IOP Conference Series: Earth and Environmental Science

PAPER • OPEN ACCESS You may also like

- Measurement and prediction of granite
Acoustic emission and digital image correlation for damage evolution in deep mine seams
using acoustic emission
damage evolution in brittle rocks under time- Sunwen Du, Guorui Feng, Zhixong Li et al.

- Performance monitoring of large-scale

dependent tensile loading autonomously healed concrete beams
under four-point bending through multiple
non-destructive testing methods
To cite this article: Omid Moradian et al 2021 IOP Conf. Ser.: Earth Environ. Sci. 833 012090 G Karaiskos, E Tsangouri, D G Aggelis et
al.

- Deep Modeling of Quasar Variability

Yutaro Tachibana, , Matthew J. Graham
et al.
View the article online for updates and enhancements.

This content was downloaded from IP address 181.169.210.179 on 20/10/2021 at 22:09

Mechanics and Rock Engineering, from Theory to Practice IOP Publishing
IOP Conf. Series: Earth and Environmental Science 833 (2021) 012090 doi:10.1088/1755-1315/833/1/012090

Acoustic emission and digital image correlation for damage

evolution in brittle rocks under time-dependent tensile loading

Omid Moradian, Rui Wu, Ying Li, Kerry Leith & Simon Loew
Swiss Federal Institute of Technology (ETH), Zurich, Switzerland

[email protected]

Abstract. Time-dependent rock deformation is considered to precede dynamic failure in many

rock-engineering projects and natural geohazards. In order to understand long-term performance
of brittle rocks, we gather mechanical load, strain, acoustic emission (AE) and digital image
correlation (DIC) data to describe the evolution of damage and cracking in response to tensile
loading during multi-stage relaxation experiments on inverted single edge notch bending
(iSENB) specimens. The source locations of the AEs correspond to a process zone ahead and
around the notch indicating the evolution of the microcracks. The results showed that the cracks
first start aseismicaly under subcritical growth until 10mm from the tip of the notch and then
they grow seismically by showing seismic signals in the form of acoustic emissions. It was
observed that the process zone obtained by DIC is smaller than the cloud of the AE locations.
This can be partially because DIC only shows surface deformation and partially because of the
errors associated with the AE source locations such as ignoring anisotropic velocity, sampling
rate, sensor locations, etc. Moment tensor analyses of the AE signals showed that both tensile
and shear cracks are involved in the micro-scale although in the macroscopic scale the damage
process is mostly considered as tensile. The results showed that microcracks start as tensile, and
then continue as shear, especially at the end of the crack where the specimen experiences
compression loading.

1. Introduction
Rocks subjected to constant load, or constant displacement boundary conditions can be observed to
deform slowly and eventually fail at stress levels less than otherwise expected from short-term strength
tests (Amitrano and Helmstetter, 2006). Similarly, rock mass failures commonly don’t occur
immediately after a forcing event (e.g. excavation or intense precipitation), but instead may occur
months or years after the disturbance, indicating failure is partly controlled by a gradual decrease in rock
strength. Understanding progressive rock deformation, and in particular the transition from sub-critical
to critical fracture velocities, is therefore important to predict the onset of dynamic and catastrophic
failure associated with landslides, earthquakes, and volcanic eruptions (Brantut et al., 2013).
Understanding this process can also provide important insight into the integrity of underground mines
and excavations, the long-term storage of hazardous wastes, effective recovery of hydrocarbon and
geothermal energy resources with controllable induced seismicity, and last but not least safe and
economic CO2 sequestration [Brantut et al., 2013]. United Nations Educational, Scientific and Cultural
Organization (UNESCO) has recently emphasized our current lack of knowledge in this area, and
“Understanding Slow Deformation before Dynamic Failure” has been suggested as one of the two

