Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303880637

The desirable and undesirable effects on stress reconstruction using the


deformation rate analysis (DRA)

Conference Paper · May 2016

CITATIONS READS

5 986

2 authors:

Ariel Hsieh Phil Dight


University of Western Australia University of Western Australia
9 PUBLICATIONS 86 CITATIONS 59 PUBLICATIONS 535 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Ariel Hsieh on 10 June 2016.

The user has requested enhancement of the downloaded file.


THE DESIRABLE AND UNDESIRABLE EFFECTS ON STRESS RECONSTRUCTION USING THE
DEFORMATION RATE ANALYSIS (DRA)

Ariel Hsieh (ariel.hsieh@uwa.edu.au)


The Australian Centre for Geomechanics, The University of Western Australia
Australia

Phil Dight
The Australian Centre for Geomechanics, The University of Western Australia
Australia

ABSTRACT

The Deformation Rate Analysis (DRA) utilizes the stress-strain curves obtained from uniaxial tests on rock
samples to reconstruct the in-situ stress the rock was subjected to. The approach to this test is to measure the
strain difference between two completed load-unload cycles under uniaxial compression. The method has been
used since 2002 to undertake in situ stress tensor determinations. In general, the method uses oriented diamond
drill core obtained for exploration purposes. The strain difference can be as small as a few microstrains, hence
an imperfect test environment could easily vary the reading of strain and as a consequence the result can be
difficult to interpret or is misleading. Another potential issue associated with the DRA method can be attributed to
the rock properties. Because the sample was originally in a confined environment (underground), it expands
after the core has been extracted. The amount of expansion (open/movement of pre-existing cracks, interfaces,
and grain boundaries) in a core is likely to be nonuniform along whole core. When applying a load to the sample,
“crushing of asperities at crack inner surface” occurs during the process of crack closure and it creates unwanted
inelastic strain which can mask the indicator of the in situ stress. This compaction effect can be observed using
acoustic sensors and observation of the stress-strain curve, and it is less pronounced in the lateral strain
difference.
In this paper we addressed some of the issues with undertaking the tests and completing the interpretation of
the results. A suggested method of conducting a test for DRA analysis is presented in an appendix. We provide
some simple solutions to improve the test condition and examine the reliability of test. Examples of the
compaction effect are shown and the suggested approach when dealing with this type of result is demonstrated.
The “bad” result is discussed and compared with “good” result. The aim is to improve or reject the result which is
unsatisfactory for stress reconstruction purposes.

KEYWORDS

In-situ stress measurement, Deformation rate analysis, compaction effect, inelastic strain, test procedure

INTRODUCTION

The Deformation Rate Analysis (DRA) [Yamamoto et al., 1990] utilizes the stress-strain curves obtained from
uniaxial tests on rock samples to reconstruct the in-situ stress the rock was subjected to. The approach is based
on examining the inelastic strain between two successive loading cycles in a uniaxial test. The loading program
is shown in Figure 1.

The pre-stress is the maximum previous stress that the specimen has been subjected. It is the first loading in the
lab experiments and the in-situ stress in the tests for stress reconstruction. The 1st and 2nd cycles are the loading
cycles applied after the pre-stress, in order to produce the DRA graph. The procedure of conducting a test is
listed in the appendix. According to the DRA the stress of the inflection point is assumed to be equal to the pre-
stress.

213
(A) Loading process (B) Stress-strain curve (C) DRA inflection point
prestress

strain difference
1st cycle

stress
2nd cycle
stress

time strain a stress


Figure 1. (A) The loading programme: the pre-stress, 1st and 2nd cycles; (B) Stress and strain curve. A black
arrow shows a strain difference between the 1st and 2nd cycles vs. stress. The stress a, is the previous maximum
stress; and (C) the inflection point which indicates previous maximum stress, a, is marked with arrow.

It is believed that the mechanism of DRA is frictional sliding over pre-existing cracks, interfaces and grain
boundaries [Wang et al., 2012]. Hence, other mechanisms (for example crack generation and compaction) which
generate irreversible inelastic strain can possibly affect the accuracy of DRA technique. Due to the fact that
frictional sliding will generate inelastic strain in any direction, the inflection point should be observed for both
axial and lateral strain differences. This is a unique aspect of this approach. Although most of stress estimations
by the DRA technique have utilizing the strain difference in the direction parallel to the sample axis (earliest DRA
report: [Yamamoto et al., 1990]; measured the in situ stress: [P Dight and Snyman, 2010; P M Dight, 2006;
Yamamoto and Yabe, 2001]), the lateral strain difference also shows a clear inflection point coincided with in-situ
stress [Hsieh et al., 2013].

