Introduction

Coal remains an irreplaceable energy for the next few decades in China, where the coal is mainly stored in Permo-Carboniferous strata. Ordovician limestone, which is usually situated below the Permo-Carboniferous strata, is a strong regional aquifer in north China, where nearly 60% of China’s coal reserves are located. Mines in this area are highly prone to major floor water inrush disasters. Floor-confined water threatens nearly 50% of China’s major coal mines, with reserves of 2.5 × 1013 kg. The probability of water inrush disasters increases sharply with high water pressure and ground pressure, which have accompanied eastern China’s increased mining depths.

Many scholars have carried out studies attempting to predict water inrush disasters induced by mining, using theoretical calculations, mathematical models, and on-site monitoring. As an example of a theoretical calculation, Siliesaliefu assumed the floor strata in a panel to be a beam under uniform load, while the mechanical problem of floor failure induced by underground mining was simplified as a simply supported beam under uniformly distributed loads (Zhang 2016). Xu et al. (2017) derived a semi-analytical solution to determine the minimum safety thickness of rock to resist water inrush disasters from filling-type karst caves. Li et al. (2019) deduced a way to calculate the critical condition of failure of an aquiclude under the presence of a fault using brittle fracture criteria and the shear failure limit equilibrium condition.

An empirical formula was proposed to calculate the water inrush coefficient, and this formula has been widely used in China (Zhang 2016). Based on this formula, a coefficient-unit inflow method of water inrush disasters was proposed by Qiao et al. (2009). Meng et al. (2012) improved the formula to calculate the water inrush coefficient by considering the height of the zone of water rising from the confined aquifer and the floor failure depth. Fan et al. (2019) adopted the water inrush coefficient to evaluate the risk of water inrush disasters in separate layers. These theoretical methods provide insight into the mechanism of water inrush disaster induced by mining activity in coal mines. However, due to the complexity of geological conditions and in situ stress conditions, there was inherent inaccuracy of their predictive abilities.

In recent years, many scholars have used mathematical models to predict water inrush disasters based on the hydrogeological conditions. These mathematical models included catastrophe theory (Hua et al. 2011), variable-weight model and unascertained measure theory (Wu et al. 2017), fuzzy mathematics (Yang et al. 2017), Bayesian networks (Wu et al. 2016), attribute mathematical theory (Li et al. 2015), intelligent machine-learning algorithm (Zhao al. 2018), and analytic hierarchy process (Du et al. 2017; Wu et al. 2015). However, these models lacked physical meaning, and the validity of predictions of water inrush disasters using these models was uncertain due to the complexity of hydrological conditions in the field, making it very difficult to use mathematical models to accurately predict inrush disasters before mining. Thus, in-site monitoring is key for more accurate predictions.

The evolution of water flow channels typically begins with a micro-fracture precursor, especially under the high-pressure conditions of deep mining. Microseismic (MS) monitoring is an important tool to monitor micro-fracture precursors. According to Liu et al. (2014) and Zhang et al. (2016), a combination of MS monitoring and numerical analysis of damaged zones can help reveal the formation process of the flow channel. Zhou et al. (2017) analyzed the spatial and temporal behaviors of density distribution, focal parameters, and cracking-type MS events and correlated them with the formation processes of a local flow channel. However, the micro-fracture precursor of water-flow channels was not clearly revealed, limiting the ability to predict an inrush disaster.

In general, there are two types of fractures that may constitute water flow channels. The first is a crack caused by the failure of rock under the action of mining stress and seepage pressure. The second is a crack formed by the activation and further failure of geological structures in the rock mass due to the mining stress and seepage pressure. Accurate monitoring of the spatial distribution and evolution process of these two kinds of fractures is the key to determining the spatial location of the water flow channels and predicting them.

In view of these two failure modes, many scholars have carried out studies at the laboratory scale by AE monitoring. Lei et al. (2003) carried out triaxial compression experiments with joints, and stated that the spatial distribution and formation process of cracks induced by the activation of primary joints were closely related to the spatial aggregation and evolution of AE events. Chen et al. (2012) carried out conventional triaxial compression experiments, Li et al. (2013) conducted direct tensile tests, and Xie et al. (2011) conducted Brazilian splitting tests. Their research results showed that the spatial distribution and formation process of shear and tensile cracks in rock specimens were consistent with the spatial aggregation and evolution of AE events. This indicated that it was feasible to use MS monitoring technology to monitor the formation of water flow channels.