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
Mechanics and Rock Engineering, from Theory to Practice IOP Publishing
IOP Conf. Series: Earth and Environmental Science 833 (2021) 012090 doi:10.1088/1755-1315/833/1/012090

priority research areas within the Natural Hazards theme of its International Year of Planet Earth
(Ventura et al., 2009, Brantut et al., 2013).
Over the last decades, researchers have studied the time-dependent or long-term performance of
rocks by creep testing and rheological modeling. As a result, numerous rheological models have been
developed to describe long-term behavior of sub-critically loaded rocks. However, the traditional
models which are mainly based on the evolution of strain with time, fail to capture the full range of
behavior from elastic deformation at the beginning of loading, viscous deformation in primary and
secondary creep and especially plastic deformation as cracking progresses toward dynamic failure
during tertiary creep. This is, in part, due to the lack of reliable experimental data for validating the
models.
Within the Griffith-Irwin framework of the linear elastic fracture mechanics (LEFM), ruptures will
propagate only when the strain energy release rate is larger than surface energy in a vacuum. However,
most near-surface fractures take place in a chemically interactive environment subjected to sub-critical
loading (e.g. surface water, moisture, temperature) and grow at rates several orders of magnitude than
otherwise expected for critical crack growth (usually >10-6 m/s vs. 103 m/s). This phenomenon can
include elements of subcritical, slow, aseismic, stable, quasi-static, or kinetic crack growth (Atkinson,
1982, 1984; Atkinson and Meredith, 1987). Subcritical crack growth (SCG) is suggested to be the key
process controlling the rate of strength degradation in near-surface brittle rocks under relatively stable
environmental conditions. Several competing micro-mechanisms could be responsible for SCG.
Nevertheless, the overwhelming body of experimental evidence suggests that the growth of pre-existing
cracks and flaws by stress corrosion is the dominant mechanism of SCG in the progressive failure of
near-surface brittle rocks (Brantut et al., 2013; Atkinson, 1982, 1984; Atkinson and Meredith, 1987).
Gaining new physical insights into the manner in which SCG progresses to dynamic failure is
important for understanding the evolution of instability in engineering projects and the assessment of
the precursory phase of natural hazards. Despite significant progress, there is still a lack of understanding
of SCG, notably the transition phase, due to the lack of quantitative analysis of progressive rock failures
in a broad range of near-surface environmental conditions. In this paper, AE monitoring and DIC are
coupled with traditional mechanical measurements in order to provide new insights into spatiotemporal
micro-crack evolution, fracture mode, and aseismic/seismic transition during SCG in brittle rocks.

2. Experimental Setup and Methodology

2.1. Sample preparation and properties

The inverted single edge notch bending (iSENB) samples were prepared with dimension of 400 × 91 ×
91 mm (length × height × depth). A notch with 10 mm depth and 3 mm width was cut orthogonally
across the center of each sample. The specimens were sourced from a fine-grain granite block from
Herrnholz quarry in Germany. The physical and mechanical properties of the tested rocks are presented
in table 1.

Table 1. Physical and mechanical properties of the tested rocks.

Properties Unit Value
Grain size range mm 0.3 – 1
Density g/cm3 2624.4 ± 3.5
P-wave Velocity m/s 3996 ± 98
UCS MPa 142.35 ± 11.20
Tensile Strength MPa 13.58
Young's Modulus GPa 30.59
Poisson's Ratio - 0.28

2
Mechanics and Rock Engineering, from Theory to Practice IOP Publishing
IOP Conf. Series: Earth and Environmental Science 833 (2021) 012090 doi:10.1088/1755-1315/833/1/012090

2.2. Loading setup

Our new rock physics and mechanics lab at ETH Zurich (www.rpml.ethz.ch) includes dual loading
frames designed to conduct thermo-hydro-seismo-mechanical observation of progressive strength
degradation of hard rock samples loaded under either uniaxial compression or induced tension during
inverted single edge notch bending configurations. Our new creep testing frames are capable of
maintaining a load of 100 kN (in iSENB configuration) for up to 100 days. Customized components
allow the software to maintain (or cycle) stress or strain, temperature (from -20° to 150° C, ± 0.1° C)
and humidity (from 20 % to 90%, ± 2.5%) using an environmental chamber surrounding the loaded
laboratory sample (Li, et al., 2019). Figure 1 shows an iSENB specimen instrumented with 16 AE
sensors, and an extensometer that measures the crack mouth opening displacement (CMOD) during
testing.

Figure 1. An inverted single edge notch-bending (iSENB) specimen during multi-stage relaxation
testing.

