International Journal of Coal Science & Technology (2022) 9:53

https://doi.org/10.1007/s40789-022-00529-6

RESEARCH

Energy evolution and water immersion‑induced weakening

in sandstone roof of coal mines
Wenjie Liu1,3,4 · Ke Yang1,2,3,4 · Shuai Zhang5 · Zhainan Zhang1,3,4 · Rijie Xu1,3,4

Received: 3 October 2021 / Accepted: 29 June 2022

© The Author(s) 2022

Abstract
The instability of underground spaces in abandoned coal mines with water-immersed rocks is one of the main hazards
hindering the geothermal energy use and ecological restoration of post-mining areas. This study conducted graded cyclic
loading–unloading tests of five groups of sandstone samples with different water contents. The evolution of input, elastic,
dissipated, damping, and plastic energies were explored, considering the damping effect. The normalized plastic energy
serves to characterize the damage evolution of sandstone samples, whose failure characteristics were analyzed from both
the macroscopic and microscopic perspectives. X-ray diffraction technique and scanning electron microscopy were used
to reveal the softening mechanism of sandstone. The results show that under graded cyclic loading, input energy, elastic
energy, and dissipated energy all increase gradually, and the fraction of elastic energy increases gradually at first and then
tends to stabilize. The variation in the fraction of dissipated energy is opposite to that of elastic energy. In each cycle, the
input energy is stored primarily in the form of elastic energy, whereas the dissipated energy is used primarily to overcome
the damping of sandstone. When the normalized number of cycles approached unity, the plastic energy fraction sharply
increases, while that of the dampening energy drops abruptly. With increasing water content, the effect of pore water on the
lubrication, the water wedge, and dissolution of mineral particles becomes more obvious, reducing the elastic-storage limit
of sandstone, meanwhile the sandstone damage factor increases significantly under the same cycle and the failure mode
changes from brittle to ductile.

Keywords Mining engineering · Energy evolution · Water–rock interaction · Weakening mechanism

1 Introduction

After the closure of coal mines due to low economic benefit

* Ke Yang
[email protected] or resource exhaustion, the surface water and groundwa-
ter contained in the aquifer rapidly fill underground spaces
1
State Key Laboratory of Mining Response and Disaster through the fissured zone, causing a drop in groundwa-
Prevention and Control in Deep Coal Mines, Anhui ter level, water pollution, surface subsidence, changes in
University of Science and Technology, Huainan 232001,
China the geological environment, and secondary disasters (Li
2 et al. 2019; Sun et al. 2021). Within the framework of
Hefei Comprehensive National Science Center, Institute
of Energy, Hefei 230031, Anhui, China clean-energy production and ecological restoration of the
3 post-mining areas, considerable efforts have been made
National and Local Joint Engineering Research Center
of Precision Coal Mining, Anhui University of Science worldwide to use geothermal energy, water, and spaces in
and Technology, Huainan 232001, Anhui, China abandoned coal mines. For example, some scholars have
4
Key Laboratory of Mining Coal Safety and Efficiently suggested constructing underground reservoirs from aban-
Constructed By Anhui Province and Ministry of Education, doned coal mines (Bian et al. 2021; Gu et al. 2015), develop-
Anhui University of Science and Technology, ing geothermal resources (Pu et al. 2021; Jardon et al. 2013),
Huainan 232001, China or building pumped storage power stations (Fan et al. 2020;
5
State Key Laboratory of Coal Resources and Safe Xie et al. 2020).
Mining, China University of Mining and Technology,
Xuzhou 221116, Jiangsu, China

