Construction and Building Materials 47 (2013) 1372–1380

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Experimental assessment of shear strength parameters on rubble stone

masonry specimens
Jelena Milosevic, António Sousa Gago ⇑, Mário Lopes, Rita Bento
ICIST, IST, Technical University of Lisbon, Portugal

h i g h l i g h t s

 The in plane shear strength of rubble stone masonry walls is studied by means of experimental tests.
 The cohesion of rubble stone masonry panels is assessed by diagonal compression tests.
 The cohesion and the friction coefficient of rubble stone masonry panels are evaluated by triplet tests.
 The obtained experimental results (shear strength parameters) are compared with other experimental works.

a r t i c l e i n f o a b s t r a c t

Article history: Rubble stone masonry walls were widely used in traditional buildings in Mediterranean countries. How-
Received 3 September 2012 ever, the mechanical behaviour of those walls is not completely characterized due to a lack of experimen-
Received in revised form 30 May 2013 tal data. This paper presents an experimental campaign carried out to characterize the shear strength
Accepted 17 June 2013
parameters of traditional Mediterranean rubble stone masonry walls. Fifteen masonry specimens were
built using old techniques and traditional materials. Two types of mortar were used in the specimens’
execution, air and hydraulic lime mortars, with intend to simulate different masonry construction peri-
Keywords:
ods. The goal of the experimental campaign was to evaluate the most important mechanical parameters
Rubble stone masonry
Cohesion
needed for numerical modelling of traditional rubble stone masonry walls, namely, the compressive
Shear strength strength and Young’s modulus through compression tests; the cohesion and friction coefficient by triplet
Diagonal compression tests tests; and the diagonal tensile strength via diagonal compression tests. The tests’ setup and load–dis-
Triplet tests placement diagrams are presented for all tests and the obtained shear strength parameters are compared
with values from the literature.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction masonry walls are still scarce. It must be highlighted the in situ
experimental tests performed by Corradi [2–4], Brignola [5,6] and
Buildings with heritage value require minimal and reversible Borri [7,8], which allowed the shear characterization of traditional
interventions, which have to be based on surveys and diagnostic Italian rubble stone masonry walls by means of diagonal compres-
studies. A full understanding of the structural behaviour and mate- sion tests. However, further tests are needed to obtain supporting
rial characteristics is essential for the restoration and no action data for safety assessment studies of old masonry buildings in
should be undertaken without detailed examination of benefits other locations. The observation of seismic damage on load-bear-
and disadvantages of the adopted methodologies [1]. ing masonry walls showed that masonry panels under in-plane
The mitigation of seismic risk in many Mediterranean urban loading may have two typical types of behaviour associated to dif-
areas requires the reinforcement of old buildings, which must be ferent failure modes [9,10]: flexural behaviour (rocking with
preceded by assessment of the actual seismic buildings’ safety. crushing) and shear behaviour (sliding shear failure and diagonal
For that purpose a complete mechanical characterization of the cracking). The occurrence of those failure modes depends on the
masonry walls is needed. In the last decades some laboratory wall’s geometry, boundary conditions, acting axial load and mate-
and in situ tests have been performed on masonry load-bearing rial’s characteristics. Different models oriented to describe those
walls, but in literature values for shear parameters on rubble stone specific failure modes are present in the literature [10,11] and
codes [12,13]. For diagonal cracking the Turnsek and Cacovic mod-
el [14] is usually accepted. In these models the strength domain is
⇑ Corresponding author. Address: IST DECivil - Av. Rovisco Pais, 1049-001 Lisboa,
Portugal. Tel.: +351 218418201; fax: +351 218418200.
defined through a single material parameter: the tensile strength
E-mail address: [email protected] (A.S. Gago). of masonry, usually obtained by diagonal compression tests. For

0950-0618/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.conbuildmat.2013.06.036
 J. Milosevic et al. / Construction and Building Materials 47 (2013) 1372–1380 1373