This focus of this study was the no. 22517 panel in the Dongjiahe coal mine. An MS monitoring system was established to study the micro-fracture precursors and the focal parameters for the micro-fracture precursor were determined. A MS monitoring system was installed to monitor floor MS events during mining activities in the no. 22517 panel. Next, the formation process of water-flow channels was identified in the panel by analyzing changes in hydrological and geological conditions, water quality, and water inflow during the mining process, along with the MS monitoring results. Finally, the way these parameters varied were studied.

Methodology

Relationship Between MS Events and Water Flow Channels

To form water flow channels, the micro-fracture events must meet two conditions. First, the micro-fractures must be near each other and second, the scale of the micro-fracture events has to be large enough to make the micro-fractures connect. Seismic energy is an important parameter for measuring the scale of micro-fracture events. If the seismic energy of each MS event is very large, only a few MS events are needed to connect the micro-fractures. If the seismic energy of each MS event is very small, a large number of MS events are needed to ensure a micro-fracture connection. Thus, we used the energy density of MS events to comprehensively determine the energy aggregation and quantity aggregation of MS events. The energy density of MS events was defined as the sum of all MS events in unit volume. The threshold of the energy density of MS events is related to the panel depth, lithology, and structural stress. Taking the no. 22517 panel in the Dongjiahe coal mine as an example, the threshold of the energy density of MS events was determined through the mutual verification of the various monitoring results, so as to provide reference for mines with similar engineering geological conditions.

Potential Precursory Indicators of Water Flow Channels

MS sensors can be used to monitor the real-time waveforms of elastic wave motion caused by the micro-fracturing of a rock mass. Using the waveforms received by different sensors, the arrival times of P or S waves can be determined. Based on the P wave and S wave velocity, the position and time of the MS events can then be calculated by the three-ball intersection principle. Finally, the focal parameters of moment magnitude, seismic moment, seismic energy, ratio of S-wave to P-wave energy, source radius, asperity radius, static stress drop, apparent stress, dynamic stress drop, maximum displacement, peak velocity parameter, and peak acceleration parameter are calculated using methods of quantitative seismology based on the waves received by different sensors, the location of MS events, and the attenuation of wave propagation in the rock mass.


Lynch et al. (2004) and Mendecki et al. (2010) used seismic energy, seismic potency, apparent stress, cumulative apparent volume, energy index, and other parameters to predict large MS events. Liu et al. (2013) used seismic moment, Gutenberg-Richter b value, spatial correlation length, and fractal characteristics to study the damage evolution characteristics of surrounding rocks during mining activities to obtain precursory information of rock-burst disasters in the Hongtoushan Copper Mine. Chen et al. (2011), Xiao et al. (2016), and Yu et al. (2015) applied apparent volume, apparent stress, energy index, and Gutenberg-Richter b value to study early warning of rock-burst disasters in a deep tunnel of the Jinping II Hydropower Station. Cheng et al. (2019) used the Gutenberg-Richter b value to study the mining-induced brittle fault activation process. Since these and other previous studies have shown that seismic energy, seismic potency, apparent stress, seismic moment, apparent volume, energy index (EI), and Gutenberg-Richter b value are important parameters for MS monitoring, these six parameters were used in this study. The physical meanings of these six parameters and specific calculations follow.

  1. 1.

    Seismic potency (m3). The seismic potency represents the inelastic volume deformation at the source of a MS event. Seismic potency can be calculated by formula 1. For a planar shear source, the seismic potency can be calculated by formula 2. At the recording site, Mendecki et al. (2010) calculated seismic potency by formula 3.

    $$P=\Delta \varepsilon V$$
    (1)
    $$P=\overline{u }A$$
    (2)
    $${P}_{P,S}=4\pi {v}_{P,S}R{\int }_{0}^{{t}_{s}}{u}_{\mathrm{corr}}\left(t\right)dt$$
    (3)

    In formula 1, \(P\) is the seismic potency, \(\Delta \varepsilon \) is the inelastic shear strain change, and \(V\) is the source volume. In formula 2, \(A\) is the area of the failure plan and \(\overline{u }\) is the relative slip distance on both sides of failure surface. In formula 3, \({u}_{\mathrm{corr}}\left(t\right)\) is the displacement pulse of the P-wave or S-wave, which is corrected for the far-field radiation pattern,\({v}_{\mathrm{P},\mathrm{S}}\) is the velocity of the P-wave or S-wave, \(R\) is the distance between a sensor and the source, and \({t}_{s}\) is the source duration, \({u}_{\mathrm{corr}}\left(0\right)=0\) and \({u}_{\mathrm{corr}}\left({t}_{s}\right)=0\).