13
Vol.:(0123456789)
 53 Page 2 of 14 W. Liu et al.

Complex water–rock interactions, physical effects, and the hysteresis effect of rocks and the energy dissipated by
chemical reactions commonly occur in abandoned coal damping. The present study thus considers uniaxial load-
mines due to tidal action and water erosion, significantly ing and cyclic loading–unloading tests of sandstone samples
deteriorating rock stability and residual strength. This issue with different water contents and determines the mechanical
was addressed by numerous scholars, who investigated the response of sandstone samples with different water contents
mechanical characteristics and failure mechanism of water- under graded cyclic loading–unloading. Considering the vis-
bearing rocks (Wong et al. 2016; Hashiba et al. 2015; Guo coelastic properties of sandstone, we determine how damp-
et al. 2021) and conducted post-peak cyclic loading–unload- ing and dissipated energy evolve. How sandstone damage
ing tests for coarse sandstone in natural and water-saturated evolves during the deformation and failure is also quantita-
conditions (Niu et al. 2018). The latter found that the water- tively described based on the energy principle. Macroscopic
saturated coarse sandstone exhibited a more significant crack propagation within sandstone samples was monitored
increase in lateral and volumetric strains compared with the using digital speckle technology. Scanning electron micros-
natural condition. In uniaxial loading–unloading tests, water copy (SEM) was used to image the microstructure of fractured
saturation had an apparent softening effect on post-peak sandstone samples. Finally, the mechanism by which water
fractured coarse sandstone strength and elastic modulus. immersion weakens the mechanical properties of the sandstone
Zhou et al. (2016) conducted static and dynamic mechani- is revealed. These research findings offer theoretical support
cal tests of sandstone samples with different water contents for evaluating the safety and stability of rocks in coal mine
under the dry–wet cycles. They found that water-bearing underground reservoirs.
sandstone could gradually restore its mechanical character-
istics in dry conditions. The tensile strength of drying ver- 2 Energy conversion theory considering
sus water-immersed sandstone with the same water content viscoelastic deformation
differs. Hua et al. (2019) compared the mechanical fracture
characteristics of sandstone under dry–wet cycling and long- Under cyclic loading, the entire process of sandstone defor-
term water immersion and reported that dry–wet cycling mation and failure is accompanied by energy accumulation,
causes more significant deterioration due to sandstone frac- release, and dissipation. To use the first law of thermodynam-
ture than long-term water immersion. Rock damage and ics, we assume that the sandstone does not exchange thermally
failure are closely related to energy accumulation, dissipa- with the environment during deformation and failure, and we
tion, and release. Therefore, some scholars have analyzed neglect any radiant energy and acoustic-emission energy. The
how water affects rocks from an energy perspective (Li et al. work done by the press machine on the sandstone samples is
2015; Wang et al. 2017). Chen et al. (2019) pointed out that the total input energy. As shown in Fig. 1 and according to
the amount of energy released upon sandstone compaction Xie et al. (2005), the energy accumulated in each cycle can be
and elastic deformation stages increases significantly. The calculated by Eqs. (1)–(4):
water content increases and the amount of energy released
after failure decreases. A higher water content leads to a
Ui = Uei + Udisi , (1)
reduction of sandstone brittleness and enhances its plastic
deformation. Geng et al. (2020) conducted uniaxial com-
pression tests on dry, water-bearing, and saturated sandstone
samples. Their results show that the water-softened sand-
stone is more prone to ductile damage and releases less elas-
tic energy. Ma et al. (2018) carried out uniaxial compression
testing of gypsum rocks under different durations of water
immersion, and the results show that the input, elastic, and
dissipated energies of gypsum rock decrease with increasing
water-immersion time. Using a constitutive damage model
based on energy dissipation, they further described the dam-
age characteristics of gypsum rock weakened by water and
under a uniaxial load.
As discussed above, the mechanical characteristics, energy
evolution, and failure modes of water-bearing sandstone dur-
ing deformation and failure processes have been extensively
studied. However, studies are scarce on the evolution of resid-
ual plastic strain in sandstone with different water content
and graded cyclic loading. In addition, few works consider Fig. 1  Stress path of cyclic loading–unloading

13
Energy evolution and water immersion‑induced weakening in sandstone roof of coal mines Page 3 of 14 53

𝜀A For a rock treated as an elastoplastic material, the shape

∫𝜀o
Ui = 𝜎i+ d𝜀i , (2) of the hysteresis loop under elastoplastic deformation is
jointly determined by rock viscosity and plasticity. Rock as
𝜀A an elastoplastic material undergoes no plastic deformation,

∫ 𝜀c
Uei = 𝜎i− d𝜀i , (3) and its unloading paths differ. Nevertheless, a closed hys-
teresis loop can still be formed for the rock. Stevens et al.
(1980) and Kuwahara et al. (1990) found that, under a load-
𝜀A 𝜀A ing–unloading action, numerous microcracks appear in rocks
∫𝜀o ∫𝜀c
Udisi = Ui − Uei = 𝜎i+ d𝜀i − 𝜎i− d𝜀i , (4) only when the stress during the second loading exceeds the
peak stress upon first loading. It was believed that, during
where Ui, Uei, and Udisi are the input, elastic, and dissipated uniaxial graded loading and unloading, no new cracks are
energies of cycle i, respectively, σi+ and σi+1+ are the stresses generated in the rocks if the loading stress is less than the
imposed on loading i and i + 1, respectively, and σi− is the stress during the previous loading. In this case, there would
stress imposed upon unloading i. be no growth in plastic energy. Based on this argumentation,
Rocks are aggregates of mineral particles with complex we assume that sandstone is an elastoplastic material along
mineral composition and structure, which gives them their the stress path from point B to point C upon unloading and
significant heterogeneity and anisotropy. In most studies back to point B upon loading. At this stage, the sandstone
on energy evolution in rocks under graded cyclic loading undergoes elastoplastic deformation, and the elastic energy
and unloading, rocks are treated as elastoplastic materials. is not dissipated. The energy lost during this cycle is the
The area enclosed by the cyclic loading–unloading curve work done by damping that arose from rock viscosity.
and the coordinate axis (Udisi) is treated as the energy that Similarly, we assume that the sandstone does exchange
causes rock damage and failure. However, such an assump- thermally with the environment (thermal radiation and
tion neglects the viscous friction between mineral particles acoustic-emission energy are neglected). Thus, the area of
and the liquid. Therefore, the plastic energy causing rock the hysteresis loop BCB is the energy dissipated by over-
damage is not differentiated from the damping energy spent coming the viscous friction between the mineral particles
on overcoming the viscous friction. In the present study, we and liquid, i.e., the damping energy Udami. The dissipated
consider the viscoelastic deformation of sand-bearing sand- energy Udi minus this portion of the damping energy is the
stone. For each cyclic loading, the dissipated energy Udisi is plastic energy Upi that causes rock damage. The relationship
split into plastic energy Upi and damping energy Udami. We between these energies and their magnitude can be calcu-
assume that damping does not cause sandstone damage and lated using Eqs. (5)–(7):
failure, and the plastic energy promotes the plastic deforma-
tion of rocks, which weakens the sandstone.
Udisi = Upi + Udami , (5)
The stress path of a single loading–unloading cycle is
𝜀B
analyzed separately, as shown in Fig. 1, where point O is
∫𝜀o
+
Udami = (𝜎i+1 − 𝜎i+ )d𝜀i , (6)
the starting point of cyclic-loading curve i, point A is the
end point of cyclic-loading curve i and the starting point of
cyclic-unloading curve i, point B is the intersection between 𝜀A 𝜀A 𝜀B