sliding, the strength domain is usually defined by two parameters (160  40  40 mm) were tested for each type of mortar. The obtained mean value
for the mortar flexural strength (i.e. tensile strength obtained by bending tests) was
[15,16]: the cohesion and friction coefficients, usually evaluated
0.35 MPa for hydraulic lime mortar and 0.25 MPa for air lime mortar. Compression
through triplet tests. However, the experimental evaluation of tests were also performed on the mortar specimens and the obtained mean values
cohesion and friction coefficients may pose some problems in for the compressive strength were 1.47 MPa for the hydraulic and 0.56 MPa for the
irregular masonry where the mortar joints are not regularly air lime mortar. Rubble masonry specimens for compression tests were made using
arranged. stone units with variable shape and dimensions, which were randomly assembled.
Specimens for triplet tests were also built with stone units with variable shape and
In this paper is described an extensive experimental campaign
dimensions but special attention was paid in order to create three horizontal layers
conducted to characterize the mechanical behaviour of load bear- of stone units (each 14 cm height – Fig. 1) with two almost horizontal bed joints.
ing rubble stone masonry walls existing in old Lisbon buildings. The specimens for diagonal compression tests were built in the position where
Three types of tests were performed: compression tests, triplet they were tested, (i.e. faces at 45° to the horizontal and vertical) starting from a bot-
tom steel-loading shoe (Fig. 1). Before the test started, a second steel-loading shoe
tests and diagonal compression tests. To characterize the compres-
was placed at the upper corner of the specimens, to allow the vertical loading. To
sive strength and Young’s modulus, compression tests were con- study the influence of the stone arrangement in the specimen’s tensile strength
ducted on small specimens. In these tests the recommendations the stones were applied in horizontal layers in first masonry panel, whereas in
of the EN 1052-1 standard [17] were followed. Other prototypes the other three panels the stone layers were applied in the diagonal direction, i.e.
were subjected to triplet tests to quantify the masonry shear parallel to the specimens edges. The specimens were built with a 70 cm thickness,
which is the typical thickness of the external masonry walls of masonry Lisbon
strength parameters cohesion and coefficient of friction. Although
buildings [25,26].
in rubble stone masonry walls the irregular stone arrangement
may prevent the shear sliding, it can happen in constructive (hor-
izontal) joints at the base and at floor levels. In triplet tests the ma-
3. Experimental tests
jor lines of EN 1052-3 standard [18] and of other works [19–21]
were only followed partially, since those publications are related As mentioned, the experimental campaign described in this paper consists of
to regular masonry. Diagonal compression tests on masonry spec- two compression tests, nine triplet tests and four diagonal tests. The tests and
imens were performed to estimate the tensile strength (diagonal the specimens are identified by a two-digit index code in which the first digit indi-
cates the type of test (C: compression test, T: triplet test, W: diagonal compression
cracking failure) of traditional masonry. The test setup and the pro-
test) and the second digit is the specimen number. Table 1 lists the specimens and
cedure followed the ASTM E519-02 standard [22] and what is sug- the corresponding tests.
gested in recent research works [3,5]. The performed tests were
intended to evaluate the rubble stone masonry mechanical behav-
iour and to compare the strength parameters obtained by the dif- 3.1. Compression tests
ferent types of tests (cohesion, friction coefficient and tensile
strength). The aim of the compression tests was to assess the strength and Young’s mod-
Due to the difficulty of performing all the tests in situ and ulus of the masonry under compression. Two masonry specimens (C1 and C2 –
hydraulic and air lime mortar specimens, respectively) were compressed by a
removing samples from old buildings, masonry specimens were
3000 kN hydraulic jack. The specimens were centered in the testing machine with
built on the laboratory using traditional materials and techniques: one displacement transducer placed on each side of the panels and were loaded
two 40  40  40 cm3 specimens were built for compression tests; continuously till the rupture. The compressive strength fc and Young’s modulus E
nine 60  40  40 cm3 specimens were made to perform triplet were evaluated from the experimental data using the following procedure [17]:
tests; and four 120  120  70 cm3 specimens were built to assess
their behaviour under diagonal compression. The specimens were fc ¼
F y;max
and E ¼
F y;max
ð1Þ
built with two different mortars (Table 1): half with air lime mor- A 3eA
tar (to simulate traditional walls in old buildings) and the other
half with hydraulic lime mortar (to simulate walls in less older where Fy,max is the maximum load reached on a specimen, A is the specimen loaded
cross-section, and e is the strain of the specimen when a load of 1/3 of the maximum
buildings). The specimens were tested 8 months after their con-
load was achieved.
struction to ensure the mortar’s hardness. The compression tests results are summarized in Table 2 and depicted in Fig. 2.
The experimental results showed an unexpected similarity between the
strength of hydraulic and air lime specimens, which can be explained by the ob-
2. Specimens execution and materials served failure modes. In both cases the failure involved stone crushing by stone
to stone contact and the mortar type had a minor influence on the specimen’s ulti-
In order to reproduce the common features of traditional Lisbon construction, mate strength. Due to the irregularity of the stone units used in the specimen’s exe-
the masonry specimens were built with one leaf of roughly cut stones. The stone cution, the mortar joints had non-uniform thickness with thin layers of mortar
used in the specimens’ manufacture was the most common stone of Lisbon build- between some of the stones’ irregularities. As a consequence of the mortar crushing
ings (‘‘Lioz’’ limestone), which has an average compressive strength of 50 MPa (ob- at an advanced stage of loading, some stone edges made contact with each other
tained in compressive tests on 10 cm edge length cubic samples [23]). During the and so the stones’ strength governed the specimens’ behaviour under compression.
execution the stones were chosen to maximize the fitting and to leave the fewest The failure by stone crushing explains the similitude between the strength of the
voids as possible. The biggest stones (with longest edge of about 30 cm) were used two specimens and, also, the unusually high compression strength obtained for
in corners and edges and the spaces among them were filled with mortar and small both types of masonry. This effect would probably be diluted in a larger specimen,
pieces of stone. The mortar was based on two types of binder (traditional air lime where it would be felt a larger influence of the mortar quality. On the other hand, in
and industrial hydraulic lime), which were mixed with a ratio binder/sand of 1/3 the compression tests reported in this paper, the lateral expansion at the top and
(in volume), following the proportions of traditional mortars. The mortars, as well bottom of the specimens was restricted by the loading plates, resulting in a confine-
as the masonry specimens, were executed by an elderly stone mason, who was ment effect that tends to increase the capacity of the specimens. Due to the dimen-
asked to follow traditional techniques. Experimental tests were performed (accord- sions of the specimens, the in-plane confinement effect that may exist in long walls
ing to the EN 1015-11 standard [24]) to evaluate the mortars’ mechanical proper- was not simulated. Due to these differences the results of the compression tests
ties, such as flexural and compressive strength: nine prismatic mortar specimens must be considered essentially indicative.

