You are on page 1of 32

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/347178393

Tension and shear anchorage systems for limestone structures

Article  in  Construction and Building Materials · November 2020


DOI: 10.1016/j.conbuildmat.2020.121616

CITATIONS READS

7 1,852

6 authors, including:

Devina Shedde Jason Ingham


University of Auckland University of Auckland
6 PUBLICATIONS   15 CITATIONS    536 PUBLICATIONS   6,703 CITATIONS   

SEE PROFILE SEE PROFILE

Ivan Giongo Dmytro Dizhur


Università degli Studi di Trento University of Auckland
64 PUBLICATIONS   616 CITATIONS    188 PUBLICATIONS   2,064 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Observations and Categorisation of Building Envelopes After Being Subjected to Earthquakes and Suggested Improvements to Existing Testing Protocols View project

Post-installed steel anchors in concrete View project

All content following this page was uploaded by Devina Shedde on 22 December 2020.

The user has requested enhancement of the downloaded file.


1 TENSION AND SHEAR ANCHORAGE SYSTEMS
2 FOR LIMESTONE STRUCTURES
3 Silvia Porcarelli1, Devina Shedde2, Ziren Wang3, Jason M. Ingham4, Ivan Giongo5, Dmytro Dizhur6
4 (1) Professional engineer, Piazza Zvanì 3, San Mauro Pascoli (FC), 47030, Italy, silvia.porcarelli@gmail.com
5 (2) Undergraduate student, Department of Civil & Environmental Engineering, University of Auckland, Auckland 1010, New
6 Zealand, dshe226@aucklanduni.ac.nz (Corresponding author)
7 (3) Undergraduate student, Department of Civil & Environmental Engineering, University of Auckland, Auckland 1010, New
8 Zealand
9 (4) Professor, Department of Civil & Environmental Engineering, University of Auckland, Private Bag 92019, Auckland 1010,
10 New Zealand, j.ingham@auckland.ac.nz
11 (5) Assistant Professor, Department of Civil, Environmental and Mechanical Engineering, University of Trento, via Mesiano 77,
12 I-38123 Trento, Italy, ivan.giongo@unitn.it
13 (6) Senior Research Fellow, Department of Civil & Environmental Engineering, University of Auckland, Private Bag 92019,
14 Auckland 1010, New Zealand, ddiz001@aucklanduni.ac.nz
15
16 Author statement:
17 Silvia Porcarelli: Formal analysis, Writing-Original draft preparation; Devina Shedde: Investigation, Data
18 curation, Writing-Reviewing and Editing; Ziren Wang: Investigation; Jason Ingham: Reviewing and Editing,
19 Supervision; Ivan Giongo: Writing-Reviewing and Editing; Dmytro Dizhur: Conceptualization, Reviewing and
20 Editing, Project administration.

21 ABSTRACT
22 Using anchorage systems to improve and repair existing buildings has been a favoured technique for many years.
23 Whilst there has been an abundance of anchor performance analysis associated with anchorage into concrete,
24 there has conversely been a comparative lack of investigation concerning the behavioural response of anchorage
25 systems in low strength materials such as sedimentary stone. In the study reported herein, 78 adhesive anchors
26 were subjected to tensile and shear testing in limestone blocks that were sourced from Oamaru, New Zealand.
27 Three nominal anchor rod diameters (ɸ12, ɸ16, ɸ20 mm) and two different bond agents (two-part epoxy
28 adhesive, cement-based grout) were employed in the test regime. 54 anchors were installed to three varying
29 depths of embedment described as a multiple of the anchor diameter (3ɸ, 5ɸ, 10ɸ), and were subjected to
30 monotonic tension loading, and an additional 24 anchors were installed at one constant depth and subjected to a
31 semi-cyclic shear loading regime. The results of these tension and shear experiments are summarised, with
32 several anchor failure mechanism identified. Under tension, epoxy anchor behaviour was governed by the
33 limestone strength, whereas grout anchor failure was governed by the bond strength. Anchor capacities in shear
34 for epoxy and grout anchors were similar, however displacements were typically larger for grout anchors due to a
35 low elastic modulus of grout. Also included is an investigation of the influence of moisture content on the
36 strength properties of the limestone. In addition, a predictive equation for anchorage capacity within limestone is

1
37 developed by applying analytical formulations originally conceived for anchorages installed into concrete, in
38 order to study their applicability to weaker natural stone. The study provides a practicable baseline for future
39 testing and the design of anchorage in sedimentary limestone. This dataset could be supplemented in future
40 studies via investigations into alternative loading regimes, and/or the construction of limestone walls to
41 investigate how local wall morphology impacts anchorage performance.

42 Keywords: Adhesive anchor; Pull-out test; Shear forces; Tensile strength; Failure mechanism; Limestone.

43 NOTATION

44 c lateral confinement pressure


45 𝑑! hole diameter
46 E Young’s Modulus
47 𝐹" yield force of the anchorage
48 𝐹# ultimate pull-out force
49 𝐹#,%&'( average value of ultimate pull-out force for each anchor series
50 𝑓) concrete compressive strength
51 𝑓)*(+,(&- limestone confined strength
52 𝑓),. limestone compressive strength
53 𝑓+,. limestone flexural strength
54 𝑓/,. limestone tensile strength
55 𝑓0,. limestone local compressive strength
56 ℎ)*(& depth of cone failure evaluated as [1]
57 ℎ&+ anchor embedment depth
58 𝑀10 plastic bending moment of the steel rod
59 W moisture content of the limestone block sample
60 x limestone stress-distribution length
61 𝜏%'2 maximum bond strength for adhesive anchor
62 𝜏! uniform bond strength
63 𝜎 stress within the limestone block samples
64 𝜆3 elastic constant evaluated as [2]
65 ɸ anchor diameter
66 ν Poisson’s ratio

67 1. INTRODUCTION
68 The global advancement of engineering knowledge and design standards has rendered many existing
69 unreinforced stone masonry (URM) structures as deficient, particularly those found in seismic regions. The low
70 tensile strength of stone masonry structures makes vertical elements significantly prone to force actions that are

2
71 orthogonal to their plane. These force actions produce out-of-plane mechanisms during earthquakes, which have
72 been observed in older as well as more recently constructed stone masonry structures [3] [4]. Attempts to
73 mitigate the risk of structural failure aim to improve connections between diaphragms and vertical elements,
74 utilising the in-plane flexural and shear strengths of the walls [5]. One of the foremost retrofitting techniques
75 used as part of a repair strategy for masonry buildings is the use of anchorage systems, which assist in
76 transferring tensile forces induced within the surrounding material [6]. In particular, injected anchors are utilised
77 to connect existing walls with new members, to stabilise cracked masonry structures, and to generally strengthen
78 a masonry building by assisting in the dissipation of energy that occurs due to seismic or other dynamic
79 excitation. [7]. However, experimental and numerical analyses performed to date largely focus on the tensile and
80 shear behaviour of anchors embedded in concrete as the substrate material [8] [9] [10] [11] and there is a
81 comparative lack of data concerning the retrofit capability of anchors within weak sedimentary stone [12]. As
82 part of the study presented herein, an experimental campaign was performed with the purpose of understanding
83 the behaviour of adhesive anchors when installed into limestone, by investigating how the failure mechanisms of
84 anchorage installed in this material are induced when subjected to tensile and shear loading [12]. To the authors’
85 knowledge, testing of anchorage within limestone was not performed within Australasia prior to the testing
86 reported herein. Therefore, the study intended to introduce a valuable ‘baseline’ dataset to this field of research,
87 and to act as a point of departure for future studies to build on. The particular limestone used in this study was
88 sourced from Oamaru, New Zealand, being a stone that is commonly used in the construction of a large
89 inventory of the heritage masonry buildings in New Zealand [13]. The objectives of the study included: (1) to
90 establish how certain anchor design parameters such as embedment depth, anchor diameter and bonding agent
91 affect anchorage performance; and (2) to conduct a comparative analysis, in terms of ultimate load, between the
92 experimental results and the results of analytical values derived from literature.
93 Included herein is a literature review concerning the types of anchorages utilised in seismic retrofit and their
94 theoretical failure behaviour when subjected to tension and shear loading. Following this review the details of an
95 experimental campaign are presented, composed of: (a) characterisation of materials; (b) a description of tension
96 and shear test apparatus and installation procedure; and (c) analysis of the test results, which consists of a
97 comparative evaluation between experimental and analytical values for the tension and shear tests.
98

99 1.1 BACKGROUND

100 Current anchorage systems are available in the form of mechanical or adhesively based anchors. Mechanical
101 anchors establish friction between the sides of the drilled hole and the fastener lugs to facilitate load transfer [14],
102 while adhesive anchors utilise a bonding agent that adheres and transmits forces by bond to the anchor and the
103 surface of the substrate material in the axial direction [6]. The nature of the adhesive system is such that no
104 additional stress is introduced to the substrate as with expansion anchors [15]. Bonded anchors can be generally
105 classified into two categories according to the bonding agent used, which may be either an epoxy or a cement-
106 based matrix. However, the use of adhesive anchors within older buildings is hindered in some countries due to
107 legislation pertaining to heritage structures, where the use of resin-based anchoring systems is prohibited [12].
108 Thus, the experimental campaign reported herein was undertaken to investigate the tension and shear load

3
109 behaviour of adhesive anchors using a chemical adhesive, as well as to investigate the performance obtained
110 when using a cement-based grout as a possible and potentially more compatible alternative for stone masonry
111 buildings.

112 1.1.1 Failure of anchorages subjected to tensile loads

113 Adhesive anchors display several modes of failure when subjected to ultimate tensile loads, as shown in Figure 1.
114 The failure mode that ultimately evolves is dependent upon a number of variables constituting the anchorage
115 system.

