On Mechanical Behavior of Metal Anchors in Historical Brick Masonry: Testing and Analytical Validation

Featured Application

This paper presents an experimental and analytical methodology for the mechanical characterization of metal anchors in ancient brick masonry walls. The presented results contribute to the identification of the most efficient anchoring systems for strengthening and retrofitting of historical brick masonry structures.

Abstract

The repair and strengthening of historical masonry buildings is a fundamental aspect in the conservation of the built cultural heritage. Temporary shoring or strengthening are often used and, usually, involve the introduction of new metallic elements. The connection between the original substrate and the new elements must be analyzed carefully to prevent further damage to the building. This paper presents a study on the mechanical behavior of metal anchors applied to brick masonry walls. An experimental campaign is developed, and a series of pull-out tests are carried out on masonry walls built in a laboratory with natural hydraulic lime mortar and low mechanical strength bricks. Two groups of tests are conducted, namely, with the actuator in the direction of the anchor axis and with the actuator inclined with respect to the fastener axis. Moreover, two types of anchoring systems are used, namely, adhesive (chemical and cementitious grout) and mechanical anchors. The experimental results are compared to predictive analytical formulas available in the literature for estimation of the ultimate load capacity, according to the type of failure. From the comparison between experimental and analytical values, it is proven that the analytical formulation originally developed for concrete substrates cannot be directly extrapolated to brick masonry cases, and specific predictive formulas should be developed. The presented research can be used to select the most efficient anchoring system for strengthening and retrofitting of historical brick masonry structures.

1. Introduction

Many historical structures built with brick masonry show signs of structural degradation and damage caused by factors such as the action of environmental loads, lack of maintenance, or the occurrence of extreme events, e.g., fires or earthquakes [1,2,3]. In numerous occasions, these buildings have lost their horizontal structural elements, remaining only with their walls. Moreover, it is common practice to preserve the masonry façades of historical buildings due to their architectural value. In order to prevent these vertical cantilevers from collapsing, temporary or permanent shoring, as well as structural stabilization through the incorporation of new structural elements are usually adopted.

The connection between the old masonry walls and the new structural elements can be achieved in different ways and using various materials. A commonly accepted choice involves the application of metal anchors or fasteners. It is noted that the use of anchoring, often with external elements, is especially suited for repair and strengthening of historical structures as it allows minimal interventions with moderate aesthetic impact [4]. Similarly, the application of anchors is equally appropriate for contemporary structures. There are two main types of anchoring systems, namely, adhesive anchors and mechanical anchors, each type with its own subcategories.

Adhesive or bonded systems can be further subdivided into chemical and grout anchors. The former type uses resin to create the bond with the substrate material, whereas grouting systems employ inorganic binders. Both systems are mostly applied through injection. Chemical anchoring usually consists of two adhesive components that are mixed during application. On the other hand, cement-based mixtures are commonly used for grout anchors. Both systems require a curing period to harden and gain strength. The curing period is shorter in the case of resins; therefore, chemical anchors are most used when a quick setting is needed. Moreover, chemical anchors require smaller drill holes, 10% to 25% larger than the diameter of the fastener, whereas grout anchors need drill holes 50% to 100% larger than the fastener diameter [5].

Mechanical anchoring is based on direct contact between the anchor and the substrate, with the corresponding transfer of loads through friction. Among the mechanical fasteners, the most common types are expansion and undercut anchors. The former type expands during installation and comprises two subcategories, namely, torque-controlled and deformation-controlled. For the torque-controlled anchors, tightening the bolt or nut activates the expansion mechanism. Alternatively, for deformation-controlled anchors, the expansion is caused by the displacement of an expander plug within a sleeve. In turn, undercut anchors establish a mechanical interlock with the substrate by means of a larger cavity at the end of the drill hole.

According to the Guidance Note by the Construction Fixings Associations [6], the types of anchoring systems recommended for solid brick masonry are adhesive as well as torque-controlled expansion anchors. These guidelines provide additional considerations for the installation of the different types of anchors with the purpose of improving the quality of the connection: drilling the holes according to the specified depths and diameters; ensuring the cleanliness of the drill holes; verifying that the anchors are centrally placed within the drill hole; and ensuring the correct preparation of the adhesives, e.g., water/binder ratio for cementitious grout considering the water absorption capacity of the substrate material; among others [4,6,7].

The connection between the substrate material and the anchors deserves special attention, since it constitutes a key component of the overall structural response. In this context, a number of experimental works have been carried out to characterize the pull-out resistance of metal anchors, e.g., [7,8,9,10,11]. Nonetheless, the amount of research focused on the application of anchors to masonry walls and historical structures is still low, see, e.g., [7,8,12,13,14,15,16].