  2. 2.

    Seismic moment (N·M). The seismic moment is used to measure the inelastic deformation at the source, and can be calculated as formula 4. For a planar shear source, the seismic moment can be calculated as formula 5. Based on this, Hanks and Kanamori (1979) defined the moment magnitude to measure the magnitude of a seismic event and established the relationship that scales seismic moment into the moment magnitude as formula 6.

    $$M=\mu P$$
    (4)
    $$M=\mu \overline{u }A$$
    (5)
    $$m=\frac{2}{3}\mathrm{log}M-6.0$$
    (6)

    where \(M\) is the seismic moment, \(P\) is the seismic potency, \(\mu \) is the shear modulus (35.2 GPa), \(A\) is the area of the failure plan, \(\overline{u }\) is the relative slip distance on both sides of the failure surface, \(m\) is the moment magnitude, and \(M\) is the seismic moment.

  3. 3.

    Seismic energy (J). The seismic energy, which is radiated as seismic waves, is the portion of the energy that is released during fracture and frictional sliding at the source. Mendecki et al. (2010) calculated the seismic energy of the P-wave or S-wave according to formula 7.

    $${E}_{P,S}=\frac{8}{5}\pi \rho {v}_{P,S}{R}^{2}{\int }_{0}^{{t}_{s}}{\dot{u}}_{\mathrm{corr}}^{2}\left(t\right)dt$$
    (7)

    where \(E\) is the seismic energy, \(\rho \) is the rock density, \({\dot{u}}_{\mathrm{corr}}^{2}\left(t\right)\) is the square of the displacement pulse of the P-wave or S-wave, \({t}_{s}\) is the source duration, \({v}_{P,S}\) is the velocity of the S-wave or P-wave, and \(R\) is the distance from the source.

  4. 4.

    Apparent stress (Pa). Aki (2012) and Wyss and Brune (1968) calculated the apparent stress as the seismic energy caused by inelastic deformation per unit volume at the source using formula 8.

    $${\sigma }_{A}=\frac{E}{P}$$
    (8)

    where \({\sigma }_{A}\) is the apparent stress, \(E\) is the seismic energy, and \(P\) is the seismic potency.

  5. 5.

    Energy index. The average energy corresponding to seismic potency can be calculated by formula 9 from the seismic potency and the seismic energy, calculated as shown in Fig. 1. The energy index was defined by Van and Butler (1993) as the ratio of the seismic energy to the average energy, and was calculated by formula 10.

    $$\mathrm{log}\overline{E }\left(P\right)=d\mathrm{log}P+c$$
    (9)
    $$EI=\frac{E}{\overline{E }(P)}=\frac{E}{{10}^{d\mathrm{log}P+c}}={10}^{-c}\frac{E}{{P}^{d}}$$
    (10)

    where \(\overline{E }(P)\) is the average energy corresponding to seismic potency, \(P\) is the seismic potency, \(E\) is the seismic energy, \(EI\) is the energy index, and \(c\) and \(d\) are constants.

  6. 6.

    Apparent volume (m3). The apparent volume was calculated by Mendecki et al. (1993) as formula 11.

    $${V}_{A}=\frac{\mu {P}^{2}}{E}=\frac{MP}{E}$$
    (11)

    where \({V}_{A}\) is the apparent volume, \(\mu \) is the shear modulus (35.2 GPa), \(E\) is the seismic energy, \(P\) is the seismic potency, and \(M\) is the seismic moment.

  7. 7.

