The interplay between moisture sensitive roof rocks and roof falls in an Illinois underground coal mine

Computers and Geotechnics 80 (2016) 152–166

Contents lists available at ScienceDirect

Computers and Geotechnics journal homepage: www.elsevier.com/locate/compgeo

Research Paper

The interplay between moisture sensitive roof rocks and roof falls in an Illinois underground coal mine Abdolreza Osouli a,⇑, Behrooz Moradi Bajestani b a b

2043 Engineering Building, Civil Engineering Department, Southern Illinois University Edwardsville, 61 Circle Dr., Edwardsville, IL 62026-1800, United States AECOM Technical Services Inc., 5609 Fairview Road Apt 5, Charlotte, NC 28209, United States

a r t i c l e

i n f o

Article history: Received 17 November 2015 Received in revised form 13 June 2016 Accepted 11 July 2016

Keywords: Roof failure Moisture sensitivity Numerical modeling Rock bedding Roof bolts

a b s t r a c t This study focuses on the performance analysis of mine roof rock due to moisture variation using a case study, which experienced roof falls. The roof layer types, thicknesses, and properties were determined using extensive lab and field data. A novel numerical model is developed using site information and observed field performance to incorporate: (1) the interaction between roof rock unit beddings, (2) the interaction of roof bolts and rock units, and (3) the effect of moisture increase on roof deformation and failure. Using the proposed methodology, the mine roof layers’ allowable maximum moisture contents prior to failure can be estimated. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Long term stability of underground mine is very critical issue [1,2]. The geomechanical properties of rocks are of great importance in construction of tunnels and underground excavations. Compressive strength is one of the most important geomechanical properties of rocks which can be evaluated using Unconfined Compressive Strength test (UCS) and Axial Point Load (APL) following American Society of Testing Methods and Materials (ASTM) guidelines [3,4]. The strength parameters of rocks are believed to be highly affected by moisture content (MC) variation [5]. It is common to observe moisture content increase in the immediate roof rocks of the mine due to water dripping from water bearing rock layers into the mine entries or flooding of abandoned mines. The effect of moisture content on strength parameters plays an important role in stability of abandoned mines [5]. As water penetrates into the rock, the electronically polar water molecules stick to the surface of rock particles. As a result, the frictional contact surface as well as the bond between the rock particles will be decreased and the strength of rocks will be reduced [6]. The effect of strength reduction due to moisture content increase has been studied and shown previously by several researches [6–10]. In a study by Mohamad et al. [10] a series of ⇑ Corresponding author. E-mail addresses: [email protected] (A. Osouli), [email protected] (B. Moradi Bajestani). http://dx.doi.org/10.1016/j.compgeo.2016.07.004 0266-352X/Ó 2016 Elsevier Ltd. All rights reserved.

96 shale rock samples were collected from Ayer Hitam, Malaysia and were immersed in water for various time periods. The APL tests were performed on weathered shale rock samples with various moisture contents. According to the results, APL indices on average decreased from 4.4 to 1.1 MPa as the MC values increased from 0.9% to 4% (i.e., approximately 24% strength reduction per 1% moisture content increase), respectively. It was also concluded in their study that the rocks with higher APL indices such as shalelimestone or gray shale, are more sensitive to moisture content increase comparing to weaker rocks such as black shale. In another study, Osouli and Moradi Bajestani [11] investigated the moisture sensitivity of various shale rocks collected from roof units of the underground mine case study, which is presented in later sections. In their study, APL and moisture content tests were performed on rock specimen with two different humidity conditions:
Natural MC values (Wn): Determined from samples being tested shortly after collected from underground. This value represents the insitu rock moisture content.
Air-dried MC values: Determined from samples being tested after being exposed to air and loosing moisture. In the Illinois studied mine, Osouli and Moradi Bajestani [11] concluded that shale rocks with APL index of higher than approximately 1000 kPa at their natural MC are more sensitive to moisture content increase comparing to weaker rocks. In that study, the gray shale specimens with APL strength indices of more than

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

1000 kPa at their natural MC state, experienced on average 29% reduction in APL per 1% increase in moisture content. However, for gray shale rocks with APL values of less than 1000 kPa at their natural MC, the rock cores experienced 21% reduction in their strength per 1% MC increase. The black shale rocks of the studied mine that had APL indices of more than 1000 kPa at natural MC, experienced on average 19% reduction in APL per 1% increase in moisture content. However, among the studied black shale rock samples, weaker specimens do not show noticeable strength sensitivity to moisture content variation. Finally, the sandy shale specimens of the studied mine show similar behavior toward moisture content variation regardless of their APL strength magnitude at their natural moisture content. The APL indices of sandy shale specimen decrease on average 17% by 1% MC increase. In many underground mines the roof is stable in active mine panels. However, during the mine life, roof failures are observed in those panels after mining due to humidity absorption or further water exposure [12]. The initial moisture content of immediate rock units is important. It is common that these moisture contents decrease during mining operation because of the mining drainage and/or ventilation, and then increase when the mine is abandoned. For example, in case of the studied mine by Osouli and Moradi Bajestani [11], the average moisture contents of roof black shale and sandy shale units prior to mining were 6.0%, and 5.7%, respectively. After mining operation, the average moisture contents of 4.5% and 3.7% was measured for black shale and sandy shale rock units in the active panels, respectively. However, in the abandoned parts of the mine, where there is no ventilation, the moisture contents of the rock units have increased to higher than pre-mining levels. A better understanding of moisture variation effect on stability of underground mines, not only result in safer and more efficient design of roof control measures but also allows to predict the potential hazardous areas and eventually prevent the fatalities. This paper aims to study the effect of moisture content variation on stability of underground openings and predict roof fall incidents in the same underground coal mine case study that investigated by Osouli and Moradi Bajestani [11]. They conducted strength tests on shale rock samples from same rock units but with different moisture contents. Using the extensive test results database, Osouli and Moradi Bajestani [11] developed correlations between MC variation and UCS index for black shale, sandy shale and gray shale rock units existing in the studied underground coal mine located in the Illinois Coal Basin. In this paper, those correlations are used to

153

determine the compressive strength of the roof rock units as the moisture content of the mine strata is increased. To study the effect of MC variation on the stability of the moisture sensitive roof rocks and to predict failure, a novel numerical analysis was conducted. The required geomechanical and rockmass parameters for the mentioned model are extracted using laboratory and field test results. The developed models were verified using underground information collected during mine visits for both stable and rooffall locations in the studied mine. Furthermore, the verified model was used to study the effect of influential parameters such as room entry width and thickness of immediate weak roof rocks on roof stability of the mine. The presented methodology can be used in the design of roof support in mines, dealing with similar roof conditions, to avoid major roof falls. Furthermore, it is discussed how the presented results could identify the timing for the development of mine panels, predict the failure incidents in advance, and keep the main entries open during the mining activity in the studied mine, which has very weak moisture sensitive roof rocks. 2. Stability of moisture sensitive roof units The Illinois Coal Basin covers most of the Illinois State extending into southwestern Indiana and western Kentucky [13]. This basin mainly contains Paleozoic sediment and dolomite rock units followed by limestone, shale and sandstone [13]. The mine roof condition for this case study was examined during a couple of underground mine inspections at two Locations of A and B, shown in Fig. 1. The mine is at depth of 73 m from the ground surface. The mining method was room-and-pillar with typically 19 m pillar sizes and entry widths of 5.5 m, which results in an extraction ratio of about 40%. Location A with stable roof is in proximity of Location B, which had massive roof failure. Fig. 2 shows the condition of roof at these locations. The mine visit locations are selected based on their proximity to the available boring locations, accessibility from underground entries and availability of rock laboratory test data related to their immediate roof units. The room height, thickness and rock type of immediate roof and floor units were determined based on the boring logs and underground inspections and are shown in Fig. 3. Several previous studies investigated on roof failure criteria. The stability and failure initiation of coal mine roof can be detected by monitoring the roof deformation. Maleki and Owens [14] has

Fig. 1. Underground coal mine map showing Locations A and B.