To ensure consistency of experimental results, several tests with similar configurations were performed
under room environmental conditions (23°C, 70% humidity). Initially, four short-term experiments (5 –
50 minutes) were conducted on four specimens (1a – 01, 02, 03, 06 – 31) to obtain the peak load (Fm).
The results were similar for the four experiments, with Fm consistently around 14.5kN (figure 2a),
suggesting a (theoretical) average fracture toughness of 1.82 MPa m1/2. Based on the measured peak
load Fm, two tests with four – hour staged loading levels (F1, F2, ... Fm) were undertaken on two
specimens (1a – 04, 07 – 31). The loading protocol for the multi-stage stress relaxation-bending test on
the latter two specimens is as follows:
a) Load each specimen to 0.12 kN (around 1% of estimated peak load) at a rate of 0,004 kN/s and then
maintain the load for around 16 hours (to allow the sample and sensors to settle in the configuration).
b) Load specimen to 6 kN (around 40% of estimated peak load) with a load point (piston) displacement
rate of 1μm/s.
c) Maintain the displacement so that the specimen experiences stress relaxation for 1800 s.
d) Increase the load by 10% of the estimated peak load by load point displacement control at a rate of
1μm/s.
e) Repeat steps c) and d) until the target load reaches 95% the peak.
f) Maintain the displacement so that the specimen undergoes stress relaxation until failure.
The load and CMOD of an iSENB sample 1a – 07 – 31 under this staged increase loading are shown in
figure 2b.

3
Mechanics and Rock Engineering, from Theory to Practice IOP Publishing
IOP Conf. Series: Earth and Environmental Science 833 (2021) 012090 doi:10.1088/1755-1315/833/1/012090

(a) (b)
Figure 2. a) Load-time data from six iSENB tests under short-term loading (samples 1a – 01, 02, 03,
06 – 31) and four-hour staged loading (1a – 04), and b) load and crack mouth opening displacement
for one iSENB test on sample 1a – 07 – 31 under multi-stage stress relaxation loading.

2.3. Digital image correlation (DIC) setup

Digital image correlation was undertaken using images taken on a Sony Alpha A7RII digital camera
with standard zoom lens mounted to the loading frame. The camera was set to time lapse mode, with
full resolution (42 MP) images taken every 20 seconds during the duration of the test. Image correlation
was undertaken using Ncorr for Matlab, with frames referenced to a source image near the end of the
1% loading period. The resolution of images at the sample was around 0.01 mm/px, and displacements
were analyzed using a subset radius of 30 px, providing a maximum resolution of observed
displacements similar to the minimum grain size.

2.4. Acoustic Emission (AE) setup

For AE monitoring of the damage processes, three types of sensors were used: KRNBB-PC point contact
broadband sensors from 100-2500 kHz, PAC resonant sensors with frequency range of 35-65 kHz, and
GmuG resonant sensors with frequency range of 300-1000 kHz. Though KRNBB-PC sensors are around
an order of magnitude less sensitive than GmuG sensors, still useful as they have a broadband frequency
range and therefore can provide information on radiated energy. A combination of broadband and
resonant sensors will work in an associated manner during our subsequent experiments, where
broadband sensors will be used for the analysis of source properties like energy budget and resonant
sensors with higher sensitivity will be used for picking small events and then doing AE location. Some
text.

3. Subcritical Crack Growth in single edge notch bending specimen

3.1. Time-series analysis

By using time-series analysis of the AE data Moradian et al. (2016) have shown that the number of AE
hits is related to the number of cracks, and the AE energy is related to magnitude of the cracking event.
Other researchers such as Lei and Satoh (2007) have employed triaxial compression tests and analysed
the statistical properties of the AE signals. They found that for the intermediate-term prediction of time-
to-failure, the decreasing b-value and increasing AE energy rate are suggested to be meaningful
indicators of transition between SCG and catastrophic failure. Figure 3 shows the time-series analysis
of the AE signals performed on the experiments. The results show that the resonant sensors, with higher
sensitivity, could detect more signals and showed that crack numbers increase logarithmically after the
4th staged loading (95% of the estimated peak load), while most signals from KRNBB-PC sensor were
recorded very close to the failure or even completely after the failure.