    Gutenberg-Richter b value. The seismic size distribution typically follows a power law by Gutenberg and Richter (1944), which can be determined by formula 12. The Gutenberg-Richter b value is the slope of this power law, and is commonly used to describe the relative proportion of small magnitude events vs. large magnitude events. A large Gutenberg-Richter b value indicates a large proportion of small seismic events. The Gutenberg-Richter b value can be calculated by fitting the seismic size distribution according to formula 12.

    $$\mathrm{log}N=a-bm$$
    (12)
Fig. 1
figure 1

Linear relationship between the logarithm of seismic potency and the logarithm of seismic energy (Xiao et al. 2016)

Full size image

where \(m\) is the magnitude of MS events, \(N\) is the number of MS events, whose magnitude is not less than \(m\), \(b\) is the slope of the power law, and \(a\) is a constant representing the seismic activity.

Micro-fracture Precursors of Water-Flow Channels in a Coal Mining Case

The no. 22517 panel in the Dongjiahe coal mine was studied using a MS monitoring system to monitor the evolution characteristics of floor MS events during mining. The Dongjiahe coal mine is located in Shaanxi Province in China (Fig. 2a). The no. 22517 panel is the panel that was being mined during MS monitoring; the lithology of the overlying and floor strata of the panel is shown in Fig. 2b. The formation process of water-flow channels was determined by analysis of hydrological and geological conditions, the change characteristics of water inflow, the water quality of the water inflow, and MS monitoring results. The change characteristics of the six MS parameters were studied during channel formation for the micro-fracture precursors of the water flow channels.

Fig. 2
figure 2

General geological situation of the no. 22517 panel in the Dongjiahe coal mine

Full size image

Hydrological and Engineering Geological Conditions

Coal mines are mainly threatened by Ordovician limestone water in the Hancheng, Chenghe, Pubai, and Fugu mining areas in Shaanxi Province. The no. 22517 panel in the Dongjiahe coal mine in the Chenghe mining area was selected to study the micro-fracture precursor of water-flow channels induced by mining. The detailed information and the rock mass strata at this site were described previously (Cheng et al. 2017) and the geological parameters are listed in Table 1.

Table 1 Geological parameters in the No. 22517 Panel
Full size table

This study focused on the hydrological and engineering geological conditions at this site (Fig. 3). The blue number in Fig. 3a indicates the pitch distance, which is defined as the distance from a location in the no. 22517 panel to the north–south roadway at the western side of that panel. As shown in Fig. 3, the pitch distance was defined as the horizontal distance from a point in the working face to the north–south roadway at the western side of the working face and was used to describe both the advancing position of mining activities and the spatial location of geological structures. The open cut is located at the eastern side of the no. 22517 panel at a pitch distance of 1217 m, and the mining stop line is located at the western side of that panel, with a pitch distance of 100 m. The north side is the haulage roadway and the south side is the orbital roadway. As shown in Fig. 3, two faults, an anticline, and a thin seam zone were exposed during the excavation of the haulage roadway and the orbital roadway and a fault was exposed by microseismic monitoring during mining of the no. 22517 panel (Cheng al. 2019). The location and parameters of the faults, anticline, and thin seam zone are shown in the Table 2. As shown in Fig. 4, the distance between the floor of the no. 5 coal and the top of the Ordovician limestone was about 31.95 m. A layer of fine conglomerate (4.8 m thick) 12.05 m below the coal seam (Fig. 4) was the stratum with the greatest strength and thickness between the floor of the no. 5 coal and the top of the Ordovician limestone, and so acted as a key strength stratum. A layer of aluminous mudstone (4 m thick), 8.6 m below the fine conglomerate (Fig. 4) had strong water resistance performance and acted as a key water-isolating stratum. Meng et al. (2012) divided the coal seam floor lithology into mudstone, mudstone-sandstone, and sandstone based on the mudstone content. As shown in Fig. 4, the mudstone content of the study region is 12.52%. The floor in the no. 22517 panel was the sandstone type, and the floor has high strength and permeability.