116
117 Figure 1. Anchor failure modes when subjected to tensile loading [16]

118 Anchor rod failure (Figure 1 (a)) involves yielding and/or fracture of the anchor. For anchor rod failure to occur,
119 the tensile strength of the anchor rod must be lower than the embedment strength n of the anchor [1]. Bond or
120 pull-out failure (Figure 1 (b)) occurs when the shear stress on the embedded surface area of the anchor exceeds
121 the adhesive bond strength prior to any other mode of failure occurring. Figure 2 shows the three possible
122 failure modes by which the bond may fail. Figure 2 (b1) shows a full bond failure, typically represented by the
123 anchor rod being fully covered by the adhesive layer, whereas Figure 2 (b2) represents a partial bond failure
124 where the anchor rod is pulled out whilst being only partially covered by the adhesive layer. A partial bond failure
125 involves mobilisation of a portion of the tested anchor. Therefore, the lateral contact area of the anchor rod
126 involved in the failure strength is reduced, which in turn reduces the attained failure strength [15]. Bond failure
127 may also occur at the interface between the tested rod and the adhesive layer, as shown in Figure 2 (b3). This
128 failure mode represents sliding between these two materials and full detachment of the adhesive cover from the
129 anchor rod. Bond failure has frequently been attributed to poor preparation of the drilled anchor hole where
130 holes that were not properly cleaned contained loose particles which obstructed contact between the adhesive
131 and the surrounding material [16], thus indicating the importance of standardising installation conditions and
132 procedures.

133
134 Figure 2. Pull-out failure classification (“covered” refers to the adhesive covering the embedded length of the anchor) [1]

4
135 Cone failure (see Figure 1 (c)) is characterised by the development and propagation of a circumferential cone-
136 shaped failure surface that originates at the bottom of the corresponding anchor [17]. Full cone failure has been
137 principally observed in anchors with a low embedment depth, whereas for anchors installed at deeper
138 embedment depths it has been observed that a combination of cone and bond failure occurs [1]. Figure 1 (d)
139 shows the combined cone-bond failure mode, characterised by the formation of a shallow cone at the top of the
140 anchor rod embedment depth, followed by sliding between the adhesive layer and the material substrate over the
141 lower portion of the embedment depth.
142 Anchor layout significantly influences the overall performance of anchorage systems. An adequate edge distance
143 to the anchor installation typically prevents brittle failure modes such as splitting failure (Figure 1 (e)) and enables
144 the development of full cone failure without reducing the load-bearing capacity of the anchorage [18], avoiding
145 the failure type shown in Figure 1 (f).

146 1.1.2 Failure of anchorages subjected to shear loads

147 Typical failure modes of adhesive or mechanical anchors when subjected to shear loading are identified as: a)
148 anchor failure, b) pull-out failure, c) pry-out failure, d) breakout failure [14] and e) splitting failure [19], as shown
149 in Figure 3. Figure 3(a) shows the anchor shear failure mode, which involves the anchor yielding in conjunction
150 with spalling and cracking of the surrounding substrate material. Pull-out failure may present itself during shear
151 loading if the anchor has displaced to such an extent that it becomes positioned at an angle to the substrate
152 surface. At this point, the applied load generates component forces within the anchor, causing the anchor to bear
153 against the stone whilst simultaneously being pulled out. This pull-out mechanism is similar to that observed
154 when subjected to tensile loading, with loss of adhesion and sliding between the bond surface and the substrate
155 as a result of a tensile stress distribution along the embedment depth of the anchor (Figure 3(b)). Pry-out failure
156 occurs for fasteners with shallow embedment depth, generating an anchor rotation in the opposite direction to
157 the applied shear load [20] and localised breakout of the substrate material, as shown in Figure 3 (c). Anchor rods
158 installed close to the free edge generally present a breakout failure characterised by the formation of a cone-
159 shape on the material surface (Figure 3 (d)), while transverse splitting predominantly occurs in the direction of
160 the closer edge, causing a brittle separation of the sample in two parts [19], as shown in Figure 3 (e). Past
161 research has established that beyond a minimum of 10ɸ, the embedment depth does not influence the shear
162 capacity of anchors [21]. This conclusion was applied in the reported experimental campaign, by considering
163 embedment depth as a fixed variable for anchors loaded in shear.

164
165 Figure 3. Anchor failure modes under shear loading

5
166 In accordance with the reviewed literature and the scope of this study, critical parameters for assessing the
167 capacity of anchorage in shear and tension were adhesive strength, strength and moisture level of the substrate
168 material, quality of hole cleaning, and the loading type/rate, such as being monotonic or cyclic loading [1].
169 Specifically concerning tension loading, variation of the embedment depth was also a critical parameter
170 influencing the tensile failure mechanisms stated above, in order to obtain a holistic range of data that would
171 help to characterise the substrate material [12]. Furthermore, the limestone utilised in this study was in the form
172 of large blocks as opposed to constructed masonry walls, thus ensuring that the aforementioned failure
173 mechanisms were induced. When inserted into a masonry wall made of Oamaru limestone and mortar, anchor
174 performance in tension and shear may vary due to local wall morphology. Parameters such as failure mode,
175 capacity and/or stiffness may be favoured by the weak compressible layers of the surrounding mortar: for
176 example, a tensile force may result in pull-out of an entire limestone block, or applied shear may generate
177 rotation of the stone block.
178

179 2. MATERIALS

180 2.1 NATURAL STONE

181 The limestone blocks used in the experimental campaign were acquired from an active stone quarry located in
182 Oamaru, New Zealand. The stone blocks were of two different sizes, 1300 x 700 x 400 mm and
183 900 x 600 x 400 mm. In order to study the mechanical properties of the limestone, three test types were
184 performed:
185 1. Uniaxial compressive tests (see Figure 4a) were undertaken for six 50 mm cubes in accordance with [22],
186 to determine the compressive strength (𝑓),. ) and the Young’s Modulus (𝐸) of the limestone.
187 2. Three-point bending tests were performed on eight 320 x 50 x 100 mm samples to evaluate the flexural
188 strength .𝑓+,. / of the stone in accordance with [23].
189 3. Indirect tensile tests (Brazilian tests) (see Figure 4b) were performed for twelve cylindrical samples of
190 60 mm diameter and 60 mm height in accordance with [24] - [25], to determine the tensile strength
191 (𝑓/,. ) of the limestone.

(a) Uniaxial compressive test (b) Loading plate and splitting failure mode for indirect tensile test
192 Figure 4. Experimental tests conducted on limestone samples at ambient moisture level

193 Table 1 shows the experimentally attained mechanical properties of the Oamaru limestone. Each test was
194 performed with samples at ambient moisture level, in order to simulate realistic moisture conditions within the

6
195 material. These properties were used in the subsequent analytical formulation to predict the pull-out and shear
196 strength of anchorage within a limestone substrate.
197 Table 1. Mechanical properties of Oamaru limestone

Mean Value St. Dev. CoV


Stone properties
(MPa) (MPa) (-)
𝒇𝒄,𝒔 5.94 1.04 0.18
𝒇𝒇,𝒔 2.64 0.19 0.07
𝒇𝒕,𝒔 0.69 0.18 0.25
E 756.40 254.37 0.33
198 Evidence in literature [26] has shown that the mechanical performance of sedimentary stones can be significantly
199 affected by the stone hygroscopic nature such that variation in moisture within the stone represents an important
200 factor when predicting the mechanical performance of the material. Correspondingly, compressive and indirect
201 tensile tests on cylindrical stone samples at several moisture levels were performed on the tested limestone.
202 Figure 5 provides a graphical representation of the test results, illustrating the effect of moisture levels on
203 uniaxial compressive and indirect tensile strengths. For both tests, the properties and the performance of the
204 limestone were significantly influenced by the moisture level, indicating a general strength decrease with an
205 increase in moisture levels. It is observed that the limestone exhibited the highest mechanical strength when it
206 was in a fully dry state. For moisture levels greater than 4% it is evident that the strength parameters were
207 relatively consistent up to the full saturation level of approximately 12% and that these parameters can be
208 approximated as 6.71 MPa and 0.69 MPa respectively.

(a) Compressive strength vs moisture content (b) Tensile strength vs moisture content
209 Figure 5. Influence of the water moisture on the mechanical properties of the limestone

210 2.2 STEEL ANCHOR RODS

211 The threaded steel anchor rods that were used for all tests undertaken in the experimental campaign were of
212 class 5.8 (ultimate tensile strength 500 MPa and yield strength 400 MPa [27]). Three nominal diameters were
213 selected for experimentation: 12 mm, 16 mm and 20 mm.

7
214 2.3 CEMENT-BASED GROUT

215 Cement-based grout was one of two adhesive systems used within this study for anchor installation. Grout has a
216 long curing period, as well as being sensitive to the water content that is present in the mixture during the
217 entirety of the curing period. Due to the high absorbency of the porous limestone, grout adhesive was difficult to
218 install despite using the manufacturer's recommendation for the maximum water-to-grout ratio of 6:25 (see
219 Table 2). The rapid loss of moisture in the grout led to a large reduction in working time, preventing full
220 installation of the anchors. Pre-saturating the drilled holes was not investigated as a possible solution due to
221 having been shown to provide minimal benefit, as well as lacking practicality with respect to real site applications
222 [4]. As such, a water-to-grout ratio of 6:22 for the grout adhesive was utilised, in order to provide a higher water
223 content than that given by the manufacturer’s maximum specified ratio of 6:25. The 2% increase in water
224 content enabled greater flow of the material, and sufficiently extended the amount of time available for anchor
225 installation. Six 50 mm grout cubes (capped with dental plaster) that were formed from a mixture with the
226 increased water content ratio were tested in uniaxial compression after 5 days of curing. Although a water
227 content that was slightly higher than the maximum recommended content was adopted, compressive strength
228 results coincided closely with the strengths stated by the manufacturer (Table 2). However, it was made apparent
229 that the absorption capacity of the stone is a significant aspect to be addressed in the development of anchorage
230 installation guidelines for limestone.
231 Table 2. Grout strength properties

Water to grout mix ratio


6:25
(L:kg)
Technical data 24-hour compressive
as per strength 10
manufacturer’s (MPa)
recommendation 5-day compressive
strength 30
(MPa)
Water to grout mix ratio
6:22
(L:kg)
Tested cube samples 5-day compressive Mean
St. Dev. CoV
according to BS EN strength - tested strength
1926:2006 [36] (MPa) (MPa) (MPa) (-)
Modified
1 30.77
technical data
2 31.52
3 34.47
32.3 4.34 0.13
4 36.96
5 35.42
6 24.87
232 2.4 EPOXY RESIN

233 Epoxy resin is specifically utilised for fastener installation into solid substrate materials such as natural stone,
234 masonry or concrete. In general, resin bonded anchors require a shorter period of curing to attain resistance and
235 are largely used when a fast setting time is required. The higher viscosity of epoxy resin allows for easy
236 installation of the anchors, resulting in epoxy resin being better suited that cement-based grout for injection in
237 deep and/or horizontally oriented holes. Epoxy resin is characterised by a superior compression strength and
238 stronger adhesion to substrate materials than for polyester or vinyl ester resins. The epoxy resin utilised in the

8
239 experimental campaign was specially designed for anchors installed into solid substrates such as concrete,
240 masonry and stone and is a two-part adhesive material which consists of the epoxy resin and an amine-based
241 hardener material. The installation system of the two-part epoxy adhesive is constituted by two compartments
242 positioned beside each other that contain the two different materials, plus an injection dispenser and a nozzle
243 both also provided by the manufacturer, with one function of the nozzle being to ensure appropriate mixing of
244 the two materials immediately prior to injection into the drilled hole. The combined resin and hardener provide
245 optimal curing speed as well as high strength and stiffness of the adhesive material within the substrate (see
246 Table 3). All the installation processes of the epoxy resin were conducted strictly in accordance with
247 manufacturer's specifications.