154

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

Fig. 2. The underground coal mine roof condition at stable roof location (left) and roof fall location (right).

3.00 m 2.0 m Gray Shale with Yellow Veins

3.75 m

Location A

2.25 m 0.75 m

Location B

2.00 m

Coal Sandy Shale

1.0 m 3.00 m

Black Shale Claystone Limestone

Immediate Roof

2.0 m 1.0 m 2.0 m

0.5 m 0.5 m

Fig. 3. The geological information related to Locations A and B obtained from the boring logs and underground inspections.

recorded the movement of the roof in two locations of a room and pillar underground coal mine located within the Blackhawk Formation in the Wasatch Plateau, Utah. It was observed that the roof has failed after it has experienced a vertical displacement of 8 cm. In another study conducted by Gale et al. [15], the interaction between the underground coal mines roof and the roof bolts during excavations were investigated. The site of the case study was at the Emerald Mine, located in Greene County, Pennsylvania which is in Northern Appalachian Coal Basin. Their study showed that the immediate 4.8 m roof was affected by mining and maximum deformation was 5.5 cm. In addition, Sarathchandran [16] studied the effect of high horizontal stresses on stability of excavation in coal rock mediums by performing a three dimensional numerical analysis. The numerical model parameters were obtained from two mine fields from UK and India. The study showed that about 4 cm displacement of the immediate roof unit initiates the failure of the mine roof. This corresponds to 1.3% vertical strain according to the thickness of the failed roof units. In the current study, after consultation with the studied mine officials and operator and considering the aforementioned investigations in other underground mines, an immediate roof deformation of 4 cm is considered excessive movement, which results in unstable roof condition. Therefore, according to the thickness of immediate roof units contributing to the mine roof failure (see black shale and sandy shale units in Fig. 3), a roof vertical strain of more than 1.3% is considered as roof failure or a roof that is about to fail.

2.1. Geological data and underground inspection The geology of the roof at Locations A and B are shown in Fig. 3 based on the boring logs and underground inspections. According to the geological data obtained from A-49 borehole (shown in Fig. 1), the coal seam thickness (room height) is 2 m and is underlied by 1 m thick claystone layer (see Fig. 3). The rock units of black shale (BSH), sandy shale (SSH), limestone (LS), claystone (CS), sandy shale (SSH) and limestone (LS) with thicknesses of 0.75, 2.25, 1, 3.75, 2 and 3 m are located above the coal seam, respectively. Similar rock units are detected during underground inspections at Location A. However, in case of Location B, a localized 0.5 m thick weak yellowish gray shale unit with yellow veins (GSH-Y) is detected during underground inspections. The mentioned unit is located 0.5 m above the coal seam and was not reported in boring logs (see Fig. 3). It is believed that the existence of a very weak and localized thin layer with high moisture content at Location B has caused the major difference between roof performance at Locations A and B. Besides boring A-49, there were more than 40 boreholes drilled at the mine site and the rock cores recovered from floor, coal, and roof layers. Therefore, an extensive dataset of rock lab tests were compiled for this case study. The structural integrity, the surface quality, moisture content and slake durability index of roof rockmass units at Locations A and B were determined based on lab tests and underground inspections. The laboratory tests included MC, unconfined compressive strength (UCS), slake durability, point

155

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166 Table 1 Summary of geological information for the stable roof Location A. Roof rock unit

Thickness (immediate unit) (m)

(Wn) (%)

Slake durability test (%)

Moisture sensitivity

Spacing of discontinuities (m)

Intensity of discontinuities

Limestone Sandy shale Black shale

1.0 2.25 0.75

1.8 3.7 4.5

95 83 59

Slightly sensitive Moderately sensitive Severely sensitive

0.6–1.8 0.06–0.2 0.06–0.2

Low High High

Table 2 Summary of geological information for the roof fall Location B. Roof rock unit

Thickness (immediate unit) (m)

(Wn) (%)

Slake durability test (%)

Moisture sensitivity

Spacing of discontinuities (cm)

Intensity of discontinuities

Limestone Sandy shale Black shale Gray shale with yellow veins

1.0 2.0 0.5 0.5

1.8 4.9 5.9 6.0

95 83 59 19

Slightly sensitive Moderately sensitive Severely sensitive Severely sensitive

0.6–1.8 0.06–0.2 0.06–0.2 Less than 0.06

Low High High Very high

load, and indirect tensile strength (ITS) tests. Tables 1 and 2 show the summary of geological data for the roof layers at these locations. The intensity of discontinuities in the rockmass was determined based on the spacing between discontinuities observed during underground inspections. Also, the surface quality of the rockmass and their moisture sensitivity were determined based on the slake durability test results. According to Hoek [17], the rocks with slake durability indices of lower than 80% and between 80% and 92% are considered severely and moderately moisture sensitive rocks, respectively. 2.2. Numerical analysis of roof stability Numerical simulation has been widely used for analyzing coal mine stability and mine roof performance [16,18–22]. Due to the jointed structure of rock masses, the generalized version of

Hoek-Brown (HB) constitutive model has been applied for numerical simulations by previous researchers for modeling coal mines [16,18,21,23]. Hoek and Brown [24] presented a failure criterion to represent the rockmass with discontinuities; therefore, it is used for modeling the rockmass units in the coal mines strata. In this paper, the numerical modeling is performed by FLAC2D Version 7.0 [25] software in order to simulate the behavior of the roof. Since the mine strata is a jointed medium, the generalized Hoek-Brown constitutive model is used. The stability of the mine roof span at various humidity conditions is simulated and performance of the roof span toward moisture content variation is analyzed. Numerical analyses were initially conducted to simulate the performance of the roof at the MC values recorded during the mine visit. For this purpose, the rock samples collected during mine visit are tested for moisture content. These obtained moisture contents

23.25 m

Roof bolt #1 Roof bolt #2 Roof bolt #3

2.0 m

Coal

Pillar

5.5 m

3.75 m 1.0 m

2.0 m

2.25 m 0.75 m 2.0 m 1.0 m

Pillar

Room

Sandy Shale Black Shale

10.0 m

0

Claystone

4 Meters

Limestone

24.5 m Fig. 4. Modeled cross section related to Location A.

8

156

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

represent insitu MC values and are referred to as ‘‘MC values during mine visit” (Wn) in this article. The analyses results were evaluated using underground observation at Locations A and B (see Fig. 1 for locations). In the next modeling step, the moisture content of the roof mine layers are incrementally changed and the behavior of the roof is analyzed. A typical cross section of room and pillar layout of the mine is modeled. The modeled cross section at Location A is shown in Fig. 4. An equivalent overburden pressure was applied at the top boundary of the modeled cross section to simulate the overburden stress. According to Sarathchandran [16], modeling 10 m of roof and floor are sufficient to eliminate the effect of top and bottom boundary conditions. In order to minimize the effect of boundary conditions on simulation results, the total thickness of mine roof and floor are selected as 33 and 10 m, respectively. Claystone and limestone layers in the mine floor are considered as 1 m and 10 m thick, respectively. The immediate roof consists of two rock layers of black shale and sandy shale with thicknesses of 0.75 and 2.25 m, respectively. The mentioned rock units are considered to be overlaid by limestone, claystone, and sandy shale layers with thicknesses of 1.0, 3.75, and 2 m, respectively (see Fig. 4). In case of Location B, the only difference in roof strata is in the immediate roof, which is formed by three rock units of BSH, GSH-Y, and SSH with thicknesses of 0.5, 0.5 and 2 m (see Fig. 3). The other layers are modeled similar to Location A. 3. Numerical modeling methodology 3.1. Rock layers modeling methodology 3.1.1. Constitutive model The generalized Hoek-Brown constitutive model was assigned to the mine roof and floor layers in the simulations [24]. The generalized Hoek-Brown constitutive model representing a non-linear correlation between the minimum and maximum confinement stresses at failure is shown in Eq. (1) [24].




r01 ¼ r03 þ rci mb

r03 þs rci

a ð1Þ

where a, s and mb are the Hoek-Brown constants depending on the rockmass, rci is the uniaxial compressive strength of the intact rock and r1 and r3 are the maximum and minimum effective stresses at failure, respectively. The constant parameters of a, s and mb are highly dependent on the characteristics and types of the rockmass. These parameters are determined using Eqs. (2)–(4) and all are based on the Geological Strength Index (GSI) [24].


