1 Analytical modeling of shear behaviors of rockbolts perpendicular

2 to joints

3 Shuqi Ma1,2, Zhiye Zhao2, Jun Peng3, Yilin Gui4*

4 1
Key Laboratory of Transportation Tunnel Engineering, Ministry of Education, School of Civil Engineering,
5 Southwest Jiaotong University, Chengdu 610031, China
6 2
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore 639789, Singapore
7 3
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan
8 430072, China
9 4
School of Civil Engineering and Geosciences, Newcastle University, Newcastle upon Tyne,NE1 7RU,United King
10 dom
11
12
13 Abstract:

14 Rockbolts have been used to minimize the deformation of underground excavations where the

15 surrounding rock masses contain weak planes such as fractures, joints or faults. Rockbolts

16 installed across rock joints are able to resist the opening and shearing movements of rock joints.

17 One of the rockbolt failure modes encountered in the field is caused by the excessive shear loads.

18 A simple analytical model based on the Beam on Elastic Foundation (BEM) method is proposed

19 in this study to predict the shear responses of a bolt installed perpendicularly to rock joint. The

20 shear load-displacement curve of a double shear test can be divided into three stages: the elastic

21 stage, the elasto-plastic stage and the plastic stage. The foundation stiffness for each respective

22 stage are varied with the curvature influencing zone L𝑐 . The pretension effects are taken into

23 account in the proposed analytical model. The model agrees well with the experimental shear

24 tests, suggesting that the analytical model has the capability to predict the shear load-

25 displacement curve of bolts crossing rock joints.

*
Corresponding author.
E-mail address: [Link]@[Link]
26 Keywords: Beam on Elastic Foundation (BEM) method; rockbolt; double shear test; dowel shear
27 behavior; analytical model.
28

29 1. Introduction

30 Rock masses surrounding underground excavations usually contain weak planes such as

31 fractures, joints or faults. Rock masses become deformable due to the presences of these weak

32 planes. Rock masses might open and slide along the fractures or joints due to the excavations.

33 When rockbolts are used to reinforce a fractured rock mass, rockbolts are subjected to tension,

34 shear and compressive forces. Fig. 1 shows several possible force types which might be

35 encountered during the rock block movement [1]. The force types of a, b and c are dominated by

36 the bolt tensile loads whereas the force types of d, e and f are dominated by the bolt shear loads.

37 Less tensile stresses were developed in the perpendicular bolts (force type of e) than the inclined

38 rockbolts (force types of d and f), as the axial displacement (see Fig. 2a) is smaller than the

39 inclined rockbolts (see Figs. 2b and 2c). The tensile stress in the bolt is calculated based on the

40 bolt axial displacement. Azuar in 1977 reported that the perpendicular bolts do not cause

41 considerable axial stress [2].

42
43 Fig. 1 The types of forces developed in rockbolts [1]

44

45 Fig. 2 The axial displacement of inclined and perpendicular rockbolts subjected to shearing loads.

46 Rockbolt failure in the field could be attributed to the tension (joint opening) or the shear along

47 the joint. Majority of rockbolts fail because of shearing loads when bolts are subjected to rock

48 burst conditions [3] or within high stress rock masses [4]. Many studies have been carried out on

49 the axial behaviors of rockbolts [5-10]. In comparison, relatively less attentions have been paid

50 to the shear behaviors of rockbolts [11-12].

51 The shear responses of a bolted rock joint are complex and influenced by many factors, such as

52 joint roughness, rock strength, grout properties, bolt properties, bolt installation angles and bolt

53 pretension. These influencing factors have been investigated by many researchers [13-24]. It is

54 generally accepted that the properties of rock and rockbolts affect the shear resistances of bolted

55 joint. Rock with higher compressive strength leads to higher shear resistances than soft rock [16,

56 22, 23]. Spang and Egger [16] showed that the increase of the bolt diameter could result in

57 higher shearing stiffness of the bolted joint. Joint roughness would cause dilation in the normal

58 direction during shearing movement and higher tensile stress could be mobilized in the bolts due

59 to the joint opening. Joint dilation increases the shear stiffness of the bolted joint [2]. Inclined

60 bolts could also increase the shear stiffness of the joint, leading to higher shear resistance than

61 the perpendicular ones [2, 13, 20, 25]. However, the shear displacement of the inclined bolts

62 prior to failure was smaller than perpendicular bolts. Chen and Li [26] conducted shear tests on

63 fully grouted rebar bolt and D-Bolt under various displacing angles. The displacing angle is

64 defined as the angle between the transversal shear displacement and axial pullout displacement.

65 In their tests, axial and shear loads were applied simultaneously. They found that for the rebar

66 bolt and D-Bolt, the ultimate load of the bolt is independent upon the displacing angle; however,

67 the ultimate displacement is affected by the displacing angles.

68 Haas [2] concluded that pretension does not have positive impact on the bolt shear behaviors.

69 McHugh and Signer [27] found that shear loads contributed greatly to the bolt failure and the

70 axial load (i.e. pretension) had little impact on the shear resistance of a joint. However, these

71 findings on pretension effects contradict the experimental results from Jalalifar and Aziz [23].

72 They found that the pretension forces increase the shear stiffness and results in higher shear

73 loads.
74 In addition to experimental studies, numerical and analytical methods have also been used to

75 study the shear responses of rockbolts. Haile [28] carried out numerical studies on the

76 contributions of rockbolts to the shear resistance of discontinuities. In his study, factors such as

77 steel types, bolt diameter, and bolt inclination angles were examined. Grasselli [20] numerically

78 modelled double shear tests using Finite Element Model code (ZSOIL_3D). Jalalifar and Aziz

79 [23] simulated the shear behaviors of rockbolts in double shear tests using FEM code ANSYS. Li

80 et al. [12] numerically investigated the influences of various factors in the double shear tests

81 using FLAC3D and found that the bolt shear behavior is influenced by concrete strength, bolt

82 installation angle and the bolt diameter. Wei et al. [29] proposed a beam-element-based rockbolt

83 model in the framework of the numerical manifold method (NMM) and this model is able to

84 simulate the shear responses of rockbolts. Bahrani and Hadjigeorgiou [30] numerically

85 investigated behaviors of bolts under pure pull and pure shear loadings using the reinforcement

86 elements in universal distinct element code (UDEC). The difference between “Rockbolt”

87 elements and “Cable” elements were investigated in the numerical experiments and they found

88 that the “Cable” elements tend to underestimate the shear capacity of the bolt. Pellet and Egger

89 [18] proposed an analytical model to simulate the shear behaviors of a bolted rock joint when

90 subjected to shearing loads. Li et al. [11] proposed an analytical model which is able to predict

91 the joint shear strength and joint shear displacement for double shear tests.