Fig. 3
figure 3

Geological information for the no. 22517 panel. a Contour line of coal seam floor in the No. 22517 Panel (The roadway in the north side was the haulage roadway, the roadway in the south side was the orbital roadway and the roadway in east side was the open-cut. X13, X17 and X19 were boreholes located within the No. 22517 Panel. The pink numbers were longitude and latitude. The blue numbers were pitch distance. The red numbers were the floor thickness between Ordovician limestone and the No. 5 coal seam. The black numbers were elevation of coal seam floor. The unit of the numbers was meter). b Side view of the haulage roadway (The X-axis was the pitch distance and Y-axis was elevation. The unit of the numbers was meter). c Side view of the orbital roadway (The X-axis was the pitch distance and Y-axis was elevation. The unit of the numbers was meter)

Full size image
Table 2 Overview of abnormal areas based on geological conditions
Full size table
Fig. 4
figure 4

Borehole lithology histogram of X17 in floor

Full size image

Because the no. 5 coal was close to the Ordovician limestone, grouting reinforcement was conducted in the panel’s floor strata. The grouting boreholes were positioned at depths from 20.7 to 29.2 m, with an average depth of 25.6 m. After the grouting reinforcement, the transient electromagnetic (TEM) and direct current (DC) measurement methods were used to detect water-rich areas in the no. 22517 panel. Three low resistivity zones and four high resistivity zones were revealed by the DC method and an anomaly zone was revealed by the TEM method. The location and shape of the three low resistivity zones and the anomaly zone are shown in Table 3.

Table 3 Geophysical prospecting results in the No. 22517 Panel
Full size table

MS Monitoring System

The MS monitoring system in this location was previously described by Cheng et al. (2017). Geophones with a sensitivity of 43.3 V/m/s, and a response frequency from 15 to 1000 Hz were arranged at the non-mining wall of the roadway and in the no. 22517 panel. Fifteen geophones were positioned in the haulage roadway and 15 geophones were positioned in the orbital roadway; the location coordinates are listed in Table 4. To firmly attach the geophones to the rock wall, a borehole length of ≈ 4 m was used to install the geophones perpendicular to the roadway direction, angled downward at 45°.

Table 4 Coordinates of geophones in the No. 22517 Panel (Cheng et al. 2019)
Full size table

The geophones were positioned ≈ 80 m apart on each roadway. The data transmission network of the MS monitoring system is shown in Fig. 5. A Geiger localization algorithm (Khadhraoui et al. 2010) method was used to locate the MS events. Use of this algorithm requires detection of the waveform of each MS event by at least four geophones. A homogeneous velocity model was also used to locate MS events. To determine the P- and S-wave velocities, some blasting events with known coordinates were analyzed. The calibration results indicated that the P-wave velocity was 2800 m/s and the S-wave velocity was 1800 m/s. Cheng et al. (2017) calculated the localization error of the MS monitoring system as within 10 m the range of 100 m below the panel’s coal seam floor.

Fig. 5
figure 5

Data transmission network of MS monitoring system in the no. 22517 panel (Cheng et al. 2017)

Full size image

Formation of Water Flow Channels

The monitoring results showed that most of the floor MS events were concentrated above the key strength stratum in the no. 22517 panel (12 m below the no. 5 coal), except for three MS events-concentrated zones (zones I, II, and III), which had the potential to form water-flow channels. The parameters and location of these three MS events-concentrated zones are introduced in Table 5. As shown in Table 5, the MS events-concentrated zone I had the largest the extended depth (about 50 m) and energy density. Thus, this zone was most likely to form a water flow channel.

Table 5 Parameters of three MS events-concentrated zones
Full size table

Formation of a water flow channel requires that the local failure area in the floor penetrates the rock mass between the Ordovician limestone roof and the coal seam floor, significantly increasing the water inflow in the panel during formation of the local failure area. Based on geological conditions in the no. 22517 panel, the elevation varied from + 230 m to + 240 m and the elevation in the Ordovician limestone roof varied from + 200 m to + 210 m in the MS events-concentrated zone I. Thus, the change characteristics in energy density of the MS events at different elevation was analyzed from the elevation of + 240 m to the elevation of + 210 and the result was shown in Fig. 6 when the threshold of the energy density of MS events was 0.05 J/m3. There are four sections in Fig. 6 (the no. 1 section at the elevation of + 240 m, the no. 2 section at the elevation of + 230 m, the no. 3 section at an elevation of + 220 m, and the no. 4 section at an elevation of + 210 m. Of these, the no. 3 section (at + 220 m) was located at the bottom of the key strength stratum. As shown in Fig. 6, the MS events-concentrated zone I gradually decreased at greater depths. The MS events in the MS events-concentrated zone I passed through the key strength stratum and extended to below + 200 m in elevation. Thus, the MS events-concentrated zone I penetrated the rock mass between the Ordovician limestone roof and the coal seam floor.