248 Table 3. Epoxy resin strength properties

Bond strength ASTM C882-91


2 day curing 12.4 (MPa)
Technical data
7 day curing 12.4 (MPa)
as per
manufacturer’s Compressive strength ASTM D-695-96 82.7 (MPa)
sheet Compressive modulus ASTM D-695-96 1493 (MPa)
Tensile strength 7 day ASTM D-638-97 43.5 (MPa)

249 3. EXPERIMENTAL TESTING


250 Experimental tensile and shear testing were undertaken to investigate the design parameters identified as being
251 critical to anchorage performance. Specimens were denoted using an alphanumeric key, with tension specimens
252 having a five-field key and shear specimens having a four-field key (see Table 4 and subsequent descriptions).

253 Table 4. Alphanumeric key for specimen identification

Denotation Field 1 2 3 4 5 (shear)


Variable T/S E/G 12/ 16/ 20 3/ 5/ 10 ϕ 1/ 2/ 3 (+ 4/ 5)
254
255 In Table 4, Field 1 denotes the type of loading being applied to the tested anchor rod, being either tension (T) or
256 shear (S) loading. The second field refers to the use of cement-based grout (G) or epoxy (E) as the anchor
257 adhesion material. Field 3 identifies the anchor diameter in millimetres. Field 4 is only applicable to tension tests
258 and identifies the embedment depth as a multiple of the anchor diameter. Field 5 identifies the specimen number
259 for the series and is thus 1-3 for tension testing and 2-5 for shear testing (for shear testing specimen 1 was used
260 to establish the shear loading regime, see Section 3.3).

261 3.1 INSTALLATION

262 Installation of all anchor rods complied with manufacturer specifications where possible and was performed with
263 diligence via the following steps: Holes were drilled into the limestone blocks using a handheld hammer drill.
264 Despite limestone being a relatively soft material, other modes of drilling were not deemed to be adequate to
265 efficiently penetrate through the material. For anchors installed with epoxy as the adhesive, the diameter of the
266 hole was 4 mm larger than the diameter of the respective anchor, as per manufacturer recommendations. For
267 holes designated for grout anchors, the diameter of the hole was 1.5 times the diameter of the respective anchor

9
268 in accordance with literature recommendations [14] [28]. To minimise the presence of loose particles within the
269 drilled holes, holes were cleaned using compressed air in conjunction with a vacuum, and were also inspected to
270 ensure that the drilling had not compromised the integrity of the substrate. Manufacturer recommendations
271 required the use of steel brushes, however during trial testing it was found that the soft nature of the stone made
272 this step counterproductive. Hence, steel brushes were not utilised for the experimental tests.
273 The cement-based grout was placed into the drilled holes using an injection dispenser. It was found to be
274 difficult to implement installation of the grout mix into the horizontally oriented holes, as the grout mix initially
275 flowed back out during its injection and the high absorbency of the limestone significantly reduced available
276 working time. As such, two steps became necessary to implement the grout installation: Firstly, a new grout mix
277 was made with a higher water content ratio in order to increase working time, and secondly, a tube was appended
278 to the end of the injection dispenser to enable deposition of the grout mix at the full depth of the drilled hole.
279 The grout was injected slowly, in conjunction with the tube being extracted from the hole, and this step was
280 followed by immediate insertion of the anchor by slowly rotating it into the hole. The epoxy resin was injected
281 into the holes using the supplied injection dispenser, utilising a narrow tube that was provided for small diameter
282 and long length drilled holes. For both adhesives, the aim was to dispense an amount of adhesive that filled 50-
283 75% of the depth of the hole.

284 3.2 TENSION TEST SETUP

285 A total of 54 anchors were installed in limestone blocks and subjected to direct tension load, taking into account
286 the following variables for the experimental programme: three nominal anchor diameters of ɸ12, ɸ16, and
287 ɸ20 mm, three embedment depths as a multiple of the respective anchor diameter (3ɸ, 5ɸ, 10ɸ), and the use of
288 either epoxy resin or cement-based grout as the bonding agent. Due to time and space constraints, three
289 specimens were installed for each series instead of five as per ASTM standards [29], which was otherwise
290 followed in close accordance with the prescribed testing specification. However, preliminary testing of anchors
291 under tension was performed in a different block of stone (and therefore not included in the paper) in order to
292 develop an initial understanding of the anchorage behaviour within the substrate, and to assist in optimising
293 anchor layouts due to the limited amount of material available. The preliminary testing produced a CoV of 16%
294 for the ultimate loads, which was deemed to be a sensible level of variation given the sedimentary nature of the
295 limestone and the influence of nearby cracks from tested anchors on one anchor in the preliminary testing.
296 Spacing of the anchors was based on an expected 45-degree cone failure base area. Based on experimental testing
297 of anchors installed into clay brick masonry, it has been shown that results from cyclic tensile loading are
298 comparable with those from monotonic loading [4]. Therefore, it was decided to undertake only monotonic
299 loading for anchors in tension in the effort to characterise anchorage tensile behaviour within limestone. For all
300 tests, the tensile load was applied monotonically to the anchors until the ultimate load was attained and
301 subsequent decline in sustained load was measured, indicating that ultimate failure had occurred.
302 The tension test apparatus (see Figure 6) consisted of a reaction rig with a 400 mm wide (clear span) footing to
303 accommodate the anticipated bearing area for cone failure, with an hydraulic jack being placed adjacent to the rig
304 to facilitate manual load application to the anchor, followed by a load cell and a plate and nut to secure the set

10
305 up. A long metal plate was fixed as close as possible to the base of the protruding section of the installed anchor,
306 upon which a linear variable differential transducer (LVDT) was mounted to measure anchor displacement. Load
307 and displacement data were collected using a compact data acquisition system.

(a) Schematic view of the apparatus (b) Typical test set-up for pull-out anchor testing

308 Figure 6. Tension test apparatus

309 3.3 SHEAR TEST SETUP

310 A total of 30 anchors tested were installed to a depth of ten times the diameter of the anchor for the shear
311 testing regime. To mitigate the risk of an anchor being influenced by those adjacent to it, a spacing distance equal
312 to the embedment depth of the anchor was provided on either side of the test anchor. As for tension tests,
313 anchors of diameter ɸ12, ɸ16, and ɸ20 mm were tested using both epoxy and grout adhesive, assigning five
314 specimens to each of the test series. Due to the size of the blocks, full cyclic loading could not be performed
315 without inducing breakout failure due to insufficient edge distance on the side of the anchor closest to the block
316 edge, which would have skewed test results. In addition, under shear cyclic loading the localised crushing of the
317 substrate material has typically governed the failure response around the loaded circumference, assuming
318 sufficient spacing exists around the full circumference [21]. This observation indicates that on the reverse cycle, a
319 mirror behaviour would be achieved. Therefore, a semi-cyclic displacement-controlled loading regime was
320 applied to the anchors (see Figure 7) as prescribed in AS/NZS-2699.2 [30], with a spacing of 2 times the
321 embedment depth between the anchor and the test apparatus as prescribed in ASTM E488-96 [29]. In order to
322 facilitate loading, the first specimen of each series was tested monotonically. Similar to the procedure proposed
323 in other standard codes e.g. EN 12612, the loading regime was established by using either the displacement at
324 peak load or at 50% of the peak load. Experimentally attained displacements at both these loads were too large
325 given the weak nature of the stone, and hence the displacement at the yield load was utilised to formulate the
326 load regime for each anchor size. Due to the limited amount of limestone available, the test to obtain the yield
327 displacement could only be performed once. However, preliminary, or trial testing of anchors under monotonic
328 shear (not included within the paper due to being performed in a different block of limestone) was performed in
329 order to develop an expectation of the anchor’s load/displacement characteristics. This allowed for comparison
330 of the specimen used to establish the loading regime with the trial anchors to ensure that it had displayed a
331 characteristic response under monotonic shear loading.

11
332
333 Figure 7. Regime for semi-cyclic shear testing, adapted from [30]

334 The shear test apparatus was assembled to apply a direct load that acted to pull the test anchor towards the rig,
335 with the reaction rig secured to the limestone block by using mechanical anchors. A metallic shear plate with a
336 coupler welded to it was placed at the base of the anchor to be tested, and a long threaded rod was screwed into
337 the coupler. This rod passed through the metal flange and engaged the hydraulic jack and load cell, which were
338 secured against the outside face of the flange as shown in Figure 8. An LVDT was positioned such that it
339 measured the displacement of the metal plate that sat at the base of the anchor being tested.

(a) Schematic view of the apparatus [29] (b) Test set-up for shear anchor testing
340 Figure 8. Shear test apparatus

341 3.4 TENSION RESULTS

342 Figure 9 shows the percentage distribution of tensile failure modes for grout and epoxy anchors from the
343 experimental campaign, as per the tensile failure modes defined in Figure 1 and Figure 2. A total of 27 tests were
344 used to classify the tensile failure modes of grout anchors and 23 tests were used for the epoxy anchors, from the

12
345 27 total tensile tests conducted for each adhesive agent. Other tensile test results were not included due to setup
346 issues that occurred, which were thought to have affected the resulting failure modes of those tests.