92 This study will focus on the shear behaviors of rockbolts which are installed at an angle of 90˚ to

93 the rock joint, i.e. the type of e as shown in Fig. 1. Jalalifar et al. [31] and Jalalifar and Aziz [23]

94 carried out experimental studies examining factors influencing the shearing behaviors of a

95 rockbolt installed perpendicularly to the joints. A specially constructed double shearing

96 apparatus as shown in Fig. 3 was used by Aziz et al. [19], Jalalifar et al. [31] and Jalalifar and
 97 Aziz [23] to study the bolt shear behavior. Jalalifar and Aziz [23] concluded that the shear

98 resistance of rock joints was influenced by the concrete strength, the bolt properties and the

99 pretension.

100

101 Fig. 3 Configuration of double shear tests used in the studies of Aziz et al. [19], Jalalifar et al.

102 [31] and Jalalifar and Aziz [23].

103 When the bolted joint is sheared, the shear movement is counteracted by the crossing bolt.

104 Dowel action is defined as the counteraction of a bolt to joint shearing movement. The Beam on

105 Elastic Foundation analogy (BEF) theory has been used as a tool to analytically model the dowel

106 shear behaviors of bolts crossing joints [32-34]. Maekawa and Qureshi [33] presented a model

107 based on the BEF theory to predict the behaviors of bolts when subjected to the combined axial
108 pullout and shear loads. In the experimental setup of Maekawa and Qureshi [33], the axial

109 displacement was induced due to the joint dilation and the tensile forces were developed due to

110 the pullout mechanism. In the theoretical studies of Maekawa and Qureshi [33] and Moradi et al.

111 [34], the overall shear loads of the bolted joint are considered as the dowel shear forces. The

112 model of Maekawa and Qureshi [33] was further extended to path-dependent cyclic loading case

113 by Soltani and Maekawa [35]. The most relevant parameter in the BEF method is the stiffness of

114 the host concrete. Marcus [36], Soroushian et al. [37] and Dei Poli et al. [38] experimentally

115 studied the foundation stiffness (k) and various empirical equations have been proposed for k.

116 However, these equations just give the elastic value for the foundation stiffness and could not

117 simulate the stiffness changes during the shear loading. Based on studies of Maekawa and

118 Qureshi [33], Moradi et al. [34] proposed the elasto-plastic formulation of spring stiffness which

119 is able to capture the local crushing nearby the joint interface.

120 In the studies [33, 34, 37, 38], the maximum shear displacement ranges from 1/5 to 1/3 of the

121 bolt diameter [38], which is much lower than that occurs in double shear tests. The maximum

122 shear displacement in double shear tests ranges from 1 to 4 times of the bolt diameter [12, 23].

123 The previous proposed models could not be simply applied to predict the shear behaviors in

124 double shear tests.

125 In this study, a simple analytical model is presented based on the Beam on Elastic Foundation

126 analogy (BEF) theory, with the objective to simulate the dowel action of rockbolts perpendicular

127 to joints. This paper focuses the dowel shear forces developed in the bolt. The pretension effects

128 are taken into account in the study while the axial force developed in the bolt during shearing is

129 not considered. The proposed analytical model as well as its experimental verification will be

130 presented in the following sections.

131 2. Analytical model for double shear tests

132 As the double shear test is symmetric (see Fig. 3), only half of the test is analyzed here as shown

133 in Fig. 4a. A crossing bolt at the joint under shear loads could be treated as a semi-infinite beam

134 on an elastic foundation. The host concrete is considered as elastic foundation which is

135 represented by springs as shown in Fig. 4. Timoshenko and Lessels [32] proposed the analytical

136 solution for the BEF theory, and the differential equation of the deflection of a beam resting on

137 an elastic foundation is expressed as:

𝑑4 𝑤
138 𝐸𝑏 𝐼𝑏 = −𝑘𝑤 (1)
𝑑𝑥 4

139 where k denotes the modulus of the elastic foundation (i.e. the spring stiffness); w denotes the

140 bolt deflection; 𝐸𝑏 is Young’s modulus of the bolt; 𝐼𝑏 is the moment of the inertia of the bolt,

𝜋𝐷 4
141 which is calculated by: 𝐼𝑏 = ; D is the diameter of the bolt.
64

142 The solution to Eq. (1) is expresses as:

143 𝑤(𝑥) = 𝑒 𝜆𝑥 (𝐴 cos 𝜆𝑥 + 𝐵 sin 𝜆𝑥) + 𝑒 −𝜆𝑥 (𝐶 cos 𝜆𝑥 + 𝐷 sin 𝜆𝑥) (2)

4 𝑘
144 𝜆 = √4𝐸 (3)
𝑏 𝐼𝑏

145 where A, B, C and D are constants of integration, which are determined by the boundary

146 conditions.
147

148 Fig. 4 (a) Half of the double shear test; (b) Semi-infinite beam on elastic foundation.

149 The half of the double shear test could be further simplified to that as shown in Fig. 4b. The
𝑃
150 shear load V0 = 2 and moment 𝑀0 are applied at the joint interface (i.e. x = 0) as shown in Fig.

151 4b. The bending moment 𝑀0 is equal to 0 at the joint interface.

152 Under this boundary condition (see Appendix), constants A, B and D are calculated to be zero

153 and C is computed by:

V
154 𝐶 = 2𝜆3 𝐸0 𝐼 (4)
𝑏 𝑏

155 Hence Eq. (2) becomes:

𝑒 −𝜆𝑥
156 𝑤(𝑥) = 2𝜆3 𝐸 V0 cos 𝜆𝑥 (5)
𝑏 𝐼𝑏

157 Substituting Eq. (3) into Eq. (5) gives:

 2𝜆V0 𝑒 −𝜆𝑥 cos 𝜆𝑥
158 𝑤(𝑥) = (6)
𝑘

159 The slope of the deflection curve can be obtained by differentiating Eq. (6) with respect to x:

𝑑𝑤 2𝜆2 V0 𝑒 −𝜆𝑥 (cos 𝜆𝑥+sin 𝜆𝑥)

160 =− (7)
𝑑𝑥 𝑘

161 Setting 𝑥 = 0 in Eq. (6), the relationship between the applied shear force V0 and local deflection

162 w0 at the joint is expressed as:

𝑘 𝑘
163 V0 = 2𝜆 w0 = 4𝜆 ∆ (8)

164 where ∆ is the shear displacement; w0 is the local deflection of the bolt at 𝑥 = 0; ∆= 2w0 ,

165 which is illustrated in Fig. 4a.

