You are on page 1of 21

International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

Journal Review Article


The Kaiser effect in rocks: principles and stress
estimation techniques$
A. Lavrov
Physical and Engineering Department, Moscow State Mining University, Leninskij prospect 6, Moscow 119 991, Russia
Accepted 2 December 2002

Abstract

Irreversible deformation of rocks is accompanied by the Kaiser effect involving acoustic emission (AE). AE activity increases
dramatically as soon as the largest previously experienced stress level is exceeded, similar to strain hardening in ductile materials.
The Kaiser effect can be considered as a basis for stress measurements in rocks. This phenomenon is characterized by complexity
which is the reason for disputable data interpretation procedures. A review of experimentally established features of the Kaiser
effect, including those under a triaxial state of stress, is given. Mechanisms and theoretical models of the phenomenon are discussed.
Stress measurement techniques based on the Kaiser effect in core specimens and in rocks around a borehole are explained; their
strengths and weaknesses are analyzed. The necessity and the directions for future research are outlined. The analysis given should
facilitate future applications of the Kaiser effect and improve the correctness of data interpretation.
r 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Kaiser effect; Acoustic emission; Rocks

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
2. Basic experimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
2.1. Indicators of the Kaiser effect; processing of AE curves . . . . . . . . . . . . . . . . . . . . . . . 153
2.2. Rock type and loading stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
2.3. What stress level is memorised by rock under cyclic uniaxial compression? . . . . . . . . . . . . . 155
2.4. Duration of load application; Kaiser effect after creep and stress relaxation . . . . . . . . . . . . 156
2.5. Influence of time delay, water saturation and heating . . . . . . . . . . . . . . . . . . . . . . . . 156
2.6. Influence of the loading rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
2.7. Stress memory formation under triaxial stress state . . . . . . . . . . . . . . . . . . . . . . . . . 157
2.8. The effect of rotation of principal axes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
2.9. Influence of the coring process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
2.10. Kaiser effect at different scales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
3. Theoretical models of the Kaiser effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
4. Practical application of the Kaiser effect for stress measurements . . . . . . . . . . . . . . . . . . . . . 164
4.1. Stress measurement using the Kaiser effect in core specimens . . . . . . . . . . . . . . . . . . . . 164
4.2. Stress measurement using the Kaiser effect in rocks surrounding a borehole . . . . . . . . . . . . 166

$
This is the first of a series of Journal Review Articles commissioned by the Editor. The series will consist of articles explaining significant or
topical subjects, or subjects requiring expert explanation. This first article falls into all three categories: the estimation of rock stress is critical for
modelling input; the International Society for Rock Mechanics (ISRM) Commission on Testing Methods is currently producing new Suggested
Methods for stress estimation; and the Kaiser effect, although apparently simple, requires expert explanation.
E-mail address: avlavrov@mail.ru (A. Lavrov).

1365-1609/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S1365-1609(02)00138-7
152 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

5. Conclusions and future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167


Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168

1. Introduction certain moment. As seen in Fig. 1, an inflexion occurs as


the largest stress level of the first cycle is exceeded in the
Design and operation of underground excavations second cycle. Hit rate, i.e. the number of AE signals
require accurate information about in situ stresses. At per second or per 1 MPa stress increment, increases
present, different stress measurement methods in rocks dramatically at that point.
are known [1]. However, most of them are time- and When rock is cyclically loaded within a certain stress
labor-consuming (e.g. the overcoring and hydrofractur- range, e.g. between stress values of s0 and s00 ; there is a
ing methods), and/or do not provide the necessary gradual decrease in the number of AE signals received in
accuracy. This is the reason why new stress measure- successive cycles [5]. The largest drop in the AE activity
ment techniques are being explored. One of the takes place between the first and second cycle (the
possibilities consists in using the so-called ‘stress Kaiser effect). As pointed out in Ref. [6], this is similar
memory’ of rocks, i.e. the ability to accumulate, to to the behaviour of irreversible strain in similar tests.
retain and, under certain conditions, to reproduce The Kaiser effect was discovered by Joseph Kaiser
information on the stresses experienced in the past. This who performed his experiments on small specimens of
ability is essentially due to irreversible damage accumu- metals, wood and sandstone in the early 1950s [7]. A first
lated in rock when it is subject to loads [2–4]. Particular attempt to investigate the Kaiser effect in a particular
manifestations of the stress memory are referred to as rock (sandstone) was made by Goodman [8]. In the later
memory effects, amongst which the Kaiser effect and work by Kurita and Fujii [2], the influence of water
strain-hardening are probably the most well-known and saturation and time delay between loading cycles on the
most interesting from the practical point of view. Kaiser effect were studied. It turned out that the stress
The Kaiser effect takes place in rocks and materials memory of rocks is similar to the human’s one: it
subjected to cyclic loading/unloading. In the simplest disappears with time.
case of cyclic, uniaxial loading with the cycle’s peak Since the early days, the Kaiser effect has been
stress increasing from cycle to cycle, the acoustic considered as a possible basis for stress measurements in
emission (AE) is zero or close to the background level rocks. One of the first attempts of such application was
so long as the stress remains below the largest previously that by Kanagawa and Nakasa [9]. Typical for those
reached stress value, sm : As this peak (‘memorised’) early applications was that the influence of the triaxial in
stress value is attained, the AE activity increases situ stress state on the damage formation was not taken
dramatically. As an example, a schematic plot of the into account. It was believed that the normal stress
dependency of cumulative AE hits versus stress in two component in each direction can be simply retrieved by
cycles of uniaxial compression is given in Fig. 1. The uniaxially reloading a specimen oriented in that direc-
cumulative hits number is the total number of AE tion. Experiments [10–16] showed that this is not
signals received from the beginning of the test to a correct. As was pointed out by Holcomb [12], a full
understanding of the role of the triaxial state of stress in
NΣ the memory formation is the main prerequisite for a
correct application of the Kaiser effect for stress
measurements. This and other fundamental questions
that should be answered to make a practical application
of the Kaiser effect feasible are summarized below (in
Second cycle the order that they are discussed in the paper):
(1) What are the most reliable indicators of the Kaiser
effect?
(2) How does the Kaiser effect differ in various rock
First cycle types (brittle/ductile; hard/soft; compact/damaged)
and at different loading stages (pre- and post-
σ dilatant; pre- and post-failure)?
Fig. 1. Cumulative AE hits, NS ; versus stress, s; in two successive (3) What stress level is memorised by rock under cyclic
cycles of uniaxial compression. uniaxial compression: the largest experienced in
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 153

recent history; the last; the longest with respect to less pronounced. In the best case, there are no AE
the exposure time; all stress levels? What is the signals so long as the stress stays below its largest
effect of palaeostresses? previously reached value, sm ; and AE suddenly emerges
(4) How does the duration of the load application at this stress level (Fig. 2a). However, sometimes AE
influence the Kaiser effect? Does the Kaiser effect activity may be present at stresses below the ‘memorised’
exist in ductile rocks subjected to creep or stress stress sm ; being probably due to the frictional movement
relaxation? along existing cracks; while at s > sm it is due to the
(5) How do factors such as time delay, heating, and crack growth [17]. In such a case, the Kaiser effect can
water saturation influence the Kaiser effect? be recognised as an inflexion (change of slope) in the
(6) How does the loading rate influence the Kaiser dependency ‘‘cumulative AE hits versus stress’’
effect? (Fig. 2b). To determine the value of sm more accurately,
(7) What stress (or stress combination) is memorised the curve may be approximated by two straight lines
by rock under a triaxial stress state, in particular (bilinear regression). The point of their intersection
under a true triaxial, i.e. polyaxial, stress state projected onto the stress axis indicates the Kaiser effect
(s1 > s2 > s3 )? stress. A pivot point method of choosing the sets of
(8) How does the rotation of principal axes in measurement points, on which regression is to be
successive loading cycles influence the Kaiser performed, was proposed in Ref. [18].
effect? An alternative to performing regression consists in
(9) How does the stress re-distribution during coring drawing tangents to the two parts of the curve (below
influence the capability of the Kaiser effect to and above the assumed sm value) and searching for their
reveal the in situ damage? intersection.
(10) Is the Kaiser effect scale-dependent? Instead of cumulative hits, the dependency of hit rate
’ means the number
versus stress can be used. Hit rate, N;
At present, answers to most of these questions are
known. These answers have shown that, at least in some
situations, stress measurements based on the Kaiser NΣ
effect are feasible, but that certain restrictions exist that
should always be taken account when performing such
measurements. The purpose of this Review Article is to
provide a concise and thorough overview of experi-
mental and theoretical results that are important for
practical applications of the Kaiser effect in rock
mechanics. The need for such a review is clear when
looking at the contradictory information on the easy
possibility or complete impossibility of such applica-
tions. The overview will hopefully contribute to a better
understanding of strengths, weaknesses and limitations σ
(a) σm
of stress measurement techniques based on the Kaiser
effect.
The paper is structured as follows. Basic experimental NΣ
results concerning the Kaiser effect under various
loading conditions are reviewed in Section 2. Theoretical
models are discussed in Section 3. In Section 4, stress
measurement techniques using the Kaiser effect in core
specimens or in rocks around a borehole are considered
followed by conclusions and discussion of further
research in Section 5.

2. Basic experimental results

2.1. Indicators of the Kaiser effect; processing of AE (b) σm σ


curves
Fig. 2. Cumulative AE hits, NS ; versus stress, s, in the second cycle of
uniaxial compression. The peak stress value attained in the first cycle
Even in the simplest case of cyclic uniaxial compres- of uniaxial compression was equal to sm : AE activity may be zero
sion, the Kaiser effect may be characterized as more or (a) or non-zero (b) at low stress values in the second cycle.
154 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

of AE signals received per second or per 1 MPa stress N


·

increment. Bilinear regression or the method of two


tangents can be applied in this case in the same manner
as for the curves ‘‘cumulative hits versus stress’’ [19].
’ 2 may be used. The
Instead of the hit rate, its square ðNÞ
I
’ 2
dependency of cumulative ðNÞ values versus stress was
reported to show a sharper Kaiser inflexion than the II
dependency of cumulative hits versus stress [19].
Another technique of AE curve processing implies
calculating the curvature of the curve ‘‘cumulative AE
hits versus stress’’. The stress value corresponding to the
maximum curvature is assumed to be the Kaiser effect σ
0 σm
stress [20]. A method similar to the maximum curvature
method was recently proposed by Villaescusa et al. [21] Fig. 3. AE hit rate, N, versus stress s in the first (I) and second (II)
who plotted the normalised slope variation of the curve reloading cycles of uniaxial compression of a rock specimen which was
‘‘cumulative AE versus stress’’. A maximum in the preloaded in uniaxial compression to the peak stress value of sm (based
on results in Ref. [17,22]).
variation occurs at the ‘memorized’ stress value.
In the tests, in which high AE activity at lower stresses
is observed in each loading cycle, a technique developed FR
by Yoshikawa and Mogi can be employed [17,22].
According to this technique, two reloading cycles are
1.0
performed, with the same maximum stress, that must be
higher than the preloading maximum stress, sm :
Dependencies of AE hit rate versus stress are plotted
for each reloading cycle. The stress value at which these
two curves separate from each other, corresponds well
with the maximum preload stress sm : At stresses less
than sm ; these two curves virtually coincide; whereas,
from this point on, the curve for the second reloading
cycle runs below that for the first reloading, and the
difference between them increases with stress (Curves I
0 1.0 σm / σUCS
and II in Fig. 3). Plotting the difference between the Fig. 4. FR in the second cycle of uniaxial compression versus peak
dependencies ‘‘hit rate versus stress’’ in the first and stress of the first cycle of uniaxial compression normalized by the
second reloading cycles yields the value of sm quite rock’s uniaxial compressive strength (based on results in Ref. [13]).
accurately, even when considerable low-stress noise is
present in each cycle [22]. towards lower stresses. This move can be quantitatively
characterised by the so-called Felicity Ratio (FR) which
2.2. Rock type and loading stage is the ratio of the AE onset stress value to the previous
peak stress. The dependency of the FR on the preloading
By now, the Kaiser effect has been observed in stress for a brittle rock is schematically shown in Fig. 4
various types of rocks, e.g. granite, granodiorite, [13,17].
quartzite, andesite, basalt, marble, schist, gneiss, tuff, In order to obtain a well-pronounced Kaiser effect,
sandstone, limestone, dolomite, siltstone, shale, rock the preload stress should be in the range from about
salt, potash rocks, cohesive and granular soils, coal 30% to about 80% of the ultimate strength. This range
[1–5,7–44]. is certainly rock-dependent. The lower boundary of this
In brittle rocks, the effect is best-pronounced if the range (20–30% of the strength) is due to the fact that
preloading stress does not exceed the stress value at below this boundary, little AE is detected anyway, even
which dilatancy begins. If during preloading the without preloading, and this ‘silence’ can be misinter-
dilatancy threshold was exceeded, the Kaiser effect preted as the Kaiser effect.
during reloading becomes less clear, because AE activity In ductile rocks, e.g. rock salt, the Kaiser effect is well
at lower stresses appears which makes difficult or totally pronounced both before and after the dilatancy onset.
inhibits the Kaiser effect recognition [2]. The closer the Moreover, the effect is quite well-pronounced in the
preloading stress to the ultimate strength of a brittle post-failure region on the descending part of the stress–
rock, the less-pronounced the Kaiser effect is during strain curve [45]. The effect takes place when the current
reloading. It can be said that, while the preload moves strain of the specimen exceeds the largest previously
towards the strength of rock, the Kaiser effect moves reached strain value. It may be said that during
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 155