In the past, we have seen many published DRA results which were poorly done and should be discarded due to
imperfect test conditions. However they were treated as “normal DRA results” and the outcome misled one’s
perspective to the reliability of this method. Before a specific standard for DRA test is made, it is important for
one to be aware of all potential problems before considering using the DRA technique as a stress measurement
method.

In this paper we discussed the difference between axial and lateral strain differences regarding the accuracy of
the stress determination. We also discussed issues which can affect the accuracy, for example bending of the
sample during testing (“bending effect”), leakage in circuit board, electronic noise, change in temperature, air
flow, and issues generated by use of LVDTs. The issues introduced by test conditions usually can be solved or
eliminated by improving the test procedure. The issues associated with material itself are harder to eliminate,
and sometimes the result should be discarded. The aim of this paper is to demonstrate the way to tell a “good
result” from a “bad result”, and the essential procedures/settings for the DRA test in order to have a successful
in-situ stress determination.

EXPERIMENTAL APPARATUS

All DRA plots presented in the following paragraphs were done by testing rock cores recovered from mine sites
when the tests were conducted by UWA. The subsamples were drilled out from cores and all samples were
prepared in accordance with Section 3. “Apparatus” presented in the appendix. The samples were loaded using
a servo-controlled loading machine of 5t capacity. The load was displacement-controlled, applied by the
movement of the upper platform, while the bottom platform was fixed. We used glued cross-type strain gauges
that ensure simultaneous measurements of the axial and lateral strains. All samples were subjected to 2 loading
cycles under a constant loading/unloading rate of 7~9MPa/min. The average strain was calculated by taking
average of readings of 4 strain gauges.

Before conducting any DRA test, a calibration test would be performed using an aluminium sample under the
same test system (load frame, wire, and logging system) and same test condition (loading rate, sample size,

214
strain gauge, and setups). It is used to estimate the calibration factor for strain gauge factor and check the load
cell/displacement control is in good working order.

Load
Upper platen

Sample Sphere
seat

Sample
Glued strain gauge

Strain gauge on a spring

Bottom platen Steel frame

Bottom plate

Figure 2. The loading frame and locations of the strain gauges on a sample.

DISCUSSION ON TEST CONDITION

The influences from test condition/procedure, such as the bending effect, leakage in circuit board, electronic
noise, changes in temperature, air flow, and LVDT, are shown in the following section.

Bending Effect

It was shown in [Hsieh et al., 2013] that the sample heterogeneity, the non-parallelness of the sample ends, and
the non-coaxiality of the applied loading can all attribute to the stress/strain non-uniformity. Although one may
believe that the load frame is perfectly leveled and the non-parallelness of the sample ends is absent, it is rarely
the reality in our experience. The test setup might appear to be perfect in the perspective of checking by naked
eye. Any minor imperfection would affect the result.

When there is only one axial strain gauge or one cross type strain gauge (one axial and one lateral gauge)
attached to a sample, the inflection point can be shifted by the bending effect without the tester’s notice. A
0.01mm non-parallelness of the samples ends can create up to 14MPa stress difference in a sample with 70GPa
stiffness and 50mm in length. The condition may not be improved by using a sphere seat suggested by [ISRM,
1979], due to the friction between sphere seat and sample. Hence it is essential to use more than one axial
strain gauge to check the bending level of each test.

Other than the bending caused by non-parallelness, non-coaxiality is another factor which contributes to the
nonuniform load. The non-coaxiality can be easily improved by using an additional frame, which can be
manufactured or modified from a Brazilian test frame (Figure 2). In Figure 2, a steel frame with two spherical
seats attached will provide minimal bending according to our experience. The sample is placed in the centre of
spherical seats and one can rotate the seat and/or sample after a little load (<300N) is applied. The action of
rotating the sample/seat helps adjusting the position of sphere and it also crushes the dirt/planarity on the
sample ends. Hence, the sample end and sphere seats are nicely attached. To help with the action of rotating
the sample, lubricant can be applied to the sample ends.