This work aims at extending the current knowledge on the mechanical behavior of metal fasteners in brick masonry walls. For this purpose, different anchoring systems, namely, mechanical, chemical, and cementitious grout anchors, were applied to brick masonry walls built in a laboratory. The conducted research included mechanical characterization tests on constituent materials and masonry specimens, as well as pull-out tests to characterize the mechanical behavior of the anchors. Moreover, two groups of pull-out tests were performed, namely, with the actuator pulling in the axial direction of the fastener (pure tensile loading) and with a 30° inclination (combined tensile and shear loading). Finally, the results obtained experimentally were compared with the estimated capacity calculated through analytical formulas available in the literature. From a comparison between experimental and analytical values, it is shown that the analytical formulation originally developed for concrete substrates cannot be directly extrapolated to brick masonry cases, and specific predictive formulas should be developed. The presented work contributes to the selection of the most efficient anchoring system for brick masonry structures according to their mechanical capacity and for the validation of existing analytical formulations.

2. Mechanical Behavior of Metal Anchors

The repair and strengthening of masonry by means of metal anchors has been carried out for centuries and is still widely accepted for the conservation and stabilization of existing masonry structures. In practice, the application of anchors in masonry walls can be associated with interventions such as: the stabilization of damaged or deformed masonry; connections between the existing walls and new structural elements; transfer of tensile stresses that cannot be sustained by the masonry alone; strengthening and retrofitting of walls and foundations; or strengthening to improve the structural dynamic behavior [7].

The present section collects a series of analytical formulas used to estimate the mechanical capacity of anchors subjected to tensile and combined tensile and shear forces. These predictive expressions have been obtained from the results of pull-out tests on adhesive and mechanical anchors, according to the type of failure mode. Most formulas have been developed for concrete, but some analogies can be made for masonry.

It is noted that the formulation presented hereafter is limited to the experimental setup developed in this study, that is, a single anchor per wall, centrally located, and at a considerable distance from the lateral edges.

2.1. Characteristic Resistance of Anchors under Tensile Loading

The analytical formulation presented hereafter was established according to the possible failure modes associated with tensile loading. In particular, the analytical formulas are used to estimate the characteristic resistance of a single anchor $N_{Rk}$ (N). The reader is referred to the relevant references for further details.

2.1.1. Adhesive Anchors

For adhesive anchors (chemical and cementitious grout), the main types of failure modes associated with tensile loading are shown in Figure 1: (a) steel failure of the anchor; (b) pull-out or interface failure (either anchor/adhesive or adhesive/substrate); (c) substrate cone failure; (d) combined pull-out and substrate cone failure; (e) pull-out of one brick; (f) lateral cone or blow-out failure. For real-life images of the specific failure modes obtained in this work, the reader is referred to Section 4.

In the case of steel failure, herein labelled T1, the characteristic resistance of an anchor can be defined as [18,19]:

$$N_{Rk}=0.75·A_s·f_{uk},$$

(1)

where $A_s$ (mm²) is the effective cross-sectional area of the anchor, and $f_{uk}$ (MPa) is the characteristic ultimate strength of steel. Alternatively, the characteristic resistance can be calculated as [20]:

$$N_{Rk}=0.90·A_s·f_{yk},$$

(2)

where $f_{yk}$ (MPa) is the characteristic yield strength of steel.

It must be noted that steel failure is rarely observed in masonry. Nonetheless, it might occur in applications with considerable embedment length and high masonry strength [21].

The second type of failure mode, i.e., pull-out or interface failure, herein called T2, may occur between the anchor and the adhesive or between the adhesive and the substrate [9]. For both cases, the tension model of uniform adhesion to describe the interface behavior is usually adopted [17,18,22]. Thus, the characteristic resistance for interface failure between the anchor and the adhesive can be estimated by [10]:

$$N_{Rk}={\tau }_k·\pi ·d·h_{ef},$$

(3)

where ${\tau }_k$ (MPa) is the adhesive bond strength of the anchor/adhesive interface, $d$ (mm) is the anchor diameter, and $h_{ef}$ (mm) is the effective embedment length. Conversely, the formula for the interface failure between the adhesive and the substrate is [11]:

$$N_{Rk}={\tau }_{k0}·\pi ·d_0·h_{ef},$$

(4)

where ${\tau }_{k0}$ (MPa) is the adhesive bond strength of the adhesive/substrate interface, and $d_0$ (mm) is the drill hole diameter.