56%

347
348 Figure 9. Classification of tensile failure modes for grout and epoxy anchors

349 Anchors installed using grout as the adhesive exhibited a full bond failure (36%), which typically presented as the
350 full length of the anchor being pulled out from the substrate and being encased in the adhesive layer (Figure 10
351 (a)); while 40% of grouted anchors with shallower embedment depth exhibited a partial bond failure as displayed
352 in Figure 10 (b). Steel failure was not observed in the grouted specimens, due to the weak properties of the
353 limestone and poor adhesion between the grout adhesive and the substrate.
354 As expected, despite the similarity between the tested grout strength and the strength stated by the grout
355 manufacturer (see Table 2), the mechanical capability of the grout was negatively influenced by the high
356 absorbency of the limestone. Consequently, the performance of the grouted anchors was hindered due to the
357 poor quality of the grout adhesion. This reduced adhesion resulted in lower capacities being obtained for the
358 grouted anchors when compared to the epoxy-set anchors, but because the grouted specimens did not exhibit
359 brittle failure they achieved higher ultimate displacement values than for the epoxy anchors. In comparison, the
360 epoxy anchors typically achieved high pull-out capacities and exhibited brittle failure modes in most samples due
361 to little or no anchor withdrawal displacement prior to failure. 56% of epoxy-set samples presented a full cone
362 failure, which includes all epoxy specimens installed to a depth of three times their diameter, along with the
363 T_E_12_5ɸ series (Figure 9).
364 All installed anchors were spaced to accommodate a 45-degree zone of influence as expected for anchors
365 installed in a concrete substrate in accordance with ACI 318 (2011) [31]. However, the angles of the cones that
366 formed for the epoxy anchors typically ranged between 18 to 22 degrees (Figure 10 (c)). The development of
367 shallower cones caused overlapping of the cone base area, consequentially resulting in interference to adjacent
368 tests. Because of detachment of a larger conical volume of substrate material, less of the material was available
369 for subsequent adjacent tests. This overlap will have negatively influenced the measured anchorage capacity of
370 anchors where the limestone surface directly around the anchor had already spalled during testing of an adjacent

13
371 anchor, and it illustrated that the spacing guidelines established for concrete are not immediately transferrable to
372 low strength materials [38].
373 Specimens in the T_E_ 20_5ɸ series and T_E_12_10ɸ series exhibited a combination of cone and splitting
374 failure (Figure 10 (d)), classified as space and edge cone failure in Figure 9. From inspection it was evident that
375 an angled failure surface had formed, with crack propagation causing full detachment of the bond surface from
376 the substrate, likely due to low lateral confinement to the anchor. This failure mechanism occurred in
377 conjunction with large chunks of limestone breaking off, instead of the idealised conical rupture surface (Figure
378 10 (d)).
379 Combined cone-bond failure with a compact, conical failure surface formation was observed for the epoxy-set
380 anchor series T_E_20_10ɸ. This was an unexpected observation given the consistent splitting of the ɸ12 and
381 ɸ16 anchors, in conjunction with the hypothesis that a larger diameter anchor may be more likely to promote
382 splitting in a weak substrate material. However, this result illustrates the influence of deeper embedment depth in
383 mobilising substrate adhesive strength, in compliance with past research [1].

(a) Typical pull-out failure of (b) Anchor partially covered (c) Epoxy full cone failure (d) Epoxy combined/splitting
grout anchors by grout (T_G_20_3ɸ_1) (T_E_12_3ɸ_3) failure (T_E_16_10ɸ_2)
(T_G_16_10ɸ_1)
384 Figure 10. Typical failure types observed following tension testing

385 Table 5 presents the maximum pull-out force achieved for each test series and the corresponding displacement
386 value at maximum force recorded during each test. As expected, regardless of the adhesive agent, the capacity of
387 the anchor increased with a greater embedment depth. For grouted anchors the ɸ12, ɸ16 and ɸ20 anchor series
388 typically displayed similar capacities at embedment depths of 3 and 5 times their respective diameters, with the
389 ɸ20 anchors at an embedment depth of 5ɸ exhibiting a minimal increase in capacity above the ɸ12 and ɸ16
390 anchors at the same depth specification of 5ɸ. This observation was due to full bond failure occurring in the ɸ12
391 and ɸ16 anchors and partial bond failure occurring in the ϕ20 anchors where only a portion of the tested anchor
392 was utilised, as evidenced by the tested anchor typically being partially covered by the grout adhesive (Figure 10
393 (b)). This partial coverage indicated that a reduced lateral contact area was utilised to resist tension loading, thus
394 reducing the ultimate tensile capacity. Partial coverage of the anchor was attributed to either: 1) the grout layer
395 disintegrating during the test as a result of significant absorption of the moisture in the grout by the limestone
396 resulting in a brittle, poor bond quality, or 2) an incomplete bond layer (Figure 2 (b2)) resulting from initial
397 difficulties encountered in installing the highly flowable grout. For a deeper embedment depth of 10ɸ a
398 significant increase in capacity was observed for the ɸ20 grout series compared to the ɸ16 anchors, wherein the

14
399 ɸ20 anchors exhibited a combined failure mode as shown in Figure 1d. Similar response was observed for the
400 epoxy anchorages, with the data reported in Table 5 indicating that for anchors installed using the epoxy
401 adhesive, increasing the anchor diameter from ɸ16 to ɸ20 for embedment depths of 5ɸ and 10ɸ had no
402 significant influence on anchorage capacity, with ɸ20 anchors exhibiting an increase in capacity of only 10%
403 when compared to ɸ16 anchors. For both the epoxy and grouted anchors this observation indicates that for ɸ16
404 and ɸ20 anchors the use of a larger anchor size has no beneficial strength increase. The lack of strength gain for
405 the T_E_20_10ɸ series may also be explained by its comparison with the brittle failure behaviour that occurred
406 for the T_E_16_10 series (Figure 10 (d)): while the latter caused the limestone block to unpredictably and
407 violently split, the ɸ20 anchor exhibited a more ductile response instead of an increase in ultimate load.

(a) Grout anchors at 3ɸ embedment depth (b) Grout anchors at 5ɸ embedment (c) Grout anchors at 10ɸ embedment
depth depth
408 Figure 11. Typical tensile force-displacement curves for grout anchors

409 Figure 11 and Figure 12 show the typical tensile force-displacement response for grout and epoxy anchors
410 installed at different embedment depths, with no data truncated at 30 mm because it was deemed that an anchor
411 which has pulled out to a displacement of 30 mm has clearly failed to be functionally effective. Comparing the
412 trend of the curves for the two different bonding agents, it is apparent that the epoxy and grout anchors
413 exhibited distinctly different failure characteristics.

(a) Epoxy anchors at 3ɸ embedment depth (b) Epoxy anchors at 5ɸ embedment (c) Epoxy anchors at 10ɸ embedment
depth depth
414 Figure 12. Typical tensile force-displacement curves for epoxy anchors

415 Table 5. Maximum load and corresponding displacement for anchors tested in tension

GROUT EPOXY
Disp. at Disp. at
Maximum Maximum
maximum maximum
load Failure mode † load Failure mode †
load load
(kN) (kN)
(mm) (mm)
𝛟12 3𝛟 1 1.8 3.1 Partially covered 5.9 0.8 Stone cone

15
2 1.5 0.7 Partially covered 6.5 0.4 Stone cone
3 4.4 1.6 Partially covered 6.5 0.2 Stone cone
1 5.2 5.7 Fully covered 10.3 4.6 Stone cone
5𝛟 2 2.1 7.2 Fully covered 7.9 4.0 Stone cone
3 3.3 3.3 Fully covered 9.0 1.0 Stone cone
1 4.8 10.7 Fully covered 24.3 2.2 Space and edge cone
10𝛟 2 7.2 1.4 Steel - Bond 25.5 1.7 Space and edge cone
3 6.4 5.7 Steel - Bond 25.2 1.7 N.C.
1 2.0 14.5 Partially covered 8.3 1.4 Stone cone
3𝛟 2 3.0 9.78 Partially covered 7.6 1.8 Stone cone
3 2.3 14.0 Partially covered 6.7 1.7 Stone cone
1 5.7 12.1 Fully covered 17.0 1.2 Stone cone
𝛟16 5𝛟 2 7.5 7.1 Fully covered 15.6 0.1 N.C.
3 5.2 23.7 Fully covered 14.3 1.5 N.C.
1 16.2 17.8 Fully covered 35.8 1.7 Stone splitting
10𝛟 2 17.3 7.7 Fully covered 40.0 1.2 Stone splitting
3 16.0 25.9 Fully covered 38.2 0.4 N.C.
1 2.0 6.0 Partially covered 10.2 0.7 Stone cone
3𝛟 2 2.5 4.6 Partially covered 9.0 0.01 Stone cone
3 2.1 5.1 Partially covered 8.3 1.5 Stone cone
1 5.7 9.9 Partially covered 18.4 0.4 Space and edge cone
𝛟20 5𝛟 2 6.5 7.6 Combined 20.5 1.0 Space and edge cone
3 7.1 9.1 Fully covered 13.5 2.9 Space and edge cone
1 29.9 24.6 Combined 46.1 0.2 Combined
10𝛟 2 20.3 44.3 Combined 38.7 0.8 Combined
3 21.11 9.9 Combined 34.5 0.6 Combined
416 † See Figure 1 and Figure 2 for illustrations of failure modes.
417 N.C. = not classified.