4
Mechanics and Rock Engineering, from Theory to Practice IOP Publishing
IOP Conf. Series: Earth and Environmental Science 833 (2021) 012090 doi:10.1088/1755-1315/833/1/012090

a)
Figure 3. a) Crack mouth opening displacement and cumulative AE events from PAC resonant
sensors (red) and KRNBB-PC wideband AE sensors (green) for the experiment performed on sample
1a – 03 – 31 and b) crack mouth displacement and cumulative number of AE events from GumG AE
sensors for the sample 1a – 07 – 31.

3.2. Spatiotemporal micro-crack evolution

Visual observations of the cracking processes provide the intuitive insights into time-dependent tensile
failure of the specimens. Slow or subcritical crack growth can be detected by using high-resolution
photography with low frame rate while for dynamic failure high-speed photography is needed. Digital
image correlation (DIC) is used to evaluate the photography data (Rastogi and Hack, 2012; Lin et al.,
2019a; Li and Einstein, 2017; Dutler et al., 2018; Koivisto et al., 2016; Moln´ar and Gravouil, 2017) for
both displacement and strain fields.
Figure 4 details calculated horizontal surface displacements for six intervals during the final creep phase
of an iSENB test (1a – 07 – 31). Times are described with respect to the final failure. The six intervals
are also marked in figure 3b. Figure 5 presents the calculated relative strain field change in the final
image of our second creep test (1a – 07 – 31) approximately 20 sec prior to dynamic failure.
The development of microcracks can be detected with AE event source locations. The Aikake
information criterion (AIC) (Maeda 1985) was used to pick arrival times of the events the same way as
discussed by Li et al (2019). Source locations were calculated using an isotropic velocity model of 4000
m/s and minimization of the residuals, in order to estimate the formation and growth of microcracks in
granite. The 2D AE locations (figure 6 and 7a) show that most microcracks distributed within an
apparent zone of nearly 4 cm width during the last relaxation stage, and extend approximately 40 mm
beyond the crack tip in the final seconds of the test (approx. 20 mm below the sample surface).
The first AE event (blue dot) happens 10mm under the notch. This indicates that that cracks start
aseismicaly under subcritical growth from the tip of the notch and then they grow seismically after
propagating for almost 10mm.

5
Mechanics and Rock Engineering, from Theory to Practice IOP Publishing
IOP Conf. Series: Earth and Environmental Science 833 (2021) 012090 doi:10.1088/1755-1315/833/1/012090

Figure 4. Surface displacements in horizontal direction (+, right, - left) at six moments before failure
for the experiment presented in figure 2b. The six intervals are also marked in figure 3b.

25 mm

Figure 5. Surface strain of sample 1a – 07 – 31 approximately 20 sec prior to dynamic failure.

Finally, source mechanisms of microcracks were determined using moment tensor inversion of the AE
waveforms. The simplified moment tensor analysis (SiGMA) (Grosse and Ohtsu 2008) is performed by
analyzing the amplitudes of the first motions. Events with a double couple (DC) component greater than
60% are considered shear, less than 40% as tensile and the rest as mixed. The evolution of the tensile
and shear cracks during the last relaxation stage of an iSENB test is presented in figure 7b. The results
show that tensile cracks happen first and then shear cracks are mobilized too but the end of the cracking
where the specimen experiences compression loading is mostly happening under shear.

6
Mechanics and Rock Engineering, from Theory to Practice IOP Publishing
IOP Conf. Series: Earth and Environmental Science 833 (2021) 012090 doi:10.1088/1755-1315/833/1/012090

Figure 6. 2D AE locations from resonant GmuG sensors during the last relaxation stage for the
experiment presented in figure 2b

a) b)
Figure 7. a) 2D AE locations from custom made resonant sensors and b) the evolution of the tensile
and shear cracks during the last relaxation stage.

4. Conclusions
In this study inverted single edge notch bending (iSENB) specimens were subjected to tensile loading,
and the AE signals were detected using resonant and wideband AE sensors. Digital image correlation
(DIC) and a mechanical crack mouth opening displacement sensor were used to observe the evolution
of the subcritical and critical crack growth during the creep process: The following conclusions can be
drawn:
1- The source locations of the AEs start showing off after 10 mm from the notch tip. This indicates
that cracks start aseismically under subcritical growth from the tip of the notch and grow
seismically only when the opening displacement rate increases at the end of the experiment.
2- The 2D AE locations show that most microcracks are distributed within an apparent zone of
nearly 4 cm width during the last relaxation stage, and extend from the crack tip 4 cm down into
the sample.