post-failure cyclic deformation, plastic rock ‘recollects’ cycle to cycle, the Kaiser effect takes place in each cycle
the maximum value of strain rather than stress [45,46]. at the stress value equal to the peak stress of the last
Displacement controlled loading should be employed cycle [40,48,49]. In such a loading regime, this is also the
in order to observe the Kaiser effect in the post-failure largest stress value reached in the recent loading history
stage. of the rock.
Interesting results were obtained by Dunning et al. If three loading cycles are performed with the
[47] in their experiments on cylindrical specimens of maximum stress in the second cycle ( sII 1 ) lower than
sandstone with a sawcut surface. The surface, inclined at in the first cycle ( sI1 ), and with the maximum stress in
351 to the s1-axis, was intended to simulate a fault in the the third cycle ( sIII III
1 ) higher than in the first one ( s1 >
I II
Earth crust. Two cycles were performed on each s1 > s1 ; Fig. 5a), the situation is more complex.
specimen. A very clear Kaiser effect was observed in Namely, if the second cycle is carried out immediately
the second cycle if the maximum load of the first cycle or very soon after completing the first cycle and the
was 60% of the yield stress, i.e. macrofailure did not specimen is not held at the peak stress in the second
occur in the first cycle. If macrofailure in the form of cycle, i.e. unloading immediately follows loading, the
stick-slip occurred in the first cycle, the quality of the Kaiser effect in the third cycle is observed at the stress
Kaiser effect in the second cycle was poor. However, if equal to the first-cycle peak stress [40,49] (Fig. 5b.)
in the first cycle macrofailure occurred in the form of In this case, the rock ‘recollects’ the largest stress
stable sliding, a very well pronounced Kaiser effect was level reached in its recent stress history, as postulated
measured in the second cycle [47]. This corresponds to in Ref. [19].
some extent to the observation made on plastic rock that If, however, there is a time delay between the first and
the Kaiser effect in the second cycle exists if the second cycle, sufficient for stress memory recovery
specimen did not loose integrity in the first cycle [45]. (Fig. 5c), the Kaiser effect in the third cycle may be
observed at the peak stress of the second cycle (Fig. 5d).
2.3. What stress level is memorised by rock under cyclic The influence of time delay on the Kaiser effect decay is
uniaxial compression? discussed in more detail in Section 2.5.
Michihiro et al. [50] carried out experiments on tuff,
If a rock specimen is loaded in cyclic uniaxial mudstone, sandstone and crystal schist in which three
compression with the cycle’s peak stress increasing from uniaxial load cycles were performed on each specimen.

σ σ

σm σm

III
I
III
I
II II

(a) (c) t
t

NΣ NΣ

III III

II II

I I

σm σ σm σ
(b) (d)
Fig. 5. Uniaxial compressive stress versus time (a, c) and cumulative AE hits versus stress (b, d) in three-cycle tests with the peak stress of the second
cycle less than the peak stress of the first cycle. The peak stress of the third cycle is larger than the peak stress of the first cycle. The time delay between
the first and second cycle is zero (a, b) or non-zero (c, d).
156 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

The relation between the largest stresses of the cycles the rock was held during the creep (FR down to 0.3 for
was as follows: sIII I II
1 > s1 > s1 : The tests have shown that mudstone) [50].
if the specimen was held at the maximum load of the
second cycle until strain recovery finished, the Kaiser 2.5. Influence of time delay, water saturation and heating
effect in the third cycle was observed at the peak stress
of the second cycle, although it was smaller than the Increasing time delay between successive loading
maximum stress of the first cycle. cycles results in decreasing quality of the Kaiser effect.
Hence, in general, it is not possible to state that rock If the second loading cycle follows, not immediately,
‘memorizes’ the maximum stress reached during recent but with a certain time delay after completing the first
stress history. The phenomenon is more complicated. cycle, AE activity appears in the second cycle at stress
The answer is dependent on the duration of the time values lower than the peak stress of the first cycle [13].
interval between preloading and reloading as well as on The inflexion becomes not so clear as in the case of zero
the duration of the load application during preloading. time delay.
This result is important for the application of the effect The extent of the Kaiser effect decay in the course of
for stress measurements. time is rock dependent. Extremely short values of the
Can palaeostresses that were acting millions of years time delay (of the order of several hours) sufficient to
ago have formed damage that could be revealed by the completely eliminate the Kaiser effect were reported by
Kaiser effect in laboratory tests? Due to the long time Goodman for sandstone and quartz diorite [8]. The
that has passed since the palaeostresses were applied, the results of Ref. [8] were disputed and explained in
Kaiser effect seems likely to indicate the maximum different ways by Holcomb [10] and Yoshikawa and
stresses applied in more recent history. Certainly the Mogi [17]. For gneiss, schist, carbonate, mudstone,
duration of stress application and the stress level limestone and sandstone, Koerner and Lord [53]
(percentage of strength) play their parts as well. reported that the Kaiser effect was well-pronounced
when the time delay between the cycles was less than
10 h. When the delay was more than 1000 h, the effect
2.4. Duration of load application; Kaiser effect after disappeared completely [53]. In rock salt, the Kaiser
creep and stress relaxation effect could be unambiguously identified after time
delays of up to 14 days [40]. In a brittle rock, i.e. granite,
Results of the experiments performed by Michihiro Kaiser effect was well-observed after a 300-day time
et al. [50] on tuff, mudstone, sandstone and crystal schist delay [54]. Another research team reported a much more
suggest that the larger the duration of stress application substantial decay of the Kaiser effect in granite: the FR
during preloading, the clearer the Kaiser effect will be in became 0.78 after a rest of 7.5 days. After a 20-day rest,
the succeeding reloading test. However, for a brittle the value of FR became 0.44, i.e. the Kaiser effect was
rock, andesite, no significant influence of the duration of lost [55]. A monotonic decrease of FR with time between
load application on the succeeding Kaiser effect could the cycles was observed in Ref. [55] and is schematically
be found ([17], where the load application duration shown in Fig. 6. The Kaiser effect degradation in the
ranged from 1 to 1000 s). The influence of the load course of time was linked by Michihiro et al. [55] to
application duration should be more pronounced in anelastic strain recovery.
ductile rocks. As pointed out by Yoshikawa and Mogi [17], the
In situ rocks may undergo complex loading regimes, dependency of the Kaiser effect on time is sensitive to
including creep and stress relaxation. In order to the value of the preloading stress: the closer the
investigate the Kaiser effect in these regimes, two-cycle
tests were performed in which the first cycle was a FR
uniaxial creep test, i.e. constant stress, or uniaxial
loading followed stress relaxation, i.e. constant strain,
1.0
followed by complete unloading of the specimen; the
second cycle was uniaxial reloading with stress increas-
ing from 0 to the ultimate strength. In ductile rocks like
clay and rock salt, a well-pronounced Kaiser effect was
observed in the second cycle in such loading regimes
[44,51,52]. In experiments by Michihiro et al. [50],
perfect Kaiser effect was observed in the second cycle, if
strain stabilization was reached during the first-cycle
creep. If rock deformation did not stop during creep, i.e. 0 t
strain did not stabilize, the Kaiser effect stress in the Fig. 6. FR in the second cycle versus time delay between the first and
following reloading is smaller than the stress at which second cycle of uniaxial compression (based on results in Ref. [55]).
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 157

preloading stress to the ultimate strength, the faster the impact processes. The dependency of the Kaiser effect
Kaiser effect decay. on loading rate should be even more pronounced in
Kaiser effect degradation in the course of time can be plastic and soft rocks.
enhanced by water saturation or heating of the specimen If the ratio of the loading rates in the first and second
inbetween the load cycles. Water saturation can impose cycle is inverse (first—slow, second—fast), the Kaiser
a severe influence on the Kaiser effect, even in a low- effect in the second cycle is well-observed. It occurs at
porosity rock like granite [2]. A comprehensive investi- the stress slightly higher than the first-cycle peak stress,
gation of the influence of water saturation and heating i.e. FR>1.0 [42]. This result obtained on a brittle
on the Kaiser effect in a brittle rock, i.e. andesite, was limestone should be additionally verified for plastic
carried out in Ref. [22]. Different combinations of rocks.
moistening and heating were studied, whereby it was
possible to moisten the specimen either during the first 2.7. Stress memory formation under triaxial stress state
cycle, or in between, or during the second cycle. Tests
have shown that the worst conditions for the Kaiser Practical applications of the Kaiser effect for stress mea-
effect observations were if the specimen was heated surements require a full recognition of the stress tensor
between the cycles, or if the second cycle was performed and the triaxial nature of the in situ stress state.
on a wet specimen. A strong influence of heating on In the Earth’s crust, rock is usually subject to a tri-
stress memory in diorite and liporite was reported by axial state of stress. This can be either axisymmetric
Panasiyan et al. [5]. In their experiments, temperature triaxial (s1 > s2 ¼ s3 ) or true triaxial, i.e. polyaxial
increase up to only 80–1001C between two load cycles (s1 > s2 > s3 ). Laboratory testing of the core
was sufficient to completely eliminate the Kaiser effect. specimens is usually carried out in uniaxial compres-
Mechanisms of the Kaiser effect degradation in the sion—in the direction which is assumed to be the
course of time as well as the influence of water direction of the in situ s1 -stress. The question about
saturation are not completely understood yet. It is directional sensitivity of the Kaiser effect will be
tempting to ascribe these phenomena to microstructural discussed in Section 2.8. But even if we assume that
changes like crack healing. However, more detailed the direction of the in situ s1 coincides with that in the
research, including microscopic analyses, is needed to laboratory test, the stress state has six independent
clarify these points. components (one way to express these being the
magnitude and orientations of the three principal
2.6. Influence of the loading rate stresses). The question is: what does the Kaiser effect
stress indicate in uniaxial compression performed after
Loading and unloading processes in situ can proceed triaxial preloading?
with various loading rates—from very slow ones during This question is linked to the fundamental problem of
geological history to very fast, e.g. when blasting is the formation of damage surface in stress space [56].
performed or when earthquakes occur. Laboratory During triaxial preloading, the damage surface of rock
testing of the rock specimens retrieved from the rock is expanded. When the specimen is reloaded, the Kaiser
mass is usually carried out with the loading rates or effect occurs when the loading path in the stress space
displacement rates accepted in rock mechanics labora- arrives at the damage surface.
tory practice. These may (and do) differ from the rate at A simple experiment allows one to demonstrate the
which stress-memory-forming loads were applied in situ. complexity of stress memory formation under triaxial
Experiments by Yoshikawa and Mogi on Shinkomat- stress state. Two loading cycles are performed. The first
su andesite revealed no or very little influence of the cycle is axisymmetric triaxial compression with the
loading rate on the Kaiser effect in that rock; loading maximum values of the principal stresses sI1 > sI2 ¼ sI3 ;
rates in the range from 0.005 to 0.3 MPa/s were studied the superscripts indicating the number of the loading
[17]. A recent study [42] performed on another brittle cycle. The second cycle is uniaxial reloading in the
rock (low-porosity limestone) has shown that if the first direction of the first-cycle s1 -stress. The test is aimed at
cycle of uniaxial compression is carried out with a higher simulating the real stress measurement procedure with
strain rate (2.4  104 s1), then the second cycle is the first cycle being similar to the in situ loading, and the
performed with a more conventional strain rate of second cycle mimicking the laboratory test of the core
1.2  105 s1, the Kaiser effect occurs at the stress specimen.
which is about 67% of the peak stress of the first cycle, Experiments show that the Kaiser effect in the second
i.e. FR = 0.67. This result indicates that the use of the cycle in such tests occurs at a stress value sII1KE which, at
Kaiser effect for estimating in situ dynamic loads may least in a certain stress range, is a linear combination of
be questionable. It should be noted also that the load the principal stresses of the first cycle [41,57]:
rate used in the first cycle in Ref. [42] is many orders
smaller than that encountered during blasting and sII I I
1KE ¼ s1  ðk1 þ 1Þs3 ; ð1Þ
158 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