Other than the bending effect on stress prediction, we found bending could make the inflection point unclear,
even when the average strain difference from 4 strain gauges was calculated. In Figure 3, three identical

215
samples was subjected to same test condition other than different level of bending. Sample B1 has lowest
bending level and clearest inflection point. The line before inflection point is almost a straight line. Sample B2
and sample B3 do not show a clear inflection point, because the line is in curve shape before and after the
inflection point. One might pick 66MPa instead of 76MPa from sample B2 and sample B3. Although a good
inflection point sometimes can still be observed from a sample with high bending levels, the bending effect could
be a major factor contributing to a bad/unclear prediction.

In terms of using an LVDT to replace the strain gauge, 4 LVDTs should be installed at 90° apart surrounding the
sample in order to check the level of bending. However, there are other issues associated with the LVDTs and it
is discussed later.

KC/E DRA plot


20
30

76
Strain (micro strain)

0 10

B1_5% bending
B2_14% bending
B3_19% bending
-20 -10
0 38 76
Stress (MPa)

Figure 3. The DRA plots of three identical samples with different bending level during testing. The higher the
bending level, the more curvature there is in the DRA response. The samples were prepared from a sub-vertical
orientation and the estimated overburden stress was 76MPa.

Issues associated with LVDT

Installing strain gauges takes time and the cost of strain gauges can be significant. A set of one axial and one
lateral gauges can cost between $14 to $20+ AUD, which equals to $56+ per test. The noise/issues generated
by bad installation are common and usually an experienced person is required when acquiring strain gauge data.
As a consequence, an LVDT becomes a popular option. However, we found using an LVDT to record the strain
difference in the DRA technique does not always work. Firstly, the accuracy of LVDT has to be better than
several microstrains, which means the error should be less than 0.0001mm in a 20mm diameter sample.
Secondly, the interface between the sample and the platens can also contribute to inelastic strain, after the first
load could flatten the planarity of the end surface. The planarity (roughness) is likely to be more than 0.0001mm
and it could mask the result.

Ideally the LVDT should be mounded on sample surface; hence the planarity issue is minimal.

Changes in Temperature and Leakage in Circuit Board

When using a strain gauge, the change in gauge resistance enables the strain to be recorded as a change in the
voltage. Since the strain gauge acts as a resistance, it generates heat. Heat then changes the resistance of the
gauge and shifts the value of voltage. This issue can be easily found by checking the stress-strain plot. In Figure
6, the same amount of offsets in the axial and lateral strains indicates the data was gradually shifted from its

216
original value during testing. The change in temperature and minor leakage in the wire connection can both give
the same effect.

(a) 16_B2 Stress-Strain Plot (b) 16_B2 DRA plot


45
60 50

Strain (micro strain)


Stress (MPa)

40
30
30
20
Axial
Lateral
0 10 15
-200 0 200 400 600 0 20 40 60
Strain (micro strain) Stress (MPa)
Figure 4. Both axial and lateral strains became same value (>0) when the load was reduced to zero, due to the
increase of temperature during test.

A simple way to solve this issue is to connect all wires and check the voltage readings. Based on our
experiments, between 10 to 30 minutes is required for the temperature increment in the gauge to reach
equilibrium. In many cases the sample could be heated up during testing as well, but a strain gauge with an
appropriate coefficient of thermal expansion and full bridge circuit board will minimize this effect. If after a certain
of time and the reading is still gradually changing, it could be a bad connection on the wire or electric leaks in
circuit board. The wire should be checked and reconnected.

Electronic Noise

It is unavoidable to have electronic noise in any measurement using strain gauges; this includes the load cell.
Due to the nature of the DRA method, very precise stress and strain records are required. The accuracy of two
micro strains is essential for a decent result if the rock is stiff. An error less than 100N is required for load cell.
Two micro-strains is equal to few micro-volts in most strain gauges and it is quite common that the noise
generated by data acquisition system is ten or hundred times higher than a few micro volts.

If the noise is generated purely by electronic interference, the range of noise amplitude will remain similar for
each test and the frequency of noise is higher than 1MHz. By recording data at very high frequency, the majority
of the electronic noise can be erased by averaging the records. In our case, 100 records per MPa are sufficient
to reduce the noise to one micro strain. However, the situation will depend on the type of load cell, strain gauge,
acquisition system and test environment.