166 In this paper, the shear displacement ∆ is treated as an input and the incremental form of the Eq.

167 (8) is derived as follows:

𝑘
168 𝑑V0 = 4𝜆 d∆ (9)

169 It can be seen from Eq. (9) that the shear behavior is directly associated with the mechanical

170 properties of the host concrete (the spring stiffness k), the applied shear displacement d∆, as well

171 as the bolt properties (𝜆: 𝐸𝑏 and 𝐼𝑏 ).

172 The values of k suggested in the literature [34, 36, 37, 38] are very scattered as different testing

173 conditions were used in their respective tests. The formula of k proposed by Moradi et al. [34] is

174 used here:

𝐸𝑏 𝐼𝑏
175 𝑘(∆) = 181 (10)
𝐿4𝑐
176 where: 𝐸𝑏 is Young’s modulus of the bolt; 𝐼𝑏 is the moment of the inertia of the bolt; L𝐶 is the

177 length of the curvature influencing zone.

178 The foundation stiffness k is a function of the curvature influencing zone (L𝐶 ). Maekawa and

179 Qureshi [33] introduced the concept of a curvature influencing zone (L𝐶 ) based on the BEF

180 theory to account for the effect of the localized curvature of the bolt. The shape of the curvature

181 profile ∅(x) within the curvature influencing zone is shown in Fig. 5.

182

183 Fig. 5 Schematic distribution profiles along the embedded rockbolt, after Maekawa and Qureshi

184 [33]

185 For a bolt crossing a joint under shear loads, a hinge point will be formed in the bolt. The hinge

186 point is defined as the point where the maximum bending moment is reached and the shear load

187 is equal to zero. According to the BEF theory, the hinge point location is expressed as:

𝜋4 4𝐸𝑏 𝐼𝑏
188 𝐿ℎ = √150𝑓′0.85 (11)
4 𝑐

189 where: 𝑓𝑐′ is the compressive strength of concrete. According to Eq. (11), the hinge point location

190 only depends on the bolt properties and the concrete strength.
191 Jalalifar and Aziz [23] studied the hinge point location and concluded that the hinge point

192 distance from the joint interface 𝐿ℎ is influenced by the concrete strength, the pretension, bolt

193 diameter and the shear displacement. In addition, the hinge point location is also affected by the

194 test setup such as the size of the concrete block and the confining conditions. The curvature

195 influencing zone L𝑐 is also dependent upon these factors in the same manner as the hinge point

196 location 𝐿ℎ .

197 In Eq. (11), the hinge point location is not associated with the axial force (pretension) and the

198 test setup. In other words, the hinge point location computed by Eq. (11) remains constant once

199 the properties of bolts and concrete are given, which is not the case in reality. As there does not

200 exist an formula which could account for the influences of all the factors, Eq. (11) is adopted in

201 this paper due to its simplicity and a coefficient a is introduced in the formula of L𝑐 to

202 incorporate the effects of the experimental conditions such as the test setup and pretension

203 effects.

204 During the initial shear loading, the shear displacement is small and the rockbolt and surrounding

205 concrete are in elastic state. The initial curvature influencing zone L𝑐0 is assumed as a times the

206 size of the hinge point location:

207 L𝑐 = L𝑐0 = 𝑎 ∙ 𝐿ℎ (12)

208 where: L𝑐0 is the initial curvature influencing zone; a is the coefficient depending on the

209 experimental conditions. The value of the coefficient a cannot be easily obtained from

210 experiments and the curve fitting technique is used to determine its value.
211 Jalalifar and Aziz [23] conducted a series of physical double shear tests and concluded that the

212 typical shear load-displacement profile consists of three stages: elastic stage, elasto-plastic stage

213 and plastic stage, which are shown in Fig. 6. During the initial loading stage, the bolt and the

214 host concrete deform elastically, which corresponds to the first section in Fig. 6. The curvature

215 influencing zone L𝑐 is assumed to remain constant as L𝑐0 and the foundation stiffness k is held

216 constant in the elastic stage. Eq. (9) coupled with a constant stiffness k have a fair accuracy in

217 describing the linear shearing behaviors in the early shear loading.

218

219 Fig. 6 Typical shear load-displacement relationship of double shear tests, after Jalalifar and Aziz

220 [23].

221 The increasing shear displacement results in local crushing of the concrete. The host concrete

222 close to the joint interface transforms from the elastic stage to the plastic stage. This stage is
223 referred as the elasto-plastic stage, which corresponds to the second stage in Fig. 6. In the elasto-

224 plastic stage, the curvature influencing zone L𝑐 increases with the concrete crushing and in the

225 meanwhile, the foundation stiffness k decreases. The non-linear shear behavior in this stage

226 could be captured by gradually changing the spring stiffness, which is a convenient method in

227 simulating the damages in the concrete nearby the joint.

228 During the elasto-plastic stage, the curvature influencing zone L𝑐 is gradually increased:

229 L𝑐 = L𝑐0 [1 + 3(𝐷𝐼(∆) − 𝑏)0.8 ] (13)

230 where 𝐷𝐼(∆) is a non-dimensional damage index. In this study, the damage index DI is defined

231 as,

∆
232 𝐷𝐼(∆) = 𝐷 (14)

233 Note that in the studies of Maekawa and Qureshi [33] and Moradi et al. [34], the damage index

∆
234 DI was defined as 2𝐷. In this study, the damage index definition in Eq. (14) is more suitable for

235 double shear tests.

236 The parameter b in Eq. (13) is dependent on the test setup and is defined as:

∆1
237 b= (15)
𝐷

238 where ∆1 is the displacement of the point at the beginning of the elasto-plastic stage as shown in

239 Fig. 6. The constant value b = 0.02 is used throughout in the studies of Maekawa and Qureshi

240 [33] and Moradi et al. [34], while in this study, the parameter b is considered as a variable

241 depending upon the experimental setup.