σ1 _ σ3 ·
N
σ1I _ σ 3I
C

k1
σ 1KE
II 1
A

B
O σ 3I σ3
0 σ 1KE
II
σ 1II
Fig. 7. Left-hand part of the damage surface formed by triaxial
axisymmetric loading in the first cycle with principal stresses sI1 ; sI3 : Fig. 8. AE hit rate versus uniaxial compressive stress in the second
Probing the damage surface in the second cycle is carried out in loading cycle of a rock specimen that was loaded in the first cycle in
uniaxial compression. The loading path in the first cycle is shown by axisymmetric triaxial compression. The Kaiser effect is recognized as a
two straight lines, OB and BC: Uniaxial reloading in the second cycle maximum.
is represented by the straight line OA: The Kaiser effect in the second
cycle takes place when the damage surface is reached in point A; i.e. at
stress value sII
1KE :

principal stresses given by (1). The dependency of AE hit


rate on stress during uniaxial reloading of triaxially
where the dimensionless coefficient of k1 is rock preloaded specimens is considerably different from that
dependent, e.g. for rock salt studied in Ref. [41,57] k1 ¼ after uniaxial preloading. The hit rate begins to increase
0:520:7: The coefficient of k1 characterizes the slope of virtually from the beginning of the test. The Kaiser
the damage surface projection onto the plane ‘‘stress effect can be recognised as a maximum in the curve ‘‘AE
difference (s1  s3 ) versus stress s3 ’’ on the left from the hit rate versus stress’’ occurring at stress value sII 1KE
point corresponding to the end-point of the first cycle defined by Eq. (1) [19,41,57]. This is schematically
(Fig. 7). illustrated in Fig. 8. The presence of AE activity from
Eq. (1) shows that a threshold value of sI1 exists that the beginning of the test represents a significant
should be exceeded in the first cycle for stress memory to difficulty with respect to using the Kaiser effect for
be formed, and thus, for the Kaiser effect to manifest in stress measurements. In Refs. [41,57] this was overcome
the second cycle. This threshold is equal to ðk1 þ 1ÞsI3 : by using deformation curves simultaneously with AE
This indicates that, if the first cycle is executed in measurements. The stress memory effects were identified
hydrostatic compression (sI1 ¼ sI2 ¼ sI3 ), no Kaiser by finding inflexions in the curves ‘‘axial strain versus
effect is observed during uniaxial compression in the uniaxial stress’’, ‘‘lateral strain versus uniaxial stress’’
second cycle. Direct experiments confirm this conclusion and ‘‘volumetric strain versus uniaxial stress’’ indicating
(e.g. [5,58]). It should be noted, however, that if both strain-hardening. Analysis of the differential coefficient
first and second cycles are conducted in hydrostatic of the lateral strain (differential Poisson’s ratio) was also
compression, the Kaiser effect is observed: AE activity used (see Ref. [41] for details). The inflexions in those
in the second cycle is remarkably lower than in the first stress–strain curves corresponded exactly to the max-
cycle, as long as the hydrostatic pressure remains less imum in the hit rate versus stress curve. It should be
than the maximum pressure reached in the first cycle noted that the experiments in Ref. [41] were performed
[11]. The mechanism behind the Kaiser effect in this case on a plastic rock, where strain-hardening is very well
might be the pore collapse rather than microcracking. pronounced. For brittle rocks, where the deformation
The result expressed by Eq. (1) shows that it is curves may hardly be used, the identification of the
impossible to determine the full in situ stress tensor Kaiser effect during uniaxial compression after triaxial
simply by uniaxially loading core specimens. At best, by preloading may represent a significant problem.
such method, it is possible to estimate the linear Another type of test which is often used in the Kaiser
combination of in situ stresses, when testing the speci- effect studies includes two loading cycles, of which the
men oriented precisely in the direction of the in situ first one is uniaxial or triaxial axisymmetric, and the
s1 -stress. second cycle is triaxial axisymmetric with the value of
There is one more difficulty, apart from the fact that s3 -stress, sII I
3 ; larger than that in the first cycle, s1 : The
the Kaiser effect during uniaxial reloading (second loading sequence in this test is thus reversed in
cycle) after triaxial preloading (first cycle) occurs at a comparison to the above-described test. Experiments
stress which is a linear combination of the first-cycle of such type were performed by Holcomb [10] as well as
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 159

Li and Nordlund [13]. Larger values of the confining


σ1 _ σ 3
pressure applied in second cycle (the confining pressure
is held constant in the second cycle while the stress
difference is increased) result in increasing stress
difference value at which the Kaiser effect takes place σ 1I _ σ 3I
in the second cycle. The Kaiser effect in the second cycle B
occurs at the moment when the damage surface formed
in the first cycle is reached. This happens when the
condition
sII II I I
1KE  ðk2 þ 1Þs3 ¼ s1  ðk2 þ 1Þs3 ð2Þ A

becomes valid during second-cycle loading. Here, k2 is a


O σ 3I σ3
rock-dependent dimensionless coefficient that charac- Fig. 10. Complete view of the damage surface formed by triaxial
terizes the slope of the damage surface projection onto axisymmetric loading with principal stresses sI1 ; sI3 : The damage-
the plane ‘‘stress difference (s1  s3 ) versus stress s3 ’’ on forming loading path is represented by two straight lines, OA and AB:
the right from the stress state of the first cycle (Fig. 9). The vertex point B corresponds to the peak stress state (based on
results in Ref. [14]).
The value of k2 was found to be equal to 2.6 for
Westerly granite [56] and to 2.8 for Kuru granite [13,16].
The slope of the damage surface projection with
the first cycle. This allowed them to test the damage
respect to the s3 -axis given by k2 ; is the bigger, the
surface on the right or on the left from the point
greater damage was introduced during the first cycle (cf.
corresponding to the first cycle. The existence of the
curves 1, 2 and 3 in Fig. 9) [13]. This corresponds well
vertex point is in agreement with the theoretical models
with the results of the theory developed in Ref. [59].
of damage surface formation developed in Refs. [59–62].
Figs. 7 and 9 are plotted on the basis of the results of
The existence of the vertex point in the damage
the tests in which the confining pressure, s3 ; in the
surface and the correspondence of this point to the stress
second cycle is smaller (Fig. 7) or larger (Fig. 9) than in
state at which the damage surface was formed allowed
the first cycle. Combining these two figures results in a
Pestman and Van Munster to propose a method for
damage surface shown in Fig. 10. The vertex point B in
determining both principal stresses based on the Kaiser
Fig. 10 corresponds to the end-point stress state of the
effect [14]. The method includes scanning the damage
first cycle. Direct experimental observation of such form
surface by performing several triaxial tests. The confin-
of the damage surface was achieved by Pestman and
ing pressure differs from test to test but remains
Van Munster [14] who preloaded sandstone specimens
constant during each test. As a result, the complete
in axisymmetric triaxial compression and then reloaded
damage surface can be plotted, with the vertex point
them, also in axisymmetric triaxial compression, with
yielding the stress state at which the stress memory was
the value of confining pressure bigger or smaller than in
formed.
The real stress environment in situ is usually
σ1 _ σ3 3 effectively true triaxial, i.e. with s1 > s2 > s3 : Experi-
mental verification of the Kaiser effect in rocks that
2 were preloaded in true triaxial compression is difficult
since it requires a special experimental equipment
1 (a polyaxial load cell). Theoretical predictions made on
σ1I _ σ 3I 1
k2 D the basis of a simple brittle fracture model using the
B Fairhurst-Cook approach (wing crack model) have
shown that the Kaiser effect after true triaxial preload-
ing should be less-pronounced than after axisymmetric
A C preloading. In particular, the amount of AE at lower
stresses during uniaxial reloading of rock should be
O σ3
I
σ 3II σ3 larger [57,63]. Experimental verification of the estima-
Fig. 9. Right-hand part of the damage surface formed by triaxial tions of Refs. [57,63] is required.
axisymmetric loading in the first cycle with principal stresses sI1 ; sI3 : An experimental study of the Kaiser effect in uniaxial
Probing the damage surface in the second cycle is carried out in triaxial tests of a brittle rock previously subjected to true triaxial
axisymmetric compression. The loading path in the first cycle is shown
stress state may be found in Ref. [64]. Disk specimens of
by two straight lines, OA and AB: Reloading in the second cycle is
represented by two straight lines, OC and CD: The Kaiser effect in the limestone were preloaded along a diameter (Brazilian
second cycle takes place when the damage surface is reached in point loading scheme) until a sufficient number of AE hits
D: was registered ensuring that sufficient damage was
160 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

introduced in the central zone of the disk. The stress in a direction cannot be retrieved simply by uniaxially
state in the central part of the disk was true triaxial, with reloading in a perpendicular direction.
one of the principal stresses being tensile (perpendicular It is interesting to draw a parallel between these
to the loaded diameter, parallel to the disk surface), the experiments and dynamic tests in a modified split
other one being approximately zero (normal to the disk Hopkinson bar performed on zirconia ceramics in
surface; thin disks were used, with the ratio thickness: Ref. [68]. Rectangular specimens were subjected to two
diameter o0:2), and the third one being compressive loading cycles each. The direction of load application in
(parallel to the loaded diameter). A prismatic specimen the second cycle was either the same as in the first cycle
was then cut from the central zone of each disk along the (this regime is further called ‘‘loading scheme A’’), or it
loaded diameter, i.e. parallel to the compressive was orthogonal to the first-cycle loading direction
principal stress. The prismatic specimens were tested in (further called ‘‘loading scheme B’’). Regular strain-
uniaxial compression with AE measurement. In all hardening was observed in the specimens of series A:
prismatic specimens tested, considerable AE activity inelastic strains in the second cycle were considerably
started from the beginning of loading. The inflexion in lower than in the first one. In the specimens tested with a
the cumulative AE versus stress curve was present in two 901-rotation (scheme B), inelastic strains in the second
of the five specimens tested, and even in those two the cycle were of the same order as in the first cycle. In each
inflexion was unclear. cycle, cracks were formed in the direction sub-parallel to
All the two-cycle experiments described so far in this the load applied.
section (first cycle triaxial, second cycle uniaxial or Notwithstanding the difference in the mechanisms of
triaxial), are still very far from what really happens inelastic deformation of zirconia ceramics and AE in
when stress measurements based on the Kaiser effect are rocks, both phenomena characterize irreversible struc-
carried out. One of the few attempts to physically tural changes. In this sense, the equality of inelastic
simulate the real stress measurement process can be strains in the first and second cycles conducted in
found in Refs. [65,66]. When a core is extracted from a mutually perpendicular directions in Ref. [68] is similar
vertical borehole, the vertical stress is first relieved. The to the absence of the Kaiser effect in rocks in case of
remaining horizontal stresses form a certain damage. y ¼ 901:
Then the horizontal stresses are decreased, and the core If a specimen is uniaxially sequentially loaded in
is extracted. This sequence was modeled in the tests several orthogonal directions, and then sequentially
described in Refs. [65,66]. Evidence was obtained that it reloaded in each of them, the Kaiser effect in
is possible to estimate horizontal in situ stresses using each direction is observed at the stress value which is
the Kaiser effect. equal to the stress applied in that direction during
preloading [54,55,69]. As concluded in Ref. [69], ‘‘the
2.8. The effect of rotation of principal axes recollection of a pre-stress level in a particular direction
appears to be unaffected by the application of stresses
The consideration in the previous section concerns the in orthogonal directions during the period before
case when the directions of principal stresses are the re-stressing’’.
same in the first and second cycle, e.g. uniaxial reloading It should be noted that sequential loading in three
is performed precisely in the direction in which s1 was orthogonal directions is completely different from a
applied during triaxial preloading. In real stress simultaneous application of stresses. The former loading
measurement situations, however, the precise orienta- program results in that anisotropic damage is created
tion of the in situ s1-stress is not known a priori. Indeed, successively in each direction. The latter case is a true
it should be obtained as a result of the stress triaxial loading. Neglecting the difference between the
measurements. In some situations, this direction can two regimes is still sometimes encountered in Kaiser
be assumed based on common sense or geological data, effect studies.
but the accuracy of such assumptions is usually quite The case of ya901 is more subtle. Holcomb and
poor. The question is: is it critical if the specimen for the Costin [56] performed an experiment in which a large
laboratory test is orientated not exactly in the direction block of Westerly granite was subjected to uniaxial
of the in situ s1-stress, but at an angle of y to it? compression. After the load was decreased to zero, sub-
Experimental verification is the easiest for the case of cores were taken from the block in different directions.
y ¼ 901: Cubic specimens are manufactured, and several Testing the sub-cores has shown that the Kaiser effect is
cycles of uniaxial compression are performed, uniaxial very sensitive to the deviation of the reloading direction
load being applied in each cycle in the direction from the direction in which the load was applied to the
orthogonal to the previous cycle. The results of the big block. The effect was observable for y less than 101.
tests indicate that there is no Kaiser effect during When the direction of the s1-axis in the second cycle
uniaxial reloading in the directions orthogonal to the deviated by more than 101 from that in the first cycle,
preloading direction [67]. Thus, a uniaxial stress applied the Kaiser effect could not be observed, i.e. the rock
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 161