∫𝜀o ∫ 𝜀C ∫𝜀o
cyclic-unloading curve i and cyclic-loading curve i + 1, point Upi = 𝜎i+ d𝜀i − 𝜎i− d𝜀i − +
(𝜎i+1 − 𝜎i+ )d𝜀i . (7)
C is the end point of cyclic-unloading curve i and the start
point of cyclic-loading curve i + 1, εo, εA, εB, and εC are the
strain at the corresponding coordinates, respectively, and
σA and σB are the stress at the corresponding coordinates,
respectively. The cyclic elastic strain εei for cycle i is defined 3 Experimental setup and test scheme
as the difference between the peak strain εA of cycle i and the
residual plastic strain εC of cycle i, the residual plastic strain 3.1 Sample preparation
εpi is the difference between the strain εo at the starting point
of the loading curve of cycle i (point O) and the strain εC at The sandstone samples were collected from the No. 4 coal
the endpoint of the unloading curve of cycle i. seam roof of the Panbei Coal Mine in Huainan and Huaibei
As shown in Fig. 1, the physical meaning of point B is Mining Area, China. Unweathered sandstone samples with
approximately the stress state during loading–unloading good integrity and uniform texture were sealed in wax at the
cycle i. Unloading starts from this stress state, followed by collection site and transported to the laboratory. According
another loading. Then, the sandstone returns to the initial to the requirements of the International Society for Rock
stress state, thus forming the closed hysteresis loop BCB. Mechanics and Rock Engineering, the sandstone was cut,

13
 53 Page 4 of 14 W. Liu et al.

drilled, and polished into standard cylindrical samples with

a diameter of 50 mm and length of 100 mm. Thirty standard
samples were subdivided into five groups and prepared for
testing. Before the mechanical tests, the sandstone samples
were dried at 105 ℃ in an oven for 24 h until attaining a con-
stant weight. Next, samples of each group were removed and
placed into distilled water for natural water-immersion tests
(Yu et al. 2019). Each sandstone sample was weighed before
and after each experimental step. According to Eq. (8), the
native water content of the sandstone samples was calcu-
lated to be approximately 0.3%, and the water content of
the water-saturated sandstone sample was about 2.07%. The Fig. 3  Sandstone samples with different water contents
water content is given by
W = {(Mw − M0 )∕M0 } × 100% (8) were selected from each group for analysis. Given that envi-
where W is water content in percent, and Mw and M0 are the ronmental factors affect the experimental results, the tests
mass of wet and dry sandstone samples, respectively. were done immediately after the preparation of each group
Sandstone quality was measured every 2 h during natural of samples.
water-immersion tests, and Fig. 1 shows the water-absorp-
tion curve of the sandstone samples over the whole pro- 3.2 Test system
cess (500 h). When water-absorption exceeded 200 h, the
water-absorption quality of the sandstone sample remained As shown in Fig. 4, the test system mainly consisted of a
essentially constant, so we considered that the sandstone had rock mechanical testing module, a digital video acquisition
reached its saturation state. At this point, the mass of the module, an SEM testing module, and an x-ray diffraction
water-absorbed sample was 10.83 g (Fig. 2). (XRD) testing module. The rock mechanical testing module
As shown in Fig. 3, sandstone samples with five levels consisted of an RMT-150B multifunction automatic rigid
of water content were prepared: 0%, 0.50%, 1.01%, 1.52%, rock servo material testing machine, which was used for
and 2.07%. Uniaxial compression and uniaxial graded cyclic high-precision uniaxial and triaxial compression tests. The
loading–unloading tests were done on the sandstone sam- maximal axial load imposed by this machine was 1000 kN,
ples. Six samples were used at each level of water content. with a loading rate of 0.01–100 kN/s. The measuring range
To avoid data contingency, three of them were used for uni- of the axial displacement sensor was 0–5 mm, with a resolu-
axial tests, and the other three were used for graded cyclic tion of 0.0015 m, which satisfied the test requirements. The
loading and unloading tests. After the experiment, valid data SEM testing module consisted of a FLEXSEM 1000, a com-
pact SEM, for high-resolution imaging in low-vacuum mode.
The magnification factor ranged from 60,000 to 300,000,
with an accelerating voltage of 0.3–20 kV and a resolution
of 4 nm. The XRD testing module consisted of a SmartLab
X-ray diffractometer for analyzing the mineral composition
of the sandstone samples. The digital video acquisition mod-
ule consisted of a Nikon digital camera, which served to
capture the fracture morphology of the sandstone samples
during the loading process.