 GSI  100 mb ¼ mi þ exp 28

ð2Þ


 GSI  100 s ¼ exp 9

ð3Þ

( a¼

GSI 0:65  200 ðfor GSI less than 25Þ

0:5

ðfor GSI equals or more than 25Þ

ð4Þ

where the GSI is an empirical index, which describes surface quality and structure of the rock unit. GSI should be determined based on guides provided by Hoek and Brown [24] and ranges from 10 to 80 based on the quality and integrity of the rockmass. The parameter of mi is a constant value depending on the rock type, which ranges from 4 to 33. Hoek and Brown [24] suggested mi based on the rock type; however, the provided guides are very general. Therefore, the proposed methods by Cai [26] was used to identify mi. Since Hoek-Brown failure criterion is highly dependent on compressive strength of rocks, Cai [26] suggested using the ratio of com-

pressive strength index to tensile strength index to determine mi (see Eq. (5)).

mi ¼

rc rt

ð5Þ

where rc and rt are the compressive and tensile strength of rock. In this study, the mi constant parameter was determined using the average UCS and ITS test results of each rock layer of the studied mine as a representative of rc and rt, respectively. Furthermore, per Hoek-Brown [24], Eq. (6) is suggested to estimate Young’s modulus of rocks. The rci is the uniaxial compressive strength index of the intact rock pieces, and GSI is the Geological Strength Index of the rockmass. According to Osouli and Moradi Bajestani [11] and Osouli et al. [27], the rock’s APL and UCS indices relies on their MC values and consequently the rock’s modulus of elasticity is dependent on MC variation. This dependency was considered in this study.

rffiffiffiffiffiffiffiffiffi

Em ðGPaÞ ¼

rci

10ð 100

GSI10Þ 40

ð6Þ

3.1.2. Hoek-Brown parameters The Hoek-Brown parameters pertinent to each roof unit based on geomechanical laboratory tests and underground field inspections are shown in Table 3. The presented parameters in this table correspond to roof rock units with the moisture contents recorded during mine visit (Wn) at Location A. According to the proximity of the A-49 borehole to the studied location, the strength test results of BSH and SSH recovered samples from borehole A-49 were used. However, there were not any available LS and CS rock core specimen from the boring logs in proximity of the studied locations. Therefore, for the LS and CS rock units, the average of the relevant strength tests performed on specimen collected from other borings across the mine site were used. The GSI indices of rock units modeled in this study were determined following the Hoek and Brown [24] procedure using the underground inspection information collected from the studied mine. For this purpose, the roof rock layers were initially characterized based on their surface quality, structure integrity, and slake durability test results (see Table 4). According to Table 4, GSI indices of 55, 40, 30 and 20 were determined for rock units of LS, SSH, BSH and GSH-Y of the studied mine, respectively, which are in accordance with the GSI indices used by previous researchers for similar rock types [28,29]. There was not adequate field information available regarding the structural integrity and the surface quality of the claystone rock unit. Therefore, a GSI index of 20 was assigned to CS rock units according to Masada and Han [28] and Marinos and Hoek [29]. The values related to constant parameter of mi were determined according to Eq. (5) using the rock’s geomechanical properties database compiled for the studied mine by Osouli and Moradi Bajestani [11]. According to Osouli and Moradi Bajestani [11], the average UCS values for black shale, sandy shale and limestone roof units of the case study coal mine are 12.9, 25.0 and 106.5 MPa, respectively. Also, the average ITS values for BSH, SSH and LS rock units are 2.1, 3.2 and 7.0 MPa, respectively [11]. Therefore, according to Cai [26] and using Eq. (5), mi values of 6.1, 7.9 and 15.2 are determined for black shale, sandy shale and limestone roof units, respectively. Due to lack of information, the constant parameter of mi for claystone rock units could not be determined using Eq. (5). Therefore, mi of 4.0 was selected per Hoek and Brown [24]. Finally, the constant parameters of s, a and mb are determined using Eqs. (2)–(4) (see Table 3). Following the same procedure, the Hoek-Brown parameters for Location B are identified and shown in Table 5. The Young’s moduli used in this case study at both Locations A and B are determined according to the laboratory

157

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166 Table 3 Hoek-Brown parameters used for simulation of roof units at the MC values recorded during mine visit (Wn) (Location A). Rock type

MC (%)

UCS (MPa)

mi

GSI

mb

S

a

E (GPa)

n

K (GPa)

G (GPa)

Limestone Sandy shale Black shale Claystone Coal [18]

3.0 3.7 4.5 8.4

106.4 39.6 28.5 18.0 24.7

15.2 7.9 6.1 4.0

55 40 30 20

3.037 0.928 0.497 0.230 1.470

0.0067 0.0013 0.0004 0.0001 0.07

0.5 0.5 0.5 0.55 0.5

25.5 7.9 1.5 5.9 3.0

0.18 0.2 0.2 0.22

13.3 4.4 0.8 3.5 2

10.8 3.3 0.6 2.4 1.2

Table 4 The GSI parameters determined based on structural integrity and surface condition of the various roof rock units. Roof rock unit

Moisture sensitivity

Spacing of discontinuities (cm)

Intensity of discontinuities

Surface condition [24]

Rock unit structure [24]

GSI index

Limestone

Slightly sensitive Moderately sensitive Severely sensitive Severely sensitive

0.6 to 1.8

Low High

0.06 to 0.2

High

Less than 0.06

Very high

Very blocky (Partially disturbed rock mass with more than four discontinuity sets) Blocky (Angular blocks formed by many intersecting discontinuity sets) Blocky (Angular blocks formed by many intersecting discontinuity sets) Disintegrated (heavily broken and poorly interlocked rock mass)

55

0.06 to 0.2

Good, slightly moisture sensitive Fair, moderately moisture sensitive Poor, severely moisture sensitive Poor, severely moisture sensitive

Sandy shale Black shale Gray shale with yellow veins

40 30 20

Table 5 Hoek-Brown parameters used for simulation of roof units at the MC values recorded during mine visit (Wn) (Location B). Rock type

MC (%)

UCS (MPa)

mi

GSI

mb

S

a

E (GPa)

n

K (GPa)

G (GPa)

Limestone Sandy shale Black shale Gray shale with yellow veins Claystone Coal [18]

3.0 5.2 5.9 6.0 8.4

106.4 39.6 28.5 28.0 18.0 24.7

15.2 7.9 6.1 5.0 4.0

55 40 30 20 20

3.037 0.928 0.497 0.288 0.230 1.470

0.0067 0.0013 0.0004 0.0001 0.0001 0.07

0.5 0.5 0.5 0.5 0.55 0.5

25.5 7.9 1.5 5.1 5.9

0.18 0.2 0.2 0.2 0.22

13.3 4.4 0.8 2.1 3.5 2

10.8 3.3 0.6 1.6 2.4 1.2

tests performed on the samples collected from the mine. In both cases of Locations A and B, the Poisson ratios are determined according to Gercek [30]. The Hoek-Brown constitutive model is also used for modeling the coal pillars. All the Hoek-Brown modeling parameters (i.e. UCS, mb, s and a) used in this study for coal pillars are determined according to Esterhuizen et al. [18]. Esterhuizen et al. [18] calibrated the coal Hoek-Brown parameters using insitu coal strength tests performed in room and pillar underground coal mines and observed coal pillar performances in Appalachian, Illinois and Western US coal fields as well as empirical methods presented by Bieniawski [31]. Since the focus of current study is on roof layers’ performance and there was limited information available about the coal properties of the studied mine, the extracted coal geomechanical parameters by Esterhuizen et al. [18], which are relevant to variety of US coal fields, are deemed representative of pillars in the studied case and used herein.