242 Afterwards, the host concrete enters into the plastic stage, during which the rock reaction is

243 assumed to be constant and the curvature influencing zone L𝑐 reaches an ultimate value and

244 remains constant throughout this stage.

245 L𝑐 = L𝑐0 [1 + 3(𝑐 − 𝑏)0.8 ] (16)

∆2
246 c= (17)
𝐷

247 where ∆2 is the displacement of the point at the beginning of the plastic stage as shown in Fig. 6.

248 The bolt failure at the joint interface occurs when the tensile and shear forces satisfy the

249 following failure criteria [18]:

2 2
𝑁 𝑉
250 (𝑁0 ) + (𝑉0 ) = 1 (18)
𝑓 𝑓

251 where 𝑁0 is the tensile force developed in the bolt; 𝑁𝑓 is the ultimate tensile force at bolt failure

252 and equal to 𝐴𝜎𝑓 ; 𝐴 is the bolt cross section area; 𝜎𝑓 is the ultimate tensile stress; 𝑉𝑓 is the

253 ultimate shear force at bolt failure; 𝑉𝑓 = 𝐴𝜏𝑓 ; according to the Von Mises criterion in plane stress

𝜎𝑓
254 state, 𝜏𝑓 = .
√3

255 As the tensile stress of bolt is neglected in this study and hence, the bolt is assumed to break

256 when 𝑉0 = 𝑉𝑓 (19)

257 Eq. (19) will overestimate the bolt shear force as the tensile stress 𝑁0 is not taken into account.

258 The calculation procedure for the proposed model is simple and is summarized below:

259 a) Incremental displacement d∆ is considered as input.

260 b) The cumulative displacement ∆ can be computed and 𝐷𝐼(∆) is obtained by Eq. (14).

261 c) The foundation stiffness k is computed by Eq. (10), in which the curvature influencing

262 zone L𝑐 is computed by Eq. (12) when 𝐷𝐼(∆) ≤ b; or by Eq. (13) when b < 𝐷𝐼(∆) < c;

263 or by Eq. (16) when 𝐷𝐼(∆) ≥ c.

264 d) The incremental shear force is computed by Eq. (9), and the cumulative shear force can

265 be then obtained.

266 3. Experimental verification by double shear tests

267 Jalalifar and Aziz [23] carried out a series of double shear tests under various concrete strength

268 (𝑓𝑐′ ) and pretensions (𝑇𝑝𝑟𝑒 ). Firstly, the double shear tests under the condition of 𝑇𝑝𝑟𝑒 = 0 are

269 used to verify the proposed model. Afterwards, the model’s ability to capture the pretension

270 effects are verified by the double shear tests under various pretension loads.

271 3.1 No pretension

272 Four types of bolts were used in the double shear tests of Jalalifar and Aziz [23]. Bolts T1, T2

273 and T3 have the same core diameter of 21.7 mm and the same Young’ Modulus of 200 GPa,

274 which are shown in Table 1. These three types of bolts have different profiles (i.e. the rib spacing

275 and rib height). Bolt T4 has smaller core diameter and lower Young’s Modulus than the other

276 three bolts.

277 Table 1. Mechanical properties of bolt materials

Bolt type T1 T2 T3 T4
Bolt core diameter, mm 21.7 21.7 21.7 10.7
Young’s Modulus, 𝐸𝑏 , GPa 200 200 200 70
Rib spacing, mm 12 12.5 25 -
Rib height, mm 0.65 1.4 1.25 -
 Yield stress, MPa 683 673 552 365
Ultimate tensile stress, MPa 862 900 942 490
Ultimate shear stress, MPa 498 520 545 283
278

279 Two concrete compressive strengths (20 MPa and 40 MPa) were used in their tests. Bolts T1 and

280 T2 were grouted in concrete with compressive strength of 20 MPa and 40 MPa. Bolts T3 and T4

281 were grouted in concrete with 𝑓𝑐′ = 40 MPa. A total number of 6 double shear tests were carried

282 out under the pretension of 𝑇𝑝𝑟𝑒 = 0. Table 2 shows the number of tests conducted for each bolt

283 type.

284 Table 2. The number of tests for each bolt type.

Bolt type T1 T2 T3 T4
Concrete, 𝑓𝑐′ = 20 MPa 1 1 0 0
Concrete, 𝑓𝑐′ = 40 MPa 1 1 1 1
285

286 The loading rate of 0.75 mm/step was used in the analytical analyses. The parameter a for the

287 four bolts was chosen as 4. For bolt types T1, T2 and T3, ∆1 and ∆2 are determined as 3 mm and

288 14 mm, respectively. For bolt T4, the used ∆1 and ∆2 in the analytical analysis are 3 mm and 7

289 mm, respectively. These values were selected to best fit the experimental data.

290 Fig. 7 shows the comparison of the analytical model and the double shear tests for bolts T1 and

291 T2 installed in concrete (𝑓𝑐′ = 20 MPa). The whole shear load-displacement curve was divided

292 into three distinct stages as shown in Fig. 7. Foundation stiffness k was computed for each

293 respective stage. It can be seen that the proposed analytical model matches well with the

294 experimental shear behaviors of bolts T1 and T2.

295 It is noted that the bolt profile could slightly affect the bolt performances (see the discrepancy

296 between the shear load-displacement curves for bolts T1 and T2 in Fig. 7). However, this effect
297 is negligible when compared to the influence of the concrete strength. The bolt profile effect is

298 not taken into account in the analytical model of this study.

299
300

301 Fig. 7 Comparison of analytical and experimental shear load-displacement curves for bolts T1

302 and T2 (𝑓𝑐′ = 20 MPa).

303 Fig. 8 shows the comparison of the analytical model and the double shear tests for bolts T1, T2

304 and T3 for 𝑓𝑐′ = 40 MPa. The analytical model produces good agreement with the experimental

305 tests. It can be seen from Figs. 7 and 8 that the concrete could significantly affect the shear

306 resistances and the increasing concrete strength leads to higher shear resistances of the joint. The

307 proposed analytical model successfully capture the concrete influence.