Table 1
Masonry specimens.

Test Hydraulic mortar Air lime mortar Dimensions (cm)

Compression C1 C2 40  40  40
Triplet T1, T2, T3, T4, T5 T6, T7, T8, T9 60  40  40
Diagonal compression W1, W4 W2, W3 120  120  70
1374 J. Milosevic et al. / Construction and Building Materials 47 (2013) 1372–1380

Fig. 1. View of a triplet test specimen with three horizontal layers (left) and of diagonal compression test specimens (right).

Table 2 as possible during the test. Since the collapse load is directly dependent of the ver-
Compression tests results. tical stress, which is due to the overhang structure and does not depend on the type
of masonry, the ratio of vertical stresses to the masonry’s resistance tends to be
Masonry Specimens Type of mortar Fy,max (kN) fc (MPa) E (GPa) smaller for hydraulic lime panels, where higher strength was expected. Therefore,
C1 Hydraulic lime 1282 8.01 1.64 the 0.5 MPa compression level was not considered in the hydraulic lime specimens
C2 Air lime 1186 7.41 0.56 to avoid extremely high horizontal forces on the setup. For specimens with hydrau-
lic lime mortar, vertical stresses of 0.1 MPa, 0.2 MPa and 0.3 MPa were applied
E is the secant value at 1/3 of the ultimate load. (Table 3).
The displacements were recorded by thirteen linear voltage displacement trans-
ducers (LVDTs) placed on the four specimens faces (Fig. 4). On the front and back
faces six transducers measured horizontal displacements and two transducers mea-
sured vertical displacements. On the face where the horizontal load was applied
one transducer was placed at the actuator and two horizontal transducers mea-
sured the middle layer horizontal displacement. On the opposite face two other
horizontal transducers were placed.
According to EN 1052-3 [18] the specimen shear strength fv is obtained from:

F h;max
fv ¼ ð2Þ
2A

where Fh,max is the maximum horizontal force (shear load) and A is the cross-sec-
tional area of the two shear surfaces. For moderate compressive stresses the shear
strength of mortar joints in regular masonry can be given by a Mohr–Coulomb for-
mulation [27–30], which can also be assumed for the present case. Therefore, the
shear strength fv of horizontal bed joints in rubble stone masonry specimens submit-
ted to a compressive stress r is given by:

fv ¼ fv o þ l  r ð3Þ

Fig. 2. Compression tests – ‘‘force–displacement’’ diagrams. where l and fvo stand for coefficient of friction and cohesion, respectively.
As stated in EN 1052-3 [18], the parameters fvoand l can be obtained from sev-
eral triplet tests performed with different compressive stress levels by means of lin-
3.2. Triplet tests ear regression.
In triplet tests, as expected, all specimens collapsed by sliding of the middle
To quantify the shear strength parameters of horizontal bed joints in rubble stone layer (Fig. 5) and higher shear strengths were obtained for higher compres-
stone masonry, triplet tests were performed on nine specimens. The rubble stone sion levels. The results are summarized in Table 3 and the ‘‘force–displacement’’
masonry specimens were subdivided into two groups, depending on the type of diagrams are depicted in Figs. 6 and 7. Transducer TSH2 recorded the horizontal dis-
mortar used in its execution: hydraulic lime mortar specimens (T1–T5) and air lime placement plotted in the ‘‘force–displacement’’ diagrams and the load cell placed
mortar specimens (T6–T9). Following EN 1052-3 [18], the specimens were submit- next to the horizontal jack measured the force magnitude. The points where the lin-
ted to a vertical compression load and then a horizontal load was applied at the ear elastic behaviour ends and the points of maximum horizontal force are also
middle stone layer. The horizontal movement of the upper and lower stone layers marked in the ‘‘force–displacement’’ diagrams.
was prevented by rigid supports (Fig. 3) and the horizontal load was gradually in- Comparing with other triplet tests, test T6 showed some peculiarities in the
creased till the sliding of the middle stone layer. The vertical load’s magnitude was specimen’s behaviour. The force–displacement diagram shows a relatively long pla-
defined by the actual state of stresses of load bearing walls found in old Lisbon ma- teau with slight hardening, registering the maximum horizontal force at relatively
sonry buildings. Three different vertical stress levels were adopted for air lime spec- large horizontal displacement. This hardening behaviour may be attributed to a
imens (Table 3): 0.1 MPa, 0.3 MPa and 0.5 MPa, which were kept constant as much stronger interlocking effect of the stones along the nearly horizontal failure surface.