Existing studies suggest that this type of failure can be influenced by several parameters, such as drill hole conditions during the application of the anchor, water absorption capacity of the substrate, adhesion effective length, injection time, or the presence of masonry joints [7].

For the substrate cone failure mode, herein labelled T3, the formulation developed for concrete is here extended to masonry. Therefore, the characteristic resistance of the anchor can be calculated as [18]:

$$N_{Rk}=k_1·{\left(f_{ck cube}\right)}^{0.5}·h_{ef}^{1.5}·\left(A_{c,N}/A_{c,N}^0\right),$$

(5)

where $k_1$ (N^0.5/mm^0.5) is a semi-empirical parameter, $f_{ck cube}$ (MPa) is the characteristic compressive strength of concrete measured for 200 mm cube specimens, and $A_{c,N}/A_{c,N}^0$ is a geometrical reduction factor defined, as shown in Figure 2. In particular, $A_{c,N}$ (mm²) is the projected area of an anchor or group of anchors on the concrete surface, whereas $A_{c,N}^0$ (mm²) is the projected area of a single anchor not limited by adjacent edges. Hence, for a single anchor not located on an edge, $A_{c,N}/A_{c,N}^0=1$.

For $k_1$, EOTA TR 029 [18] proposes 7.2 N^0.5/mm^0.5 and 10.1 N^0.5/mm^0.5 for cracked and non-cracked concrete, respectively, whereas EN 1992-4 [17] and fib Bulletin 58 [22] propose 7.7 N^0.5/mm^0.5 and 11.0 N^0.5/mm^0.5 for cracked and uncracked concrete applications, respectively.

In general, the anchor capacity for this type of failure mode is influenced by the strength of the substrate, the presence of cracks, and the proximity to other anchors and edges. Moreover, substrate cone failure is generally associated with short embedment lengths and low substrate strength [9].

Concerning the combined pull-out and substrate cone failure, herein called T4, the characteristic resistance of the anchor can be defined as [18]:

$$N_{Rk}={\tau }_k·\pi ·d·h_{ef}·\left(A_{c,N}/A_{c,N}^0\right).$$

(6)

The final type of adhesive anchor failure of interest for this study corresponds to the pull-out of one brick. For this failure mode, herein labelled T5, the characteristic resistance of an anchor can be calculated as [23]:

$$N_{Rk}=2·l_{brick}·b_{brick}·\left(0.5·f_{vk0}+0.4·{\sigma }_d\right)+b_{brick}·h_{brick}·f_{vk0}$$

(7)

where $l_{brick}$ (mm), $b_{brick}$ (mm) and $h_{brick}$ (mm) correspond to the brick length, width, and height, respectively; ${\sigma }_d$ (MPa) is the design compressive stress perpendicular to the shear stress; and $f_{vk0}$ (MPa) is the initial shear strength, which can be defined as 0.2 MPa for clay brick masonry with mortar strength M2.5 to M9, according to EN 1996-1-1 [24]. It is noted that Equation (7) is for an application for cases where the vertical joints are filled with mortar. The reader is referred to ETAG 029—Annex C [23] for further considerations regarding adhesive anchors in brick masonry applications.

2.1.2. Mechanical Anchors

The main types of failure expected for mechanical anchors are analogous to the failure modes described for adhesive anchors (see Figure 1): (a) steel failure; (b) pull-out or interface failure; (c) substrate cone; (d) combined pull-out and cone; (e) pull-out of one brick; and (f) blow-out failure.

Considering the steel failure of the anchor, Equations (1) and (2) can be applied to mechanical fasteners as well.

The pull-out or interface failure of mechanical anchors, herein called T6, may occur at moderate embedment lengths in low-strength substrate material or in cases when the drill hole has a greater diameter than the fastener [9]. The following expression can be used to calculate the characteristic resistance associated with this failure mode [17]:

$$N_{Rk}=k_2·A_h·f_{ck},$$

(8)

where $k_2$ (-) is a semi-empirical parameter, $f_{ck}$ (MPa) is the characteristic compressive strength of concrete, and $A_h$ (mm²) is the load-bearing area of the head of the fastener, which, in turn, is defined as:

$$A_h=\frac{\pi }{4}·\left(d_h^2-d_a^2\right),$$

(9)

where $d_h$ (mm) is the diameter of the anchor head, and $d_a$ (mm) is the diameter of the anchor.

The pull-out or interface failure in mechanical anchors can be associated with moderate embedment lengths in substrates with low strength or in cases where the drill hole is greater than the diameter of the anchor [9]. The standard EN 1992-4 [17] establishes $k_2$ values of 7.5 for cracked concrete and 10.5 for non-cracked concrete.