behaved like a ‘fresh’ one, not having been preloaded in differently orientated specimens may exist. This is
the first cycle. probably the reason for evidence of the Kaiser effect
In Ref. [43], this extremely high sensitivity of the in core specimens reported in the literature (Section 4.1).
Kaiser effect versus rotation of principal axes in Another consequence of the results described above is
successive loading cycles was reconfirmed in experiments that they suggest a methodology for a precise estimation
using cyclic Brazilian tests with disk rotation between of the in situ s1-direction by testing specimens cored in
loading cycles. This made it possible to avoid cutting different orientations, provided that only memory of the
variously orientated specimens from a preloaded block, most recently experienced stress state is present in the
which might introduce additional damage in the rock. rock. The large number of specimens needed for such
Disks were subjected to two loading cycles each, with or procedure can be reduced if the in situ s1-direction is
without disk rotation between the cycles. The Kaiser known approximately from some other data, e.g. from
effect was observed if the rotation angle of the disk geological evidence, anelastic strain recovery (ASR) or
between first and second cycle did not exceed 101. For differential strain curve analysis techniques.
rotation angles larger than 101, the Kaiser effect was
absent: AE activity gradually increased from the
beginning of the second cycle, not showing any inflexion 2.9. Influence of the coring process
or another anomaly. No Kaiser effect could be detected
also in the tests with disk rotation by y ¼ 901; the result The most popular stress measurement technique
agreeing well with the above-described results for cubic based on the Kaiser effect involves retrieving core
specimens. The results of Ref. [43] are in agreement with specimens from the Earth’s crust and testing them in the
those of Holcomb [12,56], although the stress state in laboratory. An important question is to what extent the
Brazilian tests is different from that in uniaxial damage introduced during the coring process influences
compression. the in situ stress memory contained in rock.
The results described so far in this section are crucial Damage can be introduced in rock during coring in
for stress measurement applications. The results show several ways. When a vertical borehole approaches a
that specimens for uniaxial laboratory compression point in the rock mass, the vertical principal stress is
should be retrieved as close as possible to the direction gradually reduced, while the horizontal stresses are
of the in situ s1-stress. The deviation of the specimen unaltered until the rock is inside the coring drill bit; then
from the in situ s1-axis should not exceed 101. Only in the horizontal stresses are reduced [66,70]. At the
this case does it make sense to search for the Kaiser bottom of the core stub, a local stress concentration
effect in the specimens undergoing uniaxial laboratory should be expected. A comprehensive finite-element
tests, and even in this case only if the threshold s1-value analysis carried out by Li and Schmitt [71] has shown
was exceeded in situ (Section 2.7). If the specimen is that, dependent on the ratio of in situ stresses, increased
oriented in the direction of in situ s1-stress, the Kaiser tensile or shear stresses can develop at the core bottom.
effect enables the estimation of the linear combination Tensile stresses develop if the horizontal far-field in situ
according to Eq. (1). Should the deviation of the stress is larger than the in situ stress parallel to the
specimen from the in situ s1-axis be more than 101, no borehole. The tensile stress direction is approximately
Kaiser effect will be observed in a uniaxial laboratory parallel to the borehole axis. After the core is inside the
test. This implies that, if the Kaiser effect in core bit, the tensile stress decreases gradually and becomes
specimens has the same mechanism as that in laboratory negligible at a distance of ca. 3R upwards from the drill
tests described in this and the preceding sections, then it bit tip along the axis of the core (R is the core radius)
is not possible to retrieve the whole in situ stress tensor [72]. A schematic in Fig. 11 illustrates the dependency of
by simply uniaxially loading specimens orientated in six the tensile stress on the distance z along the core [71,72].
different directions [12]. In fact, there will be no Kaiser Even if the tensile stresses are not high enough to
effect in most of them, if not in all. produce core disking, they may be sufficient to generate
A different situation is of course when the rock during microcracks perpendicular to the core axis. This may be
its history was subject to several loading cycles with the reason for numerous reported observations that
different orientations of the principal axes. In this case, there is a Kaiser effect in sub-cores drilled perpendicular
anisotropic damage could be formed in each loading to the direction of the in situ s1-stress. A rigorous
cycle, e.g. in the laboratory tests on cubic specimens experimental and theoretical verification of this sugges-
with sequential uniaxial loading in orthogonal direc- tion is required.
tions. If the threshold s1-value was exceeded in each of If the largest compressive in situ far-field stress is
those cycles, and if the time delay between the in situ acting parallel to the borehole axis, shear stresses
loading and the laboratory test was not sufficient for the develop in the core stub. These can produce damage,
Kaiser effect to degrade completely, then the Kaiser although, as shown in Ref. [71], they do not cause core
effect in laboratory tests on several (or even all) disking.
162 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

σtensile 2.10. Kaiser effect at different scales

Most of the Kaiser effect features described in the


preceding sections were established during laboratory
tests on relatively small rock specimens (dimensions of
the order of n  1 cm). However, stress memory and,
more particularly, the Kaiser effect can be observed in
rocks at different scales.
Sobolev et al. [23] observed the Kaiser effect in their
tests on a large (about 1 m) cubic rock specimen
0 3R z
subjected to cyclic uniaxial compresision.
Cyclic loading of borehole walls by means of a rock
Fig. 11. Tensile stress versus distance z along the axis of the borehole. jack or a pressuremeter resulted in the Kaiser effect [27].
The position of the drill bit tip is indicated by an arrow. Tensile stress
becomes negligible at a distance ca. 3R upwards from the drill bit tip; R
This is an example of an in situ Kaiser effect, at borehole
is the borehole radius (schematically; based on results of Ref. [72]). scale.
Being loaded with a pressuremeter or a rock jack in
one cycle shortly after drilling, rocks around a borehole
indicate a memory effect on in situ stresses. This was
evidenced in Refs. [27,77,78] and suggests a basis for a
The damage produced by tensile and shear stresses stress measurement technique which exploits the Kaiser
caused by the near-drill-bit stress concentration will effect of rocks around a borehole (see Section 4.2 for
have a complicated impact on the Kaiser effect more detail). Fracture mechanics validation of this
observation in the retrieved core specimens. A further technique using a two-dimensional wing crack model
experimental study and fracture mechanics analysis was given in Refs. [79–82].
of this problem is needed. Poromechanical effects, Being a characteristic feature of heterogeneous solids,
the drill string weight and the friction forces produced the Kaiser effect should be observed at large scales as
by a rotating drill bit should be taken into account well. A literature review shows that, indeed, examples of
as well. the Kaiser effect in the microseismic and seismic
The drill bit rotation should produce some additional activities can be found. For example, Simpson and
damage caused by the friction forces. The effect of the Negmatullaev reported that cyclic changes of the water
damage due to the rotation of the bit was studied in level in the Nurek reservoir resulted in seismic activity
Refs. [55,73]. Large cubic specimens were preloaded in when the water level approached the previous maximum
uniaxial compression, then unloaded, and smaller level [83]. Care should however be taken when drawing
specimens were drilled out to be tested in uniaxial parallels between reservoir-induced microseismic activ-
compression. This procedure is certainly different from ity and the Kaiser effect because of the complexity of the
a real coring procedure because the latter is performed mechanisms involved in fault dynamics, as demon-
under in situ stresses. The experiments in Refs. [55,73] strated in Ref. [84].
have shown that a perfect Kaiser effect is observed in the Seismicity in the vicinity of a water reservoir at Koyna
drilled-out specimens, i.e. the surface damage intro- (India) was explained by means of the Kaiser effect in
duced by the rotating bit is negligible for the Kaiser Refs. [85,86]. The existence of a large-scale Kaiser effect
effect observation. in situ could limit the applicability of microseismic and
A stress-free core specimen recently extracted from in AE techniques for rockburst forecast in mines, as
situ indicates an exponentially decaying AE [74]. This is pointed out by Young [87]. This is valid also for
similar to the time-dependent anelastic strain recovery precursory anomalies other than AE and foreshocks.
(ASR) and is quite a well-known phenomenon which Parallels between stress memory effects and earthquake/
was observed, e.g. in the specimens obtained from KTB rockburst precursors were drawn in Ref. [88].
boreholes [75,76]. For example, in a gneiss core an The existence of the Kaiser effect in the post-failure
exponentially decaying AE activity was measured during regime (Section 2.2) means that this phenomenon could
2 days after separating the core from the rock mass [76]. be observed even in heavily deformed Earth crust areas,
Indicating the development of a post-drilling damage, although the Kaiser effect will be complicated in this
the AE activity during ASR may to some extent obscure case because of the complex loading history of rocks.
the Kaiser effect in laboratory tests. Results of experiments by Dunning et al. [47]
Additional damage to the core can be caused by a performed on sawcut rock blocks and described in
temperature change in comparison to in situ, and, in Section 2.2 suggest that a kind of Kaiser effect may exist
low-permeable rocks, by pore pressure of the locked-in in the Earth’s crust on faults and fault-like structures.
fluid [70]. However, laboratory tests [89] on sawcut blocks
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 163

undergoing cyclic stick-slip motion indicated no or very L


poor Kaiser effect. In Ref. [89], 15 to 150 stick-clip
events were produced on each block subjected to biaxial
compression and instrumented with AE and strain
sensors. In 80% of these stick-slip events, precursory
AE was observed indicating no stress memory. More
experimental work is needed to clarify the applicability
of laboratory results to large-scale observations.

