status solidi (b) 31.1: K67-
K70.
Franssen, R.C. 1994. The rheology of synthetic rocksalt in uniaxial
compression. Tectonophysics, 233(1), 1-40.
Frost, H.J. & M.F. Ashby, 1982. Deformation mechanism maps: the
plasticity and creep of metals and ceramics. Pergamon press.
Gloyna, E.F. & T.D. Reynolds, 1961. Permeability measurements of rock
salt. Journal of Geophysical Research 66.11: 3913-3921.
Hatzor, Y.H. & Eli P. Heyman, 1997. Dilation of anisotropic rock salt:
Evidence from Mount Sedom diapir. Journal of Geophysical Research:
Solid Earth (1978–2012) 102.B7 : 14853-14868.
Heard, H.C. 1972. Steady‐State Flow in Polycrystalline Halite at Pressure of
2 Kilo bars. Flow and Fracture of Rocks, 191-209.
Heard, H.C. & F.J. Ryerson, 1986. Effect of Cation Impurities on Steady‐
State Flow of Salt. Mineral and Rock Deformation: Laboratory Studies:
The Paterson Volume, 99-115.
Houben, M.E., A. ten Hove, C.J. Peach and C.J. Spiers, 2013. Crack healing
in rocksalt via diffusion in adsorbed aqueous films: Microphysical
modelling versus experiments. Physics and Chemistry of the Earth, Parts
A/B/C 64 : 95-104.
Hudec, M.R. & M.P. Jackson, 2007. Terra infirma: understanding salt
tectonics. Earth-Science Reviews, 82(1), 1-28.
Hull, D. & D.J. Bacon, 1983. Introduction to Dislocations, 3rd ed., Int. Ser.
Mater. Sci. Technol., 37, Pergamon, New York. Kennedy, L. A., and J.
M. Logan, The role of veining and dissolution.
Hunsche, U., 1998. Determination of the dilatancy boundary and damage up
to failure for four types of rock salt at different stress geometries. In:
Aubertin, M., Hardy Jr., H.R. (Eds.), The Mechanical Behaviour of salt
IV. Proceedings of the fourth conference.
256
Hunsche, U., & A. Hampel, 1999. Rock salt—the mechanical properties of
the host rock material for a radioactive waste repository. Engineering
geology, 52(3), 271-291.
Istvan J.A., L.J. Evans, J.H. Weber & C. Devine, 1997. Rock mechanics for
gas storage in bedded salt caverns. International Journal of Rock
Mechanics and Mining Sciences 34, no. 3: 142-e1.
Jackson, M.P.A. & C.J. Talbot, 1986. External shapes, strain rates, and
dynamics of salt structures. Geological Society of America Bulletin 97.3:
305-323.
Karato, Shun-ichiro, 2008. Deformation of earth materials: An introduction
to the rheology of solid earth. Cambridge University Press.
Klinkenberg L.J. 1941. The permeability of porous media to liquids and
gases. Drill. Prod. Pract. 2, 200.
Liang, W.G., C.H. Yang, Y.S. Zhao, M.B. Dusseault & J. Liu, 2007.
Experimental investigation of mechanical properties of bedded salt rock.
International Journal of Rock Mechanics and Mining Sciences. 44(3),
400-411.
L. Wei, N. Muhammad, Y. Li, C.J. Spiers, C. Yang & H. Ma, 2014.
Experimental study of permeability of salt rock and its application to
deep underground gas storage. Yanshilixue Yu Gongcheng Xuebao, 33
(10), (pp. 1953-1961) (9 p.)
Liang, W.G., C. Zhang, H. Gao, X. Yang, S. Xu & Y. Zhao, 2012.
Experiments on mechanical properties of salt rocks under cyclic loading.
Journal of Rock Mechanics and Geotechnical Engineering. 4(1), 54-61.
McDonnell, R.D., C.J. Peach & C.J. Spiers, 1999. Flow behavior of fine‐
grained synthetic dunite in the presence of 0.5 wt. % H2O. Journal of
Geophysical Research: Solid Earth (1978–2012) 104.B8: 17823-17845.