Fig. 3. Setup for triplet test.

 J. Milosevic et al. / Construction and Building Materials 47 (2013) 1372–1380 1375

Table 3
Triplet tests results.

Specimen Type of Vertical force Vertical stress r Maximum horizontal force Shear strength fv Shear strength fv (average value)
mortar (kN) (MPa) (kN) (MPa) (MPa)
T1 Hydraulic 24 0.1 126 0.26 0.33
T2 188 0.39
T5 48 0.2 213 0.44 0.44
T3 72 0.3 267 0.56 0.57
T4 279 0.58
T6 Air lime 24 0.1 64 0.13 0.13
T7 56 0.12
T8 72 0.3 139 0.29 0.29
T9 120 0.5 161 0.34 0.34

Fig. 4. Triplet tests – transducers on the specimen’s front (a) and back (b) faces, on the restrained lateral face (c) and on the loaded lateral face (d) (dimensions in (cm)).

Fig. 5. Crack pattern of masonry specimens: (a) specimen T1 and (b) specimen T7.

As observed in test T7 (Fig. 7), without this hardening effect the maximum load the vertical displacement showed is the average of displacements recorded by
would be slightly smaller and would have been registered at a much smaller hori- LVDT’s 12 and 13 and the corresponding horizontal displacement was recorded
zontal displacement. using the transducer placed on the horizontal actuator (TSH2). In Figs. 8 and 9
An important issue regarding shear tests is the dilatancy of masonry joints. the points where the linear behaviour ends and the points of maximum force
The dilatancy (i.e. the relation between the vertical and the horizontal strains) are also marked. It can be noticed that the vertical displacements started to in-
has a significant role in numerical modelling and can be much more relevant in crease after the end of the linear behaviour. The small slope of the ‘‘vertical vs.
rubble stone masonry walls than in brick masonry walls. This is because after horizontal displacements’’ diagram around the point of maximum load detected
cracking and sliding the two sides of the cracks do not match, indicating an in- at test T6 (Fig. 9) can be explained by the occurrence of the maximum force for
crease in volume. This effect increases with the irregularity of the crack surfaces, a large horizontal displacement.
therefore it tends to be stronger in rubble stone masonry than in brick masonry. Fig. 10 depicts the relation between vertical load and horizontal displacement
Figs. 8 and 9 show the vertical displacement (that can be considered as measuring for all tests, where it can be seen that during the tests the vertical load was almost
dilatancy) as a function of the horizontal displacement. It is important to note that constant, as required.
1376 J. Milosevic et al. / Construction and Building Materials 47 (2013) 1372–1380

Fig. 6. Triplet tests – ‘‘force–displacement’’ diagrams for hydraulic lime mortar

Fig. 9. Triplet tests – vertical vs. horizontal displacements for air lime mortar
specimens.
specimens.

Fig. 10. Triplet tests – vertical load vs. horizontal displacement.

Fig. 7. Triplet tests – ‘‘force–displacement’’ diagrams for air lime mortar specimens.

Fig. 11. Triplet tests – relation between shear stress and normal stress for hydraulic
Fig. 8. Triplet tests – vertical vs. horizontal displacements for hydraulic lime mortar and air lime mortar specimens (R is correlation coefficient of the linear regression).
specimens.