3. Theoretical models of the Kaiser effect 0 σc σo σ

Application of the Kaiser effect for stress measure- Fig. 12. Crack length L versus stress s for a reversible Griffith crack
(based on Ref. [97]).
ments requires a firm understanding of the mechanism
behind the phenomenon.
Generally speaking, the Kaiser effect is a manifesta-
tion of the stress memory of heterogeneous solids. A tensile stress is relieved as a penny-shaped tensile crack
number of stress memory effects other than the Kaiser propagates, so that a stable crack growth is possible.
effect are known in rocks: e.g. the deformation memory Thus, the crack length, after completing loading,
effect, or strain hardening, on which the deformation corresponds to the peak stress reached. This gives rise
rate analysis (DRA) stress measurement technique is to the Kaiser effect. Based on this model, Holcomb [12]
based [54,90,91]; the ultrasonic memory effect [31,92]; simulated the Kaiser effect under various conditions,
the electric memory effect [24,92]; the memory effect in including some types of true triaxial loading. In
electromagnetic emission [24]; the ion emission memory particular, he found that if the first cycle is performed
effect [93]. Some curious parallels can be drawn between in the so-called ‘extensional’ mode (principal stresses s1
stress memory effects in rocks and temperature memory and s2 are compressive and equal to each other; stress s3
effects in spin glasses [94] and plexiglass [95]. is smaller than s1 ; s2 ) and the second cycle is uniaxial
The mechanism of the Kaiser effect should be sought compression in the direction of the first-cycle s1 ; then no
for in the formation of irreversible damage due to the Kaiser effect is observed. This is in agreement with
stresses applied to the rock. In brittle rocks, microcracks experimental observations [103].
are one of the most important mechanisms controlling Holcomb concluded that uniaxial reloading ‘is not
the irreversible damage and the Kaiser effect. One of the able to detect general stress histories applied in the
first attempts of micromechanical analysis of the Kaiser laboratory’ [12]. As an alternative to uniaxial reloading,
effect was undertaken in Refs. [96,97]. Stevens and he suggested the use of extensional reloading. Probing
Holcomb [96] have shown that a sliding crack model (a the damage surface in extensional mode should allow
mass on an inclined plane) accounts for stress memory the activation of microcracks with the normals parallel
only if the dynamic coefficient of friction between crack to a specified direction, thus being more selective than
faces md is equal to the static one ms : md ¼ ms : In this conventional uniaxial probing. If the extensional reload-
case, the sliding process occurs in a stable way. If the ing is used, it is possible to extract some information
dynamic friction is less than the static, i.e. md oms ; the even after a true triaxial first cycle in which all principal
motion on the sliding crack proceeds in stick-slip mode, stresses are different and compressive. In particular, the
and stress memory is violated [96]. As an alternative to orientations of principal axes can be found [12].
the sliding crack model, a reversible Griffith crack The simulations in Ref. [12] confirmed high sensitivity
model was suggested in Ref. [97]. In this model, a crack of the Kaiser effect against the non-coincidence of
can be in one of the two states: ‘closed’ or ‘opened’. A principal axes in successive loading cycles. This was later
closed crack is assumed to be healed so that its strength reconfirmed in Ref. [104] by using a Fairhurst-Cook
is restored, at least partially. The opening stress is higher wing crack model (shear crack+two tensile cracks).
than the closing stress, which gives rise to a hysteresis. Using the wing crack model for simulating the Kaiser
The behaviour of a single reversible Griffith crack is effect in uniaxial compression (second cycle) after true
illustrated in Fig. 12. Considering a population of such triaxial precompression (first cycle) has shown that the
cracks (‘mesoscopic hysteretic elements’) allowed the intermediate principal stress (s2 ) significantly affects
simulation of stress memory and other properties of the AE signature during the uniaxial second cycle
nonlinear dilatant rocks [98–101]. [57,63,105].
A micromechanical model developed by Costin [102] Using the wing crack model for the simulation of the
allowed the simulation of the damage surface expansion Kaiser effect during uniaxial reloading (second cycle)
under a triaxial state of stress. In the model, the local after axisymmetric triaxial preloading (first cycle)
164 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

[13,16,104] confirmed the experimental finding that the boundary-element displacement discontinuity method
Kaiser effect is not so well-pronounced in this case as in implemented in DIGS code [43,110] were used in the
cyclic uniaxial tests. This is due to the cracks which did Kaiser effect simulations. With PFC, the simulation of
not grow in the first cycle because of the higher overall the complete coring process was carried out in
compression (confining stress). The model has shown Refs. [66,109]. It was shown that the Kaiser effect can be
that the Kaiser effect during uniaxial reloading occurs at used for measuring horizontal in situ stresses. With
a stress value which is a linear combination of the first- DIGS, the Kaiser effect in cyclic Brazilian tests with
cycle principal stresses [13,16,104,106]. In order to disk rotation in between the cycles was modeled [43].
obtain the values of both s1 and s3 principal stresses, Moreover, the Kaiser effect in uniaxial compression
the second cycle should be carried out in proportional after triaxial axisymmetric preloading was simulated in
triaxial compression [107]. The Kaiser effect should be Ref. [110], the results being in agreement with experi-
the best-pronounced when the ratio of principal stresses ments as well as with analytical models.
in the second cycle is the same as in the first cycle. The use of numerical codes, especially of three-
Testing a series of specimens with different values of the dimensional ones, is an important tool in the Kaiser
ratio s1 =s3 should thus allow one to obtain the values of effect studies allowing investigation of the phenomenon
both principal stresses of the first cycle [107]. in rocks with different constitutive behavior and under
Generally speaking, the results of Ref. [107] suggest various loading conditions.
that the Kaiser effect should be more pronounced, the
closer the reloading path is to the preceding stress
history. The effect should be perfect when the reloading 4. Practical application of the Kaiser effect for stress
path follows the preloading path precisely, i.e. if the measurements
stress history of the rock is fully restored. Thus, we are
trapped in a vicious circle which seems to undermine the Two major stress measurement techniques based on
use of this phenomenon for stress measurements: to the Kaiser effect are known. One of them implies that
obtain the in situ stresses, the in situ loading path should core specimens are retrieved from in situ and tested in
be followed in the laboratory test. However, as shown laboratory, in the hope of finding an inflexion in
above, some useful information (directions of principal the dependency ‘‘cumulative AE hits versus stress’’—
axes, linear combination of the principal stresses et al.) indicating that the stress applied in the laboratory
can be estimated by the Kaiser effect without knowing equals the stress experienced by the rock in situ. The
the previous stress history. Moreover, experimental and other technique implies that, instead of loading the
theoretical results obtained by Pestman and Van rock in laboratory, it is loaded in situ by means of a rock
Munster [14] suggest that it is possible to determine jack or pressuremeter placed in a borehole, with simul-
both principal stresses by probing the damage surface at taneous measurement of AE.
various confining pressures in the hope to find the vertex
point. For the theoretical justification of their method, 4.1. Stress measurement using the Kaiser effect in core
Pestman and Van Munster developed an analytical specimens
model of the Kaiser effect based on the wing cracks
concept. The model allowed them to explain the The majority of stress measurements using the Kaiser
existence of the vertex point in the damage surface effect in core specimens are carried out as follows. Cores
formed under an axisymmetric triaxial stress state [14]. are retrieved from in situ. Sub-cores oriented in different
All models presented in this section so far consider directions (at least six, because there are six independent
microcrack growth. An alternative approach pointed components of the stress tensor) are manufactured from
out in Ref. [12] implies that the Kaiser effect is due to the cores. These six specimens are tested in uniaxial
the void closure instead of crack growth. This approach compression, with simultaneous measurement of AE
could provide an explanation for numerous observa- activity. The curve ‘‘cumulative AE hits versus uniaxial
tions of the Kaiser effect in rocks retrieved from rock stress’’ is then analyzed in order to find an inflexion
masses where the current stress conditions could not (change of slope). It is often believed that this inflexion
create sufficient microcrack damage [12]. occurs at the stress value equal to the normal stress
Instead of the micromechanical approach, a model of component that was acting in this direction in situ.
the Kaiser effect can be developed using principles of The incorrectness of such an approach becomes
continuous damage mechanics (CDM). This was de- evident after looking at the Kaiser effect features
monstrated by Tang et al. [108] who employed CDM to discussed in Sections 2.7 and 2.8. First, no Kaiser effect
simulate the Kaiser effect in cyclic uniaxial compression. should be expected in uniaxial compressive tests if the
There is also another way to theoretically investigate the loading axis is inclined at more than 101 to the direction
Kaiser effect using numerical codes. A discrete particle of in situ s1 -stress. Thus, no Kaiser effect should be
model implemented in PFC code [66,109] and a observed in most of the six specimens, if not in all of
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 165

them. Second, even in the specimen tested in the (2) The coring process may introduce irreversible
direction of the in situ s1-axis, the Kaiser effect occurs anisotropic damage, e.g. microcracks (Section 2.9).
at the stress defined by Eq. (1) rather than at the in situ For example, the concentration of tensile stresses
s1-stress value. parallel to the borehole axis, or horizontal stresses
If we assume that the normal stress component can be acting after the vertical stress has been relieved, may
retrieved by simply uniaxially testing the rock in various produce microcracks perpendicular to the core axis.
directions, we inevitably come to conclusion that after a These cracks may give rise to the Kaiser effect in the
hydrostatic preloading of type s1 ¼ s2 ¼ s3 there must specimens cored perpendicular to the borehole axis
be a Kaiser effect in the subsequent uniaxial reloading in and tested in uniaxial compression. In this way,
any direction. Experiments show that this is not the observations of the Kaiser effect in core specimens
case. No Kaiser effect is observed in uniaxial tests after orthogonal to the borehole axes reported, e.g. in
hydrostatic pre-compression. This reasoning was first Ref. [31], can be explained.
given in Ref. [12]. (3) During recent geological history, the rock could
The complexity of the manifestation of the Kaiser have been sequentially subjected to stress states with
effect in core specimens with different orientations was s1 oriented in various directions. Rotations of
evidenced in Ref. [111]. Watters and Roberts [111] principal axes of the stress tensor are likely to
measured AE activity during uniaxial laboratory load- occur, e.g. in the areas of the Earth’s crust with
ing of core specimens of granodiorite taken in directions enhanced tectonic activity or in the vicinity of
parallel or normal to the ground surface. Specimens underground cavities [113,114]. The stress tensor
were taken from near the surface, hence one of the may rotate gradually, or the rock may be loaded/
principal stresses (normal to surface, s3 ) was close to unloaded sequentially with a different s1-axis
zero. In situ stresses parallel to the surface were to be direction in each cycle. Each of these successively
high, as evidenced by spalling of rock sheets. No Kaiser applied and differently orientated in situ stress
effect was observed in the specimens oriented perpendi- states should form anisotropic damage responsible
cular to the surface. The Kaiser effect in the specimens for the Kaiser effect in the laboratory test on the
parallel to the surface and, thus, parallel to the in situ s1 specimen oriented in the direction of the corre-
or s2 was by far not so well-pronounced as after uniaxial sponding in situ s1-stress.
preloading. In particular, some of the specimens tested
along the assumed s1-axis showed no evidence of the These three explanations indicate that there is some
Kaiser effect [111]. uncertainty and a lack of understanding about what is
Still, there are numerous experimental observations of actually measured by the Kaiser effect. It has been
the Kaiser effect in rock specimens oriented in various clear for the last ten years at least (see Ref. [12]) that
directions. Often it exists in six or more directions. The uniaxial loading alone does not allow a determination
principal stress values inferred from such measurements of in situ stress tensor provided that microcrack
are often in good agreement with stress values obtained growth is the only mechanism responsible for the
by other techniques, e.g. overcoring. Recent examples of Kaiser effect observed in core specimens. If we admit
such measurements can be found in Refs. [21,112]. These that microcracks are produced by in situ loading
results cannot be ignored, but they require an explanation and carry information about in situ stresses, then the
in order that a correct micromechanical interpretation of Kaiser effect can be applied for estimation of the
experiments becomes possible. Microcrack models dis- directions of principal axes as well as for estimation of
cussed in Section 3 provide no explanation. At least three their values. Uniaxial testing is not sufficient in this
alternative explanations can be considered. case because it allows only the determination of a
linear combination of principal stresses given by Eq. (1),
(1) The Kaiser effect in retrieved specimens may be due and this only in the case that the specimen is oriented
to a mechanism different from microcrack growth. exactly in the direction of in situ s1-stress. All
For example, stress memory may be caused by the uniaxial probing can be used for is the precise
irreversible void closure, as pointed out by Hol- estimation of the in situ s1-direction since it is the only
comb [12]. Pore and crack closure may produce AE, direction in which the Kaiser effect should be well-
and, hence, the Kaiser effect should be expected pronounced in uniaxial tests. This technique, mentioned
similar to that caused by crack growth. The in Ref. [43], requires however a large number of
question is whether the AE signals produced by specimens.
closing voids have an amplitude sufficient for being More sophisticated techniques proposed in
detected with AE apparatus settings usually used in Refs. [14,107] allow the determination of both
laboratory practice. Experimental verification as principal stresses s1 and s3 provided the in situ state
well as theoretical investigation of the Kaiser effect of stress is axisymmetric triaxial and the specimen is
mechanism linked to the void closure are required. oriented in the direction of in situ s1-stress.
166 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