Regarding the substrate cone failure in mechanical anchors, herein referred to as T7, the characteristic resistance can be estimated through [22]:

$$N_{Rk}=k_1·{\left(f_{ck}\right)}^{0.5}·h_{ef}^{1.5},$$

(10)

where $k_1$ (N^0.5/mm^0.5) is the same semi-empirical parameter presented previously for the substrate cone failure in adhesive anchors (T3).

2.2. Characteristic Resistance of Anchors under Shear Loading

Analogously to the anchoring systems subjected to tensile loading, the analytical formulation for anchors under shear stresses was determined according to the possible failure modes. In this case, the predictive analytical formulas aim at estimating the characteristic strength of a single anchor under shear loading, $V_{Rk}$ (N).

The types of failure modes associated with shear are shown in Figure 3: (a) steel failure of the anchor without lever arm; (b) steel failure of the anchor with lever arm; (c) substrate pry-out failure; and (d) edge failure. It is noted that steel failure is usually complemented by crushing and spalling of the substrate material close to the anchor. For real-life images of the failure modes obtained in this work, the reader is referred to Section 4.

Regarding the steel failure under shear load without lever arm, herein called T8, the characteristic resistance of the anchor can be estimated as [18]:

$$V_{Rk}=0.50·A_s·f_{uk},$$

(11)

On the other hand, for the steel failure under shear loading with lever arm, herein referred to as T9, the characteristic resistance of the anchor is defined by [18]:

$$V_{Rk}={\alpha }_M·M_{Rk}/l,$$

(12)

where ${\alpha }_M$ (-) is a factor that takes into account the degree of restraint of the fastener at the side of the load application (${\alpha }_M=$ 1 if the anchor can rotate or ${\alpha }_M=$ 2 for full restraint), $M_{Rk}$ (N·mm) is the characteristic bending resistance, and $l$ (mm) is the lever arm or distance between the substrate surface and the load. The characteristic bending resistance can be calculated as [18]:

$$M_{Rk}=M_{Rk}^0·\left(1-N_{Sd}/N_{Rd}\right).$$

(13)

where $M_{Rk}^0$ (N·mm) is the characteristic bending resistance of a single anchor, $N_{Sd}$ (N) is the design value for tensile load, and $N_{Rd}$ (N) is the design resistance for tensile loading. The characteristic bending resistance of a single anchor is usually provided by the manufacturer and is defined as [18]:

$$M_{Rk}^0=1.2·W_{el}·f_{uk},$$

(14)

where $W_{el}$ (mm³) is the elastic section modulus of the fastener calculated from the effective cross-section of steel ($\pi d^3/32$ for a round section of diameter $d$).

In Equation (13), the design resistance for tension loading can be obtained by:

$$N_{Rd}=N_{Rk}/{\gamma }_{Ms},$$

(15)

where $N_{Rk}$ (N) is the characteristic strength of an anchor or group of anchors, obtained with the substrate cone failure formula, and ${\gamma }_{Ms}$ (-) is the partial safety factor of the material, which is set to 1.25 for failure with lever arm [18].

The substrate pry-out failure, herein labelled T10, is associated with the rotation of the anchor under shear loading with lever arm. For this failure mode, the characteristic resistance can be calculated as [18]:

$$V_{Rk}=k_3·N_{Rk},$$

(16)

where $k_3$ (-) is a semi-empirical factor, and $N_{Rk}$ (N) is the characteristic strength of an anchor or group of anchors, obtained from the substrate cone failure formulation. According to EOTA TR 029 [18], the parameter $k_3$ is equal to 1 for an effective embedment length $h_{ef}<$ 60 mm, or 2 in the case of $h_{ef}\ge$ 60 mm.

3. Materials and Methods

The experimental investigation was conducted in the Structures Laboratory at the University of Minho, in Portugal. Brick masonry walls were prepared to simulate historical brickwork, and mechanical characterization tests were performed on constituent materials as well as on masonry panels and anchors.

3.1. Mechanical Characterization of the Constituent Materials

The studied materials included solid bricks and natural hydraulic lime (NHL) mortar. The bricks consisted of fired-clay extruded units with dimensions 200 mm × 95 mm × 65 mm. The mortar was prepared with NHL 3.5 (following the classification in EN 459-1 [25]), which was supplied by SECIL, and sand from a nearby quarry with a proportion binder/aggregate of 1:2.5 in volume. The quantity of water for the preparation of the mortar was chosen to obtain 145–155 mm in the flow table test [26], which is a recommended value for conservation mortars [27].