418 When considering the force-displacement response of the tested anchors, grouted anchors exhibited larger
419 displacements compared to anchors installed using epoxy as the adhesive. Upon reaching the maximum failure
420 load, grouted anchors presented a gradual decrease of anchor strength (Figure 11 (a)-(b)), while epoxy anchors
421 exhibited an abrupt drop of capacity after reaching their peak load due to the brittle nature of the cone failure
422 mode, which is particularly evident for all epoxy anchors installed at shallower embedment depths. This indicated
423 that a high strength adhesive may not always be beneficial, particularly in the case of low-strength substrates, for
424 its tendency to promote brittle failure modes as was observed in the cone failure and splitting failure for the
425 epoxy anchors.
426 All grout anchors installed at 10ɸ embedment depth and the T_G_16_5ɸ series presented force-displacement
427 curves with a similar trend, wherein there was a gradual increase in the anchorage capacity and the ultimate
428 failure load was attained at greater displacement values than the displacements at ultimate capacity observed for
429 the other grout series. Figure 11 (c) shows the improvement of pull-out capacity when the anchor diameter is
430 increased for a fixed embedment depth. Considering the average peak load of each diameter series, the strength
431 at a 10ɸ embedment depth was 63% higher for ɸ16 mm than for ɸ12, and 31% higher for ɸ20 than for
432 ɸ16 mm.
433 The brittle failures exhibited by the epoxy anchors explain the general trend of the corresponding force-
434 displacement curves shown in Figure 12. The typical response of the epoxy anchors involved greater ultimate
435 pull-out forces for lower displacement values than measured for the grouted anchors. A high initial stiffness was
436 measured for all anchorage diameters, with anchor strength increasing with an increase in anchorage diameter.
437 The T_E_12_10, T_E_16_10 and T_E_20_10 series reported in Figure 12 (c) were the only samples that
438 showed a gradual decrease in strength after their peak load was attained, reaching final displacement values that

16
439 were comparable to those of the grouted specimens at termination of the test. In assessing the ultimate failure
440 load for epoxy anchors, it is evident that a greater diameter size is not directly proportional to an increase in
441 anchor capacity [38]: 16 mm anchors displayed a 34% higher strength than 12 mm anchors, but only a 4%
442 increase in strength was evaluated from 16 mm to 20 mm anchors. These results illustrate that employing a larger
443 diameter anchor is less advantageous to anchor capacity than increasing the embedment depth, and that for
444 embedment depths which exceed 10 times the anchor diameter, the tensile strength of the steel rod is expected
445 to be attained.

446 3.5 SHEAR RESULTS

447 Based on past literature, it was hypothesised that variation in shear load (within the same bonding agent) would
448 arise via steel yielding and/or localised crushing of the substrate material, and therefore minimal; this hypothesis
449 was proven correct in the experimental testing. Almost all grout and epoxy anchors loaded in shear displayed
450 bending of the steel anchor rods (Figure 13 (a)). Localised crushing of the limestone substrate was observed in
451 front of the anchor that was being loaded, but did not negatively influence the strength of the nearby tested
452 specimens (Figure 13 (b)). This bending of the steel anchor rods led to the formation of a single plastic hinge
453 within the limestone block, which caused the propagation of several cracks adjacent to the adhesive layer along
454 the embedment depth of the tested anchor as shown in Figure 13 (c) (note that following completion of the test
455 the stone was partially removed to expose the anchor rod connection). The test set up was designed with a
456 spacing distance equal to the embedment depth of the anchors in order to prevent a test being influenced by
457 previous testing of an adjacent anchor. For the ɸ12 and ɸ16 anchor series, no splitting of the limestone blocks
458 or influence due to testing of adjacent anchors was observed. Two of the ɸ20 anchors installed in the grout
459 adhesive generated splitting of the limestone block, with an immediate drop of the anchorage shear capacity. The
460 propagation of the splitting cracks occurred in the direction of the nearest free edge, involving the orthogonal
461 side of the limestone block and the nearby installed specimens (Figure 13 (d)). Such brittle behaviour was only
462 observed for ɸ20 grout anchors, highlighting the importance of providing sufficient distance between anchors,
463 as well as sufficient edge distance, for larger rod diameters which generate larger bearing stresses.

(a) Yielding of anchor rod (b) Local crushing in front of (d) Splitting caused by ɸ20
(S_G_16_2) anchor (S_G_12_2) (c) Crushing of the adhesive grout anchors (S_G_20_5)
layer along the anchorage
depth (S_E_20_2)
464 Figure 13. Typical failure types observed following shear testing

17
465 Table 6. Peak shear load and displacement at peak load for anchors installed in grout and epoxy
GROUT EPOXY
Ultimate Ultimate
Disp. Vd,mean CoV Disp, mean CoV Disp. Vd,mean CoV Disp, mean CoV
Load Load
(mm) (kN) (-) (mm) (-) (mm) (kN) (-) (mm) (-)
(kN) (kN)
2 7.92 12.20 10.7 10.82
3 6.67 3.19 10.5 6.09
ϕ12 8.77 0.30 10.35 0.62 10.29 0.04 7.69 0.32
4 4.49* 1.14* 9.6 5.43
5 11.72 15.67 10.3 8.41
2 18.00 10.05 15.1 8.18
3 15.81 13.41 15.5 9.56
ϕ16 16.15 0.09 11.96 0.12 16.01 0.05 7.14 0.41
4 14.59 12.38 16.9 2.89
5 16.21 12.01 16.6 7.95
2 21.47 25.21 21.0 13.33
3 23.10* 14.02* 28.6 13.26
ϕ20 21.34 0.01 23.49 0.10 24.39 0.13 11.26 0.21
4 21.20 21.78 25.8 8.55
5 16.10* 15.92* 22.17 9.92
466 * - splitting of the limestone block affected the shear capacity of the anchor

467 The average ultimate load and displacement values reported in Table 6 were evaluated excluding the anchor
468 samples that failed due to splitting of the limestone. For grouted anchors the ɸ16 series had mean peak loads
469 approximately 46% greater than for the ɸ12 series, and the ɸ20 series presented approximately 24% higher
470 capacity than the ɸ16 series. Epoxy anchors loaded in shear exhibited similar capacities to the grout anchors.
471 Performing a comparison of the average value between epoxy and grout anchors, an increased shear strength of
472 approximately 15% was achieved by the ɸ12 epoxy anchors and 10% strength increased occurred for the ɸ20
473 epoxy anchors. The ɸ16 epoxy series presented 1% lower shear capacity than the ɸ16 grouted anchors. In Table
474 6 larger displacement values are reported for grouted anchors than for epoxy anchors, mainly attributed to the
475 low modulus of elasticity of the grout adhesive material.

(a) Grout anchor series (b) Epoxy anchor series


476 Figure 14. Typical semi-cyclic load displacement curves for anchors under shear load

477 Figure 14 shows the response of the grout and epoxy anchor series for different diameters when subjected to
478 semi-cyclic shear loading, indicating that the epoxy and grout anchors performed similarly. The two graphs in
479 Figure 14 display a positive trend between shear strength and anchor diameter, with ductile response up to the
480 ultimate displacement of 8 times the yield displacement. Although the loading regime varied with each anchor

18
481 series due to being established by the displacement at yield load for each series, this observation was consistently
482 observed across all tests.
483 The envelope curves of force-displacement response were obtained for each anchor series by joining the first
484 cycle peak points for each displacement amplitude. Figure 15 and Figure 16 show the backbone curves of grout
485 and epoxy anchors respectively, highlighting in red the mean curve obtained for each anchor diameter. The
486 results in Figure 15 are indicative of where splitting of the limestone block affected the shear capacity of the
487 anchor, as evidenced by the low loads and displacements for S_G_12_4, S_G_20_3 and S_G_20_5 specimens.
488 The S_G_12_2, S_G_20_2 and S_G_20_4 series reported in Figure 15 (a),(b) are the only samples that showed a
489 sudden increase of 30% in stiffness at the last cycle load which was attributed to the anchor rod initiating a
490 tension action and exhibiting a ‘rope’ effect at higher levels of lateral displacement.

(a) ɸ12 grout anchor series (c) ɸ20 grout anchor series
(b) ɸ16 grout anchor series

491 Figure 15. Backbone curve of grout anchor series subjected to shear loading
492 * - splitting of the limestone block affected the shear capacity of the anchor

(a) ɸ12 epoxy anchor series (b) ɸ16 epoxy anchor series (c) ɸ20 epoxy anchor series

493 Figure 16. Backbone curve of epoxy anchor series subjected to shear loading

494 To facilitate the comparison of results for the two adhesive types, the behaviour of the tested anchors was
495 modelled using a bilinear idealisation based on the mean envelope curves using the equivalent energy elastic-
496 plastic (EEEP) approach, as shown in Figure 17. An ultimate displacement of 10 mm was set, recognising that

19
497 experimental response was approximately constant from this displacement onwards. Bilinearisation of the mean
498 envelope curve was performed without considering the rope effect that affected the ɸ20 grout anchors series,
499 such that the bilinear curve was obtained considering the shear load before an increase in stiffness was observed.
500 No significant differences were observed between the responses for the grout and epoxy adhesives, with the
501 most significant difference being evident for the ɸ20 epoxy anchors, which exhibited a 64% increase in initial
502 stiffness and a 17% increase in yield strength when compared to the ɸ20 grout anchors. It was identified that the
503 mean curve for the grouted ɸ20 anchors presented a lower stiffness than for the ɸ12 and ɸ16 counterparts,
504 which was attributed to brittle failure of two tested anchors for the ɸ20 grout series.

505 Figure 17. Bilinear force-displacement curves for grout and epoxy anchor series loaded in shear

506 4. PREDICTION OF PULL-OUT STRENGTH


507 Analytical formulations originally conceived for anchors installed into concrete were applied to the tested
508 anchors set into limestone in order to study their applicability to the substrate. The study reported herein
509 involved a comparative analysis of the ultimate loads from the experimental campaign and from theoretical
510 expressions. Eight theoretical formulations as reported in Table 7 were used to establish the anchorage strength
511 within a limestone substrate, taking into account the different failure modes obtained from the experimental
512 campaign previously described.