Fig. 6
figure 6

Change characteristics of the energy density of MS events at different elevation in the MS events-concentrated zone I. (Note: The No. 1 Section is at the elevation of +240 m. The No. 2 Section is at the elevation of +230 m. The No. 3 Section is at the elevation of +220 m. The No. 4 Section is at the elevation of +210 m. The No. 5 Section is at the elevation of +200 m)

Full size image

On this basis, the relationship between the formation process of the MS events-concentrated zone I (Fig. 7) and the change characteristics of water inflow in the no. 22517 panel during the mining process (Fig. 8) was analyzed. As shown in Fig. 7, the number (energy) of the MS events in the MS events-concentrated zone I started to increase on June 11, 2015 (the MS events-concentrated zone I was 85 m from the mining at the time), which meant that the MS events-concentrated zone I began to form. The number (energy) of MS events in the floor peaked on July 10, 2015 (the MS events-concentrated zone I was 45 m from mining activities at that time), which meant that the MS events-concentrated zone I had formed. The number (energy) of MS events in the floor descended to a stable value on July 23, 2015 (the MS events-concentrated zone I was 37 m from mining at that time), which meant that the MS events-concentrated zone I remained unchanged. As shown in Fig. 8, the working face inflow remained unchanged at 60 m3/h before July 10, 2015. There was a significant increase in water inflow in the no. 22517 panel from June 11, 2015 to July 10, 2015 and it peaked at 82 m3/h on July 10, 2015. After July 10, 2015, the working face inflow began to decrease until July 23, 2015. The working face inflow was stable at 70 m3/h after July 23, 2015. Previous results showed that the formation of the MS events-concentrated zone I corresponded with increasing water inflow in the no. 22517 panel.

Fig. 7
figure 7

Formation of the MS events-concentrated zone I

Full size image
Fig. 8
figure 8

Change characteristics of water inflow in the no. 22517 panel during the mining process

Full size image

The source of the water inflow in the no. 22517 panel is required to further verify the formation of a water channel in this area, so the water flowing into the no. 22517 panel was analyzed (Table 6). There were four hydro-chemical types of Ordovician limestone water in the Weibei Mining Area (Tao 1999). The first type was \(HCO_{3} - Ca\), the second type was\({SO}_{4}\bullet {HCO}_{3}-Na\bullet Ca\), the third type was \({SO}_{4}\bullet {HCO}_{3}-Na\bullet Ca\bullet Mg\) and the forth type was\({SO}_{4}\bullet {HCO}_{3}\bullet Cl-Na\bullet Ca\bullet Mg\). The salinity of the Ordovician limestone water in the Weibei mining area was about 1 g/L. Because the Dongjiahe coal mine was located in the Weibei mining area, it could be inferred that the hydro-chemical type of the water inflow was \({SO}_{4}\bullet {HCO}_{3}\bullet Cl-Na\bullet Ca\bullet Mg\) and most of the water inflow came from the Ordovician limestone water in the no. 22517 panel.

Table 6 Content analysis result of the water inflow in the No. 22517 Panel
Full size table

As previously discussed, zone I penetrated the rock mass between the Ordovician limestone roof and the coal seam floor, which meant that zone I had the necessary conditions to form a water flow channel. Indeed, since the formation of zone I corresponded with the increase in water inflow and most of the water came from the Ordovician limestone, it is clear that Ordovician limestone water entered the no. 22517 panel through a water flow channel formed in zone I and the threshold of the energy density of MS events in the in the no. 22517 panel was 0.05 J/m3. The same method was used to analyze the MS events-concentrated zone III and revealed that a small water flow channel also formed there.