3.1.3. Roof layer discontinuities The studied mine has layered shale roof units with weak bonding between the beddings. In order to properly represent the observed discontinuities, the immediate BSH and SSH roof units are divided into 3 and 2 sublayers, respectively. Interaction surfaces are defined to simulate the interaction between the consecutive beddings. Fig. 5 shows the bedding layers defined within the BSH and SSH immediate roof units. Furthermore, the interfaces between the BSH and SSH units (i.e., Interfaces 6 and 7) and between the SSH and LS units (i.e., Interfaces 10 and 11) are shown in Fig. 5. Simulating the interaction between various roof layers as well as sublayers, allows to monitor the displacement of each bedding layer individually and to detect the separation of roof units

from each other. According to FLAC2D manual, the interaction between adjacent interfaces can be defined based on the bulk (kn) and shear (ks) moduli of the weaker rock layer using Eq. (7).



kn ¼ ks ¼ 10 

K þ 43 G Dzmin



ð7Þ

where the G and K are the shear and bulk modulus of weaker layer and can be calculated according to Eqs. (8) and (9) [32].

K¼

E 3  ð1  2mÞ

ð8Þ

LS Layer SSH Second Bedding Layer SSH First Bedding Layer BSH Third Bedding Layer BSH Second Bedding Layer BSH First Bedding Layer

Interface # 11 Interface # 10 Interface # 9 Interface # 8 Interface # 7 Interface # 6 Interface # 5 Interface # 4 Interface # 3 Interface # 2 Interface # 1

1.00 m 1.12 m 1.12 m 0.25 m 0.25 m 0.25 m

Room/Coal

Fig. 5. The bedding layers and interfaces modeled for simulation of the mine roof at Location A.

158

G¼

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

E 2  ð1 þ mÞ

ð9Þ

where the E and m are the Young’s moduli and Poisson ratio of the weaker layer in contact with the interface and Dzmin is the width of the smallest zone in contact with the interface, which in this study is 0.25 m. Tables 3 and 5 show the assigned shear and bulk modulus of the rock units related to Locations A and B. 3.2. Rock bolts modeling methodology The roof support system plays a significant role in supporting the immediate roof units and preventing roof discontinuities from being detached. The most challenging part of modeling is simulation of the bolt and rock interaction. Rock bolts are the common roof support system used in underground coal mines including this case study. Fig. 6 shows the exaggerated sketch of the simulated mine roof strata with and without roof bolts. As shown in this figure, the interaction between black shale beddings is not strong enough and these layers typically fail in absence of roof bolts. For modeling the roof rock bolts, the strength of the steel, the diameter and length of the bolts are assigned according to as-built drawings as well as field observations. The interaction of the grouted rock bolts with surrounding rock medium, are typically represented by the rock bolt pull-out tests [33,34]. In this case study, due to lack of the pull-out tests results, the empirical correlation suggested by St. John and Van Dillen are used for characterizing and calibrating the parameters necessary for simulating the interaction between grouted rock bolts and surrounding rock medium [33]. In the studied mine, five typical 1.8 m long and 22 mm (7/8 in.) diameter fully grouted tension rebar SRD Grade 75 rock bolts per row with 1.2 m spacing were utilized at Locations A and B. The utilized roof support is simulated by modeling five roof rock bolts in the entry (see Fig. 4). Two sets of input parameters are required to simulate the roof bolts. The first set of parameters are related to the rock bolts such as bolt nominal diameter (a), elastic modulus (E), spacing (s) and tensile yields strength (tyield). The second set of parameters is related to the interaction between the grout and surrounding rock medium including shear stiffness (kbond) and cohesion of the interaction (sbond). The shear stiffness of the grout and rock interaction (kbond) is determined according to Eq. (10) [32].

kbond ¼

2pG 10  lnð1 þ 2t=DÞ

ð10Þ

speak ¼ sl  Q B

ð11Þ

sbond ¼ pðD þ 2tÞspeak

ð12Þ

where sl is approximately one-half of the weaker uniaxial compressive strength of rock or grout and QB is the quality of the bond between the grout and rock and it is considered one if the bond is perfectly intact [33]. The underground inspections from the roof fall areas of the studied mine suggest that the failure of the bolt-grout-rock system has occurred along the grout-rock interface as the grouted rock bolts remained intact even after roof failure. In the case of the studied mine, assumption of a perfectly intact bond between grout and the surrounding rock (i.e., QB = 1) in the simulations, results in no failure at Location B of the mine. Therefore, in order to calibrate the numerical model, bond quality parameter (QB) had to be reduced to match the field performance. After considering the pull out test results conducted by Gale et al. and Zipf [15,35] on similar rock types, the QB was changed and the simulation results were compared with field observations at Location A and B. This process was repeated until the simulation results were in agreement with field observations. Consequently, a representative Sbond parameter of 77 kN/m for roof rock units at Wn moisture contents was determined. It should be mentioned that since Eq. (12) is dependent on the compressive strength of the intact rock, the Sbond is indirectly dependent on the moisture content. Therefore as the moisture content increases, the Sbond decreases. 4. Roof performance results 4.1. Roof performance at Location A 4.1.1. Performance of roof rocks at natural MC In order to analyze the performance of the mine at Location A, the moisture contents recorded for roof rock units during the mine visit were initially used. Fig. 7 shows the vertical displacement of interfaces along the roof span. According to Fig. 7, the underground coal mine roof experiences a maximum deformation of approximately 7 mm (approximately 0.2% strain) at the center of very first BSH bedding layer. Also, it is observed that the Interface No. 2 (see Fig. 5 for locations) has experienced a vertical deformation of approximately 6 mm, nearly 1.5 mm larger than Interface No. 3, which means that the first and second bedding layers of BSH have been detached. On the other hand, the relative vertical deformation of Interface Nos. 8 and 9 (overlapped on each other in Fig. 7) are minimal indicating that the SSH bedding layers are not separated from each other. Similarly, the relative vertical deformation of

Roof bolt

where G is the minimum shear modulus of grout and rock, D is reinforcing diameter and t is annulus thickness [33]. The borehole diameter is 35 mm (1–3/8 in.), which results in annulus thickness (t) of 6.4 mm (2/8 in.) [15]. The parameter of kbond represents the shear stiffness of the grout or rock, whichever is less. Since the shear modulus of resin grout is more than 2.25 GPa [34] and also more than the shear modulus of black shale in the studied mine, the shear modulus of black shale was used to obtain kbond of 1.32 ⁄ 109 N/m/ m, which is in the range suggested by Zipf [35] for similar conditions.

According to FLAC manual [32], the maximum shear strength of the grout-rock interaction per unit length is a function of the cohesive strength of the interaction (i.e., Sbond). This value can be determined using pull-out tests. In case that such test result is not available, the maximum shear force per unit length (kN/m) can be estimated from the peak shear strength and using Eqs. (11) and (12) [33].

Coal Sandy Shale Black Shale

Pillar

Pillar

Pillar

Pillar

Claystone Limestone

Floor

Floor

Fig. 6. A sketch showing the exaggerated vertical displacement of roof rock beddings without (left) and with (right) roof rock bolts.