513
514 Figure 18. Theoretical models for the prediction of anchor strength when loaded in tension

515 Model 1 represents the elastic bond stress model, which considers the non-linear distribution of bond stress
516 along the embedment depth when the ultimate load is developed [32], reaching a peak value 𝜏%'2 at the surface
517 of the sample block and having the lowest value depth 𝜏! at the end of the anchor. The maximum bond stress
518 𝜏%'2 was taken to be equal to the tensile strength of the stone 𝑓/,. as reported in Table 1, while the uniform

20
519 bond stress 𝜏! was set equal to 0.85 ∙ 𝜏%'2 as reported in [2]. Model 2 represents the uniform bond stress model,
520 which considers a uniform distribution of bond stress 𝜏! along the entire anchor embedment depth. This model
521 has been found to be accurate in predicting the anchorage strength for an embedment depth lower than 25 times
522 the anchor diameter [1]. Model 3 represents an interface bond model, which is based on sliding at the interface
523 between the anchor and the adhesive layer [33]. Models 4 and 5 take into account the combination of cone and
524 bond failure, with Model 4 being a function of the uniform bond stress 𝜏! and Model 5 being dependent on the
525 elastic bond stress distribution 𝜏%'2 [1]. Models 6, 7 and 8 represent full cone failure modes, and differ from
526 each other by the experimental coefficients used in their analytical formulations. These models correlate the
527 resistance of the break-out failure to the tensile strength of the material and consequently to the cone area
528 involved in the failure. Model 7 also accounts for the size effect by assuming that the ultimate load is
529 proportional to the embedment depth to the power of 1.5 instead of raising to the power of 2 as for the surface
530 area of a cone [34]. Model 7 represents the concrete capacity design (CCD) model, that takes into account a 35°
531 break-out cone failure, while Model 8 is called the concrete cone method (CCM) and is based on 45° break-out
532 cone failure [34]. In all equations, the compressive strength used in the expression was the compressive strength
533 of the limestone derived from the aforementioned material characterisation tests (see Table 1).
534 Table 7. Analytical formulations for the prediction of the anchor tensile strength [2]

Model Author Ultimate axial force

Doerr et al. 4-! 5" 6#$


1 Elastic bond stress
(1989) [32] 𝐹# = 𝜏%'2 ∙ 𝜋 ∙ 𝑑! 3 5"
∙ tanh 8
4-!

Uniform bond Cook et al.


2 𝐹# = 𝜏! ∙ 𝜋 ∙ 𝑑! ∙ ℎ&+
stress (1993) [1]
Steel / bond Marti et al.
3 𝐹# = 𝜏 ∙ 𝜋 ∙ 𝑑 ∙ ℎ&+
interface failure (1993) [33]
Combined cone +
Cook et al.
4 uniform bond
(1993) [1] 𝐹# = 0.92 ∙ ℎ)*(& 7 ∙ =𝑓) + 𝜏! ∙ 𝜋 ∙ 𝑑! ∙ ℎ&+
model
Combined cone + Cook et al. 4-! 5" (6#$ 96%&'# )
5
elastic bond model (1993) [1] 𝐹# = 0.92 ∙ ℎ)*(& 7 ∙ =𝑓) + 𝜏%'2 ∙ 𝜋 ∙ 𝑑! 3 5"
∙ tan 8
4-!

Cook et al.
6 Cone model
(1993) [1] 𝐹# = 0.92 ∙ ℎ&+ 7 ∙ =𝑓)

Concrete capacity ACI 349-06


7
design “CCD” [34] 𝐹# = 𝑘 ∙ =𝑓) ∙ ℎ&+ <,= 𝑘 = 7 𝑓𝑜𝑟 𝑝𝑜𝑠𝑡 𝑖𝑛𝑠𝑡𝑎𝑙𝑙𝑒𝑑 𝑎𝑛𝑐ℎ𝑜𝑟 𝑖𝑛 𝑐𝑜𝑛𝑐𝑟𝑒𝑡𝑒

Concrete cone ACI 349-06 -


8 𝐹# = 0.96 ∙ =𝑓) ∙ ℎ&+ 7 31 + 6 ! 8
method “CCM” [34] #$

535 4.1 GROUT ANCHORS

536 Figure 19 displays the strength predictions derived using various analytical models for grout anchors installed in
537 limestone, in conjunction with average values obtained from experimental testing. The comparison of ultimate
538 load between the experimental data and the analytical formulations was performed by taking the average value of
539 each anchor series for each diameter. The hollow blue circle represents the average experimental value for each

21
540 anchor series, without considering the failure mode of each anchor. The blue line represents the maximum and
541 the minimum failure load obtained from the experimental campaign, showing the load deviation from the mean
542 value for each anchor diameter.

543
544 Figure 19. Prediction of failure load for grout anchors installed at 3-5-10ɸ embedment depths

545 For grouted anchors the most accurate prediction of ultimate load was obtained when using the theoretical
546 models that characterised bond failure, whereas the three breakout cone models, 6, 7 and 8 typically produced
547 higher anchor strength capacities than those attained from experimental testing. Therefore, only the theoretical
548 models which represent the bond failure mode have been shown in Figure 19, in order to focus on the models
549 which produced the closest approximation to the experimentally obtained failure loads. Almost all experimental
550 strength values for grouted anchors were greater than their theoretical counterparts, with only the ɸ20 anchor
551 series installed at 3ɸ embedment depth exhibiting lower failure loads. This observation is likely due to the
552 particular combination of a shallow embedment depth and a large diameter anchor. The lower failure loads
553 would have been further exacerbated by the general poor curing of the grout adhesive, which was a result of the
554 highly porous nature of the limestone (as discussed in Section 3.4). ɸ12 and ɸ16 anchors installed at 5ɸ
555 embedment depth displayed large variability in peak load. However, when considering the average value for both
556 anchor diameters it was identified that none of the theoretical models exceeded the experimental anchorage
557 capacity. Anchors installed at 10ɸ embedment depth most consistently exhibited the greatest elevation in
558 strength above the theoretical models, with the ɸ20 anchor samples achieved ultimate loads that were
559 approximately 38% higher than the capacities predicted by the theoretical formulations.
560 Table 8 presents a summary of the most accurate prediction model obtained for each configuration of grouted
561 anchors, accounting for both anchor diameter and anchor embedment depth. The results demonstrate that
562 different bond failure models routinely underpredicted the average peak strength of grouted anchors and that the
563 most accurate theoretical models correctly described the behaviour of the anchor failure modes that occurred
564 during testing. From Table 8 it is evident that the theoretical models better predict the experimental capacity of

22
565 grouted anchors installed at shallower embedment depths, while anchors installed at greater embedment depths
566 achieved capacities approximately 30% higher than that predicted by the theoretical formulations.

567 Table 8. Most accurate pull-out prediction models for grouted anchors (see Table 7 for model details)

ɸ 12 16 20
4. Combined
Theoretical Model 1. Elastic bond 2. Uniform bond
Cone+Uniform
Actual Failure Mode
Partially covered Partially covered Partially covered
(see Fig. 2)

𝐹#,%*-&0 (kN) 1.38 2.34 2.32
𝐹#,/&./ (kN) 1.65 2.43 2.20
Model under or over Underpredicted Underpredicted Overpredicted
Error (%) 16% 4% 5%
2. Uniform
Theoretical Model 1. Elastic bond 1. Elastic bond
bond
Actual Failure Mode
Fully covered Fully covered N.C.
(see Fig. 2)

𝐹#,%*-&0 (kN) 2.25 4.00 6.25
𝐹#,/&./ (kN) 3.53 6.08 6.43
Model under or over Underpredicted Underpredicted Underpredicted
Error (%) 36% 34% 3%
5. Combined 5. Combined 5. Combined
Theoretical Model
Cone+Elastic Cone+Elastic Cone+Elastic
Actual Failure Mode
N.C. Fully covered Combined
(see Fig. 2)
10ɸ
𝐹#,%*-&0 (kN) 5.33 9.48 14.81
𝐹#,/&./ (kN) 6.13 16.50 23.96
Model under or over Underpredicted Underpredicted Underpredicted
Error (%) 13% 43% 38%
568 N.C. = not classified.

569 4.2 EPOXY ANCHORS

570 For epoxy anchors the theoretical models that best approximated the anchor failure load were related to cone
571 failure modes, which coincides with the typically observed behaviour for epoxy anchor failure (see Figure 9),
572 with the theoretical models corresponding to bond failure generally displayed a lower strength than that achieved
573 experimentally. Thus, these bond failure models were not incorporated into Figure 20 in order to focus on
574 models that achieved closer prediction to the experimentally attained capacities.
575

23
576
577 Figure 20. Prediction of failure load for epoxy anchors installed at 3-5-10ɸ embedment depths

578 The capacities of anchors installed at the 3ɸ embedment depth were adequately predicted using different cone
579 models. Model 8 represents the concrete cone method (CCM) and produced an underestimation of strength for
580 ɸ12 and ɸ16 anchors but a strength overestimation for ɸ20 anchors, whereas Model 7 (CCD) provided an
581 underestimation of the anchor strength for all anchor diameters and also presented larger errors in the capacity
582 predictions. The capacity prediction for anchors installed at 5ɸ embedment depth were more accurately
583 represented by the three cone models 6, 7 and 8, all of which presented similar strength prediction results for
584 ɸ12 and ɸ16 anchors, whereas ɸ20 anchor samples installed at 5ɸ embedment depth displayed a lower capacity
585 than ɸ16 anchors due to splitting failure that occurred during the tests. For the ɸ20 anchors having an
586 embedment depth of 5ɸ the CCD model was the only model that accurately predicted the average failure load of
587 the test set, underestimating the anchorage capacity by 4%. For the deepest embedment of 10ɸ, one of the
588 theoretical models characterising bond failure was expected to be the most closely-correlated prediction model,
589 as has been promulgated in past research [1], whereas the most common experimental failure mode for this
590 embedment depth was found to be characterised by a combined cone-bond failure mode that occurred for 13%
591 of samples, albeit with a shallow cone formation (see Figure 9 and Figure 18b). Model 5 that represents the
592 combined cone and uniform bond model was the only theoretical model that underpredicted the average peak
593 load of the ɸ20 series. However, the ɸ12 and ɸ16 anchors installed at 10ɸ embedment depth failed with a
594 modified cone and space edge failure. Although sufficient edge spacing was provided, it is likely that a wider
595 volume of limestone was engaged, as was observed for the anchors installed at 3ɸ embedment depth, resulting in
596 the modified cone and space edge failure. The most accurate capacity prediction for these anchors was obtained
597 using Model 7, with an error of 12% for ɸ12 anchors and 10% for ɸ16 anchors respectively.
598 Table 9. Most accurate pull-out predictions for epoxy anchors (see Table 7 for model details)