3.3 Experimental scheme and results

Each group of the sandstone samples was further split

into subgroups U and C, which were used for the uniaxial
compression and uniaxial graded loading–unloading tests,
respectively. To improve the experimental accuracy and
reduce the randomness of the experimental results, no less
than three parallel tests were conducted for each loading
scheme and for each water content. The uniaxial compres-
Fig. 2  Spontaneous imbibition curve of sandstone samples sion tests were first conducted to obtain the strain–stress

13
Energy evolution and water immersion‑induced weakening in sandstone roof of coal mines Page 5 of 14 53

Fig. 4  Test system

Fig. 6  Stress path of the cyclic loading–unloading test

Fig. 5  Stress–strain curve under uniaxial loading

curves for different water contents, as shown in Fig. 5. Next, For example, C-25%-1 refers to the uniaxial graded cyclic
the axial pressure at each unloading point was determined loading–unloading test for the first group of samples with a
to design the subsequent cyclic loading–unloading test. The water content of 0.50%, and U-75%-3 refers to the uniaxial
stress-controlled mode was implemented in the uniaxial compression testing of the third group of samples with a
cyclic loading–unloading tests, as shown in Fig. 6. water content of 1.52%.
The loading and unloading rates were both 0.5 MPa/s. Figure 7 shows the XRD results for the mineral composi-
The stress gradient during the loading stage was 5 MPa. The tion of the three groups of sandstone samples with different
target stress of unloading for the unloading stage was set water contents. The sandstone samples were composed of
to 1.25 MPa. The stress path of the uniaxial cyclic load- mineral crystals, including quartz, kaolinite, albite, and cal-
ing–unloading test was 0, 5, 1.25, 10, 1.25, 15, 1.25, 20, cite. The quartz content was the highest of all the minerals
1.25, 25, 1.25, 30 MPa, sample failure, test termination. in the three groups of samples, accounting for about 50%
To facilitate the analysis of the test results, we normalized by mass. The sandstone samples with different water con-
the water content. The saturated water content was defined tents varied in mineral composition and content. The sand-
as 100%. Next, the samples in each group were numbered. stone samples with zero water content contained mainly two

13
 53 Page 6 of 14 W. Liu et al.

minerals, whereas those with a water content of 2.07% con-

tained mainly three minerals. Variations in mineral content
did not correlate significantly with water content.

4 Analysis of mechanical characteristics

of sandstone samples under cyclic loading

The softening coefficient was used to more intuitively

characterize the softening properties of sandstone samples
with different water contents. The softening coefficient λ is
defined as the ratio of the compressive strength Rw of the
water-bearing sandstone samples with different water con-
tents to the compressive strength Rd of the dry sandstone:

0 < 𝜆 ≤ 1.
Rw
𝜆= , (9)
Rd

The smaller the softening coefficient, the stronger the sof-

tening and the greater the impact of water on the rocks. As
shown in Table 1, water significantly affects the mechani-
cal properties of sandstone. Compared with dry samples,
the average uniaxial compressive strength of water-bearing
sandstone significantly dropped with water content from the
initial 110.51 to 18.23 MPa. The average elastic modulus
of the sandstone also decreased continuously from 19.61 to
3.53 GPa as the water content increased. Conversely, Pois-
son’s ratio increased.
Figure 8 shows the stress–strain curves of different
groups of sandstone samples subjected to graded cyclic
loading–unloading tests. All sandstone samples underwent
recoverable plastic deformation during each loading–unload-
ing cycle and unrecoverable plastic deformation. Before
failure, the stress–strain curve deviates from the historical
stress–strain curve. In addition, as the irreversible plastic
deformation accumulates, the cyclic loading–unloading
curve gradually shifts toward larger strain. Under graded
cyclic loading–unloading, the sandstone strength and the
number of loading–unloading cycles before failure gradu-
ally decrease as the water content increases.
For the sake of convenience, we normalized the number
of cycles before failure. Figure 9 shows the curves of elastic
strain εei or residual plastic strain εpi versus the normalized
number of cycles. These results show that the sandstone
undergoes plastic strain and residual plastic strain in each
cycle. Since the rock is an inhomogeneous anisotropic mate-
rial, the primary pores and cracks within the sandstone are
gradually compacted under initial cyclic loading. At this
stage, the elastic strain curve versus the normalized number
of cycles increases nonlinearly, and the residual plastic strain
is larger than the other stage. The sandstone enters the plas-
tic deformation stage in the subsequent cycles as the number
Fig. 7  XRD patterns of sandstone samples of cycles and stress increase. The elastic strain increases

13
Energy evolution and water immersion‑induced weakening in sandstone roof of coal mines Page 7 of 14 53

Table 1  Mechanical parameters Sample Water con- Uniaxial compressive Elastic modu- Poisson's ratio Average
under different water contents tent (%) strength (MPa) lus (GPa) softening
coefficient

U-0%-1 0.00 114.09 19.70 0.11 1

U-0%-2 0.00 110.35 20.36 0.11
U-0%-3 0.00 107.09 18.79 0.12
U-25%-1 0.50 83.12 13.91 0.16 0.72
U-25%-2 0.53 82.89 15.41 0.14
U-25%-3 0.48 71.93 12.43 0.17
U-50%-1 1.01 61.52 10.58 0.21 0.52
U-50%-2 1.04 54.60 10.84 0.19
U-50%-3 1.02 57.70 9.27 0.22
U-75%-1 1.52 44.47 6.21 0.26 0.38
U-75%-2 1.49 40.56 6.65 0.24
U-75%-3 1.53 42.12 5.84 0.24
U-100%-1 2.07 16.39 3.47 0.31 0.16
U-100%-2 2.03 20.14 3.67 0.32
U-100%-3 2.08 18.17 3.45 0.29