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

159

Distance along the roof span width from left pillar (m) 0.5

1.5

2.5

Vertical deformation (mm)

0 -1 -2 -3 -4 -5 -6 -7 -8

3.5

4.5 Interface #1 Interface #2 Interface #3 Interface #4 Interface #5 Interface #6 Interface #7 Interface #8 Interface #9 Interface #10 Interface #11

Fig. 7. Vertical deformation of interfaces modeled inside the mine roof at Location A at MC values recorded during mine visit (Wn).

Fig. 8. The axial tensile stress inside the roof bolts Nos. 1, 2 and 3 at Location A with Wn MC values recorded during the mine visit (see Fig. 4 for bolt locations).

Interface Nos. 10 and 11 are negligible showing that the upper boundary of SSH layer is deforming as much as the lower boundary of LS layer. The small displacement obtained from numerical results shows that the Location A is stable and vertical strain is less than 1.3%. Fig. 8 shows the axial tensile stress inside roof bolts Nos. 1, 2 and 3. Roof bolt No. 1 is the closest to the coal pillar and roof bolt No. 3 is at the center of the roof span (see Fig. 4 for locations). The maximum and minimum stresses are obtained for the rock bolts Nos. 3 and 1, respectively. The maximum tensile stress of 110 MPa is recorded for the roof bolt No. 3, which is considerably less than the tensile yielding stress 517 MPa for SRD grade 75 steel. It is also worth noting that, major tensile stress concentration is observed at a roof bolt length between 0.6 and 0.8 m from the coal seam. The reason is that there is a transition from BSH layer to SSH at about 0.75 m top of the coal seam. Therefore, all the axial force due to weight of the BSH unit will be transferred to the rock bolt anchored in SSH. 4.1.2. Effect of MC increase on strength parameters It is commonly observed that the immediate roof layers of Illinois underground coal mines experience moisture content increase after mining. According to Fig. 3, at Location A the first two immediate roof rock layers are BSH and SSH. These layers protect the upper layers from moisture exposure that comes from the entry. In order to represent the moisture content increase in the simulations, the MC values of weak and moisture sensitive SSH and BSH rock units are increased incrementally while the moisture content of other roof layers can be assumed constant. Also, the moisture content of the coal pillars and mine floor is assumed to be constant since the scope of this study is to analyze the moisture sensitivity effects of just the immediate roof layers on roof stability. The MC values of BSH and SSH roof layer are increased incrementally until the mine roof shows severe vertical strain or displacement (i.e., more than 1.3% strain or 40 mm displacement).

As it was discussed before, APL index of rock units decrease as their MC values increase [11]. According to Osouli et al. [27], the APL index of rocks has a linear relationship with their UCS index. Therefore, the uniaxial compressive strength of the intact rock (rci) is affected by moisture content variation. Fig. 9 shows the effect of moisture content variation on the UCS strength index (rci) of black shale and sandy shale layers existing in the case study coal mine based on the experimental results suggested by Osouli and Moradi Bajestani [11] and Osouli et al. [27]. The corresponding UCS values for each MC increment from this figure are used to modify the assigned geomechanical properties and simulate roof performance at various moisture conditions. According to Fig. 9, the rate of strength reduction decreases for sandy shale and black shale layers which have UCS values of less than 16 MPa. As the UCS index changes, the Young’s modulus is updated per Eq. (6). The parameters of mi, GSI and Poisson ratio are independent from the MC variation since they rely on the type of rock, the structure and surface condition of rock layer. 4.1.3. Effect of increasing MC on performance of roof at Location A Fig. 10 shows the vertical displacements of roof span when MC contents of the immediate BSH and SSH roof units are increased from 4.5% and 3.7% (MC values during mine visit) to 9.5% and 8.7%, respectively. As expected, in all cases the maximum roof deformation happens at the middle of the roof span. The roof span experiences a larger vertical displacement as the moisture content of roof rock unit increase. However, for all cases shown in Fig. 10, the vertical displacement of roof span is less than 4 cm and the vertical strain is less than the strain criteria of 1.3%. Fig. 11 shows the maximum vertical deformation of the roof span at various moisture conditions. The vertical displacement of 30 mm equivalent to about 1% strain was obtained when BSH and SSH have moisture contents of 9.5% and 8.7%, respectively. These moisture contents are approximately 5% more than the MC values recorded during the mine visit (Wn). If moisture contents of more than 9.5% and 8.7% are assigned to BSH and SSH, the roof will experience severe vertical strains and continuous displacements.

Fig. 9. The effect of moisture content variation on the UCS strength index (rci).

160

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

Fig. 10. Vertical displacements along the roof span for various modeled moisture conditions (Location A).

Fig. 11. Maximum vertical displacement of roof span and maximum tensile strength in middle roof bolt from simulations conducted for Location A.

The maximum moisture content values that the BSH and SSH roof units can have without causing roof failure are referred to as ‘‘ultimate MC values” (Wu) herein. According to the results presented in Fig. 11, the ultimate MC values are 9.5% and 8.7% in case of the studied mine at Location A. The ultimate MC values (Wu) should be taken into consideration for design and operation of the mine. It is essential to make sure that the moisture content of BSH and SSH roof units do not exceed ultimate MC values in order to ensure the stability of the mine and the safety of operators. Using the methodology discussed herein, the mine roof layers’ ultimate moisture contents can be estimated. This information along with periodic moisture content check of roof rocks can be used as a monitoring or prediction tool to identify the roof failure incidences in advance as well as come up with alternative moisture control systems. The roof failure occurs because the rock bolts detach from the surrounding rock medium [36]. The capacity of bonding between the rock bolt grout and surrounding rock unit is affected by the rock strength. As the roof units are exposed to moisture, their compressive strength decrease. Therefore, the bonding between the rock bolt and roof unit weakens which results in detachment of rock and bolts and progressive mine roof failure. In order to study the discussed failure mechanism, the maximum tensile stress inside the middle rock bolt (i.e., Bolt No. 3) is shown in Fig. 11. According to this figure, the maximum tensile

stress of middle rock bolt reaches its peak at 270 MPa when the MC values of BSH and SSH units are increased to 8.5% and 7.7%, respectively. At greater MC values, the rock medium becomes weaker and the connection between the roof bolt and surrounding rock medium deteriorates. Therefore, less load is transferred to the stable roof causing significant increase in downward deformation of roof span and finally failure. Furthermore, according to Fig. 11, the vertical deformation rate of roof span increases significantly when BSH and SSH MC values become larger than 8.5% and 7.7%, respectively. 4.2. Roof performance at Location B The numerical study of Location A showed that the mine roof will be stable if the moisture content of the immediate BSH and SSH roof units are less than 9.5% and 8.7%, respectively. The moisture content tests conducted during mine visits suggest that the immediate BSH and SSH roof units at the failure location (i.e., Location B) have MC values of 5.9% and 4.9%, respectively. Therefore, it was expected to have stable roof at this location. The geometry of mine layout and geological layers at Locations A and B are very similar, except a localized and thin gray shale layer with yellow veins (GSH-Y) was observed in the immediate roof at Location B (see Fig. 12). The moisture content of the GSH-Y samples collected from Location B was 6%. In order to identify the reasoning of failure

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

161

Fig. 12. Cross section showing the black coal seam at the bottom and yellowish gray shale layer with yellow veins (GSH-Y) at the top at roof fall Location B.