As stated in EN 1052-3 [18], the linear regression of results from several triplet It must be mentioned that the two shear strength values obtained for the
tests performed with different compressive stress levels provides the shear strength hydraulic lime masonry panels, tested with the lower vertical stress level
parameters of the Coulomb’s friction model, namely the initial shear strength fvo (0.1 MPa), are quite different (D = 33%) and a third triplet test with this vertical
(cohesion) and the coefficient of friction l. Fig. 11 shows the relation between stress would have increased the accuracy of the calculations. The low shear strength
the vertical compression stress and the shear strength for all tests. Two straight value obtained in test T1 (0.26 MPa) that may induce a high value for the calculated
lines, one for each type of mortar specimen, evaluated by linear regression are also hydraulic lime specimen’s friction coefficient, must be confirmed.
presented in the graph. It is worth mentioning the good correlation between the Table 4 summarizes some results published in the literature obtained by triplet
experimental results and the linear regression lines, which confirms the initial tests on regular masonry (regular units with horizontal bed joints). Different values
assumption of Coulomb’s friction law for the shear strength of horizontal bed joints for shear strength parameters were obtained for different types of masonry. For
in rubble stone masonry specimens. For hydraulic lime mortar specimens the val- specimens with hydraulic lime mortar and stone bricks the cohesion and the fric-
ues obtained by linear regression for cohesion and coefficient of friction were tion coefficient values are around 0.3 MPa and 0.7, respectively. Even though that
0.20 MPa and 1.23, respectively. For air lime mortar the obtained values were the tests on rubble and regular stone masonry are not directly comparable, the pre-
0.08 MPa for cohesion and 0.56 for coefficient of friction. vious values give some confidence to the obtained results for rubble stone masonry.
 J. Milosevic et al. / Construction and Building Materials 47 (2013) 1372–1380 1377

Table 4
Triplet tests – literature review.

Masonry type Cohesion fvo (MPa) Coefficient of friction l

Tuff units, pozzolanic mortar 0.3 2.0 Prota, et al. [19]
Hollow units, cement mortar 1.6 0.9 Gabor, et al. [31]
Solid brick units, cement mortar 0.7 0.7 Amadio and Rajgeli [32]
Solid brick units, hydraulic lime mortar 0.2 0.8 Magenes and Calvi [33]
Sand stone units, hydraulic lime mortar 0.3 0.7 Binda, et al. [34]
Bricks, hydraulic lime mortar 0.2 0.6 Roberti, et al. [35]

Fig. 12. Setup for diagonal compression tests.