Fig. 8  Cyclic stress–strain curves

approximately linearly versus the normalized number of interconnected and penetrate the samples. The residual plas-
cycles. At this stage, the residual plastic strain decreases tic strain increases slightly versus the normalized number of
rapidly and stabilizes. When the residual strain accumulates cycles. For the same number of normalized cycles, the elas-
to a certain level, microcracks within the sandstone become tic strain of sandstone decreases gradually with increasing

13
 53 Page 8 of 14 W. Liu et al.

as a function of water content. The results in Fig. 10 lead to

the following observations:

(1) The input, elastic, and dissipated energies of sandstone

samples with different water contents increase with the
normalized number of cycles because a greater number
of cycles implies more work done on the sandstone
sample by the external force. As a result, the input
energy increases. Part of the input energy is stored in
the sandstone in the form of elastic energy, and the
remaining part is dissipated via rock damage.
(2) Comparing the shares of elastic energy and dissi-
pated energy as a function of water content shows
that, under the initial cyclic loading, the input energy
mainly takes the form of dissipated energy, which sig-
nificantly exceeds the share of elastic energy. Under
initial loading, the sandstone is damaged and fails due
to the compression of primary defects, leading to more
significant energy dissipation. Besides, the larger the
water content, the higher the dissipated energy share
and the more severe the sandstone damage under initial
loading.
(3) As the normalized number of cycles increases, the
elastic energy gradually increases, and the dissipated
energy decreases until finally stabilizing. These results
indicate that, at this stage, the input energy is stored
in the sandstone mainly in the form of elastic energy.
The fraction of elastic energy ranges from 0.7 to 0.9.
In addition, a greater water content leads to less elastic
energy and more severe sample damage under the same
stress.
(4) When the normalized number of cycles approaches
unity, the input, elastic, and dissipated energies con-
tinue to increase. The elastic energy decreases, and the
dissipated energy increases, which indicates that the
microcracks within the sandstone gradually propagate
Fig. 9  Elastic strain and plastic strain curves and penetrate the sandstone sample, resulting in its fail-
ure. Given the above, we predicted the sandstone failure
based on the energy evolution.
water content. With a saturated water content, the residual
plastic strain increases significantly, which shows that with Using Eqs. (5)–(7), we calculate the evolution of the
the increase of water content, the degree of plastic damage of damping energy, plastic energy, and their shares in sandstone
sandstone under the same external load gradually increases. samples as a function of water content. The results appear
in Fig. 11 and show that, with increasing cycle number, the
damping energy and the plastic energy in the sandstone
increase gradually for all water contents. The plastic energy
5 Energy evolution analysis first decreases rapidly, then stabilizes, and later increases
abruptly. The damping energy follows the opposite trend.
5.1 Energy evolution law Such variations are considered relevant to rock compression
and peak damage. The dissipated energy in the sandstone
According to theoretical background and Eqs. (1)–(4) in samples decreases with increasing water content. Most of
Sect. 2, we derived the evolution of the elastic and dissi- the dissipated energy in each cycle is spent on overcoming
pated energies and plot their fraction in sandstone samples the work done by viscosity. As the water content increases,

13
Energy evolution and water immersion‑induced weakening in sandstone roof of coal mines Page 9 of 14 53

Fig. 10  Energy distribution diagram

Fig. 11  Dissipated energy distribution diagram

13
 53 Page 10 of 14 W. Liu et al.

the plastic energy gradually overcomes the damping energy,

which indicates that water immersion significantly reduces
the work done by frictional damping between the mineral
particles. As a result, the bonding strength between the min-
eral particles decreases, promoting sandstone damage and
failure. The variation of the elastic energy share is like that
of the dissipated energy. When the normalized number of
cycles approaches unity, the plastic energy share increases
abruptly. In contrast, the damping energy share drops
sharply, portending the impending failure of the sandstone
sample.

5.2 Damage mechanism based on plastic energy

Based on the results detailed above, the input energy under

cyclic loading–unloading conditions was stored or released
mainly in the form of elastic energy or dissipated in the form Fig. 12  Damage evolution curve
of damping energy. Only a small share of the input energy
was spent on sandstone damage and failure. In addition, the
higher the water content, the larger the plastic energy share The damage generated in the previous two to three cycles
under the same stress gradient. To characterize the degree accounts for nearly 24% of the total damage, resulting in
of rock damage and failure, Liu et al. (2018) normalized the instability and failure. These results indicate that, as the test
dissipated energy. They analyzed the cumulative damage of proceeds, the pores and cracks propagate and become inter-
rocks under cyclic loading. Plastic energy was a primary connected. After the cracks propagate to a certain degree,
factor causing rock damage, whereas damping energy did crack development and damage accumulation accelerate. As
not lead to residual deformation or damage. Therefore, we the water content increases, the curve of the damage factor
considered it more accurate and reasonable to use normal- as a function of cycle number shifts towards the y axis, and
ized plastic energy to characterize rock damage. The damage the region corresponding to the decelerating growth stage
factor D was defined as the ratio of cumulative damage to gradually disappears. This implies that under the graded
total damage: cyclic loading conditions, a higher water content corre-
sponds to faster growth of the damage factor. In the water-
saturated sandstone sample (C-100), the damage generated
∑
Upi
D= , (10) in the first cycle accounts for about 34% of the total damage.
Up