at Location B, a separate simulation was conducted at Location B to include the effect of about 0.5 m thick GSH-Y layer. The HoekBrown parameters used for simulation of roof rock units present at Location B are shown in Table 5. The Hoek-Brown parameters used for GSH-Y rock unit are determined based on the geomechanical tests and underground visit information. The slake durability index related to GSH-Y rock unit is measured as low as 19% which suggests that the mentioned rock unit is severely moisture sensitive. Also, the observed spacing between the discontinuities in the GSH-Y rock unit was less than 6 cm and the length of the discontinuities were measured more than 10 m. Therefore, according to the slake durability test results as well as the underground inspections data, GSI index of 20 is assigned to GSH-Y unit. The Ball Peen Hammer tests are performed on the GSH-Y samples collected from the Location B and the UCS index of 28 MPa is determined for this unit. Based on the above mentioned parameters and using Eqs. (2)–(9), the Hoek-Brown parameters related to GSH-Y unit are determined and presented

in Table 5. Similar to the BSH and SSH roof units, the strength reduction of GSH-Y roof units due to incremental moisture increase is determined according to the experimental results presented by Osouli and Moradi Bajestani [11]. Fig. 13 shows the maximum displacement of the roof at Location B when various moisture conditions were assigned to roof layers in the numerical analysis. According to this figure, the ultimate MC values that the mine roof can take prior to failure is 5.7%, 5.7% and 4.9% for BSH, GSH-Y, and SSH roof units, respectively. These values were less than the MC values recorded during the mine visit which were 5.9%, 6.0% and 4.9% for BSH, GSH-Y, and SSH, respectively. By assigning MC values higher than the ultimate ones to the roof rock units, the immediate roof rock units are detached from roof bolts and the roof experiences continuous displacement and vertical strains of more than 1.3%. Fig. 11 shows that at Location A, the rate of roof vertical deformation of the studied mine increases significantly when moisture contents of BSH and SSH are more than approximately 8.5% and

Fig. 13. Maximum vertical displacement of roof span at Location B for various moisture conditions of the roof.

162

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

7.8%, respectively. However, at Location B, the increase in rate of roof vertical deformation is detected when moisture content of BSH and SSH layers exceed 5.5% and 4.7%, respectively (see Fig. 13). The MC values at which the deformation rate of mine roof will have a sudden increase are defined as critical MC values herein and are further discussed in Section 6.2. By monitoring and detecting the critical MC values throughout the mine, the areas prone to failure can be detected well in advance and necessary roof fall prevention measures can be implemented. 5. Numerical model and field observations

Vertical stress inside the pillar (MPa)

The results of the numerical models are compared with the field observations. In refer to coal pillars, Fig. 14 shows the vertical stress inside the pillar at 0.5, 1 and 1.5 m above the floor level. According to this figure, in all cases the location at which the pillar is experiencing the maximum vertical stress is not at the immediate face of the pillar but is located inside the pillar. At 0.5 and 1 m above the room floor, the maximum vertical stress in pillars occurs within a distance of approximately 1.5 m from the face of the pillar. The maximum vertical stress inside the pillar at 1.5 m above floor level is observed within a distance of nearly 0.75 m from the pillar face. This finding is in agreement with the field observations reported by Esterhuizen et al. [18] for underground coal mine case studies located in Southern Appalachian Basin and Blackhawk Formation in Utah. According to field observations reported by Esterhuizen et al. [18], the maximum vertical stress occurs not at the face of pillar but up to 2 m inside the pillar. This stress concentration will result in spalling of coal pillars in long term and this phenomenon was observed around many pillars of the studied mine. In refer to roof layers stability, the modeled performance of the roof at Locations A and B of the studied mine matches reasonably with the roof performance observed in the field. In case of Location B, the ultimate MC values estimated by numerical analyses are less than the recorded MC values during mine visit; therefore, the roof was expected to fail. For better comparison, it would have been helpful if the MC values of the roof units were periodically measured prior to failure. On the other hand in case of the Location A, the recorded MC values of immediate BSH and SSH were measured to be 4.5% and 3.7%, respectively, while the ultimate MC values obtained from the numerical model were 9.5% and 8.7%, respectively. The fact that the ultimate MC values obtained from developed model are about twice more than the recorded MC values indicates that the roof should not experience roof fall. The field observation at Location A confirms that the roof is in stable condition and has not shown any indication of failure. Furthermore, the

ultimate MC values of the BSH and SSH at Location B (i.e., 5.7% and 4.9%) are almost 4% less than the ones corresponding to Location A (i.e., 9.5% and 8.7%). This indicates that the mine roof at Location B is more vulnerable to moisture content increase and more prone to failure due to existence of the GSH-Y rock unit in the immediate roof. It is worth mentioning that there are many sources, which may result in uncertainties in roof analyses of the mine. These include the mischaracterization of roof layers, misinterpretation of rock thicknesses, lack of geomechanical properties of the rock or the bond between rocks and the bolts, and erroneous laboratory test results. However, the most challenging one to overcome is the mischaracterization of the roof layers because the geology of the roof is significantly varied across a mine area and it is very costprohibitive to characterize all roof areas prior to mining. The uncertainty in regards to geomechanical properties of a roof layer can be overcome by conducting sufficient laboratory test result on quality core samples. In case of scattered results, the representative parameters can be extracted using statistical analyses. The effect of geometry of the mine or roof layers are discussed in next section. 6. Parametric study 6.1. Roof thickness and room entry width factors The roof rock layer’s thicknesses are different across the mine site. The variation in rocks layers’ thickness is very common in rock deposits and is expected in most underground or tunneling work. The entry widths may also vary in different mines and even in different panels of a single mine. Therefore, a series of additional numerical simulations were conducted to identify the effect of the variation of the first (BSH) and second (SSH) immediate layers’ thicknesses and different room entry widths on the stability of the mine roof. The results of these analyses can be useful for identifying the ultimate moisture content and interpreting the roof performance under various roof thicknesses and room entry widths at the studied mine and the mines with similar geology. The numerical models previously discussed had entry widths of 5.5 m, a BSH unit of 0.75 m and SSH unit of 2.25 m thick. However, according to the boring logs from across the mine area, the immediate mine roof consists of a 0.3–1.3 m thick BSH and 0.1–2.6 m thick SSH rock units. Therefore, thicknesses of 0.25, 0.75 and 1.25 m were considered for BSH and 0.25, 1.25, 2.25 and 2.50 were considered for SSH layers. It is common practice for underground coal mines to use 5.5 m to 6.1 m room entry widths in coal room and pillar mines. In order

2.8 0.5m above the Floor 2.7

1.0m above the Floor 1.5m above the Floor

2.6 2.5 2.4 2.3 2.2 0

1

2

3

4

5

6

7

8

9

10

Distance into the pilar from pillar face in the entry (m) Fig. 14. Vertical stress distribution from the face of the pillar into the pillar at various levels above the floor (Location A).

163

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

BSH and SSH roof units are then increased until the mine roof becomes unstable. Fig. 15 shows the maximum deformation of roof as the moisture content of the two immediate layers (BSH and SSH) increase and also ultimate MC values for various room entry widths and BSH thicknesses. The dashed lines show the projection of movements. The comparison of simulated scenarios with similar BSH thicknesses but different room entry widths show the mine room with a wider roof span is more sensitive to variation of moisture content and resultantly more prone to failure. As the width of the mine increases, more surcharge pressure is applied on the mine roof span and consequently the middle of the roof span will experience greater horizontal stresses due to beam effect. Higher horizontal stress and weak roof layers result in more deformation of the roof as the moisture content increases. In addition, it can be observed that the maximum vertical deformation of the roof span is larger in case of a roof with thicker BSH rock unit. For example, according to Fig. 15, Scenarios 1, 2 and 3, the mine roof fails at ultimate moisture content values of 10.3%, 9.5% and 8.8% for BSH, respectively. The corresponding moisture values for SSH are 9.5%, 8.7% and 8.0%, respectively. Therefore, an appropriate knowledge about the thickness of rock layers existing in the mine roof is critical for designing of the roof support and roof control plans. The scenarios that demonstrate the effect of SSH thickness variations are shown in Table 7. According to this table, in Scenarios 7, 8 and 9, the thickness of BSH layer is constant while the thickness of SSH layer is changing from 0.25 to 2.25 m null. In regard to Scenarios 7 and 10, however, the thickness of both BSH and SSH roof units are changing in a way that the ratio of SSH to BSH thickness remains constant. In all mentioned simulation scenarios, the geomechanical data of roof units, the thickness of roof units above immediate SSH layer and the room entry width are similar to the ones used for simulation of the mine at Location A. Fig. 16 shows the effect of moisture sensitivity of roof rocks with various BSH and SSH immediate roof unit thicknesses. According to Scenarios 7, 8 and 9, while the BSH layer thickness

to investigate the effect of room entry width on roof performance, the room entry widths of 5.5 and 6.1 m was considered in these series of simulations. Except the thickness of immediate roof units and the room entry width, all other modeling parameters such as depth of mining, geomechanical properties of rocks and properties of roof bolts are the same as the ones used for simulation of the studied mine at Location A. Table 6 summarizes the scenarios that investigate the effect of BSH thickness and room entry width variation. The SSH thickness in Scenarios 1–6 was 2.25 m thick. Table 7 summarizes the scenarios that investigate the effect of variation in the first and second immediate roof layer, i.e., BSH and SSH, thicknesses. In all simulation scenarios, the moisture contents of BSH and SSH roof units are initially considered to be the same as the MC values recorded during mine visits at Location A, i.e., 4.5% and 3.7% for BSH and SSH, respectively. The MC values of the