The noise in Figure 5 shows a typical pattern of electronic noise (and maybe other sources of noise, too). The
fluctuation of the readings remains in a similar range and the frequency of noise is high. The graph on right in
Figure 5 also shows other source of noise with lower frequency. It could be unstable power source, vibration or
air flow issue.

Air Flow

Exposed wire/solder terminals can be quite sensitive to air flow because it reduces/increases the temperature of
wire. The change in voltage is very small but it is big enough to shift several micro strains. A plastic cover to stop
air flow through the test sample/wire or any type of coating on the exposed wire/solder terminal can stop the shift
in data (Figure 6).

217
Figure 5. The noise in the DRA plot is larger than 200 microstrains. The result should be discarded as invalid.
(after [Chan, 2008; Chang, 2007])

GT14 A1
5
Strain (micro strain)

-5

-10
0 10 20 30 40 50 60 70
Stress (MPa)

Figure 6. The black line is the DRA plot is from a test without a plastic cover to stop air flow. The grey line is the
plot when same sample was tested again few days later with plastic cover.

DISCUSSION ON MATERIAL PROPERTY

Because the core was originally in a confined environment (underground), it expanded after the core has been
extracted. Due to the distribution of in situ stress, the influence of drilling induced stress, and the likelihood that
the rock has anisotropy deformation, the amount of expansion (open/movement of pre-existing cracks, interfaces,
and grain boundaries) in a core is likely to be non-uniform among whole core. The direction of open/movement
of pre-existing cracks, interfaces, and grain boundaries can be highly variable. Hence, when a load is applied
along the axis of the sample, the asperities (sharp points) on one side of crack surface might not align with the
other side of crack surface in the direction of loading (Figure 7). The asperities on the crack surface would crush
during the process of crack closure and the crack would not fully re-open when the load is removed. The amount
of this “did not re-open” deformation has similar phenomenon as compaction.

218
Compaction caused by the alignment gives similar characteristics to frictional sliding, because they are both
irreversible and the amount of deformation they have generated in each cycle would be different from that of the
following cycle. The inelastic deformation caused by compaction could mask the DRA inflection point.
In-situ extracted applied load

Figure 7. The carton graphs show how the compaction phenomenon was formed. First, the process of extracting
core/sample from underground means the confinement has been removed and the crack can open. Second, the
core/sample is then subjected to a load in the direction of interest. The load leads to the process of crack closure,
and as a consequence some asperities on the inner crack surface would be crushed. The crack with a crushed
inner surface may not be able to fully re-open when the applied load is removed; hence the permanent
irreversible strain is formed and indistinct from the irreversible strain causing by sliding.

Since the crush of asperities was caused by the compression at the axial (loading) direction, the majority of
inelastic strain from compaction is likely to be in the axial direction. This means the lateral strain difference is
possibly not as sensitive as axial strain difference to the compaction effect. Moreover, because frictional sliding
would not change the volume of a sample, the amount of “frictional sliding inelastic strain” in the axial direction
would be proportional to the amount in the lateral direction. Hence, the lateral strain difference should be able to
indicate the same inflection point as the axial strain difference, while it is not subject to the unwanted compaction.

According to the more than 700 samples we have tested in the past, the compaction effect is more significant in
the rock with higher porosity or lower modulus (for example <50GPa). In this type of material, the lateral strain
difference sometimes provides better stress prediction (inflection point) than the axial strain difference. In stiff
rock (>80GPa) with low anisotropy, both axial and lateral strain difference can provide a good/clear inflection
point. However, if the rock stiffness is very high (>100GPa), the strain difference could be very small (<5 micro
strains). As a consequence, the error/noise in the strain readings has to be better than one micro strain in order
to observe the inflection point.

Figure 8 shows an example of good and bad inflection points. The lateral strain difference for sample SF47E2 is
less than two micro strains but the inflection point is clear due to very accurate strain readings. The axial strain
difference in the same sample is not clear. Both axial and lateral inflection points indicate the in-situ stress in
sample SF47E1, although some compaction effect was shown in axial strain difference.