where Upi is the plastic energy generated during cycle i, and

Up is the total plastic energy. 6 Fracture morphology and failure modes
The damage variable estimated by this formula varies
from zero to unity, satisfying the principle of damage irre- Rock damage accumulation is the process whereby microc-
versibility. Taking a dry sandstone sample (C-0) as an exam- racks initiate, propagate, and become interconnected within
ple, the damage factor evolution is plotted in Fig. 12 and the rocks. Fracture morphology characteristics vary at dif-
may be subdivided into three stages: decelerating growth, ferent stages of crack development. The fracture surface is
stable growth, and accelerating growth. where the microcracks and pores appear in large numbers.
At the stage of decelerating growth, the damage factor Microcracks and pores propagate along the weakest plane
grows at a saturating rate under the initial cyclic loading until the penetrating fracture surface is formed, leading
due to rock anisotropy and heterogeneity. The damage factor to varying failure modes in the sandstone. We studied the
reaches 35% at the stage of decelerating growth. At the sta- microscopic fracture characteristics of cracks along with
ble growth stage, the damage generated per cycle gradually the macroscopic failure morphology of sandstone. There-
approaches a constant as the graded cyclic loading increases. fore, the impact of water on the sandstone failure process
This stage accounts for the majority of the loading–unload- was investigated on both the microscopic and macroscopic
ing cycles. These results indicate that damage accumulates scales. This approach was conducive to disclosing the mac-
stably, whereas microcracks initiate and propagate without roscopic mechanism that weakens the mechanical strength
coalescing and penetrating the samples’ bulk. When the and the energy-dissipation mechanism in water-bearing
damage reaches about 80%, its accumulation accelerates. sandstone.

13
Energy evolution and water immersion‑induced weakening in sandstone roof of coal mines Page 11 of 14 53

6.1 Fracture morphology samples gradually evolve from dense structures to lose struc-
tures. The water–rock interaction and the loading–unloading
A Flex SEM1000 was used to image typical fractures in cycles promote damage accumulation and microstructural
dry, semi-saturated, and saturated sandstone samples under changes in sandstone samples.
graded cyclic loading–unloading. Figure 13 shows images
of the microstructural variations in the sandstone samples 6.2 Failure mode
at different magnifications. The images show that the frac-
tures in the dry sandstone samples are serrated, the mineral With the development of monitoring technologies, digital
particles form dense structures, and the cementing materi- imaging methods have become instrumental for studying
als between the particles are compacted. Some micropores macroscopic-crack propagation in rocks. The preparation
and cracks appear only along the mineral particle bounda- of artificial speckles on the surface of the water-bearing
ries and in the cementing materials. In the semi-saturated sandstone samples increases the test duration. To mitigate
sandstone samples, the microstructure patterns vary sig- this problem, we used the PhotoInfor and PostViewer image
nificantly. Corrosion pits appeared on the particle surfaces. analysis software packages to determine sandstone defor-
Numerous secondary minerals and residues adhere to the mation and failure during the tests (Li et al. 2006), which
mineral particle surfaces. In the saturated sandstone sam- show the evolution of strain at the surface of the sandstone
ples, significant dissolution and corrosion of the calcareous samples as a function of water content.
cement are apparent in loose structures. The cracks develop Figure 14 shows the cloud maps of the strain field at dif-
and propagate, and the mineral particles are more rounded. ferent time points for the corresponding number of cycles
As analyzed above, the degree of cementation decreases in in dry, semi-saturated (water content 1.01%), and water-
sandstone due to water immersion. The frictional damping saturated sandstone samples under graded cyclic load-
between the particles weakens. Consequently, the sandstone ing–unloading. According to Fig. 14a, after twenty cycles,

Fig.13  Micro-structural characteristics of sandstone samples

13
 53 Page 12 of 14 W. Liu et al.

Fig. 14  Cloud maps of sandstone strain field with different water contents under graded cyclic loading–unloading conditions

cracks successively initiate from the upper and lower edges some ductile features. Meanwhile, more and more second-
of the dry sandstone sample. Later, the cracks propagate in ary fissures are generated near the main fracture, resulting
the principal stress direction and develop into three tensile in increasingly complicated patterns in the strain field cloud
cracks growing parallel to the loading direction. The sand- maps. The most probable explanation is that the sandstone
stone sample then undergoes a splitting failure. According bulk becomes damaged due to lubrication, wedging, and cor-
to Fig. 14b, cracks successively initiate from the upper-right rosion by water. As a result, a shear zone appears. In addi-
and the lower-middle parts of the semi-saturated sandstone tion, as the water content increases, the shear zone within
sample. The cracks first propagated along the loading direc- the sample became increasingly prominent. The presence
tion and then merge into a single crack. The sandstone sam- of the shear zone contributes to the shear slip failure of the
ples fail in tensile-shear mixed mode. According to Fig. 14c, sandstone samples under stress.
cracks initiate from the middle of the saturated sandstone
sample and gradually propagate. Inclined cracks are gener-
ated in the upper and lower parts of the sample. The sand- 7 Discussion
stone sample has a single inclined-plane shear failure mode.
As the water content increases, the failure mode changes According to the above analyses, the mechanical properties
from tensile-brittle cleavage fracture in the dry state to of sandstone samples permanently deteriorates as the water
tensile-shear mixed-mode failure. In the water-saturated content increases, whereas the damage accumulates gradu-
state, sandstone samples undergo single shear failure with ally. Both the compressive strength and the energy-storage