Mehrer, H., 2011. Diffusion in Solids under Pressure. DIMETA-82 Defect
and Diffusion Forum (Vol. 309, pp. 91-112).
Mohamed, F.A. & T.G. Langdon, 1974. Method of estimating stacking‐fault
energies in alkali halide crystals using creep data. Journal of Applied
Physics, 45(5), 1965-1967.
257
Muhammad, N., C.J. Spiers, C.J. Peach & J.H.P. de Bresser, 2012. Effect of
confining pressure on plastic flow of salt at 125 oC. In: Bérest, P.,
Ghoreychi, M., Hadj-Hassen, F., and Tijani, M. eds. Mechanical
behaviour of salt VII, CRC press, pp. 57-64.
Muhammad, N., C.J. Spiers, C.J. Peach, J.H.P. de Bresser & W. Liu, 2015.
Permeability of layered rock salt at different stresses and geometries. In:
Roberts, Mellegard & Hansen (Eds). Mechanical Behavior of Salt VIII,
CRC press, pp. 23-33.
Peach, C.J., 1991. Influence of deformation on the fluid transport properties
of salt rocks. Geologica ultraiectina, 77, 1-238.
Peach, C.J. & C.J. Spiers, 1996. Influence of crystal plastic deformation on
dilatancy and permeability development in synthetic salt rock.
Tectonophysics, Vol.256, pp101-128.
Peach, C.J., C.J. Spiers, & P.W. Trimby, 2001. Effect of confining pressure
on dilatation, recrystallization, and flow of rock salt at 150 C. Journal of
Geophysical Research: Solid Earth (1978–2012) 106.B7: 13315-13328.
Pereira, J.M. & C. Arson, 2013. Retention and permeability properties of
damaged porous rocks. Computers and Geotechnics, 48, 272-282.
Poirier, J.P. 1976. On the symmetrical role of cross-slip of screw
dislocations and climb of edge dislocations as recovery processes
controlling high-temperature creep. Revue de Physique Appliqu'e, 11 (6),
731-738.
Poirier, J.P. 1985. Creep of crystals: high-temperature deformation
processes in metals, ceramics and minerals. Cambridge University Press.
Popp, T., W. Minkley, K. Salzer & O. Schulze, 2012. Gas transport
properties of rock salt—synoptic view. In: Berest, P., Ghoreychi, M.,
Hadj-Hassen, F. & Tijani, M.: Mechanical Behavior of Salt VII. Taylor
& Francis group, London N, ISBN 978-0-415-62122-9, 143-153.
Raj, Rishi, 1982. Creep in polycrystalline aggregates by matter transport
through a liquid phase. Journal of Geophysical Research: Solid Earth
(1978–2012)87.B6: 4731-4739.
258
Renner, Jörg & Brian Evans, 2002. Do calcite rocks obey the power-law
creep equation? Geological Society, London, Special Publications 200.1:
293-307.
Roedder, E. & R.L. Bassett, 1981. Problems in determination of the water
content of rock-salt samples and its significance in nuclear-waste storage
siting. Geology 9, no. 11: 525-530.
Rouby, D., S. Raillard, F. Guillocheau, R. Bouroullec & T. Nalpas, 2002.
Kinematics of a growth fault/raft system on the West African margin
using 3-D restoration. Journal of Structural Geology, 24(4), 783-796.
Rutter, E.H. & D.H. Mainprice, 1978. The effect of water on stress
relaxation of faulted and unfaulted sandstone. Pure and Applied
geophysics 116.4-5: 634-654.
Schulze, Otto, Till Popp & Hartmut Kern, 2001. Development of damage
and permeability in deforming rock salt. Engineering Geology 61.2: 163-
180.
Schutjens, P.M.T.M., 1991. Intergranular pressure solution in halite
aggregates and quartz sands: an experimental investigation. Geologica
Ultraiectina, 76, 1-233.