7
Mechanics and Rock Engineering, from Theory to Practice IOP Publishing
IOP Conf. Series: Earth and Environmental Science 833 (2021) 012090 doi:10.1088/1755-1315/833/1/012090

3- Surface strain of the last frame from DIC results reveals a small process zone comparing to the
cloud of the AE events. This is because DIC shows only surface cracks while many of the
microcracks can happen in the specimen volume.
4- Moment tensor analyses of the AE signals showed that both tensile and shear cracks are involved
in the micro-scale although the macroscopic processes is considered as tensile. The results
showed that microcracks start as tensile, then both tensile and shear cracking happen and finally
microcracks continue as shear, especially at the end of the crack where the specimen experience
compression loading.

Acknowledgment
This work was supported financially by the SNSF R’Equip project SNF_170746 “Long-term damage
evolution in brittle rocks subject to controlled climatic conditions” and China Scholarship Council
(CSC). The authors would also like to thank Elsys Company, especially Roman Bertsch for their
hardware and software support for this research.

References
[1] Armitrano D, and A Helmstetter 2006 Brittle creep, damage, time to failure in rocks, J. Geophys.
Res., 111, B11201, doi:10.1029/ 2005JB004252.
[2] Brantut N, M J Heap, Meredith P G and Baud P 2013. Time-dependent cracking and brittle creep
in crustal rocks: a review. J. Struct. Geol. 52: 17–43.
[3] Grosse C U, and Ohtsu M 2008, Acoustic emission testing: Springer Science & Business Media.
[4] Dutler N, Nejati M, Valley B, Amann F, and Molinari G 2018 On the link between frac- ture
toughness, tensile strength, and fracture process zone in anisotropic rocks: Engineering
Fracture Mechanics, 201, 56–79.
[5] Koivisto J, Ovaska M, Miksic A, Laurson L, and Alava M J 2016 Predicting sample lifetimes in
creep fracture of heterogeneous materials: Physical Review E, 94, no. 2, 23002.
[6] Lei X, and Satoh T 2007 Indicators of critical point behavior prior to rock failure inferred from
pre-failure damage: Tectonophysics, 431, no. 1-4, 97–111.
[7] Li Y, Leith K, Moradian O, Loew S, & Perras M A 2019 A new laboratory to undertake
climatically controlled static loading and constant strain tests: design and preliminary results.
In 53rd US Rock Mechanics/Geomechanics Symposium. American Rock Mechanics
Association.
[8] Li B Q, and Einstein H H 2017 Comparison of visual and acoustic emission observations in a
four point bending experiment on Barre Granite: Rock Mechanics and Rock Engineering, 50,
no. 9, 2277–2296.
[9] Li B Q, da Silva B G, and Einstein H 2019 Laboratory Hydraulic Fracturing of Granite: Acoustic
Emission Observations and Interpretation: Engineering Fracture Mechanics.
[10] Lin Q, Wan B, Wang S, Li S, and Fakhimi A 2019 Visual detection of a cohesionless crack in
rock under three-point bending: Engineering Fracture Mechanics, 211, 17–31.
[11] Maeda N 1985 A method for reading and checking phase times in autoprocessing system of
seismic wave data. J Seismol Soc Japan 38:365–379
[12] Moln´ar G, and Gravouil A 2017 2D and 3D Abaqus implementation of a robust staggered phase-
field solution for modeling brittle fracture: Finite Elements in Analysis and Design, 130, 27–
38.
[13] Detection of cracking levels in brittle rocks by parametric analysis of the acoustic emission
signals. Rock Mechanics and Rock Engineering Journal, Volume 49, Issue 3, pp 785-800
[14] Rastogi P K, and Hack E 2012 Optical methods for solid mechanics: a full-field approach: John
Wiley & Sons.
[15] Ventura G, Vinciguerra S, Moretti S, Meredith P H, Heap M J, Baud P, ... & Kummerow J 2009
Understanding slow deformation before dynamic failure. In Geophysical hazards (pp. 229-
247). Springer, Dordrecht.