308

309 Fig. 8 Comparison of analytical and experimental shear load-displacement curves for bolts T1,

310 T2 and T3 (𝑓𝑐′ = 40 MPa).

311 Fig. 9 shows the results of the analytical model and the shear test for bolt T4 installed in concrete

312 with 𝑓𝑐′ = 40 MPa. The proposed analytical model agrees well with the experimental results. It

313 can be seen from Figs. 8 and 9 that under the same concrete strength (𝑓𝑐′ = 40 MPa), bolt T4

314 generates much smaller shear resistances than bolts T1, T2 and T3 as bolt T4 has much smaller

315 Young’s Modulus and diameter. In addition to the concrete strength, the bolt diameter and

316 Young’s Modulus could also affect the bolt shear behaviors. The proposed model is able to

317 simulate the effects of the bolt properties.

318

319 Fig. 9 Comparison of analytical and experimental shear load-displacement curves for bolts T4

320 (𝑓𝑐′ = 40 MPa).

321 The stiffness evolutions of analytical solutions in Figs. 7-9 are shown in Fig. 10. It can be seen

322 that foundation stiffness remains still in the initial loading stage, which is followed by a gradual

323 decrease in the elasto-plastic stage. Afterwards, the foundation stiffness reached an ultimate

324 value in the plastic stage. For T1, T2 and T3 having the same diameter and Young’s Modulus,

325 the chosen parameters ∆1 and ∆2 are the same for both concrete strengths of 20 and 40 MPa. For

326 T4 which has smaller diameter and lower Young’s Modulus, the parameter ∆1 is the same as

327 other bolts while ∆2 has a smaller value. These parameters are selected based on trial and error

328 methods in order to best model the experimental curves.

329

330 Fig. 10 The foundation stiffness evolutions.

331 3.2. Pretension effects

332 Up until now, the complete shear load-displacement curve of the double shear test can be

333 properly represented by the proposed analytical model. However, the BEF theory does not take

334 into account the axial forces developed in the bolt and hence the proposed model could not

335 reflect the effects of pretension on the shear behaviors.

336 Jalalifar [21] and Jalalifar and Aziz [23] found that the hinge point location is also affected by

337 the applied pretension loads. Smaller Lℎ will be formed when bolts are subjected to higher

338 pretension forces. In the same manner, the curvature influencing zone L𝑐 is supposed to decrease

339 with the increasing pretension forces. In this paper, the pretension effects are simply represented

340 by varying the curvature influencing zone L𝑐 . It means that the parameter a in Eq. (12) depends

341 not only on the test setup but also on the pretension forces.

342 In the double shear tests of Jalalifar and Aziz [23], pretensions of 0 kN, 20 kN, 50 kN and 80 kN

343 were used. In the verification of pretension effects, the used bolt was bolt T1 as shown in Table 1.
344 For the pre-tensioned bolts, double shear tests were also conducted in two concrete strength of 20

345 and 40 MPa. Table 3 shows the parameters used in the analytical analysis. The used parameters

346 in the following analytical analysis are the same for both concrete strength ( 𝑓𝑐′ =

347 20 𝑀𝑃𝑎 𝑎𝑛𝑑 40 𝑀𝑃𝑎). In addition, the parameters for the case of 𝑇𝑝𝑟𝑒 = 0 kN are the same as

348 those used for bolts T1, T2, and T3 in Section 3.1.

349 The parameter ∆2 is different for un-pretensioned bolt (𝑇𝑝𝑟𝑒 = 0 kN) and pretensioned bolt

350 (𝑇𝑝𝑟𝑒 =20, 50 and 80 kN). The curvature influencing zone L𝑐 (represented by the parameter a in

351 Table 3) is decreased with the increasing pretension forces.

352 Table 3. The parameters used in modeling the behaviors of bolts in 20 MPa and 40 MPa concrete.

Parameters a in Eq. (12) ∆1 (mm) ∆2 (mm)

𝑇𝑝𝑟𝑒 = 0 kN 4 3 14

𝑇𝑝𝑟𝑒 = 20 kN 3.8 3 8

𝑇𝑝𝑟𝑒 = 50 kN 3.5 3 8

𝑇𝑝𝑟𝑒 = 80 kN 3.2 3 8

353

354 Fig. 11 shows the comparison between the analytical model and the double shear tests for

355 concrete strength of 20 MPa. It can be seen that the analytical model is able to predict the shear

356 load-displacement curves of bolts with different pretensions. This model successfully takes into

357 account the pretension effects. The pretension increases the shear stiffness, leading to higher

358 shear resistance than un-pretensioned bolts at a same displacement.

359

360 Fig. 11 Comparison of analytical model and the double shear tests under the concrete strength of

361 20 MPa.

362 The shear behaviors of bolts in 40 MPa concrete are also predicted by the analytical model where

363 the used parameters are the same as those used for the 20 MPa concrete. It can be seen from Fig.

364 12 that the analytical model has good agreement with the shear behaviors of bolts subjected to

365 various pretensions in 40 MPa concrete. This indicates that the analytical model accounts for the

366 effects of pretension as well as the effects of concrete strength.

367

368 Fig. 12 Comparison of analytical model and the double shear tests under the concrete strength of

369 40 MPa.

370 4. Experimental verification by single shear tests

371 In the above section, the proposed analytical model is verified by the double shear tests. Except

372 for double shear tests, single shear tests have also been used to study the shear behaviors of rock

373 bolts. In this section, single shear tests conducted by previous researchers are used to verify the

374 analytical model. The parameters used in the analytical modelling under various conditions are

375 discussed later.

376 Note that when analytically modelling the double shear tests in Section 3, the overall shear load

377 is equal to two times the cumulative shear load obtained via Eq. (9); however, in this section, the

378 overall shear load of the single shear tests is equal to the computed shear load by Eq. (9), as there

379 is only one bolted joint in the single shear test.

380 Schubert [39] carried out a single shear test on bolts installed perpendicular to the rock joint. The

381 compressive strength of the concrete is 40 MPa, the bolt diameter is 24 mm and the Young’s

382 modulus of the bolt is 210 GPa. The three parameters are selected as: a = 4; ∆1 = 3 𝑚𝑚; and

383 ∆2 = 7 𝑚𝑚. Fig. 13 shows the comparison between the analytical model and the experimental

384 shear test.

385

386 Fig. 13 Comparison of analytical model and the shear test of Schubert [39]

387 Stjern [40] carried out a single shear test on bolts installed in concrete with compressive strength

388 of 65 MPa. The bolt is 18 mm in diameter and the Young’s modulus is 200 GPa. The three

389 parameters are selected as: a = 4; ∆1 = 3 𝑚𝑚; and ∆2 = 7 𝑚𝑚. Fig. 14 shows the comparison of

390 the analytical model and the experimental shear test.