The technique described in Ref. [14] implies that the equipped with an AE transducer is placed in the
damage surface has a vertex point corresponding to the borehole at the point desired. The borehole walls are
in situ state of stress. This point is sought for by probing loaded, isotropically or directionally, if a pressuremeter
the damage surface in triaxial compression tests with or a rock jack, respectively, is employed. AE activity is
different s3-stress values (being constant in each test). A measured continuously during the loading process. The
conventional triaxial load cell is needed for implement- dependency of AE cumulative hits versus pressure is
ing this technique. examined in the hope of finding an inflexion indicating
The technique proposed in Ref. [107] is based on the that the pressure on the hole walls reached the original
results of numerical simulations using wing crack model in situ stress, thus giving rise to the Kaiser effect.
carried out in Refs. [57,107]. The technique implies that The in situ loading may be considered as the first
core specimens oriented in the direction of in situ s1- loading cycle, after which the rock is unloaded by the
stress are tested in proportional axisymmetric compres- borehole. Then the rock is reloaded by pressuremeter/
sion, the ratio s3 =s1 being constant during the test. rock jack.
According to simulation results, the Kaiser effect should Some of the difficulties/weaknesses inherent in the
be most pronounced in the test with the s3/s1 ratio equal borehole technique should be mentioned. Acoustic noise
to that in situ. The values of s1 and s3 at which the effect coming from the friction processes between loading
takes place should be equal to in situ stress values. A device and the borehole wall as well as from technolo-
testing machine allowing simultaneous proportional gical sources might obscure the AE from the rock to
increase of the axial stress and the confining pressure some extent. The ‘acoustic jacks’ described, e.g. in
is needed for implementing this technique. Refs. [77,78], are not available on the market and need
The method of extensional loading proposed in to be specially designed and manufactured. Costs of
Ref. [12] allows the estimation of all principal stresses their manufacturing and of conducting in situ tests are
even in the most general case of a true triaxial in situ relatively high, in comparison to those of laboratory
state of stress (s1 > s2 > s3 ). The extensional loading tests required by the core specimen technique considered
means that two principal stresses are equal to each other in Section 4.1. Some of the interpretation problems
and higher than the third principal stress. By conduc- typical for the core specimen technique remain valid also
ting extensional loading of specimens oriented in for the borehole technique. Interpretation should take
different directions the whole damage surface formed the triaxial type of the in situ stress state into account.
in situ can be scanned in detail. A polyaxial testing Moreover, if the directions of the in situ principal axes
machine is needed for these tests. are not known a priori, the orientation of the borehole is
The techniques proposed in Refs. [12,14,107] are in no arbitrary with respect to them. This should introduce
way simple and have not been probed widely so far. uncertainty in the data interpretation.
Comparison of their results with the results of indepen- Among the strengths of the borehole technique is the
dent stress measurements by overcoring or hydrofrac are fact that probing of rock is performed at a larger scale,
needed to make conclusions about the applicability of thus reducing the frequently discussed question of
these Kaiser effect techniques. representativeness of stress measurement results ob-
If the predominant mechanism of the Kaiser effect is tained on core specimens. The borehole is drilled
void closure or something else, as it seems to be from the without taking a core, which makes the drilling process
stories of success mentioned above, these mechanisms easier and cheaper in comparison to that required for
should be carefully investigated both experimentally and the core specimen technique. The rock around borehole
theoretically. It is not improbable that several mechan- is loaded by pressuremeter/rock jack in conditions close
isms, e.g. microcrack growth and irreversible void to the original in situ environment (temperature,
closure, play simultaneous roles. The clarification of moisture content). The pore pressure, however, can
the mechanism of what has been observed in the tests on change after the borehole has been drilled, and the effect
core specimens is a prerequisite for correct data of this change should be investigated. Other parameters
interpretation of stress measurements using the Kaiser (e.g. temperature) can change as well in the course of
effect. time, which may result in a partial recovery of the
Kaiser effect like that observed on rock specimens.
4.2. Stress measurement using the Kaiser effect in rocks Experimental observations of stress memory recovery in
surrounding a borehole rocks around a borehole were described in Ref. [78] and
simulated numerically using a wing crack model in
This technique, referred to as ‘borehole technique’ Refs. [15,82,115].
further in this Section, is implemented as follows. A One of the most serious weaknesses mentioned above
borehole is drilled in the rock mass where in situ stresses is that a full restoration of the stress state is impossible if
are to be determined. A loading device (pressuremeter loading devices like a pressuremeter or a rock jack are
for soils and soft rocks, rock jack for hard rocks) used. Even in the simplest case of a borehole drilled in
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 167

the direction of an in situ principal stress, e.g. s2 ; it is loading tests in boreholes was simulated using the wing
not completely clear how the Kaiser effect stress is crack model in Refs. [81,82].
related to in situ stresses. This problem is similar to that The stress measurement technique using memory
of the core specimen technique. In order to clarify this effects in rocks around a borehole has considerable
question, two-dimensional numerical simulations using potential. It could provide a way of obtaining the values
a wing crack model were carried out in Refs. [15,79–82]. of in situ stresses by direct in situ measurements. For
Their results show that if borehole walls are loaded by successful application of the Kaiser effect in rocks
means of a pressuremeter creating uniform pressure, the surrounding a borehole, the circumstamces should be
Kaiser effect pressure linearly depends on one of the clarified, especially in the case when the direction of a
principal stresses, s1 or s3 ; orthogonal to the hole axis, borehole is not parallel to any of the in situ principal
provided that the other principal stress, s3 or s1 ; is stresses.
constant. For a constant ratio of in situ principal
stresses, s1 =s3 ; the Kaiser effect pressure is proportional
to one of them [15,79–82].
These modeling results need to be verified experimen- 5. Conclusions and future research
tally. They show that a great deal of complexity is
involved in the borehole technique, probably as much as This review has shown that a great number of
in the core specimen technique. questions important for stress measurement applications
Despite these difficulties, some encouraging results of the Kaiser effect have been solved or partially solved
have been obtained. In Ref. [78], a horizontal borehole at present. However, the solution of these questions has
was drilled from an underground salt mine and tested revealed some fundamental limitations on the stress
with a pressuremeter equipped with an AE transducer measurement applications. The mechanisms of the
and an electromagnetic emission receiver. Pressurization Kaiser effect should be studied further, including those
was carried out at different distances from the mine wall which are not crack growth.
(i.e. from the borehole opening). Both AE and electro- One of the questions which is still open concerns the
magnetic emission measurements indicated stress mem- Kaiser effect manifestation in rocks previously subjected
ory effects at the same pressure values. The dependency to a true triaxial state of stress (i.e. most rock masses).
of this pressure value on the distance along the borehole The few theoretical analysis efforts already undertaken
axis resembled the expected dependency of the vertical in this field are clearly not sufficient. The question is
in situ stress in the excavation near-field zone, with a formulated as follows: how does the damage surface
maximum at a certain distance from the excavation. evolve under a true triaxial state of stress, especially if
In Ref. [116], pressurization tests were carried out in a rocks undergo loading paths with ‘kinks’ in the stress
vertical borehole drilled in an open pit sand mine. AE space and with occasional continuous or discrete
was measured during pressurization. An AE transducer rotation of principal axes?
was attached to the end of a waveguide installed in the The importance of answering this question was
soil parallel to the testing borehole. The analysis of the postulated by Rudnicki [118] who emphasized the need
AE cumulative curves showed that the Kaiser inflexion to conduct multi-axial experiments and develop con-
corresponded well to the expected horizontal soil stitutive and micromechanical models of rock behaviour
pressure that should have been present in the soil before under true triaxial stress conditions. The use of the AE
excavation started. technique and the Kaiser effect in the experiments
Watters and Soltani [117] performed directional should enable a deeper insight into the damage surface
loading of borehole walls with a rock jack. The hole formation in rocks under a true triaxial stress state.
was drilled in a granite rock mass. AE activity showed a The investigation into the Kaiser effect in rocks
well-pronounced anisotropy: the number of AE hits was subjected to a triaxial stress state has shown the
several times as high in one of the directions as in the incapability of uniaxial testing procedures to estimate
other one. The Kaiser effect pressure was anisotropic, in situ stresses, even in the simplest case of the
too. axisymmetric triaxial in situ state of stress. Another
Lord and Koerner [27] performed loading of borehole difficulty results from the extremely high sensitivity of
walls with a rock jack as described in Ref. [77]. The the Kaiser effect with respect to non-coincidence of the
boreholes were drilled in a rock mass or in a large rock principal axes in situ and in laboratory tests.
block. Inflexions in the dependencies of cumulative AE The stress measurement technique based on the
hits versus pressure indicated the presence of memory on Kaiser effect in rocks around a borehole seems to be a
in situ stresses. Cyclic pressurization of borehole walls promising one. This attractiveness may, however, result
delivered a very well-pronounced Kaiser effect, when the from our underestimation of the fundamental limita-
pressure reached the maximum previously reached tions, as was the case for the core specimen technique a
pressure value. The Kaiser effect taking place in cyclic few decades ago.
168 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

Even if the Kaiser effect works well, it should be References


recommended for use in combination with other stress
memory effects. For example, for ductile rocks such as [1] Sugawara K, Obara Y, editors. Rock Stress. Proceedings of the
rock salt, strain measurements can be used efficiently International Symposium on Rock Stress, Kumamoto, Japan,
October 7–10, 1997. Rotterdam: A.A. Balkema, 1997.
[41]. The Kaiser effect was experimentally observed, [2] Kurita K, Fujii N. Stress memory of crystalline rocks in acoustic
together with anomalies (e.g. inflexions) in other emission. Geophys Res Lett 1979;6:9–12.
physical parameters like inelastic strain [25,26,41,53, [3] Yamshchikov VS, Shkuratnik VL, Lavrov AV. Memory effects
54,116,119,120], velocity of ultrasonic waves [31,54,69], in rocks (review). J Min Sci 1994;30:463–73.
electromagnetic emission activity [24,73]. Measuring [4] Yu XF, Ge SG, Yu J, Shang YJ. Rock memory and intelligent
computing of the excavation process in rock. Int J Rock Mech
these parameters, as well as some others, e.g. perme-
Min Sci 2000;37:549–54.
ability and electric resistivity, simultaneously with AE [5] Panasiyan LL, Kolegov SA, Morgunov AN. Stress memory
should provide more reliable indicators of stress studies in rocks by means of acoustic emission. In: Proceedings
memory in rocks. of the Interenational Conference Mechanics of Jointed and
It should be noted that stress measurement is not the Faulted Rock (MJFR). Rotterdam: A.A.Balkema, 1990.
p. 435–9.
only possible application of the Kaiser effect. Other
[6] Gatelier N, Pellet F, Loret B. Mechanical damage of an
applications, although not well-studied so far, include anisotropic porous rock in cyclic triaxial tests. Int J Rock Mech
the investigation of physico-chemical structural changes Min Sci 2002;39:335–54.
in rocks, and seismicity analysis. The decay of the Kaiser [7] Kaiser J. Erkenntnisse und Folgerungen aus der Messung von
effect in the course of time or induced by moistening is Ger.auschen bei Zugbeanspruchung von metallischen Werkstof-
believed to be due to crack healing or other physico- .
fen. Archiv Eisenhuttenwesen 1953;24:43–5.
[8] Goodman RE. Subaudible noise during compression of rocks.
chemical processes. The Kaiser effect and FR provide Bull Geol Soc Am 1963;74:487–90.
indirect means of quantifying these phenomena [9] Kanagawa T, Nakasa H. Method of estimating ground pressure.
[13,15,115]. USPatent No. 4107981, 1978.
Large scale loading of rocks may be (and sometimes [10] Holcomb DJ. Using acoustic emission to determine in situ stress:
is) accompanied by the Kaiser effect (Section 2.10). problems and promise. In: Geomechanics, vol. 57, ASME,
AMD, 1983. p. 11–21.
Further investigation of the large scale manifestation of
[11] Holcomb DJ, Martin RJ III. Determining peak stress history
stress memory should provide a better understanding of using acoustic emissions. In: Proceedings of the 26th US
natural and human-induced seismicity patterns. Symposium on Rock Mech. Rotterdam: A.A.Balkema, 1985.
The following future research directions may be p. 715–22.
outlined: [12] Holcomb DJ. General theory of the Kaiser effect. Int J Rock
Mech Min Sci Geomech Abstr 1993;30:929–35.
(1) investigation of the damage formation and the [13] Li C, Nordlund E. Experimental verification of the Kaiser effect
Kaiser effect in rocks subjected to a true triaxial in rocks. Rock Mech Rock Eng 1993;26:333–51.
(polyaxial) state of stress; [14] Pestman BJ, Van Munster JG. An acoustic emission study of
damage development and stress-memory effects in sandstone. Int
(2) further investigation into the influence of damage
J Rock Mech Min Sci Geomech Abstr 1996;33:585–93.
introduced during coring on the Kaiser effect in [15] Shkuratnik VL, Lavrov AV. Memory effects in rocks. Physical
subsequent laboratory tests; regularities and theoretical models. Moscow: Academy of
(3) further investigation of the in situ Kaiser effect and Mining Sciences Press, 1997.
its potential for stress measurements, the borehole [16] Li C. Theory for the Kaiser effect and its potential applications.
technique described in Section 4.2, as well as other In: Proceedings of the Sixth Conference. AE/MA in Geologic
Structures and Materials. Clausthal-Zellerfeld: Trans Tech
techniques, e.g. using flatjacks, should be verified; Publications, 1998. p. 171–85.
(4) investigation of the Kaiser effect mechanisms other [17] Yoshikawa S, Mogi K. Experimental studies on the effect
than microcracking. of stress history on acoustic emission activity—a possibility
for estimation of rock stress. J Acoust Emission 1989;8(4):
113–23.
[18] Shen HW. Objective Kaiser stress evaluation in rock. In:
Acknowledgements Proceedings of the Fifth Conference. AE/MA in Geologic
Structures and Materials. Clausthal-Zellerfeld: Trans Tech
Publications, 1995. p. 177–96.
The author is grateful to Prof. J.A. Hudson for his
[19] Hardy Jr. HR, Zhang D, Zelanko JC. Recent studies
encouragement to write this review article and for of the Kaiser effect in geologic materials. In: Proceedings of
raising some of the questions discussed herein. Numer- the Fourth Conference. AE/MA in Geologic Structures and
ous discussions with Prof. V.L. Shkuratnik during many Materials. Clausthal-Zellerfeld: Trans Tech Publications, 1989.
years of the author’s work at the Moscow State Mining p. 27–55.
[20] Momayez M, Hassani FP, Hardy Jr. HR. Maximum curvature
University are highly appreciated. The support of the
method: a technique to estimate kaiser-effect load from acoustic
Russian Foundation for Basic Research for this work is emission data. J Acoust Emission 1991/1992;10(3–4):61–5.
gratefully acknowledged (Projects No. 01-05-64105 and [21] Villaescusa E, Seto M, Baird G. Stress measurement from
No. 00-15-98590). oriented core. Int J Rock Mech Min Sci 2002;39:603–15.
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 169