Micro-fracture Precursors of Water Flow Channels

The focal parameters were calculated to characterize their evolution over time as the no. 22517 panel was mined (Fig. 9). The cumulative apparent volume increased significantly as the large and small water flow channels formed. With a month as the time window, the power law was used to fit the moment magnitude distribution of floor MS events, and the Gutenberg-Richter b value of each month was calculated (Fig. 10a) and the change of Gutenberg-Richter b value with time was obtained (as shown in Fig. 10b). From Fig. 10b, it can be seen that the Gutenberg-Richter b value generally showed a downward trend before the formation of the big water-flow channel, dropped to 0.7 at the start of channel formation, dropped to the lowest value of 0.5 during channel formation, and then increased. The calculations indicate that the cumulative apparent volume increased significantly and the rapid decrease of the Gutenberg-Richter b value can be regarded as the micro-fracture precursor of the formation of a water-flow channel, with a value of 0.7 as an early warning threshold for channel formation.

Fig. 9
figure 9

Change of focal parameters during formation of water flow channels. a Evolution characteristics of seismic energy with time during mining process (Y-axis is seismic energy and the unit is J. X-axis is the date). b Evolution characteristics of seismic potency with time during mining process (Y-axis is seismic potency and the unit is m3. X-axis is the date). c Evolution characteristics of apparent stress with time during mining process (Y-axis is apparent stress and the unit is Pa. X-axis is the date). d Evolution characteristics of apparent volume with time during mining process (Y-axis is apparent volume and the unit is m3. X-axis is the date). e Fitting formula between the logarithm of seismic potency and the logarithm of seismic energy can be fitted (Y-axis is the logarithm of seismic energy. X-axis is the logarithm of seismic potency). f Evolution characteristics of cumulative EI (red line) and cumulative apparent volume (black line) with time during mining process (Y-axis in the left is cumulative EI. Y-axis in the right is cumulative apparent volume. X-axis is the logarithm of seismic potency)

Full size image
Fig. 10
figure 10

Evolution of Gutenberg-Richter b value with time

Full size image

Discussion

Feasibility Analysis on Early Warning for Floor Water Inrush Disasters

Meng et al. (2012) improved the empirical formula (Zhang 2016) to calculate the water inrush coefficient by considering the height of the zone of water rising from the confined aquifer and the floor failure depth, presented as formula 13. In China’s Regulations on Water Prevention and Control in Coal Mines, the threshold of water inrush coefficient is set as 0.06 MPa/m in regions with geological structures (such as areas of abnormal water richness, fault areas, and fold areas) and 0.1 MPa/m in other regions.

$${T}_{S}=\frac{P}{M-{C}_{P}-{Z}_{0}}$$
(13)

where T is the coefficient of water inrush, with units of MPa/m, P is the water pressure caused by the confined aquifer in floor, M is the thickness of the floor aquifer, with units of m, Cp is the floor failure depth, with units of m, and Z0 is the height of the zone of water rising from the confined aquifer.

Calculation of the water inrush coefficient in the no. 22517 panel required determination of the depth of the floor failure and the height of the zone of water rising from the confined aquifer. Gu et al. (2011) used elastic wave velocity monitoring in a borehole to monitor floor failure depth at this location and reported a floor failure depth of 12 m. Because aluminous mudstone has good water-resisting property, the height of the zone of water rising from the confined aquifer is 0 m. The water inrush coefficient at different locations of the no. 22517 panel was calculated using formula 2, allowing construction of a contour map of the water inrush coefficient, presented in Fig. 11. As shown in Fig. 11, the water inrush coefficient was large (0.95 MPa/m) at the northwest end and small (0.04 MPa/m) at the southeast end. In the inverted trapezoidal area from the pitch distance of 0 m to that of 900 m at the haulage roadway and the pitch distance of 0 m to that of 600 m at the orbital roadway of the no. 22517 panel, the water inrush coefficient in floor was greater than 0.06 MPa/m.

Fig. 11
figure 11

Distribution of floor water inrush coefficient in the no. 22517 panel

Full size image

The region from 190 m to that of 720 m (the pitch distance) in the no. 22517 panel includes many geological structures (as shown in Table 2) and four water-rich abnormal areas (Table 3). According to Tables 5 and 3, the location of the MS events-concentrated zone I, where a big water-flow channel formed, corresponded with the location of the 2nd low resistivity zone. The location of the MS events-concentrated zone III, where a small water-flow channel formed, corresponded with the location of the 1st low resistivity zone and the anomaly zone. It could be seen that water flow channels are often generated in areas where water-rich geophysically abnormal areas and abnormal areas determined by the water inrush coefficient overlap, but not all overlapping areas form water flow channels. However, our work has shown that the formation of water flow channels can be quantitatively analyzed by using MS monitoring to determine the micro-fracture precursors of water-flow channels.