The tests performed for the mechanical characterization of bricks included: elastic modulus, following the standard EN 12390-13 for hardened concrete [28]; uniaxial compression, according to EN 772-1 [29]; and three-point bending tests, as specified in ASTM C67/C67M [30]. Similarly, the mechanical characterization tests on mortar specimens included: elastic modulus, according to EN 12390-13 [28]; uniaxial compression and three-point bending, following EN 1015-11 [31]. Molded mortar specimens were prepared with the same mixture used for the construction of the walls. The mortar specimens remained in the same location as the walls to ensure analogous curing conditions, namely, temperature and relative humidity. The average results obtained from the mechanical characterization tests on constituent materials are presented in

Table 1. Mechanical characterization of the constituent materials (CoV between parenthesis).

Material	Compressive Strength (MPa)	Tensile Strength (MPa)	Elastic Modulus (GPa)
Brick	19.90 (4.5)	2.13 (16.3)	9.74 (6.9)
Mortar (28 days)	1.42 (10.5)	0.41 (14.1)	2.15 (4.7)

. In the table, the coefficient of variation (CoV) is given between parenthesis, expressed in percentage.

3.2. Masonry Walls

Various sets of brick masonry walls were built to characterize the mechanical behavior of the material in terms of compressive, tensile, and shear strength. The dimensions and number of specimens employed for each test are presented in

Table 2. Dimensions and number of walls built for experimental characterization.

Dimensions (Length × Width × Height) (m)	Number of Specimens	Test
0.40 × 0.20 × 0.40	5	Axial compression (A)
0.40 × 0.40 × 0.50	3	Axial compression (B)
0.80 × 0.30 × 0.80	3	Diagonal compression
0.80 × 0.40 × 1.20	12	Pull-out

. Compressive strength was obtained with masonry specimens with thicknesses of 0.20 m (labelled A) and 0.40 m (labelled B). The mechanical characterization tests for masonry were performed following the recommendations provided in LUM B6 [32], EN 1052-1 [33], ASTM C1314-22a [34], and ASTM E519/E519M [35]. All the specimens were stored in controlled laboratory conditions before testing.

Additionally, two brick walls were built for each type of anchorage system, namely, mechanical, chemical, and cementitious grout. Thus, six test walls were prepared for the pull-out test under tension (labelled BW_1 to BW_6), and six walls were built for the pull-out test under combined tension-shear (labelled BW_7 to BW_12). The geometrical configuration of the walls prepared for the pull-out tests is presented in Figure 4. It must be noted that none of the available guidelines specify the dimensions of masonry specimens for pull-out tests. The height of the wall was set to 1.20 m to be compatible with the laboratory equipment. In turn, a thickness of 0.40 m was considered to be representative of a load-bearing historical wall [36]. The width of 0.80 m was defined from the expected diameter of the cone failure produced by the anchor pull-out, as indicated in Figure 2 [18,23,37]. Finally, the arrangement of bricks was chosen to replicate a commonly used bond system in historical masonry structures [36].

3.3. Mechanical Characterization of the Masonry Walls

Axial compression and diagonal compression tests were performed 60 days and 40 days after the preparation of the walls, respectively. The axial compression tests were used to evaluate the compressive strength of the masonry prisms, whereas diagonal compression was used to assess the tensile strength and shear strength. Both tests were carried out under displacement control with an imposed deformation rate of 3 µm/s. For the walls subjected to axial loading, failure resulted in a conical shape, as illustrated in ASTM C1314-22a [34]. In turn, the failure observed in the specimens under diagonal compression occurred progressively and in a stepped manner.

The average results obtained from the mechanical characterization tests on masonry specimens are presented in

Table 3. Mechanical characterization of the masonry walls (CoV between parenthesis).

Compressive Strength (A) (MPa)	Compressive Strength (B) (MPa)	Tensile Strength (MPa)	Shear Strength (MPa)
9.84 (2.63)	8.82 (7.91)	0.10 (12.82)	0.21 (12.81)

, together with the respective coefficients of variation (CoV) expressed in percentage. Again, note that the first two columns refer to the compressive strength of masonry specimens with thicknesses of 0.20 m (labelled A) and 0.40 m (labelled B), respectively.

3.4. Pull-Out Tests

The metal anchors used for the pull-out tests in the present study were supplied by HILTI. The anchors had 10 mm diameter and different lengths according to the type of anchoring system, namely, mechanical or adhesive (chemical and grout). The technical specifications in terms of length ($l$) and effective embedment ($h_{ef}$) are presented in

Table 4. Specifications of the anchoring systems applied on the masonry walls.