3.3. Diagonal compression tests

The shear elastic modulus G is obtained by:
In the diagonal compression tests the masonry specimens were placed in the test
s1=3
rig with a diagonal axis in the vertical direction (Fig. 12) and loaded in compression G¼ ð8Þ
c1=3
along this direction (causing a diagonal tension failure with the specimen splitting
apart along the vertical diagonal). The vertical load was applied by an 800 kN
where s1/3 is the shear stress for a load of 1/3 of the maximum load Pmax and c1/3 is
hydraulic jack acting on the steel-loading shoe placed at the top specimens corner.
the corresponding shear strain.
Nine linear displacement transducers were placed on each specimen (Fig. 12):
As mentioned, four specimens were tested under diagonal compression – two
five transducers on the front face, three on the back face and one to measure verti-
built with hydraulic lime mortar (W1 and W4) and two with air lime mortar (W2
cal displacement under the hydraulic jack. To avoid damages on the instrumenta-
and W3). In all tests the specimens’ collapse was caused by a main crack developed
tion, the transducers were removed (except the one placed on the hydraulic jack)
in the middle of the specimen, continuously propagating from the center towards
when the behaviour of the specimen indicated that it could be close to failure. After
the upper and bottom specimen’s corners. It should be mentioned that in all tests
that the vertical displacement was continuously increased at the top of the speci-
the stones were not damaged and the crack appeared only through the mortar,
mens until the collapse.
dividing the specimen in two almost symmetrical parts (Fig. 13).
The analysis and interpretation of the diagonal compression tests differ from
Despite of the collapse quasi-brittle nature in all cases, the specimens showed
author to author and from standard to standard. The test was introduced to simu-
different behaviour after the collapse: the specimens with air lime mortar (W2
late a pure shear stress state and the ASTM standard [22] assumes that the diagonal
and W3) disintegrated, while the specimens with hydraulic mortar (W1 and W4)
compression test produces a uniform shear stress and a Mohr’s circle centered in
broke in two parts, each remaining in one piece. The ‘‘force–displacement’’ dia-
the origin of the Cartesian system of axis. In that case the tensile strength of ma-
grams (where the vertical displacement is the average value of the measurements
sonry ft is equal to the shear cohesion fvo and is obtained by:
recorded using LVDTs 3 and 7), for hydraulic lime and air lime mortar specimens are
0:707  P presented in Figs. 14 and 15, respectively. The results are also summarized in Ta-
ft ¼ fv o ¼ ð4Þ ble 5. The dotted parts of the curves in Figs. 14 and 15 were obtained using the mea-
An
surements of the transducers under the hydraulic jack, instead of the average values
where P is the maximum load applied by the vertical jack and An is the net area of the for all measurements. Some unloading and loading branches are visible on the
specimen, calculated as follows: ‘‘force–displacement’’ diagrams, which were due to pauses during which the ap-
  plied load decreased as a consequence of the damages development. The pauses
wþh
An ¼ tn ð5Þ could not be avoided, as it was necessary to stop the loading to analyse the crack
2 patterns or to remove transducers.
For specimen W1 the maximum applied load was 372 kN with a vertical short-
where w is the specimen width, h is the specimen height, t is the thickness of the
ening of 1.55 mm (Point 1 – Fig. 14). The collapse occurred at a larger vertical dis-
specimen and n is the percentage of the unit’s gross area that is solid, expressed
placement (5.29 mm) with a load of 268 kN (Point 2 – Fig. 14). In the case of the
as a decimal (in the present work the value n = 1 was adopted).
specimen W4 the collapse occurred when the maximum load was applied:
The RILEM code [36] considers that the stress field is not uniform and proposes
306 kN with a vertical displacement of 3.47 mm (Point 3 – Fig. 14). The more
the following expression to evaluate the tensile strength of masonry:
(apparent) ductile behaviour of specimen W1 is related to the stone arrangement,
0:5  P since the W1 specimen was built with horizontal stone layers (at 45° to the external
ft ¼ ð6Þ
An inclined surfaces), whereas, to be representative of real masonry walls, the other
three specimens were built with diagonal layers (at 45°). As expected, the speci-
Both interpretations are able to represent the experimental crack pattern, lo- mens built with air lime mortar showed much lower strength than the specimens
cated along the diagonal, at 45°. based on hydraulic lime mortar. The ultimate load for specimen W2 was 29 kN,
Brignola et al. [5] proposed a new and more accurate methodology for the eval- with a vertical shortening of 1.58 mm (Point 1 – Fig. 15), and for the specimen
uation of the tensile strength of masonry from diagonal compression tests with the W3 the ultimate load was 28 kN, with a vertical displacement of 1.52 mm (Point
following expression: 2 – Fig. 15).
aP In the diagonal compression tests the specimens’ collapse was achieved without
ft ¼ ð7Þ damage to the stones, i.e., the cracks propagated through the mortar joints. That fact
An
is in accordance with the observed major influence of the mortar type in the
where a is a coefficient dependent on masonry typology. For irregular masonry a specimen’s strength. In addition, the values for tensile strength obtained by diago-
should be lower than 0.5 (a = 0.35 for rubble stone masonry) and for regular ma- nal compression tests (Table 5) are much lower than the values obtained for the
sonry a can be assumed equal to 0.5, coherently with the RILEM proposal. mortar’s tensile strength by bending tests (0.35 and 0.25 MPa for hydraulic and
1378 J. Milosevic et al. / Construction and Building Materials 47 (2013) 1372–1380

Fig. 13. Diagonal compression tests – main crack at the middle of the specimens.

Table 5
Diagonal compression tests.

Masonry Pmax ASTM RILEM Brignola et al. [5]

specimen (kN) a = 0.35
ft G ft ft (MPa)
(MPa) (MPa) (MPa)
W1 372 0.313 389 0.220 0.155
W2 29 0.024 58 0.017 0.012
W3 28 0.024 93 0.017 0.012
W4 306 0.258 252 0.182 0.128

Table 6
Diagonal compression (in situ) tests – literature review.

Traditional Masonry (lime mortar and ft = a  p/An

Fig. 14. Diagonal compression tests – ‘‘force–displacement’’ diagrams for hydraulic calcareous stone)
lime mortar specimens. a ft (MPa)
Double-leaf roughly cut stone masonry 0.707 0.05–0.07 Corradi [3]
Random rubble stone masonry 0.707 0.06–0.16 Chiostrini
[37]
Random rubble stone masonry 0.5 0.04–0.11 Chiostrini
[37]
Random rubble stone masonry 0.35 0.02–0.04 Brignola
[5]
Double-leaf roughly cut stone masonry 0.707 0.02–0.04 Borri [8]
Triple-leaf roughly cut stone masonry 0.707 0.02–0.04 Borri [8]
Random rubble stone masonry 0.5 0.04–0.07 Brignola
[6]

Table 7
Reference values of the mechanical parameters (maxima and minima) NTC08, 2008
[13]ft = average tensile strength; E = average value of the elastic modulus; G = average
value of the shear modulus.