SF47E1 DRA plot SF47E2 DRA plot


10 12

8
Strain (microstrain)
Strain (microstrain)

0 4

-10 -4
0 10 20 30 40 50 0 10 20 30 40 50
Stress (MPa) Stress (MPa)

219
Figure 8. The volcanic sediment samples from 788m depth below surface with the in situ stress of 23MPa (dash
line arrow). Both axial (in black) and lateral (in grey) inflection points are clear and close to the in situ stress in
the sample SF47E1 (left). The axial inflection point in the sample SF47E2 (right) is not as clear as lateral
inflection point.

CONCLUSIONS

The DRA method is relevantly new and it still requires further development to improve the understanding of this
technique. In our experience this technique is able to give good stress estimation in medium to stiff rock, if the
proper test conditions are provided. A successful test requires having noise levels less than few micro strains in
the stress-strain plot and a proper setting of the strain gauge/LVDT/data acquisition. In this paper, economic
solutions have been provided to improve the bending effect, leakage in the circuit board, electronic noise,
changes in temperature, air flow, and issues with use of LVDTs.

REFERENCES

Chan, S. C. (2008), Investigating the laboratory experiments to estimate pre-stress on Changchikeng Sandstone,
Master thesis thesis, 144 pp, National Cheng Kung University, Tainan.

Chang, J. F. (2007), Investigating the laboratory experiments to estimate pre-stress on black schist, Master
thesis thesis, 127 pp, National Cheng Kung University, Tainan.

Dight, P. M., and L. A. Snyman (2010), Stress measurement for St Barbara mines Gwalia Deeps preject- one of
the world's deepest underground haulage mines, Mining Technology, 119(4), 246-254.

Dight, P M. (2006), Determination of in-situ stress from oriented core, in In-Situ Rock Stress - Measurement,
Interpretation and Application, edited by M. Lu, C. C. Li, H. Kjorholt and H. Dahle, pp. 167-175.

Hsieh, A., A. V. Dyskin, and P. M. Dight (2013), The influence of sample bending on deformation rate analysis
stress reconstruction, Int. J. Rock Mech. Min. Sci., 64, 90-95.

ISRM (1979), Suggested Method for Determining the Uniaxial Compressive Strength and Deformability of Rock
MaterialsRep., 137-140 pp, ISRM, International Society of Rock Mechanics.

Wang, H. J., A. V. Dyskin, A. Hsieh, and P. M. Dight (2012), The mechanism of the deformation memory effect
and the deformation rate analysis in layered rock in the low stress region, Computers and Geotech., 44, 83-92.

Yamamoto, K., and Y. Yabe (2001), Stresses at sites close to the Nojima Fault measured from core samples,
The island Arc, 10, 266-281.

Yamamoto, K., Y. Kuwahara, N. Kato, and T. Hirasawa (1990), Deformation Rate Analysis: A new method for in
situ stress estimation from inelastic deformation of rock samples under uni-axial compressions, Tohoku Geophys.
J., 33(2), 127-147.

220
APPENDIX

SUGGESTED METHOD FOR STRESS MEASUREMENT FROM A ROCK CORE USING THE DEFORMATION
RATE ANALYSIS (DRA) TECHNIQUE

SCOPE

The purpose of this test is to measure the strain difference between two completed load-unload cycles under
uniaxial compression. The maximum gradient change (inflection point) in graph of stress vs. strain difference
curve indicates the level of in situ stress, according to [Yamamoto et al., 1990]. The orientation of the stress
determined by the inflection point is parallel to the axis of loading to specimen.

TERMINOLOGY

(a) Pre-stress: the maximum previous stress that the specimen has been subjected, before applying DRA
technique (Figure 9 (A)). In the case of applying this technique on the core extracted from underground, the pre-
stress is assumed to be the in situ stress.

(b) Strain difference: the difference in strain between ith and jth cycles (Figure 9 (B) and (C)). This can be
expressed as strain difference Δεij(σ) = εj (σ) – εi(σ), j>i . It is usually calculated by averaging the axial strain
difference from two or four strain gauges. It can also be averaged from the lateral strain difference. The lateral
strain difference is less disturbed by the process of compaction than axial strain difference. But it requires much
higher accuracy in strain recording, due to the fact the value of lateral strain is much less than the value of axial
strain.

(c) Inflection point: the maximum gradient change in the graph of stress vs. strain difference curve (Figure 9 (C)).
The axial strain difference curve bends downwards at the stress level of in situ stress. The lateral strain
difference bends upwards.