Mechanical Anchor			Chemical Anchor			Cementitious Grout Anchor
Type	l (mm)	hef (mm)	Type	l (mm)	hef (mm)	Type	l (mm)	hef (mm)
HUS3-H 10 screw anchor	100	85	Resin HIT-RE 500; HIT-V-8.8 M10 anchor rod	190	150	Mapefill P Grout; HIT-V-8.8 M10 anchor rod	190	150

, as provided by the manufacturer. For the application of the anchors (including drilling for adhesive fasteners) as well as for the pull-out tests, a compressive stress of 0.2 MPa was applied on top of the walls in order to reproduce the confinement effect induced by the vertical loads in a real façade.

The resin used for chemical anchoring was injected using specific equipment provided by HILTI (HDE 500-A22 battery-operated dispenser), whereas the cementitious grout (Mapefill P from MAPEI) was injected manually using a syringe. For both cases, holes 20 mm in diameter (twice the dimension of the fastener) were drilled into the walls. A schematic description of the system used for the cementitious grout anchor is shown in Figure 5. In particular, a screw nut was welded at the end of the anchor to ensure a centered fit within the drill hole and allow the placement of the injection and air outlet tubes.

The pull-out tests were performed after the application of the mechanical anchors and after the curing period in the case of adhesive anchors. The setup employed for the pull-out tests is shown in Figure 6. During the tests, a compressive stress of 0.2 MPa was applied on top of the walls, as previously mentioned. Moreover, the walls were fixed to the testing frame to prevent their displacement. A hinged joint was used to connect the head of the anchor to the actuator. Linear variable differential transformers (LVDTs) were used to measure displacements in the areas outside and inside the expected cone failure induced by the pulling force.

4. Results

The results of the pull-out tests with the horizontal actuator are presented in

Table 5. Results from pull-out test with horizontal actuator.

Wall	Anchor Type	Maximum Axial Load in Pull-Out Test Np (kN)	Horizontal Anchor Displacementat at Np (mm)	Most Common Failure Mode (see Section 2)	Estimated Strength Value NR (kN)	Np/NR
BW_1	Mechanical	12.55	2.16	T7	17.23	0.73
BW_5	Mechanical	12.88	2.05	T7	17.23	0.75
BW_2	Chemical	27.00	3.06	T3	40.40	0.67
BW_6	Chemical	23.11	2.84	T2	10.84	2.13
BW_3	Grout	26.29	10.02	T2	10.84	2.43
BW_4	Grout	17.06	2.02	T4	10.84	1.57

, whereas the results obtained from the tests with the inclined actuator are collected in

Table 6. Experimental results from pull-out test with inclined actuator.

Wall	Anchor Type	Max. Load in Pull-Out Test Fp (kN)	Max. Axial Component Np (kN)	Max. Shear Component Vp (kN)
BW_9	Mechanical	8.57	7.42	4.29
BW_10	Mechanical	8.55	7.40	4.28
BW_11	Chemical	21.01	18.20	10.51
BW_12	Chemical	19.33	16.74	9.67
BW_7	Grout	16.42	14.22	8.21
BW_8	Grout	19.86	17.20	9.93

and

Table 7. Failure mode and comparison between experimental and analytical results from pull-out test with inclined actuator.

Wall	Failure Mode in Tension and Shear	Estimated Strength Value NR (kN)	Np/NR	Estimated Strength Value VR (kN)	Vp/VR
BW_9	T7 and T9	17.23	0.43	4.43	0.97
BW_10	T6 and T9	5.39	1.37	4.43	0.97
BW_11	T2 and T9	10.84	1.68	29.08	0.36
BW_12	T2 and T9	10.84	1.54	29.08	0.33
BW_7	T2 and T9	10.84	1.31	29.08	0.28
BW_8	T2 and T9	10.84	1.59	29.08	0.34

. The tables also show the strength estimated for each anchor, according to the type of failure, as discussed in Section 2. The identification of failure modes was based on the displacement data registered by the LVDTs as well as on visual inspection. The observed failure modes are shown in Figure 7 and Figure 8 for the tests with horizontal and inclined actuator, respectively. It is noted that different displacement and failure modes were observed even among the same types of fasteners, which emphasizes the heterogeneous nature of masonry.