391

392 Fig. 14 Comparison of analytical model and the shear test of Stjern [40]

393 Goris et al. [41] carried out single shear tests on cable bolts installed in concrete with

394 compressive strength of 68.9 MPa. The cable bolt is 15.24 mm in diameter and the Young’s

395 modulus is estimated as 200 GPa. The three parameters are selected as: a = 4; ∆1 = 3 𝑚𝑚; and

396 ∆2 = 5 𝑚𝑚. Fig. 15 shows the comparison of the analytical model and the experimental shear

397 test.
398

399 Fig. 15 Comparison of analytical model and the shear test of Goris et al. [41]

400 McHugh and Signer [27] carried out single shear tests on bolts installed in concrete with

401 compressive strength of 85.5 MPa. The bolt is 22 mm in diameter. As two 6.4-mm-wide by 3.2-

402 mm-deep slots were milled on the two sides of the bolt, the actual diameter of the bolt should be

403 smaller than 22 mm and is chosen as 19 mm in the analytical modelling. The bolt has the

404 Young’s modulus of 200 Gpa. The three parameters are selected as: a = 4; ∆1 = 3 𝑚𝑚; and ∆2 =

405 7 𝑚𝑚. Fig. 16 shows the comparison of the analytical model and the experimental shear test.

406
407

408 Fig. 16 Comparison of analytical model and the shear test of McHugh and Signer [27]

409 Based on the above four single shear tests shown in Figs. 13-16, it can be seen that the proposed

410 model is able to predict the global shear load evolution of the single shear tests. The parameters

411 used in the analytical modelling are listed in Table 4. The parameters such as the bolt diameter D,

412 the bolt Young’s modulus 𝐸𝑏 and the concrete compressive strength 𝑓𝑐′ can be obtained directly

413 from the test setup. The parameters a, b (∆1) and c (∆2 ) need to be solved via the curve-fitting

414 technique. As can be seen, the values of a and b (∆1) remain constant for the double shear tests

415 and single shear tests under various bolt properties and concrete strength. However, the

416 parameter ∆2 of double shear tests tend to have larger values than those in single shear tests.

417
418 Table 4 The parameters used in the analytical modelling for the experimental shear tests with the

419 pretension 𝑇𝑝𝑟𝑒 = 0 kN.

Parameters Concrete Bolt Young’s a ∆1 (mm) ∆2 (mm)

compressive diameter Modulus
strength, 𝑓𝑐′ (mm) (GPa)
(MPa)
Jalalifar and Aziz [23], 20 21.7 200 4 3 14
𝑇𝑝𝑟𝑒 = 0
Jalalifar and Aziz [23], 40 21.7 200 4 3 14
𝑇𝑝𝑟𝑒 = 0
Jalalifar and Aziz [23], 40 10.7 70 4 3 7
𝑇𝑝𝑟𝑒 = 0
Schubert (1984)[39] 40 24 210 4 3 7
Stjern [40] 65 18 200 4 3 7
Goris et al. [41] 68.9 15.24 200 4 3 5
McHugh and Signer 85.5 22 200 4 3 7
[27]
420

421 5. Discussion

422 The proposed analytical model is for bolts installed perpendicularly to the rock joint. This model

423 provides a simple but effective way to predict the overall shear load-displacement curves of bolts.

424 The limitation of the proposed model lies in that the model does not take into account the axial

425 forces developed in bolts. The factors such as bolt surface profiles (rib spacing and rib height),

426 grout strength, the interaction between bolt and grout, and the interaction between grout and host

427 concrete, which could affect the axial behaviors of the bolt, are not considered in the proposed

428 model.

429 The failure mechanism of a bolt crossing a joint is complex as it includes the damage of the host

430 concrete which provides bearing support to the bolt, the yielding of the bolt, the formation and

431 evolution of the hinge point along the bolt, and the deterioration of grout between the bolt and
432 the concrete. Few studies in the available literature have discussed the failure mechanisms of the

433 two materials. The experimental tests mainly focused on the global shear load-displacement

434 curves. Even for the shear tests where strain gauges are attached on the bolt, strain gauges close

435 to the joint interface are usually damaged due to the shear displacement and consequently, the

436 bolt failure mechanism cannot be measured properly. The failure mechanism of these two

437 materials are unclear. In this study, the analytical model is based on the Beam on Elastic

438 Foundation (BEM) method. The global shear load-displacement curve is simply described by the

439 foundation stiffness degradation, which represents the crushing and damage of the concrete

440 nearby the joint interface.

441 According to Eq. (19), the total failure shear load of the double shear test for T4 bolts is

442 predicted as: 2𝑉𝑓 = 50 kN for T4 bolts. Most tests of Jalalifar and Aziz [23] were terminated

443 before the bolt failure. The failure shear load of the T4 bolt is 75 kN as shown in Fig. 9, which is

444 larger than 50 kN. Hence, in reality Eq. (19) underestimates the failure shear load. The predicted

445 failure load for the T1 bolt is 2𝑉𝑓 = 378 kN. Fig. 12 shows that the T1 bolt does not fail as the

446 applied shear load reaches 500 kN. It can be seen that the failure criteria defined by Eq. (19)

447 tends to give a conservative prediction of the failure shear load. Eq. (18) gives an even more

448 conservative value as the axial stress is accounted for. In comparison, Eq. (19) provides a closer

449 prediction on the failure load of bolts under double shear tests.

450 6. Conclusions

451 A simple analytical model is proposed in this study based on the Beam on Elastic Foundation

452 (BEM) method, with the objective to predict the shear load-displacement curve of the bolt

453 subjected to shearing loads. The shear load-displacement curve is divided into three distinct
454 stages: the elastic stage, the elasto-plastic stage and the plastic stage. The foundation stiffness is

455 associated with the curvature influencing zone L𝑐 . The foundation stiffness is a constant value in

456 elastic stage whereas the stiffness in the elasto-plastic stage is mobilized downwards with the

457 increasing shear displacement. The non-linear shear behavior of rockbolts is successfully

458 modelled by gradually changing the foundation stiffness in the BEM. In the plastic stage, the

459 foundation stiffness reaches the ultimate value and remains constant afterwards. The analytical

460 model is in an incremental form and its capability of representing the dowel shear behaviors of

461 bolts is evaluated by shear tests. Physical shear tests show that pretension could decrease the

462 curvature influencing zone. The pretension effects are simply incorporated in the proposed

463 model by varying the size of the curvature influencing zone.