13
Energy evolution and water immersion‑induced weakening in sandstone roof of coal mines Page 13 of 14 53

characteristics deteriorate significantly. This phenomenon is physical action. The area of rock-water interaction increases,
primarily attributed to physical interactions between water and the area and degree of bonding between mineral lparti-
and sandstone, which are briefly summarized below. cles decreases. The mineral-particle skeleton is thus gradu-
“Coal measures” is a lithostratigraphic term for the coal- ally weakened. The sandstone samples become looser and
bearing part of the upper carboniferous system, which con- weaker in structure, the brittleness decreases, and the failure
sists of rocks such as claystone, shale, siltstone, and sand- mode changes from brittle to ductile.
stone, interstratified with the beds of coal. Coal measures
sandstone is a sedimentary rock composed of mineral par-
ticles of varying sizes and shapes. It has many bulk defects,
including microcracks, fissures, and pores. In water-immer- 8 Conclusions
sion tests, the water first rapidly fills the pores and fissures
between the mineral particles. If the water immersion lasts (1) The content and composition of rock-forming miner-
for a long period, the pore water gradually permeates to the als in sandstone have no clear relationship with water
surface of the mineral particle to form bound water. Under content. However, with increasing water content, the
the external force action, the sandstone samples undergo uniaxial compressive strength, elastic modulus, and
frictional slip between the mineral particles. As the amount elastic storage limit of sandstone all gradually decrease,
of bound water increases on the mineral particle surfaces, its and the elastic strain of sandstone decreases under the
lubricating and softening effects are enhanced, which weak- same normalized cycle.
ens the bonding between particles, decreasing the frictional (2) Under graded cyclic loading and unloading, the input
coefficient and the cohesive force within the sandstone. energy of sandstone is stored mostly in the form of elas-
According to SEM and XRD analysis of typical fractures, tic energy, whereas the dissipated energy is mostly in
the sandstone is mainly composed of quartz, feldspar, and the form of work to overcome damping. The larger the
clay minerals; on the microscopic scale, the fracture sur- water content of sandstone, the faster the growth of the
face gradually rounds. The cracks and fissures within the damage factor, and the water accelerates the mechanical
sandstone samples are the primary sites where the physi- damage of sandstone. With increasing cycle number,
cal and mechanical interactions occurred between water the proportion of damping energy increases gradually
and rocks. As the water content increases, the calcareous first and then tends to stabilize. When the normalized
cement is more likely to dissolve and corrode upon contact cycle number approaches unity, the fraction of damp-
with water, and the soluble substances enter the immersion ing energy plummets, and that of the plastic energy
solution due to the movement of the water molecules. There- increases sharply.
fore, the pores and fissures grow and become more rounded. (3) With increasing water content, the effect of pore water
In addition, the clay minerals expand due to water absorp- on the lubrication, water wedge, and dissolution of
tion with the water–rock interaction. Expansion stress then mineral particles is enhanced, mineral particles are
concentrates significantly at the crack tip, promoting crack gradually rounded, and friction damping between parti-
initiation and propagation. The mineral particles on the sur- cles is weakened, which makes the sandstone gradually
faces of new cracks then undergo another round of physical evolve from a dense structure to a weak structure. More
reactions, including dissolution and expansion due to water and more secondary fractures appear around the main
absorption. These interactions promote the continuous dam- fractures in sandstone, and the strain field becomes
age and failure of sandstone. Under the joint action of these more and more complex. The failure mode of sandstone
factors, the macro-mechanical parameters and the energy- gradually changes from tension-splitting brittle failure
storage characteristics of sandstone deteriorate as the water to shear failure.
content increases.
The analysis of the test results reveals that the deteriora-
tion of the mechanical parameter and the failure modes of Acknowledgements The authors are grateful for the financial sup-
port from the key scientific research project of Shanxi Province (No.
sandstone samples vary with water content, which implies 57820191101016), the bidding project of Shanxi Province of China
that the physical impact of water on sandstone depends on (No. 20191101016), and the Doctoral Innovation Fund of Anhui Uni-
the stage of water immersion. At the initial stage of water versity of Science and Technology (No. 2021CX1003).
immersion, the water–rock interaction primarily features the
absorption of water molecules by mineral particles. With Availability of data and materials The datasets used and/or analyzed
in the current study are available from the corresponding author upon
the extension of soaking time, the water–rock interaction reasonable request.
is dominated by the swelling purification of clay minerals,
and the physical reaction is gradually enhanced. The second-
ary porosity increases within the sandstone samples under

13
 53 Page 14 of 14 W. Liu et al.