Traditional Masonry (lime mortar and ft (MPa) E (MPa) G (MPa)

calcareous stone)
Fig. 15. Diagonal compression tests – ‘‘force–displacement’’ diagrams for air lime
mortar specimens. Irregular stone masonry (pebbles, 0.03–0.05 690–1050 230–350
erratic and irregular stone)
Uncut stone masonry with facing walls 0.05–0.08 1020–1440 340–480
air lime mortars, respectively) and the relative differences between the tensile of limited thickness and infill core
strength of hydraulic and air lime mortar specimens is much more marked in the Cut stone masonry with good bonding 0.08–0.11 1500–1980 500–660
diagonal compression tests. This can be due to the fact that the phenomena that
conditions rupture is different in both cases, i.e.: in the flexural test it is the tensile
strength of the mortar and in the case of the diagonal compression test it is the ten- aged stage, with small displacements, where measurement errors may have an
sile resistance of the interface between the mortar and the stones. important influence. For specimens built with hydraulic lime mortar (W1 and
Moreover, from the differences in tests W1 and W4, whose specimens were W4) the variation of the shear modulus results can also be explained by the differ-
built with different stone arrangements, and from the similarities obtained in tests ent stone arrangement adopted on the specimens.
W2 and W3, specimens of which were built with the same stone arrangement, it Table 6 summarizes some published results obtained by in situ diagonal com-
can also be concluded that the stone arrangement influences (by a moderate de- pression tests on traditional masonry and in Table 7 are shown the values of tensile
gree) the masonry strength and its deformation capacity. strength and Young’s modulus proposed by the Italian Standard NTC08, 2008 [13]
For the air lime mortar specimens (W2 and W3) the maximum compression for traditional masonry. Comparing the literature values (Tables 6 and 7) with the
loads were similar but the shear elastic modulus G varies significantly. The reported
variation can be due to the fact that the shear modulus is evaluated on the undam-
 J. Milosevic et al. / Construction and Building Materials 47 (2013) 1372–1380 1379

obtained values in the performed diagonal compression tests it can be concluded Acknowledgments
that the obtained results deserve some confidence, even though that the number
of tested specimens of each type of mortar was limited.
The authors acknowledge the financial contribution of FCT
(Fundação para a Ciência e a Tecnologia) to the project SEVERES:
‘‘Seismic Vulnerability of Old Masonry Buildings’’
(www.severes.org).