Seeger, A., 1956. The mechanism of glide and work hardening in face-
centered cubic and hexagonal close-packed metals, in Dislocations and
Mechanical Properties of Crystals, edited by J. C. Fisher, W. G.
Johnston, R. Thomson, and T. Vreeland, pp. 243–329, John Wiley, New
York.
Senseny, P.E., F.D. Hansen, J.E. Russell, N.L. Carter & J.W. Handin, 1992.
Mechanical behaviour of rock salt: phenomenology and
micromechanisms. International journal of rock mechanics and mining
sciences & geomechanics abstracts, vol. 29, no. 4, pp. 363-378.
Pergamon.
Sherby, O.D. & J. Weertman, 1979. Diffusion-controlled dislocation creep:
a defense. Acta Metallurgica 27, no. 3: 387-400.
Shimizu, I., 2008. Theories and applicability of grain size piezometers: The
role of dynamic recrystallization mechanisms. J. Struct. Geol. 30, 899-
917.
259
Simmons, G. & H. Wang, 1971. Single Crystal Elastic Con-stants and
Calculated Aggregate Properties: A Handbook, 2nd edition, 370 pp, MIT
Press, Cambridge, MA.
Skrotzki, W., O. Frommeyer, & P. Haasen 1981. Plasticity of
polycrystalline ionic solids. physica status solidi (a), 66 (1), 219-228.
Skrotzki, W. & Z.G. Liu, 1982. Analysis of the cross slip process in alkali
halides. physica status solidi (a), 73 (2), K225-K229.
Skrotzki, W. & P. Welch, 1983. Development of texture and microstructure
in extruded ionic polycrystalline aggregates. Tectonophysics 99.1: 47-61.
Spiers, C.J., J.L. Urai, G.S. Lister, J.N. Boland & H.J. Zwart, 1986. The
influence of fluid-rock interaction on the rheology of salt rock.
Commission of the European Communities, Luxembourg.
Spiers, C.J., P.M.T.M. Schutjens, R.H. Brzesowsky, C.J. Peach, J.L.
Liezenberg & H.J. Zwart, 1990. Experimental determination of
constitutive parameters governing creep of rocksalt by pressure solution.
Geological Society, London, Special Publications 54, no. 1: 215-227.
Spiers, C.J. & N.L. Carter, 1998. Microphysics of Rocksalt Flow in Nature,
The Mechanical Behaviour of Salt IV, Proceedings of the Fourth
Conference. M. Aubertin,H. R. Hardy Jr.,115–128, Trans. Tech.,
Clausthal-Zellerfeld,Germany.
Spiers, C.J. & P.M.T.M. Schutjens, 1999. Intergranular pressure solution in
NaCl: Grain-to-grain contact experiments under the optical microscope.
Oil & Gas Science and Technology, 54(6), 729-750.
Stocker, R.L. & M.F. Ashby, 1973. On the rheology of the up-per mantle.
Reviews of Geophysics, 11(2), 391-426.
Stormont, J.C. & J.J.K. Daemen, 1992. Laboratory study of gas permeability
changes in rock salt during deformation. International journal of rock
mechanics and mining sciences & geomechanics abstracts. Vol. 29. No.
4. Pergamon.
Stormont, J.C. 1997. Conduct and interpretation of gas permeability
measurements in rock salt. International Journal of Rock Mechanics and
Mining Sciences 34.3 : 303-e1.
260
Sutherland, H.J. & S.P. Cave,1980. Argon gas permeability of New Mexico
rock salt under hydrostatic compression. Int. J.Rock Mech. Min. Sci.
Geomech., Abstr., 17: 281-288.
Talbot, C.J. Fold trains in a glacier of salt in southern Iran. Journal of
Structural Geology 1.1 (1979): 5-18.
Tasker, P.W.& T.J. Bullough, 1981. An atomistic calculation of extended
planar defects in ionic crystals Application to stacking faults in the alkali
halides. Philosophical Magazine A 43.2: 313-324.