[22] Yoshikawa S, Mogi K. A new method for estimation of the use for stress measurement in a rock mass. Rock Mech Rock
crustal stress from rock samples: laboratory study in the case of Eng 2001;34:275–91.
uniaxial compression. Tectonophysics 1981;74:323–39. [42] Lavrov A. Kaiser effect observation in brittle rock cyclically
[23] Sobolev GA, Semerchan AA, Salov BG, Spetzler HA, Sonder- loaded with different loading rates. Mech Mater 2001;33:669–77.
geld KH, Badanov AV, Koltsov AV, Los VF, Nasimov RM, [43] Lavrov A, Vervoort A, Wevers M, Napier JAL. Experimental
Ponomarev AV, Stakhovsky IR, Terentyev VA, Turetsky IM. and numerical study of the Kaiser effect in cyclic Brazilian tests
Precursors of destruction in a large rock sample. Izv Akad Nauk with disk rotation. Int J Rock Mech Min Sci 2002;39:287–302.
SSSR Fiz Zemli (Phys Solid Earth) 1982;8:29–43 [in Russian]. [44] Lavrov A, Vervoort A, Filimonov Y, Wevers M, Mertens J.
[24] Rzhevskii VV, Iamshchikov VS, Shkuratnik VL, Lykov KG, Acoustic emission in host rock material for radioactive waste
Farafonov VM. Emission effects of memory in rocks. Dokl disposal: comparison between clay and rock salt. Bull Eng Geol
Akad Nauk SSSR 1983;273:1094–7 [in Russian]. Environ, in print.
[25] Koerner RM, Lord AE, Deutsch WL. Determination of [45] Filimonov Y, Lavrov A, Shafarenko Y, Shkuratnik V.
prestress in granular soils using AE. J Geotech Eng ASCE Observation of post-failure Kaiser effect in a plastic rock. Pure
1984;110:346–58. Appl Geophys (PAGEOPH) 2002;159:1321–31.
[26] Koerner RM, Lord AE, Deutsch WL. Determination of [46] Qin S, Wang S, Long H, Liu J. A new approach to estimating
prestress in cohesive soils using AE. J Geotech Eng ASCE geo-stresses from laboratory Kaiser effect measurements. Int J
1984;110:1537–48. Rock Mech Min Sci 1999;36:1073–7.
[27] Lord AE, Koerner RM. Field determination of prestress [47] Dunning JD, Leaird JD, Miller ME. The Kaiser effect and
(existing stress) in soil and rock masses using acoustic emission. frictional deformation. In: Proceedings of the Fourth Conference
J Acoust Emission 1985;4(2–3):S22–5. on AE/MA in Geologic Structures and Materials. Clausthal-
[28] Leaird JD, Dunning JD, Miller ME. Kaiser experiment in Zellerfeld: Trans Tech Publications, 1989. p. 3–13.
sawcut rock. J Acoust Emission 1985;4(2–3):S11–6. [48] Konecny P. Acoustic emission during cyclic loading of
[29] Reymond M. Acoustic emission following hydraulic fracturing carboniferous rocks and the manifestation of Kaiser effect. In:
in calcareous and sandstone rock. Acoust Lett 1989;13:95–8. Proceedings of the EUROCK 2000 Symposium. Essen: Verlag
[30] Mottahead P, Vance JB. A/E determination of salt behavior Gluckauf, p. 649–52.
under stress. In: Proceedings of the Fourth Conference. AE/MA [49] Lavrov AV, Filimonov YL, Shafarenko YM, Shkuratnik VL.
in Geologic Structures and Materials. Clausthal-Zellerfeld: Experimental investigation of memory effects in rock salt at
Trans Tech Publications, 1989. p. 465–74. different regimes of cyclic loading. In: Physics and Mechanics
[31] Kuwahara Y, Yamamoto K, Hirasawa T. An experimental and of Geomaterials. Moscow: Vuzovskaya Kniga, 2002. p. 73–93
theoretical study of inelastic deformation of brittle rocks under [in Russian with abstract in English].
cyclic uniaxial loading. Tohoku Geophys J (Sci Rep Tohoku [50] Michihiro K, Yoshioka H, Hata K, Fujiwara T. Strain
Univ; Ser 5; Geophys) 1990;33:1–21. dependence of the Kaiser effect for various rocks. In: Proceed-
[32] Nordlund E, Li C. Acoustic emission and the Kaiser effect in ings of the Fourth Conference on AE/MA in Geologic
rock materials. In: Rock Mechanics Contributions and Chal- Structures and Materials. Clausthal-Zellerfeld: Trans Tech
lenges. Proceedings of the 3first US Rock Mechanics Sympo- Publications, 1989. p. 87–95.
sium. Rotterdam: A.A.Balkema, 1990. p. 1043–50. [51] Filimonov YL. Patterns of acoustic emission during deformation
[33] Friedel MJ, Thill RE. Stress determination in rock using the of salt rocks. PhD thesis, Moscow State Mining University,
Kaiser effect. US Bureau of Mines Report of Investigations Moscow, 2002. 203pp.
No. 9286, 1990. 20pp. [52] Lavrov AV, Filimonov YL, Shkuratnik VL. Features of acoustic
[34] Yamshchikov VS, Shkuratnik VL, Lykov KG. Stress measure- emission in plastic rocks. In: Proceedings of the 11th Session
ment in a rock bed based on emission memory effects. Sov Min of the Russian Acoustical Society, vol. 2. Moscow: GEOS, 2001.
Sci USSR 1990;26(2):122–7. p. 150–4 [in Russian].
[35] Friedel MJ, Thill RE. US Bureau of Mines research on the [53] Koerner RM, Lord AE. AE detection of prestress in soil and
Kaiser effect for determining stress in rock. J Acoust Emission rock. In: Proceedings of the Fourth Conference on AE/MA in
1991/1992;10(1–2):S77–89. Geologic Structures and Materials. Clausthal-Zellerfeld: Trans
[36] Hardy Jr. HR. Evaluation of in situ stresses in salt using acoustic Tech Publications, 1989. p. 73–86.
emission techniques. In: Proceedings of the Seventh Symposium [54] Shin K, Kanagawa T. Kaiser effect of rock in acousto-elasticity,
on Salt, vol. 1. Amsterdam: Elsevier, 1993. p. 49–58. AE and DR. In: Proceedings of the Fifth Conference on AE/MA
[37] Seto M, Nag DK, Vutukuri VS. Application of acoustic emission in Geologic Structures and Materials. Clausthal-Zellerfeld:
technique to evaluation of rock mass damage. In: Trends in Trans Tech Publications, 1995. p. 197–204.
NDE science and technology. Proceedings of the 14th World [55] Michihiro K, Hata K, Yoshioka H, Fujiwara T. Determination
Conference on Non-Destructive Testing, vol. 4, New Delhi: of the initial stresses on rock mass using acoustic emission
Ashgate Publishing Company, 1996. p. 2451–4. method. J Acoust Emission 1991/1992;10(1–2):S63–76.
[38] Hardy Jr. HR, Shen HW. Acoustic emission in salt during elastic [56] Holcomb DJ, Costin LS. Detecting damage surfaces in
and inelastic deformation. In: Proceedings of the Sixth brittle materials using acoustic emissions. J Appl Mech 1986;
Conference on AE/MA in Geologic Structures and Materials. 53:536–44.
Clausthal-Zellerfeld: Trans Tech Publications, 1998. p. 15–28. [57] Lavrov AV. Formation and manifestation of memory effects in
[39] Barr SP, Hunt DP. Anelastic strain recovery and the Kaiser rocks. D.Sc. (Dr. Eng. Habil.) thesis, Moscow, Moscow State
effect retention span in the Carnmenellis granite. UK Rock Mining University, 2001. 582pp [in Russian].
Mech Rock Eng 1999;32:169–93. [58] Filimonov YL, Lavrov AV, Shafarenko YM, Shkuratnik VL.
[40] Filimonov YL, Lavrov AV, Shkuratnik VL. Prospects of Experimentelle Untersuchung des Steinsalzes mittels einaxialem
memory effects for stress measurement applications in rock salt. Drucktest nach hydrostatischer Vorbelastung. Gluckauf- .
In: Proceedings of EUROCK’2001 Symposium. Rotterdam: Forschungshefte 2000;61:80–83 [in German].
A.A.Balkema, 2001. p. 59–63. [59] Holcomb DJ, Rudnicki JW. Inelastic constitutive properties and
[41] Filimonov YL, Lavrov AV, Shafarenko YM, Shkuratnik VL. shear localization in Tennessee marble. Int J Numer Anal Meth
Memory effects in rock salt under triaxial stress state and their Geomech 2001;25:109–29.
170 A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171