The estimated capacity of the anchors was calculated, as described in Section 2, according to the types of failure modes. In particular, the characteristic strength of the fasteners was calculated based on Equations (1)–(10) for the cases with horizontal actuator (as well as for the tensile component in the cases with shear loading) and according to Equations (11)–(16) for the cases with inclined actuator. In the formulas originally developed for concrete, the compressive strength of masonry was used instead of the characteristic resistance of concrete. In this sense, the average strength obtained from axial compression tests on masonry specimens was used, i.e., $f_k=$ 9.33 MPa (average value of the first two columns, A and B, in Table 3). The effective embedment lengths presented in Table 4 were used for the corresponding calculations. In addition, the diameter and load-bearing area of the head of the anchors were defined as $d=$ 10 mm and $A_h=$ 77 mm², respectively, according to the specifications of the manufacturer. Moreover, $k_1=$ 7.2 N^0.5/mm^0.5 was assumed, while the adhesive bond strength ${\tau }_k$ was taken as 2.3 MPa [18]. In turn, $k_2=$ 7.5 was defined for pull-out failure cases in mechanical anchors [17]. Considering all the fasteners were applied isolated, located at the center of the wall panel at a distance greater than 3 × $h_{ef}$ from the nearest edges, the factor $A_{c,N}/A_{c,N}^0$ was assumed equal to 1.

With regard to the tests with the inclined actuator, the axial and shear components were calculated by decomposition of the applied force considering an inclination angle of 30°. As shown in Figure 8, the most common failure mode observed for the cases with shear loading corresponds to steel bending with consequent crushing/spalling of the substrate material close to the fastener (T9). Furthermore, the axial component of these tests is also related to anchor pull-out (T2 and T6, depending on the type of anchor) and substrate cone failure (T7 for mechanical anchors). For the calculation of the characteristic shear strength, according to Equation (12), ${\alpha }_M$ was taken equal to 1, and a lever arm $l=$ 2 mm was considered.

5. Discussion

A summary of the experimental results obtained from the pull-out tests is presented in

Table 8. Summary of $N_p$ and $V_p$ values obtained from the pull-out tests.

Wall	Type of Anchor	${\mathit{N}}_{\mathit{p}}\left(\mathbf{kN}\right)$	${\mathit{V}}_{\mathit{p}}\left(\mathbf{kN}\right)$
BW_1	Mechanical	12.55	-
BW_5	Mechanical	12.88	-
BW_9	Mechanical	7.42	4.29
BW_10	Mechanical	7.40	4.28
Average value of mechanical anchors (CoV in %)		10.06 (30.4)	4.28 (0.2)
BW_2	Chemical	27.00	-
BW_6	Chemical	23.11	-
BW_11	Chemical	18.20	10.51
BW_12	Chemical	16.74	9.67
Average value of chemical anchors (CoV in %)		21.26 (22.1)	10.09 (5.9)
BW_3	Grout	26.29	-
BW_4	Grout	17.06	-
BW_7	Grout	14.22	8.21
BW_8	Grout	17.20	9.93
Average value of grout anchors (CoV in %)		18.69 (28.1)	9.07 (13.4)

. It is noted that the average values presented in the table were calculated for each type of anchor, regardless of the specific failure mode, since the main purpose was to draw comparisons between the three different anchoring systems. This generalization is acceptable, bearing in mind that masonry properties usually present a significant scatter, and similar loads may lead to different failure modes. Therefore, by comparing the average capacity for the three systems, it was observed that chemical anchors performed better in terms of resistance, followed by the cementitious grout type, and, finally, by the mechanical fasteners. In particular, the average load obtained for chemical anchors was 14% higher than that of the grout, and more than twice the average value obtained for mechanical anchors.

Although the difference, with respect to the mechanical fasteners, is reasonably large, it must be recalled that the effective embedment length of mechanical fasteners was lower, i.e., $h_{ef}=$ 85 mm instead of $h_{ef}=$ 150 mm for adhesive anchors. By normalizing the results by a factor 150/85, the tensile and shear components are $N_p^{*}=$ 17.75 kN and $V_p^{*}=$ 7.55 kN, respectively, which are closer to the values obtained for the chemical and grout anchors. Nonetheless, the performance of adhesive fasteners is still superior, in addition to allowing longer effective embedment lengths.

It is further noticed that the introduction of a shear component led to an overall reduction in the tensile capacity for all the anchoring systems.

Table 9. Relationship between experimental values obtained from the pull-out tests and estimated values.