Ter Heege, J.H., J.H.P. De Bresser & C.J. Spiers, 2005a. Rheological
behaviour of synthetic rocksalt: The interplay between water, dynamic
recrystallization and deformation mechanisms. J. Struct. Geol. 27, 948-
963.
Ter Heege, J.H., J.H.P. De Bresser & C.J. Spiers, 2005b. Dynamic
recrystallization of wet synthetic polycrystalline halite: dependence of
grain size distribution on flow stress, temperature and strain.
Tectonophysics 396.1: 35-57.
Twiss, R.J., 1977. Theory and applicability of a recrystallized grain size
paleopiezometer. Pure and applied Geophysics 115, 227-244.
Urai, J.L. 1983. Water assisted dynamic recrystallization and weakening in
polycrystalline bischofite. Tectonophysics 96.1: 125-157.
Urai, J.L., 1985. Water-enhanced dynamic recrystallization and solution
transfer in experimentally deformed carnallite. Tectonophysics, 120(3),
285-317.
Urai, J.L., C.J. Spiers, H.J. Zwart & G.S. Lister 1986. Weakening of rock
salt by water during long-term creep.
Urai, J.L., C.J. Spiers, C.J. Peach, R.C.M.W. Franssen & J.L. Liezenberg,
1987. Deformation mechanisms operating in naturally deformed halite
rocks as deduced from microstructural investigations. Geologie en
Mijnbouw 66, no. 2: 165-176.
Van Eekelen, H.A., T. Hulsebos & J.L. Urai, 1984. Creep of bischofite, The
Mechanical Behavior of Salt: Proceedings of the First Conference Held
261
at the Pennsylvania State University, University Park, Pennsylvania,
November 9-11, 1981. Gulf Pub Co.
Vendeville, B. Ci & M.P.A. Jackson, 1991. The rise of diapirs during thin-
skinned extension. Marine and Petroleum Geology 9, no. 4 : 331-354.
Verrall, R.A., R.J. Fields & M.F. Ashby, 1977. Deformation‐Mechanism
Maps for LiF and NaCl. Journal of the American Ceramic Society 60,
no. 5‐6: 211-216.
Walsh, J.B., 1981. Effect of pore pressure and confining pressure on fracture
permeability. Int. J. Rock Mech. Min. Sci. and Geomech. Abstr., Vol. 18,
pp. 429-435.
Watanabe, T. & C.J. Peach, 2002. Electrical impedance measurement of
plastically deforming halite rocks at 125 oC and 50 MPa. Journal of
Geophysical Research: Solid Earth (1978–2012), 107(B1), ECV-2.
Wawersik, W.R. & D.H., Zeuch, 1986. Modeling and mechanistic
interpretation of creep of rock salt below 200 oC. Tectonophysics, 121(2),
125-152.
Weertman, J., 1955. Theory of Steady‐State Creep Based on Dislocation
Climb. Journal of Applied Physics 26, no. 10: 1213-1217.
Weertman, J., 1957. Steady‐State Creep of Crystals. Journal of Applied
Physics, 28(10), 1185-1189.
Weertman, J., 1968. Dislocation climb theory of steady-state creep. ASM
TRANS QUART 61.4: 681-694.
Werner M. & H. Mehrer, 1983. Diffusion in Metals and Alloys, F.J.
Kedves, D.L. Beke (Eds.), Trans Tech Publications, Diffusion and
Defect Monograph Series No. 7, 392.
Wolf, H. 1960. Die Aktivierungsenergie für die Quergleitung aufgespaltener
Schrauben versetzungen, Z. Naturforsch, 15A, 180– 193.
Zhou, H., J. He & Z. Wu, 2009. Permeability and meso-structure
characteristics of bedded salt rock. Chinese Journal of Rock Mechanics
and Engineering, 28(10), 2068-2073.