Table 6 Simulation scenarios for various room entries. Modeling scenario

Thickness of first roof units (BSH) (m)

Thickness of second roof unit (SSH) (m)

Room entry width (m)

Scenario Scenario Scenario Scenario Scenario Scenario

0.25 0.75 1.25 0.25 0.75 1.25

2.25 2.25 2.25 2.25 2.25 2.25

5.5 5.5 5.5 6.1 6.1 6.1

1 2 (Location A) 3 4 5 6

Table 7 Simulation scenarios for various roof structures. Modeling scenario

Thickness of first roof units (BSH) (m)

Thickness of second roof unit (SSH) (m)

Scenario Scenario Scenario Scenario

0.25 0.25 0.25 1.25

2.50 1.25 0.25 1.25

Moisture content of immediate SSH roof unit (%) 1.2 Maximum vertical deformation at the center of roof span (mm)

7 8 9 10

2.2

3.2

4.2

5.2

6.2

7.2

8.2

9.2

10.2

10

11

40 Modeling scenario

35

Scenario 1 Scenario 2 (Location A) Scenario 3

30 25 20

Room Thickness Thickness entry Symbol of BSH of SSH width (m) (m) (m) 0.25 2.25 5.5 0.75

2.25

5.5

1.25

2.25

5.5

Scenario 4

0.25

2.25

6.1

Scenario 5

0.75

2.25

6.1

Scenario 6

1.25

2.25

6.1

15 10 5 0

2

3

4 5 6 7 8 9 Moisture content of immediate BSH roof unit (%)

Fig. 15. Maximum vertical displacement of roof span in case of various entry widths and roof structures.

164

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

Maximum Roof Vertical Deformation (mm)

Moisture Content of Immediate BSH Roof Unit (%) 0.2

1.2

2.2

3.2

4.2

5.2

6.2

7.2

8.2

9.2

10.2

6

7

8

9

10

11

40 Thickness Thickness Symbol of BSH of SSH (m) (m)

Modeling scenario

35 30

Scenario 7

0.25

2.5

25

Scenario 8

0.25

1.25

Scenario 9

0.25

0.25

Scenario 10

1.25

1.25

20 15 10 5 0 1

2

3

4

5

Moisture Content of Immediate BSH Roof Unit (%) Fig. 16. Effect of immediate SSH layer thickness on stability and moisture sensitivity of mine roof.

6.2. Roof span vertical deformation rate

Maximum vertical deformation at the center of roof span (mm)

According to Fig. 15, the deformation rate of roof span increases as the moisture content of the roof units increases. This figure shows that the rate of increase in vertical deformation relative to increase in moisture content varies depending on the moisture content level and the dimension of the room entry. The sketch of this variation is shown in Fig. 17. In case of mines with room entry widths of 5.5 m, three distinct increase in defor-

mation rates of low, medium and high, represented by mild (Zone A), transition (Zone B) and steep (Zone C) slopes in Fig. 17a, relative to MC increase are observed. The slope of the lines related to mild, transition and steep deformation rates are labeled as S1, S2 and S3. The moisture content at which the slope of deformation increase relative to MC increase changes from mild to transition is called the transition moisture content (MCtransition). While the MC content at which the Zone C is initiated and the deformation rate becomes significant is defined as critical MC value (MCcritical). For mines with 6.1 m entry widths, i.e., Scenarios 4, 5 and 6 in Fig. 15, the low deformation rates suddenly change to high deformation rates without transitioning zone. Therefore, for Scenarios 4, 5 and 6 the behavior of the roof span is idealized with Zone A and C, with mild and steep slopes, respectively, as shown in Fig. 17b. According to Fig. 15, the MCcritical values for Scenarios 1, 2 and 3 are approximately 1% greater than the ones related to Scenarios 4, 5 and 6. According to this figure, the deformation increase per 1% moisture content increase in Zone A is about 1.5–3 mm for both 5.5 and 6.1 m entry widths. However, the deformation increase per 1% moisture content increase in Zone C is about 16 mm to 20 mm and 10–12 mm for room entry widths of 5.5 m (see Scenar-

(a) 1 S3

S2 1

S1

Zone A (Mild slope)

MCinitial

1

Zone B Zone C (Transition (Critical slope) slope)

MCtransition

MCcritical MCultimate

Moisture content of immediate roof unit (%)

Maximum vertical deformation at the center of roof span (mm)

remains constant, the performance of roof does not vary significantly as the SSH layer thickness changes. However, the change in the thickness of BSH layer from 0.25 to 1.25 m (see Scenarios 7 and 10) results in significant differences in the magnitude of roof deformation as well as the moisture content at failure. Furthermore, according to Fig. 16, it can be observed that the performance of the roof in two Scenarios of 7 and 10 are significantly different from each other, although the ratio of SSH to BSH thicknesses is 1.0 in both scenarios. Therefore, the deformation and moisture sensitivity of mine roof is not directly dependent on ratio of SSH and BSH layer thicknesses.

(b) 1 S3

1

S1

Zone A (Mild slope)

MCinitial

Zone C (Critical slope)

MCcritical MCultimate

Moisture content of immediate roof unit (%)

Fig. 17. A sketch showing the nomenclature for maximum roof deformation versus moisture content variation plots for (a) 5.5 m entry width and (b) 6.1 m entry width.

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

ios 1–3) and 6.1 m (see Scenarios 4–6), respectively. This observation indicates that the continued monitoring of moisture content and roof deformation can be used to identify the critical moisture content. It is also observed that the MCultimate is about 1% more than MCcritical for the simulated Scenarios 1–6. Therefore, inadvance detection of MCcritical or MCultimate is possible and this information can be used to save lives and cleaning cost and allow mine operational planning. 7. Conclusion In this study, the experimental results and field observations were integrated with numerical simulation to analyze the observed roof failure at an underground coal mine located in Illinois Coal Basin due to moisture content increase. The HoekBrown constitute model is used in this study to represent roof behavior. The interaction between the bedding layers as well as the interaction between the roof bolts and roof units is considered in the developed numerical model. The results obtained from the numerical model regarding the performance of the mine at two stable and failed roof locations are in agreement with the field observations. It was shown that by increasing the moisture content of the mine roof rocks, the roof of the case study mine will become unstable. The developed model shows the failure of the mine roof is initiated by the separation of immediate roof bedding due to high horizontal stresses in the immediate roof along with the detaching of the roof bolt and the rock units. At Location A, by increasing the MC values of the roof units as high as twice the insitu values, the mine roof will still be stable. Furthermore, the existence of a very thin and localized, but weak and moisture sensitive rock layer, exacerbate the roof failure potential. These layers may not be detected in the subsurface investigation drilling and detection of these layers is crucial in order to determine the mining areas prone to failure. Finally, the effect of entry width and roof layers’ thicknesses on performance of the case study mine roof was analyzed considering the changes in moisture content. It is shown that the immediate roof unit thickness and entry width have important influence on the stability of the roof. In the studied mine, the entry widths of 5.5 m results in gradual roof vertical deformation rate changes from low to mild and then to high with MC increase. For entry widths of 6.1 m, the roof vertical deformation rate changes suddenly from low to high. The results found from the numerical model suggest that the MCcritical can be detected prior to failure by monitoring the deformation rates and moisture contents at failure prone areas of the mine. The presented methodology can be useful for roof support design and development of roof control plans which incorporates moisture content measurements. It should be noted that the methodology discussed herein was developed based on a case study. Therefore, I should be applied cautiously to other mines with similar rock and conditions. Acknowledgement The authors would like to acknowledge the support of Dr. Joseph Hirschi from Illinois Clean Coal Institute. This research was made possible with support, in part, by the Office of Coal Development of the Illinois Department of Commerce and Economic Opportunity through the Illinois Clean Coal Institute. References [1] Mortazavia A, Hassanib FP, Shabania M. A numerical investigation of rock pillar failure mechanism in underground openings. Comput Geotech 2009;36:691–7.