Type of Anchor	Average ${\mathit{N}}_{\mathit{p}}\left(\mathbf{kN}\right)$	Average ${\mathit{N}}_{\mathit{R}}\left(\mathbf{kN}\right)$	${\mathit{N}}_{\mathit{p}}/{\mathit{N}}_{\mathit{R}}$	Average ${\mathit{V}}_{\mathit{p}}\left(\mathbf{kN}\right)$	Average ${\mathit{V}}_{\mathit{R}}\left(\mathbf{kN}\right)$	${\mathit{V}}_{\mathit{p}}/{\mathit{V}}_{\mathit{R}}$
Mechanical	12.72	17.23	0.74	-	-	-
Mechanical	7.41	11.31	0.66	4.28	4.43	0.97
Chemical	25.06	25.62	0.98	-	-	-
Chemical	17.47	10.84	1.61	10.09	29.08	0.35
Grout	21.68	10.84	2.00	-	-	-
Grout	15.71	10.84	1.45	9.07	29.08	0.31

provides a comparison between the results obtained from the pull-out tests and the values estimated from the analytical formulas presented in Section 2. Considering the predictive formulas for the tensile loading cases (horizontal actuator), the average estimated value for chemical anchors was comparable to the average resistance obtained from the pull-out tests. Conversely, the experimental values for cementitious grout and mechanical anchors were 100% above and 26% below the estimated values, respectively. The higher value estimated for mechanical anchors is especially problematic, since the formulation seems to overestimate the performance of the system.

For the cases with combined tension and shear (inclined actuator), the predictive formulas for the tensile component did not match the experimentally observed values. In particular, the performance of chemical and grout anchors was underestimated by around 61% and 45%, respectively, whereas the capacity of the mechanical fasteners was overestimated by 34%. With respect to the shear component, the analytical formulas were appropriate for the mechanical anchors but largely overestimated the resistance of the adhesive cases.

The comparison between experimental and analytical values indicates that the analytical formulation originally developed for concrete substrates cannot be directly extrapolated to brick masonry cases, and specific predictive formulas should be developed. In particular, the higher capacity predicted for mechanical anchors under tensile loading and for adhesive anchors under shear loading suggests a safety issue. It is recalled that most of the formulas found in the literature were originally developed for concrete. Therefore, further research is necessary for specific applications in masonry.

Finally, it is noted that Muñoz et al. [16] performed an analogous study on metal anchors applied in stone masonry walls. Their experimental results from pull-out tests revealed slightly higher average loads for all the anchoring systems for both tensile and shear loading. With respect to the analytical results, these authors also pointed out the inadequacy of the existing predictive formulas for their application to stone masonry. In their case, the estimated capacity obtained from the formulas tended towards conservative results.

6. Conclusions

Metal anchors are commonly used for the stabilization of masonry structures, either as a temporary solution for shoring purposes or for strengthening interventions in historical buildings. Nonetheless, the mechanical behavior of the anchoring systems needs further investigation, both in the experimental field and in the theoretical aspects with the development of more consistent predictive models, especially for historical brick masonry.

This paper presented the results of an experimental campaign to study the load capacity of three types of metal anchors, namely, mechanical, chemical, and cementitious grout fasteners, applied on brick masonry walls built in a laboratory. The experimental works included mechanical characterization tests for constituent materials and masonry specimens, as well as pull-out tests. Moreover, two sets of pull-out tests were performed, namely, with the actuator pulling in the axial direction of the anchors (tensile loading) and with a 30° inclination, with respect to the horizontal plane (combined tensile and shear loading).

Considering the results of the pull-out tests, the chemical anchors showed better performance, with an average ultimate load about 14% higher than that of the grout and 111% greater than the value obtained for mechanical anchors. Conversely, the mechanical fasteners showed the lowest capacity. An overall reduction in the tensile capacity of all the anchoring systems was observed when the shear component was introduced. Furthermore, different failure modes and loading capacities were obtained even within the same type of anchors, which emphasizes the heterogeneous nature of masonry.

In general, the predictive analytical formulas were proven inadequate to estimate the ultimate loading capacity of metal anchors applied in brick masonry. In particular, the capacity predicted for mechanical anchors under tensile loading and for adhesive anchors under shear loading was overestimated, which supposes a safety issue. It was recalled that most of the predictive formulas available in the literature were originally developed for concrete. Further experimental and analytical studies are, therefore, necessary to develop specific predictive models for masonry.

It must be noted that the conclusions drawn from the presented work are limited to the specific studied cases, namely, the constituent materials and bond arrangement used for the walls, the curing conditions, and the age at which the tests were performed. The intrinsic heterogeneity of masonry as a substrate material requires further research and more experimental investigation is needed to establish a richer, more robust database. In particular, additional studies are suggested with a special focus on different types of connectors, anchor diameter and embedment length, wall material and dimensions, confinement stresses, and actuator position, among other aspects.