262
263
Samenvatting
264
De aanwezigheid van evaporitische lagen in sedimentaire bekkens,
vaak gedomineerd door het zoutmineraal haliet, heeft grote invloed op
de ontwikkeling van structuren tijdens tektonische activiteit. Ook zijn
op veel plekken in evaporitische zoutlagen cavernes te vinden die
ontstaan zijn door oplossingsmijnbouw. Deze cavernes worden
steeds meer gebruikt voor strategische opslag van energiebronnen in
de vorm van gasvormige of vloeibare brandstoffen, en voor de opslag
van energie tijdens daluren in de vorm van perslucht. Daarnaast
worden conventionele mijngangen gebruikt voor lange termijn opslag
van gevaarlijk afval. De lage permeabiliteit van de meeste
zoutgesteenten, met lage mechanische sterkte en de neiging tot zelf-
afdichting, vormt de belangrijkste reden voor ingenieurs om
zoutformaties te gebruiken voor dergelijke opslagprojecten. De
Zechstein zoutafzettingen in de ondergrond van Nederland bestaan uit
zowel de oorspronkelijke afzettingen als gemigreerde zout
diapieren/pijlers die kunnen zijn uitgegroeid tot op kilometer-schaal.
Sommigen zoutafzettingen bevatten economisch waardevolle
magnesiumrijke zouten welke worden gewonnen door middel van
oplossingsmijnbouw. De cavernes die op deze manier ontstaan
midden in deze gemakkelijk oplosbare en mechanisch zwakke
materialen brengen verdere uitdagingen met zich mee met betrekking
tot lange-termijn opslag van afval. Zorgvuldig beheer van de met
water gevulde cavernes vereist een uitgebreide kennis van het
reologisch- en transportgedrag van de omliggende zout formaties, ten
einde de veiligheid van toekomstige exploitatie te garanderen.
265
Dit proefschrift adresseert een aantal tekortkomingen in de huidige
kennis over het mechanisch gedrag van zout, met name met
betrekking tot kruip van haliet en van mengsels van haliet met
magnesium zouten. In het bijzonder is de afhankelijkheid van kruip
van de alzijdige druk onderzocht, gebruik makend van synthetisch en
natuurlijk halietgesteente, met als doel de samengestelde
mechanismen te identificeren welke bijdragen aan het totale
deformatie proces. Dit alles met het oog op het beter begrijpen van de
natuurkunde van het kruipen van zout. Tevens is de reologie van
magnesium bevattende zouten, carnalliet en bischofiet, onderzocht,
met speciale aandacht voor het mechanische gedrag van mengsels van
deze zouten met haliet. Naast de reologische aspecten zijn ook de
effecten van de samengestelde gelaagdheid op de doorlaatbaarheid
(permeabiliteit) van door uitgraving beschadigde zones rondom
gemijnde holtes bestudeerd, gebruik makend van natuurlijk gelaagd
materiaal uit mijnen in China. Doel hierbij was om beter inzicht te
krijgen in de effecten van de oriëntatie van de gelaagdheid ten
opzichte van de deformerende differentiaalspanning in holtes en
mijngangen op diepte.
Hoofdstuk 1 omvat een literatuurstudie met betrekking tot zout
reologie, en beschrijft de hoofddoelen van deze studie en de planning
van het proefschrift.