Declarations Li YH, Jing HW, Zeng QY (2006) Development and application of

digital photographic measurement software system for geotechni-
cal engineering. Chin J Rock Mech Eng S2:3859–3866
Conflict of interest The authors declare that they have no competing
Li TB, Chen ZQ, Chen GQ et al (2015) An experimental study of
interests.
energy mechanism of sandstone with different moisture contents.
Rock and Soil Mech 36(S2):229–236
Open Access This article is licensed under a Creative Commons Attri- Li W, Wang DH, Li HJ (2019) Environmental engineering issues
bution 4.0 International License, which permits use, sharing, adapta- induced by abandoned coal mine hidden disasters. IOP Conf Ser
tion, distribution and reproduction in any medium or format, as long Earth Environ Sci 237:022039
as you give appropriate credit to the original author(s) and the source, Liu Y, Dai F, Dong L, Xu NW (2018) Experimental investigation on
provide a link to the Creative Commons licence, and indicate if changes the fatigue mechanical properties of intermittently jointed rock
were made. The images or other third party material in this article are models under cyclic uniaxial compression with different loading
included in the article's Creative Commons licence, unless indicated parameters. Rock Mech Rock Eng 51(1):47–68
otherwise in a credit line to the material. If material is not included in Ma CQ, Li HZ, Niu Y (2018) Experimental investigation on the
the article's Creative Commons licence and your intended use is not mechanical behavior and damage evolution mechanism of water-
permitted by statutory regulation or exceeds the permitted use, you will immersed gypsum rock. Environ Earth Sci 77:23
need to obtain permission directly from the copyright holder. To view a Niu SJ, Ge SS, Yang DF et al (2018) Mechanical properties and
copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. energy mechanism of saturated sandstones. J Cent South Univ
25(6):1447–1463
Pu H, Bian ZF, Zhang JX et al (2021) Research on a reuse mode of
References geothermal resources in abandoned coal mines. J China Coal Soc
46(2):677–687
Stevens JL, Holcomb DJ (1980) A Theoretical investigation of the slid-
Bian ZF, Zhou YJ, Zeng CL et al (2021) An discussion of the basic ing crack model of dilatancy. J Geophys Res 85(B12):7091–7100
problems for the construction of underground pumped storage Sun K, Fan LM, Xia YC et al (2021) Impact of coal mining on ground-
reservoir in abandoned coal mines. J China Coal Soc. https://​doi.​ water of Luohe Formation in Binchang mining area. Int J Coal Sci
org/​10.​13225/j.​cnki.​jccs.​YG21.​0392 Technol 8(1):88–102
Chen GQ, Li TB, Wang W et al (2019) Weakening effects of the pres- Wang H, Yang TH, Liu HL et al (2017) Mechanical properties and
ence of water on the brittleness of hard sandstone. Bull Eng Geol energy evolution of dry and saturated sandstones under cyclic
Environ 78:1471–1483 loading. Rock Soil Mech 38(6):1600–1608
Fan JY, Xie HP, Jie C et al (2020) Preliminary feasibility analysis of Wong LNY, Maruvanchery V, Liu G (2016) Water effects on rock
a hybrid pumped-hydro energy storage system using abandoned strength and stiffness degradation. Acta Geotech 11:713–737
coal mine goafs. Appl Energy 258:114007 Xie H, Ju Y, Li L (2005) Criteria for strength and structural failure of
Geng JS, Cao LW (2020) Failure analysis of water-bearing sandstone rocks based on energy dissipation and energy release principles.
using acoustic emission and energy dissipation. Eng Fract Mech Chin J Rock Mech Eng 24:3003–3010
231:107021 Xie H, Zhao JW, Zhou HW et al (2020) Secondary utilizations and
Gu DZ (2015) Theory framework and technological system of coal perspectives of mined underground space. Tunn Undergr Space
mine underground reservoir. J China Coal Soc 40:239–246 Technol 96:103129
Guo PY, Gu J, Su Y et al (2021) Effect of cyclic wetting–drying on Yu Y, Tang SB, Tang CA (2019) Experimental study on the influence
tensile mechanical behavior and microstructure of clay-bearing of water content on instantaneous and creep mechanical properties
sandstone. Int J Coal Sci Technol 8(5):956–968 of red sandstone. J China Coal Soc 44(2):473–481
Hashiba K, Fukui K (2015) Effect of water on the deformation Zhou ZL, Cai X, Cao WZ et al (2016) Influence of water content on
and failure of rock in uniaxial tension. Rock Mech Rock Eng mechanical properties of rock in both saturation and drying pro-
48:1751–1761 cesses. Rock Mech Rock Eng 49:3009–3025
Hua W, Li JX, Dong SM et al (2019) Experimental study on mixed
mode fracture behavior of sandstone under water-rock interac- Publisher's Note Springer Nature remains neutral with regard to
tions. Processes 7:70 jurisdictional claims in published maps and institutional affiliations.
Jardon S, Ordonez A, Alvarez R, Cienfuegos P et al (2013) Mine water
for energy and water supply in the central coal basin of Asturias
(Spain). Mine Water Environ 32(2):139–151
Kuwahara Y, Kyamamoto I, Hirasawa T (1990) An experimental and
theoretical study of inelastic deformation of brittle rocks under
cyclic uniaxial loading. Tohoku Geophys J 33(1):1–21

13