[60] Rudnicki JW, Rice JR. Conditions for the localization of [77] McElroy JJ, Koerner RM, Lord AE. An acoustic jack to
deformation in pressure-sensitive dilatant materials. J Mech assess in situ rock behavior. Int J Rock Mech Min Sci 1985;
Phys Solids 1975;23:371–94. 22:21–9.
[61] Kachanov ML. A microcrack model of rock inelasticity. Part I: [78] Yamshchikov VS, Shkuratnik VL, Lykov KG, Farafonov VM.
frictional sliding on microcracks. Mech Mater 1982;1:19–27. Evaluation of the stressed-state of a bed based on emission
[62] Kachanov M. Effective elastic properties of cracked solids: memory effects of rocks in near-well space. Sov Min Sci USSR
critical review of some basic concepts. Appl Mech Rev 1991;27:100–3.
1992;45:304–35. [79] Yamshchikov VS, Shkuratnik VL, Lavrov AV. Monitoring the
[63] Lavrov A. Theoretical investigation of the Kaiser effect stress in a rock column on the basis of memory effects around a
manifestation in rocks after true triaxial pre-loading. Arch Min borehole. J Min Sci 1995;31:104–8.
Sci 2001;46:47–65. [80] Shkuratnik VL, Lavrov AV. Simulation modeling for monitor-
[64] Lavrov A, Vervoort A, Wevers M. Anisotropic damage ing the plane stressed state of a rock mass based on the acoustic
formation in brittle rock: experimental study by means of emission memory effect around a borehole. J Min Sci
acoustic emission and Kaiser effect. In: Proceedings of the Sixth 1995;31:355–9.
Euromech-Mecamat Conference on Non-Linear Mechanics of [81] Chkouratnik VL, Lavrov AV. Numerical 2D-simulation of
Anisotropic Materials. University of Li!ege, 2002. p. 385–92. memory effects in rocks around a borehole. In: Rock stress.
[65] Pestman BJ, Kenter CJ, Van Munster JG. Core-based determi- Proceedings of the International Symposium on Rock Stress.
nation of in situ stress magnitudes. In: Proceedings of the DC Rotterdam: A.A.Balkema, 1997. p. 193–6.
Rocks 2001, US Symposium on Rock Mechanics, July 7–10 [82] Lavrov AV. Development of theoretical models of emission
2001, Washington, DC. Rotterdam: A.A.Balkema. memory effects in rocks. PhD thesis, Moscow, Moscow State
[66] Holt RM, Pestman BJ, Kenter CJ. Use of discrete particle model Mining University, 1998. 202pp.
to assess feasibility of core based stress determination. In: [83] Simpson DW, Negmatullaev SK. Induced seismicity at Nurek
Proceedings of the DC Rocks 2001, US Symposium on Rock reservoir, Tadjikistan, USSR. Bull Seismol Soc Am
Mechanics, July 7–10 2001, Washington, DC. Rotterdam: 1976;71:1561–86.
A.A.Balkema. [84] Roeloffs EA. Fault stability changes induced beneath a reservoir
[67] Stuart CE, Meredith PG, Murrell SAF, Van Munster JG. with cyclic variations in water level. J Geoph Res B 1988;
Anisotropic crack damage and stress-memory effects in rocks 93:2107–24.
under triaxial loading. Int J Rock Mech Min Sci Geomech Abstr [85] Gupta HK, Mandal P, Rastogi BK. How long will triggered
1993;30:937–41. earthquakes at Koyna, India continue? Curr Sci 2002;82:202–10.
[68] Subhash G, Nemat-Nasser S. Dynamic stress-induced transfor- [86] Mandal P, Rastogi BK, Gupta HK. Recent Indian earthquakes.
mation and texture formation in uniaxial compression of Curr Sci 2000;79:1334–46.
zirconia ceramics. J Am Ceram Soc 1993;76:153–65. [87] Young RP. Seismic propagation in rock masses: implications for
[69] Stuart CE, Meredith PG, Murrell SAF, Van Munster H. AE/MA. In: Procedings of the Fourth Conference on AE/MA in
Influence of anisotropic crack damage development on the Geologic Structures and Materials. Clausthal-Zellerfeld: Trans
Kaiser effect under true triaxial stress conditions. In: Proceed- Tech Publications, 1989. p. 531–50.
ings of the Fifth Conference on AE/MA in Geologic Structures [88] Shkuratnik VL, Lavrov AV. Memory effects and variations
and Materials. Clausthal-Zellerfeld: Trans Tech Publications, of the physical fields and properties of rocks with the
1995. p. 205–19. origination of dynamic phenomena in a rock mass. J Min Sci
[70] Holt RM, Brignoli M, Kenter CJ. Core quality: quantification of 1999;35:471–5.
coring-induced rock alteration. Int J Rock Mech Min Sci [89] Mostryukov AO, Lykov VI, Patonin AV, Petrov VA, Sasorova
2000;37:889–907. EV. Acoustic emission in experiments ‘‘stick-slip’’ as clone of
[71] Li Y, Schmitt DR. Effects of Poisson’s ratio and core stub length low-frequency predictors of earthquakes. In: Physics and
on bottomhole stress concentrations. Int J Rock Mech Min Sci Mechanics of Geomaterials. Moscow: Vuzovskaya Kniga,
1997;34:761–73. 2002. p. 94–104 [in Russian with abstract in English].
[72] Sakaguchi K, Iino W, Matsuki K. Damage in rock core caused [90] Yamamoto K, Kuwahara Y, Kato N, Hirasawa T. Deformation
by induced tensile stress and its relation to differential strain rate analysis: a new method for in situ stress estimation from
curve analysis. Int J Rock Mech Min Sci 2002;39:367–80. inelastic deformation of rock samples under uni-axial compres-
[73] Yamshchikov VS, Shkuratnik VL, Lykov KG. Stress measure- sion. Tohoku Geophy J (Sci Rep Tohoku Univ; Ser 5; Geophys)
ment in a rock bed based on emission memory effects. Sov Min 1990;33:127–47.
Sci USSR 1990;26:122–7. [91] Yamamoto K, Yamamoto H, Kato N, Hirasawa T. Deforma-
[74] Teufel LW. Acoustic emissions during anelastic strain recovery tion rate analysis for in situ stress estimation. In: Proceedings of
of cores from deep boreholes. In: Proceedings of the 30th US the Fifth Conference on AE/MA in Geologic Structures and
Symposium on Rock Mech. Assoc Eng Geol, p. 269–76. Materials. Clausthal-Zellerfeld: Trans Tech Publications, 1995.
[75] Zinke JC. The relationship between open, dry microcracks and P p. 243–55.
wave velocity of crystalline rocks: application to in situ stress [92] Fujii N, Hamano Y. Anisotropic changes in resistivity and
estimates on KTB drill cores after stress relaxation. J Geoph Res velocity during rock deformation. In: High-pressure research:
B 1999;104:10863–81. applications in geophysics. London: Academic Press, 1977.
[76] Berckhemer H, Rauen A, Winter H, Kern H, Kontny A, Lienert p. 53–64.
M, Nover G, Pohl J, Popp T, Schult A, Zinke J, Soffel HC. [93] Reed LD, McDowell GM. A fracto-emission memory effect and
Petrophysical properties of the 9-km-deep crustal section at subharmonic vibrations in rock samples stressed at sonic
KTB. J Geoph Res B 1997;102:18337–61. [A rich collection of frequences. Rock Mech Rock Eng 1994;27:253–61.
data on acoustic emission during anelastic strain recovery and [94] Jonason K, Nordblad P, Vincent E, Hammann J, Bouchaud J-P.
other data measured on KTB core specimens can be found Memory interference effects in spin glasses. Eur Phys J
at the web site of GeoForschungsZentrum Potsdam: http:// 2000;13:99–105.
icdp.gfz-potsdam.de/html/ktb/ektbdata.htm or http://icdp.gfz- [95] Bellon L, Ciliberto S, Laroche C. Advanced memory effects in
potsdam.de/html/ktb/ktb.htm]. the aging of a polymer glass. Eur Phys J 2002;25:223–31.
A. Lavrov / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 151–171 171

[96] Stevens JL, Holcomb DJ. A theoretical investigation of Proceedings of the Fourth North American Rock Mechanics
the sliding crack model of dilatancy. J Geoph Res B 1980;85: Symp., July 31–August 3 2000. Rotterdam: A.A.Balkema, 2000.
7091–100. p. 1217–24.
[97] Holcomb DJ, Stevens JL. The reversible Griffith crack—a viable [110] Lavrov A, Vervoort A, Napier JAL. Damage development in
model of dilatancy. J Geoph Res B 1980;85:7101–7. rock subjected to cyclic loading. In: Proceedings of the Second
[98] Holcomb DJ. Memory, relaxation, and microfracturing in Biot Conference on Poromechanics. Lisse, Swets & Zeitlinger,
dilatant rock. J Geoph Res B 1981;86:6235–48. 2002. p. 911–6.
[99] Guyer RA, McCall KR, Boitnott GN. Hysteresis, discrete [111] Watters RJ, Roberts K. The Kaiser effect and its application to
memory, and nonlinear-wave propagation in rock—a new slope instability. In: Proceedings of the 5th Conference on AE/
paradigm. Phys Rev Lett 1995;74:3491–4. MA in Geologic Structures and Materials. Clausthal-Zellerfeld:
[100] Guyer RA, McCall KR, Boitnott GN, Hilbert LB,Plona. Trans Tech Publications, 1995. p. 233–42.
Quantitative implementation of Preisach-Mayergoyz space to [112] Wang HT, Xian XF, Yin GZ, Xu J. A new method of
find static and dynamic elastic moduli in rock. J Geophys Res B determining geostresses by the acoustic emission Kaiser effect.
1997;102:5281–93. Int J Rock Mech Min Sci 2000;37:543–7.
[101] Guyer RA, Johnson PA. Nonlinear mesoscopic elasticity: [113] Kaiser PK, Yazici S, Maloney S. Mining-induced stress change
evidence for a new class of materials. Physics Today and consequences of stress path on excavation stability—a case
1999;52(4):30–6. study. Int J Rock Mech Min Sci 2001;38:167–80.
[102] Costin LS. A microcrack model for the deformation and failure [114] Eberhardt E. Numerical modelling of three-dimension stress
of brittle rock. J Geophys Res B 1983;88:9485–92. rotation ahead of an advancing tunnel face. Int J Rock Mech
[103] Holcomb DJ. Observations of the Kaiser effect under multiaxial Min Sci 2001;38:499–518.
stress states: implications for use in determining in situ stress. [115] Lavrov AV. Computer simulation of memory relaxaion in rocks
Geophys Res Lett 1993;20:2119–22. around a borehole. Arch Min Sci 1997;42:353–65.
[104] Lavrov AV. Three-dimensional simulation of memory effects in [116] Chodyn L, Zuberek WH. Effect of the discrete stress memory in
rock samples. In: Rock Stress. Proceedings of the International acoustic emission in soils. Acta Geophys Pol 1992;40:139–58.
Symposium on Rock Stress. Rotterdam: A.A.Balkema, 1997. [117] Watters RJ, Soltani AM. Directional acoustic emission activity
p. 197–202. in response to borehole deformation in rock masses. In:
[105] Lavrov AV, Yasinski MV. On the influence of intermediate Proceedings of the 26th US Symposium on Rock Mechanics
principal stress upon acoustic emission in cyclically loaded rocks. (Research and Engineering Applications in Rock Masses), vol. 2.
In Proceedings of the 10th Session of the Russian Acoustical Rotterdam: A A Balkema, 1985, p. 723–30.
Society. Moscow: GEOS 2000, p. 211–4 [in Russian; extended [118] Rudnicki JW. Geomechanics. Int J Solids Struct 2000;37:349–58.
abstract in English at URL: http://www.akin.ru/e main.htm]. [119] Kudo R, Yokoyama T, Ito H, Kuwahara Y, Nishizawa
[106] Shkuratnik VL, Lavrov AV. Memory effects in rock. J Min Sci O,Yamamoto K. Stress measurements with core samples by
1995;31:20–8. AE-DRA methods in the 1995 Hyogoken-nanbu earthquake
[107] Shkuratnik VL, Lavrov AV. Numerical simulation of the Kaiser source region. In: Rock Stress. Proceedings of the International
effect under triaxial stress state. Proceedings of the Third Symposium on Rock Stress. Rotterdam: A.A.Balkema, 1997.
International Conference on Mechanics of Jointed and Faulted p. 359–62.
Rock (MJFR-3). Rotterdam: A.A.Balkema, 1998. p. 381–5. [120] Utagawa M, Seto M, Katsuyama K. Determination of in situ
[108] Tang CA, Chen ZH, Xu XH, Li C. A theoretical model for stress using DRA and AE techniques. In: Rock Stress.
Kaiser effect in rock. Pure Appl Geophys (PAGEOPH) Proceedings of the International Symposium on Rock Stress.
1997;150:203–15. Rotterdam: A.A.Balkema, 1997. p. 187–92.
[109] Holt RM, Brandshaug T, Cundall PA. Discrete particle and
laboratory modelling of core mechanics. In: Pacific Rocks 2000.

You might also like