Hoofdstuk 2 gaat over het snelheidsbepalende mechanisme van de
kruip van droog polykristallijn zout. Het temperatuurbereik dat dit
deel van de studie omvat is 22 tot 350 °C. De gebruikte
266
vervormingssnelheden liggen tussen de 4x10-7
s-1
en 10-4
s-1
. Een
aantal testen werden uitgevoerd bij constante vervormingssnelheid,
andere testen omvatten systematische stappen in de
vervormingssnelheid. De gedachte achter dit gedeelte van de studie is
om onderscheid te maken tussen verschillende mechanismen van
dislocatiekruip, zoals het klimmen van dislocaties (“dislocation
climb”), het schuiven van dislocaties van het ene glijvlak naar het
andere (“dislocation cross-slip”) en het glijden van dislocaties binnen
één vlak (“dislocation glide”), gebruik makend van de druk-
afhankelijkheid van de vloeispanning van droog zout. De theorie stelt
dat er een atomistische activeringsenergie of -volume is, welke
geassocieerd is met vastestofvloei door dislocatie kruip. Volgens
theoretische modellen zou het zout in het geval van een positief
activeringsvolume, bij hogere druk een grotere sterkte moeten
hebben, gerelateerd aan het snelheidsbepalende mechanisme van
kruip. Voor deze analyse werd de omvattende (alzijdige) druk
gevarieerd tussen 50 en 600 MPa, iets wat nog niet eerder
systematisch was gedaan voor volledig omvat zout. Het zout werd
sterker bevonden bij hogere omvattende druk, en op basis van de
experimentele gegevens werd geconcludeerd dat het mechanisme dat
de snelheid van kruip bepaalt, onder de geteste omstandigheden, in
het overgangsgebied ligt van dislocatie glij naar dislocatie klim.
Hoofdstuk 3 gaat over de vloeiwet bij in situ condities van water
houdend haliet, gebruik makend van synthetische en natuurlijke zout
monsters. De toegepaste temperatuur en omvattende druk waren
267
125 °C en 50 MPa, respectievelijk. De monsters werden getest in
multi-stap experimenten met constante vervormingssnelheid gedeeltes
(5x10-5
– 5x10-8
s-1
), gevolgd door spanningsrelaxatie na
geselecteerde stappen. De relaxatie data laten zien dat het
snelheidsbepalende mechanisme niet hetzelfde blijft, maar verandert
van kruip gecontroleerd door dislocatie activiteit (bij hogere
spanningen en vervormingssnelheden), naar korrelgrootte-
afhankelijke kruip (waarschijnlijk drukoplossing) tegen het einde van
de relaxatie (i.e. bij lagere spanningen en reksnelheden).
Hoofdstuk 4 is een compilatie van testen gedaan op bischofiet,
carnalliet en hun mengsels met haliet, uitgevoerd onder in situ
condities met een omvattende druk van 40 MPa en een temperatuur
van 70 °C. Deze testen werden uitgevoerd op polykristallijne
monsters verkregen uit boorkernen van natuurlijk materiaal. Hierbij
werden vochtabsorberende condities toegepast rondom de monsters.
De monsters werden op dezelfde manier getest als de haliet
gerapporteerd in Hoofdstuk 3, namelijk in multi-stap experimenten
met constante vervormingssnelheid gedeeltes (10-5
– 10-8
s-1
), gevolgd
door spanningsrelaxatie na geselecteerde stappen. De resultaten
toonden aan dat de carnalliet sterker is dan de bischofiet, en dat
mengsels met haliet op hun beurt sterker zijn dan carnalliet. De
sterkte van het mengel lijkt direct gerelateerd te zijn aan het massa
percentage van haliet, i.e. hogere massa percentages haliet maken het
mengsel sterker. De samenstelling van de verschillende monsters
werden getest met behulp van een micro-XRF techniek.
268
Spanningsrelaxatie toonde aan dat een verandering in het
vloeimechanisme plaatsvond, van kruip gecontroleerd door dislocatie
activiteit bij hogere spanningen en reksnelheden (aan het begin van
de relaxatie) naar korrelgrootte-afhankelijke kruip bij lagere
spanningen en reksnelheden (tegen het einde van de relaxatie). Op
basis van de resultaten werden samengestelde vloeiwetten voor
bischofiet en carnalliet opgesteld, met een combinatie van
korrelgrootte-onafhankelijke dislocatie kruip en korrelgrootte-
afhankelijke (drukoplossing) kruip die het mechanisch gedrag bij
hogere en lagere spanningen beschrijven, respectievelijk.