A Real-Time Early Warning Method

Based on the above results, a real-time early warning method for water inrush disasters in the floor was established (Fig. 12). This method has three steps. In the first step, the overall risk assessment of water inrush disasters in the floor from a panel is evaluated. If the risk of water inrush is high, the panel should not be mined until it meets safe mining standards by grouting reinforcement of the floor strata. If the risk of water inrush disasters is low and the overall stability of the panel is good, local dangerous areas in the floor are identified (the second step). In the third step, MS monitoring is installed in the panel and the localized, potentially dangerous areas are monitored in real time for early detection of water inrush disasters.

Fig. 12
figure 12

A real-time early warning method for water inrush disasters in floor

Full size image

In the first step, hydrologic and engineering geological information is acquired by geological survey and geophysical prospecting. Geological information (coal seam thickness, floor lithology distribution, and thickness, water-proof layer distribution and spatial thickness, key stratum spatial location and thickness, Ordovician limestone aquifer spatial location, and water pressure distribution) is acquired using geological exposures, such as boreholes, roadway excavation, and mining activities. Geophysical prospecting is performed using seismic wave, ground penetrating radar, audio-frequency electrical penetration, transient electromagnetics, and high-density resistivity methods. Geophysical prospecting provides the spatial location of failure zones, faults, folds, and collapse columns, as well as the water content of these features. Next, the floor failure depth should be determined and the overall risk assessment of water inrush disasters through the floor from the work surface can be evaluated by assessing the potential correlation between the spatial location of the key stratum and the floor failure zone. If the floor failure zone in a panel passes through the key stratum in floor, then grouting of this stratum is needed to reduce the failure depth. However, the floor in the no. 22517 panel was found to be stable.

In the second step, the water inrush coefficient is calculated by formula 2 based on the height of the zone of water rising from the confined aquifer and the floor failure depth. The coefficient data is then presented as a contour map of the area of interest. Using the water inrush coefficient thresholds stipulated by the Regulations on Water Prevention and Control in Coal Mines for normal regions and regions with geological structures, potentially dangerous areas can be identified.

In the third step, a MS monitoring system is established to monitor in real time the evolution characteristics of floor MS events in areas with a risk of a water inrush disaster. These collected data allow calculation of the cumulative apparent volumes and the Gutenberg-Richter b values. In this way, monitoring is performed in real time for the early detection of water inrush disasters induced by mining activity in coal mines.

Conclusion

  1. 1.

    Using MS monitoring technology to monitor the formation process of water flowing channels was shown to be feasible. The energy density of MS events were used to comprehensively determine the energy aggregation and quantity aggregation of MS events to identify the spatial location and formation of water flow channels. This information, along with the MS focal parameters of seismic energy, seismic potency, apparent stress, seismic moment, apparent volume, energy index (EI) and Gutenberg-Richter b value were used in this study to detect micro-fracture precursors of water-flow channels induced by mining activity.

  2. 2.

    In the no. 22517 panel in the Dongjiahe coal mine, the formation of water-flow channels was described by analysis of the hydrological and geological conditions, changes in water inflow during the mining process, the water quality of the water inflow, and MS monitoring results. The detected MS data was temporally and spatially consistent with formation of a water channel, and was used to identify the spatial location and formation of the water flow channels in the no. 22517 panel, where the threshold of the energy density of MS events was 0.05 J/m3, which can provide reference for mines with similar engineering geological conditions. The change characteristics of MS focal parameters were studied during channel formation, and the significant increase in cumulative apparent volume and rapid decrease of the Gutenberg-Richter b value correspond to the micro-fracture precursor of channel formation. The data suggest that a Gutenberg-Richter b value of 0.7 should be an early warning threshold for channel formation.

  3. 3.

    The feasibility for a real-time early warning for water inrush disasters through the coal seam floor was studied based on previous research results and after determining it was feasible to establish a real-time early warning method using MS monitoring, geophysical prospecting, and water inrush coefficients, a real-time early warning method for water inrush disasters through floor was established using these methods.