165

[2] Suchowerska AM, Merifield RS, Carter JP, Clausen J. Prediction of underground cavity roof collapse using the Hoek-Brown failure criterion. Comput Geotech 2012;44:93–103. [3] ASTMD7012-10. Standard test method for compressive strength and elastic moduli of intact rock core specimens under varying states of stress and temperatures. Annual Book of ASTM Standards 408. Philadelphia (PA, USA); 2010. [4] ASTMD5731-08. Standard test method for determination of the point load strength index of rock and application to rock strength classifications. Annual Book of ASTM Standards 408. Philadelphia (PA, USA); 2008. [5] Li Z, Reddish DJ. The effect of groundwater recharge on broken Rocks. Int J Rock Mech Mining Sci 2004;41(3):409–17. [6] Chenevert ME. Shale alteration by water adsorption. J Petrol Technol 1970;22 (9):1141–8. [7] Vásárhelyi B, Ván P. Influence of water content on the strength of rock. J Eng Geol 2006;84(2):70–4. [8] Romana M, Vásárhelyi B. A discussion on the decrease of unconfined compressive strength between saturated and dry rock samples. Proceedings of the 11th congress of the international society for rock mechanics, Lisbon, Portugal, vol. 3 [Chapter 31]. [9] Lashkaripour GR, Passaris EKS. Correlations between index parameters and mechanical properties of shales. Proceedings of the 8th international congress on rock mechanics, Tokyo, Japan, vol. 1. p. 257–61. [10] Mohamad ET, Mohd Fridaus M, Aziz A, Maiye OM, Liang M. The effect of moisture content on the strength and anisotropy index of tropically weathered shale. Electron J Geotech Eng 2013;18. [11] Osouli A, Moradi Bajestani B. Geomechanical properties of coal mine roof rocks in Illinois. In: Proceedings of the IFCEE conference. San Antonio, Texas, USA. [12] Poulsen BA, Shen B, Williams DJ, Huddlestone-Holmes C, Erarslan N. Strength reduction on saturation of coal and coal measures rocks with implications for coal pillar strength. Int J Rock Mech Min Sci 2014;71:41–52. [13] Swann DH. A summary geologic history of the Illinois Basin. In: Geology and petroleum production of the Illinois Basin. Illinois Geologic Society; 1968. p. 3–21. [14] Maleki HN, Owens J. Analysis of the interaction between mobile roof supports and mine strata. Spokane, USA: Spokane Research Laboratory, National Institute for Occupational Safety and Health; 2008. [15] Gale WJ, Mark C, Chen J. Computer simulation of ground behavior and rock bolt interaction at Emerald Mine. In: Proceedings 23rd international conference on ground control in mining. Morgantown, USA. p. 27–34. [16] Sarathchandran A. Three dimensional numerical modelling of coal mine roadways under high horizontal stress fields. Dissertation Towards the Degree of Master of Science Submitted to the University of Exeter. Exeter, United Kingdom; 2014. [17] Hoek E. Rock mechanics laboratory testing in the context of a consulting engineering organization. Int J Rock Mech Mining Sci Geomech 1977;14:92–101. [18] Esterhuizen E, Mark C, Murphy MM. Numerical model calibration for simulating coal pillars, gob and overburden response. In: Proceedings of 29th international conference on ground control in mining. Morgantown, USA. [19] Goel RK, Swarup A, Sheorey PR. Bolt length requirement in underground openings. Int J Rock Mech Min Sci 2007;44:802–11. [20] Zamiran S, Salam S, Osouli A, Ostadhassan M. Floor and pillar stability considerations underground disposal of fine coal waste. In: Proceedings of 49th US rock mechanics/geomechanics symposium. San Francisco, USA. [21] Zamiran S, Osouli A. Subsidence and stability analyses of an Illinois coal mine. In: Proceedings of ISRM 13th international congress on rock mechanics. Montréal, Canada. [22] Gu R, Ozbay U. Numerical investigation of unstable rock failure in underground mining condition Original Research Article. Comput Geotech 2015;63:171–82. [23] Li AJ, Merifield RS, Lyamin AV. Effect of rock mass disturbance on the stability of rock slopes using the Hoek-Brown failure criterion. Comput Geotech 2011;38(4):546–58. [24] Hoek E, Brown ET. Practical estimates of rock mass strength. Int J Rock Mech Mining Sci 1997;34(8):1165–86. [25] Itasca Consulting Group. Fast Lagrangian Analysis of Continua in 2 Dimension (FLAC-2D V7.0). Minneapolis, USA; 2011. [26] Cai M. A simple method to estimate tensile strength and hoek-brown strength parameter mi of brittle rocks. In: Proceedings of the 3rd CANUS rock mechanics symposium. Toronto, Canada. [27] Osouli A, Moradi Bajestani B, Shafii I. Conversion of point load test results to conventional strength tests for shale, claystone and limestone. In: Proceedings of the 48th US American Rock Mechanics Association (ARMA) geomechanics symposium. Minneapolis, USA. [28] Masada T, Han X. Rock mass classification system: transition from RMR to GSI. Final report prepared in cooperation with the Ohio Department of Transportation. Athens, USA; 2013. [29] Marinos P, Hoek E. GSI: a geologically friendly tool for rock mass strength estimation. GeoEng 2000. In: Proceedings of international conference on geotechnical and geological engineering. Melbourne, Australia. [30] Gercek H. Poisson’s ratio values for rocks. Int J Rock Mech Min Sci 2007;44:42017. [31] Bieniawski ZT. A method revisited: coal pillar strength formula based on field investigations. In: Proceedings of the workshop in coal pillar mechanics and design. U.S. Bureau of Mines; 1992. p. 158–65.

166

A. Osouli, B. Moradi Bajestani / Computers and Geotechnics 80 (2016) 152–166

[32] Itasca Consulting Group. Manual for Fast Lagrangian Analysis of Continua in 2 Dimension (FLAC-2D V7.0). Minneapolis, USA; 2011. [33] St. John CM, Van Dillen DE. Rockbolts: a new numerical representation and its application in tunnel design. In: Proceedings of the 24th U.S. symposium on rock mechanics. College Station, USA. [34] Farmer IW. Stress distribution along a resin grouted rock anchor. Int J Rock Mech Mining Sci Geomech 1975;12:347–51.

[35] Zipf RK. Numerical modeling procedures for practical coal mine design. In: Proceedings of the international workshop on rock mass classification in underground mine design, NIOSH IC 9.98. p. 53–162. [36] Liu B, Yue ZQ, Tham LG. Analytical design method for a Truss-Bolt system for reinforcement of fractured coal mine roofs—illustrated with a case study. Int J Rock Mech Min Sci 2005;42:195–218.