Hoofdstuk 5 gaat over de transporteigenschappen van gelaagde zout
uit zout mijnen in China. De monsters werden getest op hun
permeabiliteit langs het contact van twee verschillende lagen, haliet
en glauberiet, gebruik makend van de argon gas permeametrie
techniek, onder toenemende differentiaalspanningen. De
experimenten werden geclassificeerd in twee groepen: groep-I
omvatte relatief grote omvattende druk (20 MPa, releatief grote
diepte), en groep-II omvat kleinere omvattende druk (10 MPa,
kleinere diepte). Verschillende oriëntaties van het contact ten
opzichte van de richting van deformatie werden getest (verticaal,
onder een hoek en horizontaal) waarbij de differentiaalspanning werd
vergroot in stappen van 10 MPa. De axiale en volumetrische
vervormingen werden tijdens de experimenten gemeten, wat
aantoonde dat alle monsters compacteerden, ondanks en toename of
afname in permeabiliteit. Microstructurele analyse toonde aan dat
269
lokale dilatatie plaatsvond op de contacten, wat een lokale toename
van de permeabiliteit moet hebben veroorzaakt, maar welke werd
verborgen door de bulk compactie van de samples.
270
271
Acknowledgements
272
I am really grateful to all my teachers for their endless support,
encouragement and feedback throughout my studies. Special thanks to
Professor Chris Spiers for his loving nature and generosity that he arranged
all the funds and enabled me to continue my studies. I have learnt many
teaching skills while assisting him during practical classes. Dr. Colin Peach
is thanked for sharing his knowledge and explaining the complications of
state of art equipment in HPT laboratory, and solving the technical issues in
a wink of an eye. The efforts of Dr. Hans de Bresser are highly appreciated,
for his timeless, prompt and constructive feedbacks on write-ups, which
helped the work go smoothly.
All staff of the Higher Education Commission (HEC) of Pakistan, especially
former chairman Dr. Atta-Ur-Rahman is heartily thanked for initiating PhD
scholarships. The programme is very useful for upgrading the education
level in our country. The support of my alma mater GC University Lahore is
also highly appreciated.
I also want to thank to my “paranimfen”, Amir Raoof and Martijn Van den
Ende, who, other than taking care of my thesis, were very helpful in making
the computer code for complex data analysis.
While staying in Netherlands, I felt like at home. Especially at HPT
laboratory, all members were very co-operative. I want to thank all of you.
My room fellows Anne Pluymakers, Amir Raoof, Tim Wolterbeek and Miao
Zhang were all very helpful in creating a good environment for studies.
Bart, Andre, Jon, Suzanne, Ross, Chen, Elisenda, Michiyo, Yu, Luuk,
George, Ayumi, Evangelos, Mariska and Ronald were all very caring
fellows. We had great time throughout my research period at HPT. Caring
273
Magda will always be remembered for her kindness and sincerity. All the
technical staff, Peter Van Krieken, Gert Kastelein, Eimert de Graaff and
Thony Van de Gon Netscher is thanked for their fantastic support. All of my
country fellows in Netherlands; Imran, Bilal, Fraz, Qamar, Mehboob, Zahid
are thanked for their moral support and hospitality.
Deep respect for my sincere teachers at all levels of my studies. My wife
Nida and our little angel like daughter Khadeejah were very supportive in
boosting up my morale. At the end, I want to say special thanks to my
parents, who were patient enough to allow me to have higher education
abroad. My brother Eijaz, sister Humaira, students Attique and Tahir and
sincere friend Qaiser Butt were looking after back home in my absence. This
milestone was not being possible to achieve without their support.
Nawaz Muhammad
274
275
Curriculum vitae Full Name Nawaz Muhammad
Date of birth 15th December 1975
Place of birth Lahore Pakistan
Matriculation Government Iqbal Hussain
high school Lahore.
Intermediate and graduation Government Islamia
college Railway road
Lahore
Masters (Physics) Government College
University Lahore.
Teaching/job experience at CASP GC University Lahore:
Research Assistant 2001-2002
Lecturer 2002-2006
Assistant professor 2006 to date
276