464 The experimental verifications suggest that the analytical model is able to predict the complete

465 shear load-displacement curve of bolts. In addition, the model successfully takes into account the

466 effects of concrete strength and bolt properties, as well as pretension effects.

467 Appendix:

468 By differentiation of Eq. (2), we have:

469 𝑤 ′ (𝑥) = 𝐴1 𝑒 𝜆𝑥 cos 𝜆𝑥 + 𝐵1 𝑒 𝜆𝑥 sin 𝜆𝑥 + 𝐶1 𝑒 −𝜆𝑥 cos 𝜆𝑥 + D1 𝑒 −𝜆𝑥 sin 𝜆𝑥 (A1)

470 where 𝐴1 = 𝐴𝜆 + 𝐵𝜆; 𝐵1 = −𝐴𝜆 + 𝐵𝜆; C1 = −𝐶𝜆 + 𝐷𝜆; D1 = −𝐶𝜆 − 𝐷𝜆.

471 Differentiating Eq. (A1) leads to

472 𝑤 ′′ (𝑥) = 𝐴2 𝑒 𝜆𝑥 cos 𝜆𝑥 + 𝐵2 𝑒 𝜆𝑥 sin 𝜆𝑥 + 𝐶2 𝑒 −𝜆𝑥 cos 𝜆𝑥 + D2 𝑒 −𝜆𝑥 sin 𝜆𝑥 (A2)

473 where 𝐴2 = 𝐴1 𝜆 + 𝐵1 𝜆; 𝐵2 = −𝐴1 𝜆 + 𝐵1 𝜆; C2 = −𝐶1 𝜆 + 𝐷1 𝜆; D2 = −𝐶1 𝜆 − 𝐷1 𝜆.

474 Differentiating Eq. (A2) gives

475 𝑤 ′′′ (𝑥) = 𝐴3 𝑒 𝜆𝑥 cos 𝜆𝑥 + 𝐵3 𝑒 𝜆𝑥 sin 𝜆𝑥 + 𝐶3 𝑒 −𝜆𝑥 cos 𝜆𝑥 + D3 𝑒 −𝜆𝑥 sin 𝜆𝑥 (A3)

476 where 𝐴3 = 𝐴2 𝜆 + 𝐵2 𝜆; 𝐵3 = −𝐴2 𝜆 + 𝐵2 𝜆; C3 = −𝐶2 𝜆 + 𝐷2 𝜆; D3 = −𝐶2 𝜆 − 𝐷2 𝜆.

477 At x → ∞, 𝑤(x → ∞) = 𝑒 𝜆𝑥 (𝐴 cos 𝜆𝑥 + 𝐵 sin 𝜆𝑥) = 0. Hence, we have 𝐴 = 𝐵 = 0.

𝑀(𝑥) 𝑉(𝑥)
478 Knowing that 𝑤 ′′ (𝑥) = and 𝑤 ′′′ (𝑥) = 𝐸 , at x = 0,
𝐸𝑏 𝐼𝑏 𝑏 𝐼𝑏

479 𝑤 ′′ (0) = 𝑀0 = −𝐶1 𝜆 + 𝐷1 𝜆 = −(−𝐶𝜆 + 𝐷𝜆)𝜆 + (−𝐶𝜆 − 𝐷𝜆)𝜆. (A4)

480 The bending moment 𝑀0 is equal to 0 at the joint interface (i.e. 𝑥 = 0) and hence, 𝐷 = 0.

𝑉0
481 At x = 0, 𝑤 ′′′ (0) = 𝐸 =𝐴2 𝜆 + 𝐵2 𝜆 − 𝐶2 𝜆 + 𝐷2 𝜆 = (2𝐶 + 2𝐷)𝜆3 (A5)
𝑏 𝐼𝑏