ɸ 12 16 20
Theoretical Model 8. CCM 8. CCM 7. CCD
Actual Failure Mode Stone Cone Stone Cone Stone Cone
3ɸ 𝐹#,%*-&0 (kN) 4.32 7.53 7.96
𝐹#,/&./ (kN) 6.30 7.53 9.17
Model under or over Underpredicted Accurate Underpredicted

24
prediction
Error (%) 31% 0% 13%
Theoretical Model 6. Cone Model 6. Cone Model 7. CCD
Space and edge
Actual Failure Mode Stone Cone N.C.
cone
5ɸ 𝐹#,%*-&0 (kN) 7.96 14.16 16.83
𝐹#,/&./ (kN) 9.07 16.94 17.47
Model under or over Underpredicted Underpredicted Underpredicted
Error (%) 12% 16% 4%
5. Combined
Theoretical Model 7. CCD 7. CCD
Cone + Elastic
Actual Failure Mode N.C. N.C. Combined
10ɸ 𝐹#,%*-&0 (kN) 22.12 34.06 24.72
𝐹#,/&./ (kN) 25.0 38 39.77
Model under or over Underpredicted Underpredicted Underpredicted
Error (%) 12% 10% 38%
599 N.C. = not classified.
600 Table 9 provides a summary of the closest prediction models obtained for the epoxy anchors, indicating that in
601 the majority of cases the most accurate results were obtained for specimens where the capacity of the tested
602 anchor was underestimated. The theoretical formulations applied for epoxy anchors presented a lower error than
603 was obtained for the prediction of grout anchors, with an average of 15%. The largest error of 38% was obtained
604 for the ɸ20 anchors at 10ɸ embedment depth. However, considering the large variability of the mechanical
605 properties of the limestone, this level of error may be considered an acceptable result. Improved predictions
606 were obtained for 3ɸ and 5ɸ embedment depths using the three theoretical cone Models 6, 7 and 8, each of
607 which account for failure modes that occurred during the tests for this anchor series. The obtained results
608 indicate that the formulations are conservative but provide a reasonable approximation for anchors installed in a
609 weak substrate.
610 It is recommended that further testing of anchorage within limestone is performed to more comprehensively
611 understand the failure modes encountered for both epoxy and grouted anchor systems. This is in order to
612 determine a confident range of pull-out prediction strengths as well as to augment existing theoretical
613 formulations such that they cater to limestone.

614 5. PREDICTION OF SHEAR STRENGTH


615 The prediction of the anchor shear capacity was performed by applying the limit analysis approach to the tested
616 anchor rods within a limestone substrate [21]. Figure 21 shows the idealisation of the failure mode observed
617 experimentally for grout and epoxy anchors loaded in shear, displaying the activation of a single plastic hinge
618 inside the limestone block due to bending of the steel anchor. The equilibrium equation (1) was applied to the
619 ideal scheme represented in Figure 21, where the local compressive strength (𝑓0,. ) is assumed as a rectangular
620 stress-block distribution whose length is 0.8 times the distance of the plastic hinge from the block surface, in
621 order to take into consideration the fact that the lower strain value close to the plastic hinge does not engage the
622 full material strength.

25
⎧ 𝑉 = 𝑓&,' ∙ 𝑑( ∙ (0.8 ∙ 𝑥)
⎪𝑀)& = 𝑉 ∙ 𝑥 − 𝑓&,' ∙ 𝑑( ∙ (0.8 ∙ 𝑥) ∙ (𝑥 − 0.4 ∙ 𝑥)
623 → 𝑉 = =2 ∙ 𝑀)& ∙ 𝑓&,' ∙ 𝑑( (1)
⎨ *!"
⎪ 𝑡ℎ𝑒𝑟𝑒𝑓𝑜𝑟𝑒 𝑥 = ;(.,-⋅/
",$ ∙ 2%

(a) Theoretical scheme of anchor failure (b) Bending of the steel anchor rod
624 Figure 21. Idealization of the failure mode for an anchor loaded in shear

625 The local compressive strength of the confined limestone block (𝑓0,. ) was established by assuming that the
626 portion of limestone substrate in front of the anchor is subjected to passive confinement due to the surrounding
627 material (𝑓0,. = 𝑓)*(+,(&- ), and the confined strength of the limestone was evaluated using equation (2) that
628 relates the strength at a bi-axial stress state to the confinement pressure (𝑐) using the ratio of the uniaxial
629 compressive strength (𝑓),. ) to the flexural strength (𝑓+,. ) [35].
+
630 𝑓)*(+,(&- = 𝑓),. + +%,) 𝑐 (2)
$,)

631 Where, the confinement pressure (𝑐) is assumed proportional to the stress (𝜎) applied by the anchor rod,
632 through the Poisson coefficient (𝜈), as shown in Figure 22. The proportional correlation of the confinement
633 pressure is assumed until the stress within the limestone blocks reaches the uniaxial compressive strength (𝑓),. ).

634 𝑐 = 𝑓),. ∙ 𝜈 (3)

635
636 Figure 22. Theoretical scheme of the confinement pressure (c)
637 The values utilised for the above equations were retrieved from Table 1, and a Poisson value of 0.2 was assumed
638 [37]. Once the limestone confined strength was evaluated the predicted shear capacity was dependent on two
639 equations with two unknowns (V, x). Table 10 provides a comparison of the shear capacity between the

26
640 experimental campaign and the results obtained from the two equilibrium equations for grout and epoxy
641 anchors.

642 Table 10. Prediction of the shear failure


GROUT EPOXY
Max. Max.
Disp. Vd,mean CoV Vpred. Err. Disp. Vd,mean CoV Vpred. Err.
ϕ No. Load Load
(mm) (kN) (-) (kN) (%) (mm) (kN) (-) (kN) (%)
(kN) (kN)
2 7.92 12.2 10.7 10.82
3 6.67 3.19 10.5 6.09
ϕ12 8.77 0.3 7.5 -14.8% 10.29 0.04 7.0 -31.6%
4 4.49* 1.14* 9.6 5.43
5 11.72 15.67 10.3 8.41
2 18 10.05 15.1 8.18
3 15.81 13.41 15.5 9.56
ϕ16 16.15 0.09 13.3 -17.8% 16.01 0.05 12.1 -24.3%
4 14.59 12.38 16.9 2.89
5 16.21 12.01 16.6 7.95
2 21.47 25.21 21 13.33
3 23.10* 14.02* 28.6 13.26
ϕ20 21.34 0.01 20.8 -2.8% 24.39 0.13 18.6 -23.9%
4 21.2 21.78 25.8 8.55
5 16.10* 15.92* 22.17 9.92
643 * - Mean values obtained by excluding tests where splitting (*) was observed.

644 From Table 10 it is evident that the limit analysis method under-estimated the shear capacity of all anchor
645 samples. Comparing the average ultimate load for the grout anchor samples, only for ɸ20 mm grout anchors was
646 the shear capacity close to the experimental result with an underprediction error of approximately 3%, while for
647 ɸ12 mm and ɸ16 mm grout anchors the shear prediction was under-estimated by approximately 15% and 18%,
648 respectively. For all epoxy anchors the shear capacity was underestimated by approximately 24%, with a higher
649 error of 32% registered for the ɸ12 mm epoxy anchor. Overall, the limit analysis approach applied to the
650 idealised failure mode results produced conservative predictions compared to the experimental evidence in terms
651 of ultimate anchor shear capacity. However, the formulations represent a valuable tool for calculating the
652 strength of the reinforcement and may be considered as a base point for future analyses to refine.

653 CONCLUSIONS
654 In this study 78 tests were conducted on Oamaru limestone in order to experimentally determine the attainable
655 capacities for post-installed anchors set into soft sedimentary stone, while also investigating the experimental
656 versus analytical correlation between cementitious grout and epoxy anchors. The investigation has resulted in a
657 database of experimental values for adhesive anchorages set into limestone. The following conclusions were
658 drawn from the experimental research reported herein:
659 - Generally, under tension load the capacity increased with increasing anchor embedment depth. Epoxy
660 anchors obtained significantly higher capacities than grout anchors, but failed in a brittle manner via
661 stone splitting or cone failure which caused greater damage to the limestone block. Grout anchors
662 typically failed at lower load, and did not exhibit brittle failure.
663 - The failure modes exhibited were archetypal:
664 o Grouted anchors loaded in tension primarily exhibited bond failure.
665 o Epoxy anchors loaded in tension exhibited a range of full cone failure, splitting failure in
666 conjunction with cone failure, and combined failure.
667 o When loaded in tension there were no cases where yielding of steel anchors was observed.

27
668 o The behaviour of epoxy anchors was governed by the strength of the limestone, whilst the
669 behaviour of grouted anchorage was regulated by the grout bond strength.
670 o The predominant mode of failure for anchors tested in shear was yielding of the steel.
671 - There was one unexpected observation for epoxy anchors loaded in tension: when cone-only failure
672 occurred, a significantly larger surface spalling area was observed due to the formation of cones having a
673 shallower angle (18 - 22°) than the expected 45° or 35° angle that is typically assumed in various
674 theoretical models.
675 - Epoxy and grout anchors exhibited similar capacities under shear load. The difference in performance
676 was found in the ultimate displacement, wherein grout anchors displaced further than epoxy anchors.
677 This result was attributed to a low modulus of elasticity of grout, as well as increased variability of the
678 grouted anchor capacity.
679 - Under shear load, anchor displacements were primarily governed by the limestone material properties,
680 and/or the size of the anchor diameter, with a larger diameter typically promoting splitting of the stone.
681 Anchor spacing and edge distance (in the direction of shear loading) is recommended to be greater than
682 2 times the embedment depth to avoid splitting failure.
683 - M20 sized anchors with epoxy adhesive and 10ɸ embedment depth attained the highest capacities in
684 tension and shear. This was an unexpected result given that M12 and M16 epoxy anchors induced
685 splitting failure, and it was thought that the M20 anchors would be likely to fail in this manner as well,
686 albeit at lower loads.
687 - With respect to anchors installed in epoxy there was no beneficial advantage in using larger diameters at
688 embedment depths of 5ɸ and 10ɸ because of the failure being governed by the cone failure mode.
689 - Experimentally obtained tensile failure loads were compared against predictions from theoretical models
690 originally conceived for anchorage in tension within concrete. Although the models could reasonably
691 describe the failure mode and were conservative in their failure load prediction by approximately 30%,
692 the high variability in prediction error suggests that stronger understanding of limestone and its
693 performance as an anchor substrate is required.
694 - For grouted anchors subject to tension loading the most accurate prediction of ultimate load was
695 obtained when using the theoretical models that characterised bond-only failure. For epoxy anchors the
696 theoretical models that best approximated anchor failure load were related to cone-only failure.
697 - The prediction of anchor shear strength based on the application of limit analysis to idealise the failure
698 mode procedure led to an over-estimation of the shear capacity. There was a 9% difference between
699 theoretical and experimental values of anchorage shear capacity for the ɸ16 diameter rods and a 25%
700 difference for both ɸ12 and ɸ20 diameter anchors.
701 - Anchorage strength is significantly influenced by the mechanical properties of the stone substrate which
702 in turn are affected by moisture content, undergoing an approximately 50% reduction in strength with a
703 moisture increase from 0% to 10-12%.