(d) Bending effect: the nonuniform stress distribution in the specimen caused by non-parallelness of the sample
ends and non-coaxiality of the applied loading.

(A) Loading process (B) Stress-strain curve (C) DRA inflection point
prestress
strain difference

1st cycle
stress

2nd cycle
stress

time strain a stress


Figure 9. (A) The loading programme: the pre-stress, 1st and 2nd cycles; (B) Stress and strain curve. A black
arrow shows a strain difference between the 1st and 2nd cycles vs. stress. The stress, a, is the previous
maximum stress; and (C) the inflection point which indicates previous maximum stress, a, is marked with arrow.

APPARATUS

(a) Specimen: the requirement of specimen should follow [ISRM, 1979], other than the size of sample. The size
of sample depends on the available cores at the location of interest.

221
The end flatness (planarity) and parallelism: The planarity of sample end was suggested as <0.02mm at each
side of sample end in [ISRM, 1979]. It might be insufficient for testing small samples. A difference of 0.02mm in
length will create a nonuniform stress up to 0.02 ÷ 𝐿𝐿 × 𝐸𝐸 , where L is sample length and E is stiffness. It is
equivalent 40 MPa in a sample with 100GPa elastic modulus and 50mm in length. Hence, a small sample with
planarity less than 0.01mm and stress correction by the equation in [Hsieh et al., 2013] is necessary.

(b) Loading system: the requirement of the loading machine and specimen should follow [ISRM, 1979]. The peak
stress in the DRA test should be between 5% - 95% of the load cell capacity.

(c) Stress rate: Since there is insufficient information to indicate the relationship between quality of test and the
stress rate, the stress rate can be any constant rate and depends on the user. However, the stress rate has a
direct impact on the sampling frequency, which is one of the keys to reducing error caused by electric
interference. The suitable sampling frequency is explained in (d).

(d) Error in the strain reading: The strain difference could be a few micro-strains to hundreds of micro-strains. If
the amount of error is close to the amount of strain difference, the error introduced by the electric interference is
sufficient to mislead the determination of inflection point or to mask up the inflection point. Hence, it is essential
to reduce the error in the strain reading to less than 1 micro-strain and as a consequence, the noise/error in the
strain difference would be less than 2 micro-strain.

To achieve such accuracy, the noise from the environment (for example: temperature, vibration, power
fluctuation) should be monitored and eliminated. The sampling frequency should be high enough to record at
least 100 readings per MPa. This 100+ readings/MPa will allow one to reduce electric noise by taking the
average of certain amount of data (ie every 20 readings) without smoothing the strain difference curve too much.

For example, if the maximum sampling frequency allowed by the acquisition system is 1 reading/second, the
stress rate should be less than 0.01 MPa/second to allow at least 100 readings/MPa recorded by the machine.

METHODS OF MEASURING STRAIN

(a) Displacement bar (linear variable differential transformer, LVDT)

A displacement bar attached to the loading platform or platen measures the displacement from one end of the
bar to the other end it sits on. It usually comes as part of the loading frame configuration so it does not require
an additional installation when conducting each test. It is economical, fast and requires no further skill other than
operate the loading system. Hence, it is a popular choice for measuring strain.

However, it is not suitable for measuring strain difference in certain cases. For example, the displacement bar is
usually attached to a stationary spot on the loading frame and allows the other end to move with the upper
platform, or the displacement bar is attached to the loading bar (moving while loading) and the other end is
against bottom platform or machine frame (statinary spot). This type of the configuration records the deformation
from both loading bar and specimen. If the sample is stiff/short compared with loading bar, the deformation of
loading bar cannot be neglected.

Another issue associated with recording strain by displacement bar is generated by the end contact. Because
the stress distribution at the planarity/roughness of sample end can concentrate at the “rough points”, these
“rough points” are possible to deform and crush. To reduce the crush at an asperity or high point, the sample
end should be polished with abrasive sand finer than grit 1200. No buffer material should be inserted between
sample and platen. In the case of conducting the DRA test and acoustic emission together, one should not use
the displacement bar because the lack of buffer material between sample and platen would allow acoustic noise
generated from sample end to contaminate the AE result.