V
482 Hence, 𝐶 = 2𝜆3 𝐸0 𝐼
𝑏 𝑏

483 References:

484 [1] Windsor. Rock reinforcement systems. Int J Rock Mech Min Sci Geomech Abstr. 1997;34:919-951
485 [2] Haas CJ. Analysis of rockbolting to prevent shear movement in fractured ground Mining engineering.
486 1981;33:698-704
487 [3] Haile A. Observation of the dynamic support performance of South African tunnel support systems. In:
488 In: Proceedings rock support and reinforcement practice in mining, Kalgoorlie, Australia, 1999a.
489 pp 335-341
490 [4] Li CC. Field Observations of Rock Bolts in High Stress Rock Masses Rock Mechanics and Rock
491 Engineering. 2010;43:491-496 doi:10.1007/s00603-009-0067-8
492 [5] Deb D, Das KC. Modelling of fully grouted rock bolt based on enriched finite element method.
493 International Journal of Rock Mechanics and Mining Sciences. 2011; 48:283-293
494 doi:[Link]
495 [6] Ma S, Nemcik J, Aziz N. An analytical model of fully grouted rock bolts subjected to tensile load
496 Construction and Building Materials. 2013;49:519-526
497 doi:[Link]
498 [7] Ma S, Nemcik J, Aziz N, Zhang Z. Analytical model for rock bolts reaching free end slip Construction
499 and Building Materials. 2014;57:30-37 doi:[Link]
500 [8] Nie W, Zhao ZY, Ning YJ, Sun JP. Development of Rock Bolt Elements in Two-Dimensional
501 Discontinuous Deformation Analysis. Rock Mechanics and Rock Engineering. 2014b;47:2157-
502 2170 doi:10.1007/s00603-013-0525-1
503 [9] He L, An XM, Zhao ZY. Fully Grouted Rock Bolts: An Analytical Investigation. Rock Mechanics
504 and Rock Engineering. 2014;48:1181-1196 doi:10.1007/s00603-014-0610-0
505 [10] Chen J, Saydam S, Hagan PC. An analytical model of the load transfer behavior of fully grouted
506 cable bolts. Construction and Building Materials. 2015. 101, Part 1:1006-1015
507 doi:[Link]
508 [11] Li X, Nemcik J, Mirzaghorbanali A, Aziz N, Rasekh H. Analytical model of shear behaviour of a
509 fully grouted cable bolt subjected to shearing International Journal of Rock Mechanics and
510 Mining Sciences. 2015;80:31-39 doi:[Link]
511 [12] Li L, Hagan PC, Saydam S, Hebblewhite B, Li Y. Parametric Study of Rockbolt Shear Behaviour by
512 Double Shear Test Rock Mechanics and Rock Engineering. 2016a; 49:4787-4797
513 doi:10.1007/s00603-016-1063-4
514 [13] Bjurstrom S. Shear strength of hard rock joints reinforced by grouted untensioned bolts. In:
515 Proceedings of the Congress of the International Society for Rock Mechanics, 1974. pp 1194-
516 1199
517 [14] Dight PM. Improvements to the stability of rock walls to open pit mines. 1983. Ph.D Thesis, Monash
518 [15] Dulacska H. Dowel action of reinforcement crossing cracks in concrete. Journal of the American
519 Concrete Institute. 1972; 69:754-757
520 [16] Spang K, Egger P. Action of fully-grouted bolts in jointed rock and factors of influence. Rock
521 Mechanics and Rock Engineering. 1990;23:201-229 doi:10.1007/bf01022954
522 [17] Ferrero AM. The shear strength of reinforced rock joints International Journal of Rock Mechanics
523 and Mining Sciences & Geomechanics Abstracts. 1995;32:595-605
524 doi:[Link]
525 [18] Pellet F, Egger P. Analytical model for the mechanical behaviour of bolted rock joints subjected to
526 shearing. Rock mechanic and rock engineering. 1996;29:73-97
527 [19] Aziz N, Pratt D, Williams R. Double shear testing of bolts. In: Aizz N (ed) Coal 2003: Coal
528 Operators' Conference, University of Wollongong & the Australasian Institute of Mining and
529 Metallurgy, 2003. pp 154-161
530 [20] Grasselli G. 3D Behaviour of bolted rock joints: experimental and numerical study. International
531 Journal of Rock Mechanics and Mining Sciences. 2005;42:13-24
532 doi:[Link]
533 [21] Jalalifar H. A new approach in determining the load transfer mechanism in fully grouted bolts.
534 University of Wollongong. 2006
535 [22] Jalalifar H, Aziz N. Analytical Behaviour of Bolt–Joint Intersection Under Lateral Loading
536 Conditions. Rock Mechanics and Rock Engineering. 2010a;43:89-94 doi:10.1007/s00603-009-
537 0032-6
538 [23] Jalalifar H, Aziz N. Experimental and 3D Numerical Simulation of Reinforced Shear Joints. Rock
539 Mechanics and Rock Engineering. 2010b;43:95-103 doi:10.1007/s00603-009-0031-7
540 [24] Li X, Aziz N, Mirzaghorbanali A, Nemcik J. Behavior of Fiber Glass Bolts, Rock Bolts and Cable
541 Bolts in Shear Rock Mechanics and Rock Engineering. 2016b;49:2723-2735 doi:10.1007/s00603-
542 015-0907-7
543 [25] Ludvig B Shear Tests on Rockbolts. In: Proceedings of the International Symposium on Rockbolting,
544 Abisko, Balkema, 1983. pp 113-123
545 [26] Chen Y, Li CC. Performance of fully encapsulated rebar bolts and D-Bolts under combined pull-and-
546 shear loading. Tunnelling and Underground Space Technology. 2015; 45:99-106
547 doi:[Link]
548 [27] McHugh E., Signer SD. Roof bolt response to shear stress: laboratory Analysis. In: In: Proceedings
549 18th international conference on ground control in mining, Morgantown, WV, 1999. pp 232-238
550 [28] Haile A. A mechanistic evaluation and design of tunnel support systems for deep level South African
551 mines. University of Natal, Durban, South Africa. 1996b.
552 [29] Wei W, Jiang Q, Peng J. New Rock Bolt Model and Numerical Implementation in Numerical
553 Manifold Method. International Journal of Geomechanics. 2016;0:E4016004
554 doi:doi:10.1061/(ASCE)GM.1943-5622.0000669
555 [30] Bahrani N, Hadjigeorgiou J. Explicit reinforcement models for fully-grouted rebar rock bolts. Journal
556 of Rock Mechanics and Geotechnical Engineering. 2017;9:267-280
557 doi:[Link]
558 [31] Jalalifar H, Aziz N, Hadi M. The effect of surface profile, rock strength and pretension load on
559 bending behaviour of fully grouted bolts. Geotechnical and Geological Engineering.
560 2006;24:1203-1227 doi:10.1007/s10706-005-1340-6
561 [32] Timoshenko S, Lessels J. Applied elasticity. Pennsylvania:Westinghouse Technical Night School
562 Press. 1925.
563 [33] Maekawa K, Qureshi J. Computational model for reinforcing bar embedded in concrete under
564 combined axial pullout and transverse displacement. Proceeding of JSCE. 1996;31:227-239
565 [34] Moradi AR, Soltani M, Tasnimi AA. A Simplified Constitutive Model for Dowel Action across RC
566 Cracks Journal of Advanced Concrete Technology. 2012;10:264-277 doi:10.3151/jact.10.264
567 [35] Soltani M, Maekawa K. Path-dependent mechanical model for deformed reinforcing bars at RC
568 interface under coupled cyclic shear and pullout tension. Engineering Structures. 2008;30:1079-
569 1091 doi:[Link]
570 [36] Marcus H. Load carrying capacity of dowels at transverse pavement joints SCI Journal Proceedings.
571 1951; 48:169-184
572 [37] Soroushian P, Obaseki K, Rojas MC. Bearing strength and stiffness of concrete under reinforcing
573 bars. ACI Materials Journal. 1987;84:179-184 doi:10.14359/1885
574 [38] Dei Poli S, Di Prisco M, Gambarova PG. Shear response, deformation, and subgrade stffness of a
575 dowel bar embedded in concrete. ACI Journal Structural Journal. 1992; 89:665-675
576 [39] Schubert P. Das Tragvermögen des mörtelversetzten Ankers unter aufgezwungener
577 Kluftverschiebung. PhD Thesis, Montan-Universität, Leoben. 1984.
578 [40] Stjern G. Practical performance of rock bolts. Trondheim, Norway: University of Trondheim. 1995.
579 [41] Goris JM, Martin LA, Curtin RP. Shear behavior of cable bolt supports in horizontal, bedded
580 deposits. In: 15th International Conference on Ground Control in Mining, Colorado School of
581 Mines, 1996. pp 511-521
582