28
704 - The testing reported herein has demonstrated the importance of developing design guidelines which
705 specifically cater to limestone, as anchorage design recommendations that are based on other materials
706 are not directly transferrable.
707 - Future testing of anchorage in Oamaru limestone should look to consider scenarios where parameters
708 such as mortar, overburden weight, or wall configurations are involved.
709 - Numerical analysis could be utilised to develop a theoretical model of the anchorage failure behaviour in
710 a weak substrate material such as Oamaru limestone.

711 ACKNOWLEDGEMENTS
712 The authors are grateful to the project partners for their extensive contributions and guidance throughout the
713 entirety of the study. Particular thanks are given to Samuel Cheng, who was integral to the project’s development
714 and the success of the experimental undertaking, and to Francisco Galvez for his continued and valuable
715 guidance throughout the study. The authors thank Parkside Quarries in Oamaru, New Zealand, for the provision
716 of limestone blocks, and Sika Ltd. for providing cement-based grout. Further thanks are extended to the
717 University of Auckland laboratory technicians who were of extensive help in facilitating experimental testing, and
718 to the University of Auckland for providing the funding and facilities for this project. This project was also
719 (partially) supported by QuakeCoRE, a New Zealand Tertiary Education Commission-funded Centre. This is
720 QuakeCoRE publication number 0630.

721 REFERENCES
722 [1] R. A. Cook, "Behavior of chemically bonded anchors," Journal of Structural Engineering, vol. 119, pp. 2744-
723 2762, 1993.
724 [2] L. Contrafatto and R. Cosenza, “Prediction of the pull-out strength of chemical anchors in natural stone,”
725 Frattura ed integrità strutturale, vol. 29, pp. 196-208, 2014.
726 [3] F. Ceroni, R. Cuzzilla and M. Pecce, “Assessment of performance of steel and GFRP bars as injected
727 anchors in masonry walls,” Construction and Building Materials, vol. 123, pp. 78-98, 2016.
728 [4] D. Dizhur, A. Schultz and J. M. Ingham, “Pull-out behavior of adhesive connections in unreinforced
729 masonry walls,” Earthquake Spectra, vol. 32(4), p. 2357–2375, 2016.
730 [5] R. Felicetti, N. Gattesco and E. Giuriani, “Local phenomena around a steel dowel embedded in a stone
731 masonry wall,” Materials and Structures, vol. 30, pp. 239-246, 1997.
732 [6] B. Gigla, “Structural design of supplementary injection anchors inside masonry,” in 15th International Brick
733 and Block Masonry Conference, Florianópolis, Brazil, 2012.
734 [7] B. Gigla, "Field pull-out tests of supplementary injection anchors in historic masonry," Transactions on the
735 Built Environment, vol. 39, pp. 1743-3509, 1999.
736 [8] R. A. Cook and C. K. R., “Factors influencing bond strength of adhesive anchors,” ACI Structural Journal, p.
737 98, 2001.
738 [9] A. Colak, “Parametric study of factors affecting the pull-out strength of steel rods bonded into precast
739 concrete panels,” International Journal of Adhesion & Adhesives, vol. 21, pp. 487-493, 2001.
740 [10] T. S. Bickel and A. F. Shaikh, “Shear Strength of Adhesive Anchors,” PCI Journal, pp. 92-10, 2002.

29
741 [11] R. Eligehausen, R. A. Cook and J. Appl, “Behavior and Design of Adhesive Bonded Anchors,” ACI
742 Structural Journal, pp. 103, 822-831, 2006.
743 [12] L. Contrafatto and R. Cosenza, “Behaviour of post-installed adhesive anchors in natural stone,” Construction
744 and Building Materials, pp. 68, 355–369., 2014.
745 [13] M. Giaretton, D. Dizhur, F. D. Porto and J. M. Ingham, “ Unreinforced stone masonry buildings in New
746 Zealand: Inventory and material characterisation,” in 9th International Masonry Conference, 2014.
747 [14] R. Muñoz, P. B. Lourenço and S. Moreira, “Experimental results on mechanical behaviour of metal anchors
748 in historic stone masonry,” Construction and Building Materials, vol. 163, p. 643–655, 2018.
749 [15] D. Dudek and M. Kadela, “Pull-Out Strength of Resin Anchors in Non-cracked and Cracked Concrete and
750 Masonry Substrates,” In Procedia Engineering, vol. 161, p. 864–867, 2014.
751 [16] F. González, J. Fernández, G. Agranati and P. Villanueva, “Influence of construction conditions on strength
752 of post installed bonded anchors,” Construction and Building Materials, vol. 165, p. 272–283, 2018.
753 [17] M. McVay, R. A. Cook and K. Krishnamurthy, “Pullout Simulation of Postinstalled Chemically Bonded
754 Anchors,” Journal of Structural Engineering, vol. 122(9), p. 1016–1024, 1996.
755 [18] R. Eligehausen and J. Appl, "Behavior and design of fastenings with bonded anchors: Numerical analysis
756 and experimental verification," 6th International Conference on Fracture Mechanics of Concrete and Concrete Structures, vol.
757 2, pp. 643-654, 2007.
758 [19] E. Giuriani, N. Gattesco and M. D. Piccolo, «Experimental tests on the shear behaviour of dowels
759 connecting concrete slabs to stone masonry walls,» Materials and Structures, vol. 26(5), p. 293–301, 1993.
760 [20] M. Hoehler, “Behavior and Testing of Fastenings to Concrete for use in Seismic Applications,” Ph.D.Thesis,
761 Institut für Werkstoffe im Bauwesen, Universität Stuttgart, Stuttgart, Germany, 2006.
762 [21] I. Giongo, D. Dizhur, R. Tomasi and J. M. Ingham, “In-situ testing of wall-to-diaphragm shear transferring
763 connections in an existing clay brick URM,” in 9th International Masonry Conference 2, Guimarães, 2014.
764 [22] British Standards, “Natural stone test methods - Determination of uniaxial compressive strength,” (BS-
765 EN1926:2006), 2006.
766 [23] ASTM-(C880/C880M), “Standard tet method for flexural strength of dimension stone,” 2015.
767 [24] ASTM-(D3967-2016), “Standard test method for splitting tensile strength of intact rock core specimens,”
768 ASTM International,West Conshohocken, USA, 2016.
769 [25] ISRM, "Suggested methods for determining tensile strength of rock materials," p. 99–103, 1978.
770 [26] F. Cherblanc, J. Berthonneau, P. Bromblet and V. Huon, “Influence of Water Content on the Mechanical
771 Behaviour of Limestone: Role of the Clay Minerals Content,” Rock Mechanics and Rock Engineering, vol. 49, no. 6,
772 pp. 2033-2042, 2016.
773 [27] Hilti, “Technical data file (Page 9),”
774 https://www.hilti.co.nz/medias/sys_master/documents/h1e/h4b/9342310940702/ETA-12-0084-for-HIT-
775 HY-200-R-injection-mortar-and-standard-element-for-anchoring-application-ETAG-001-05-Option-1-
776 Approval-document-ASSET-DOC-APPROVAL-0197.pdf.
777 [28] R. A. Cook, J. Burtz and M. Ansley, “Design Guidelines and Specifications for Engineered Grouts,” Report
778 of University of Florida, Gainsville, Florida,, 2003.

30
779 [29] ASTM-(E488-96), “Standard test methods for strength of anchors in concrete and masonry elements,”
780 2003.
781 [30] AS/NZS-2699.2, “Built-in components for masonry construction,” Standards New Zealand, 2000.
782 [31] ACI-318, “Building code requirements for structural concrete and commentary,” American Concrete Institute,
783 2011.
784 [32] G. T. Doerr, A. R. Cook and R. E. Klingner, “Adhesive Anchors: Behaviour and Spacing Requirements,”
785 University of Texas, Austin, no. 1126-2, 1989.
786 [33] Marti, P., “Anchoring of Concrete Reinforcement Using HIT-HY 150,” Hilti Development Corporation,
787 Technical Report, 1993.
788 [34] ACI-349, "Code Requirements for Nuclear Safety-Related Concrete Structures (ACI 349-06) and
789 Commentary," American Concrete Institute, 2007.
790 [35] M. Como and R. Luciano, “A new multi-axial failure criterion for concrete,” 2007.
791 [36] BS-EN 1926: 2006, "Natural stone test methods - Determination of uniaxial compressive strength", British
792 Standards, 2006.
793 [37] H. Gercek, “Poisson’s ratio value for rocks”, International Journal of Rock Mechanics and Mining Sciences, vol.
794 44(1), pp. 1-13, 2007.
795 [38] F. Silveri, P. Riva, G. Profeta, E. Poverello and C. Algeri, “Experimental Study on Injected Anchors for the
796 Seismic Retrofit of Historical Masonry Buildings”, International Journal of Architectural Heritage, vol. 10, no. 2-3, pp.
797 182-203, 2016.

31

View publication stats

You might also like