The displacement bar is also insensitive to the bending effect. If the displacement bar is the only option for strain
reading, an aluminium sample with four glued strain gauges (Figure 2) should be subjected to the load in order
to check the uniformity of the applied stress. The relationship (calibration factor) between the deformation of

222
aluminium sample and recorded displacement by the displacement bar should also be obtained prior to the DRA
test.

(b) Glued strain gauge (electrical resistance strain gauge)

A strain gauge is one of most accurate methods to measure strain because it measures the deformation on the
specimen directly. It can detect the level of bending and provide the information on deformation anisotropy when
four axial strain gauges and four lateral strain gauges are placed on the specimen [Hsieh et al., 2013].
Compared with a displacement bar which only gives the axial strain difference, strain gauges can measure both
axial and lateral strain differences.
The length of the strain gauge should be longer than ten grain diameters in magnitude and the gauge should not
encroach within half specimen diameter of the specimen end [ISRM, 1979]. The downside of using strain gauges
is the additional cost and the gauge installation requires some experience.

(c) Glued strain gauge on a spring or LVDT between upper and bottom platforms

In order to eliminate the unwanted deformation recorded by the displacement bar, a spring with glued strain
gauge or LVDT is placed between upper and bottom platforms to measure the deformation (see “strain gauge on
a spring” in Figure 2). This configuration provides better measurement compared with displacement bar attached
to the machine, because it is free from the deformation of machine itself. However, it is still affected by the
planarity of specimen end or the “buffer material” between specimen and platform. If the planarity of specimen
end undertakes plastic deformation (crush) or the plastic deformation/ irreversible strain was formed in the buffer
material at the 1st loading cycle of DRA test, this “end effect” will affect the strain difference between two cycles.
Alternatively, the “legs” of spring can be fixed on the sample instead of platen. In this way the unwanted
deformation created by sample end will have no influence on the strain reading. However, to locate the spring
firmly on the sample surface with no movement at contact spot is another challenge. A delicate
design/arrangement is necessary for quality measurement.

Depending on the configuration, the glued strain gauge on a spring or LVDT might be insensitive to the bending
effect even when four devices are installed at 90º apart. It is essential to check the non-coaxiality of the applied
loading and the calibration factor by aluminium sample according to the procedure mentioned in (a).

PROCEDURE

(a) To examine the reliability of the DRA technique on recovering laboratory applied pre-stress, three loading
cycles are subjected to the specimen at a constant rate. The first cycle is applied as pre-stress, the second and
third cycles should be at least 10MPa or 2 times higher than the pre-stress.

(b) To extract the value of in situ stress, two loading cycles (load-unload-load-unload) are subjected to the
specimen at a constant stress rate. Although the value of in situ stress might be unknown, one should try to
make the maximum applied stress at least 10 MPa or 2 times higher than the expected in situ stress.

(c) Most rock (lithology) is able to remember the laboratory applied stress right after the rock is drilled from core;
even the laboratory applied stress is lower than the in situ stress. The reason that rock records “the most recent
stress” instead of “previous maximum stress” might be due to the mechanism of long-term memory (geological
time scale) and is different from the mechanism of short-term memory (laboratory time scale). The specimen
prepared for in situ stress measurement should not be subjected any stress before conducting the DRA
technique, because the short-term memory is likely to mask up or affect the inflection point of the long-term
memory, and the interaction between these two mechanisms is unclear.

CALCULATION AND INTERPRETATION OF RESULTS

(a) The calculation of axial strain, lateral (diametric) strain, and compressive stress could be found in section 4
(a) to (d) in [ISRM, 1979]. The strain difference between two continue loading cycle is calculated by

223
Δεij(σ) =εj (σ) –εi(σ), j>i.
The strain difference vs. stress should be plotted in the way that the change in the gradient is easy to visualise.
(b) The maximum gradient change in the graph of stress vs. strain difference curve is called the inflection point.

REFERENCES

Hsieh, A., A. V. Dyskin, and P.M. Dight (2013), The influence of sample bending on deformation rate analysis
stress reconstruction, Int. J. Rock Mech. Min. Sci., 64, 90-95.

ISRM (1979), Suggested Method for Determining the Uniaxial Compressive Strength and Deformability of Rock
MaterialsRep., 137-140 pp, ISRM, International Society of Rock Mechanics.

224
View publication stats

You might also like