4. Conclusions References

[1] ICOMOS: Recommendations for the analysis, conservation and structural

The goal of the present research programme was to assess the restoration of architectural heritage. ICOMOS, International Scientific
most important mechanical parameters of specimens similar to Committee for Analysis and Restoration of Structures of Architectural
Heritage; 2003.
traditional load bearing walls on rubble stone masonry. Fifteen
[2] Corradi M, Borri A, Vignoli A. Strengthening techniques tested on masonry
masonry specimens were specially built for the present experi- structures struck by the Umbria-Marche earthquake of 1997–1998. Constr
mental program using traditional techniques and materials (lime Build Mater 2002;16:229–39.
stones, air lime and hydraulic lime mortars). Samples made with [3] Corradi M, Borri A, Vignoli A. Experimental study on the determination of
strength of masonry walls. Constr Build Mater 2003;17:325–37.
hydraulic lime mortar were built to simulate the latter stage of tra- [4] Corradi M, Tedeschi C, Binda L, Borri A. Experimental evaluation of shear and
ditional structural masonry construction in Portugal. compression strength of masonry wall before and after reinforcement: Deep
The experimental program discussed in this paper focused on repointing. Constr Build Mater 2008;22:463–72.
[5] Brignola A, Frumento S, Lagomarsino S, Podestà S. Identification of shear
the three types of tests available to characterize the masonry parameters of masonry panels through the in-situ diagonal compression test.
mechanical behaviour: compression, triplet and diagonal compres- Int J Arch Heritage 2008;3:52–73.
sion tests. [6] Brignola A, Podestà S, Lagomarsino S. Experimental results of shear strength
and stiffness of existing masonry walls. New Delhi: Structural Analysis of
Compression tests allowed the evaluation of compressive Historical Constructions; 2006.
strength and the Young’s modulus for traditional stone masonry [7] Borri A, Castori G, Corradi M. Shear behavior of masonry panels strengthened
and triplet tests were used to estimate the masonry cohesion by high strength steel cords. Constr Build Mater 2010;22:463–72.
[8] Borri A, Castori G, Corradi M, Speranzini E. Shear behaviour of unreinforced and
and coefficient of friction. The obtained results on compressive
reinforced masonry panels subjected to in situ diagonal compression tests.
tests (fc = 7.41 MPa; E = 0.56 GPa for air lime mortar specimens Constr Build Mater 2011;25:4403–14.
and fc = 8.01 MPa; E = 1.64 GPa for hydraulic lime mortar speci- [9] Calderini C, Cattari S, Lagomarsino S. In-plane strength of unreinforced
masonry piers. Earthquake Eng Struct Dyn 2009;38:243–67.
mens) must be regarded as indicative values due to the limited
[10] Magenes G, Calvi GM. In-plane seismic response of brick masonry walls.
number of performed tests and to the test boundary conditions. Earthquake Eng Struct Dynam 1997;26:1091–112.
Two contradictory effects of the test boundary conditions may [11] Calderini C, Cattari S, Lagomarsino S. The use of the diagonal compression test
have affected the results, i.e.: the dimensions of the specimens to identify the shear mechanical parameters of masonry. Constr Build Mater
2010;24:677–85.
do not reproduce the in-plan confinement of real long loadbear- [12] EC 6: Eurocode 6 Design of masonry structures, part 1–1: general rules for
ing walls; and the confinement imposed by the bearing plates buildings – rules for reinforced and unreinforced masonry 1995. ENV 1996–1-
at the top and bottom of the specimens does not exist in real 1:1995.
[13] (NTC08, 2008). Ministero delle Infrastrutture e dei Transporti – Nuove Norme
situations. Tecniche per le Costruzioni. Italian Ministry of Infrastructures and
Typical failure modes were observed on the triplet tests and the Transportation. G.U. n.29 – S.O. n.30, February, 4th 2008, Rome, Italy [in Italian].
obtained results for cohesion and coefficient of friction were, [14] Turnšek V, Čačovič F. Some experimental results on the strength of brick
masonry walls. In: Proc of the 2nd international brick masonry conference,
respectively, 0.08 MPa and 0.56, for air lime mortar specimens Stoke-on-Trent; 1970. p. 149–56.
and 0.20 MPa and 1.23, for hydraulic lime mortar specimens. The [15] Calderini C, Cattari S, Lagomarsino S. Identification of shear mechanical
high value obtained for the friction coefficient of hydraulic lime parameters of masonry piers from diagonal compression test. In: Proc. 11th
Canadian Masonry symposium, Toronto, Ontario; 2009.
mortar specimens could have been a result of the low strength va- [16] Mann W, Müller H. Failure of shear-stressed masonry – an enlarged theory,
lue obtained in test T1. However, the coefficient of friction on those tests and application to shear-walls. In: Proc of the international symposium
specimens should have been significantly higher than that ob- on load bearing brickwork, London; 1980. p. 1–13.
[17] EN 1052-1 Methods of test for masonry-Part 1: determination of compressive
tained for air lime mortar specimens.
strength; 1998.
To obtain the tensile strength that bounds the strength do- [18] EN 1052-3 Methods of test for masonry-Part 3: determination of initial shear
main of the (diagonal cracking) Turnsek and Cacovic model strength; 2002.
[14], diagonal compression tests on four specimens were also per- [19] Prota A, Marcari G, Fabbrocino G, Manfredi G, Aldea C. Experimental in-plane
behaviour of tuff masonry strengthened with cementitious matrix–grid
formed. Two specimens were built with hydraulic mortar and the composites. J Compos Constr 2006;10(3):223.
other two with air lime mortar. The tested specimens showed [20] Lourenço PB, Barros JO, Oliveira JT. Shear testing of stack bonded masonry.
brittle behaviour with low values for tensile strength, especially Constr Build Mater 2004;18:125–32.
[21] Vasconcelos G, Lourenço PB. Experimental characterization of stone masonry
in the case of the air lime mortar specimens (ft = 0.024 MPa, in shear and compression. Constr Build Mater 2009;23:3337–45.
according to ASTM [22] methodology). The specimen built with [22] ASTM. ASTM E 519-02 Standard test method for diagonal tension (shear) in
hydraulic mortar and stone layers parallel to the faces reached masonry assemblages. West Conshohocken (PA): ASTM International; 2002.
[23] Carvalho J. Mechanical characterization of loadbearing masonry stone through
a medium value for the initial shear strength of about ft = 0.258 - non-destructive testing (in Portuguese), Master’s thesis. Lisbon: Instituto
MPa (according to ASTM [22] methodology). It was noted in the Superior Técnico; 2008.
shear tests that the mortar composition has an important influ- [24] CEN.: 1015–11: 1999 Methods of test for mortar for masonry-Part 11:
determination of flexural and compressive strength of hardened mortar; 1999.
ence on shear strength. The tests also showed that the stone [25] Cardoso R, Lopes ML, Bento R. Seismic evaluation of old masonry buildings.
arrangement could influence the shear strength, as the specimen Part I: method description and application to a case-study. Eng Struct
with layers at 45° to the faces showed shear strength 21% above a 2005;27:2024–35.
[26] Mendes N, Lourenço PB. Seismic assessment of masonry ‘‘Gaioleiro’’ buildings
similar specimen with stone layers parallel to the faces (both
in lisbon, Portugal. J Earthquake Eng 2010;14:80–101.
with hydraulic mortar). It must be said that the representative- [27] Vasconcelos G. Experimental investigations on the mechanics of stone
ness of the obtained results depends on the limited number of masonry: characterization of granites and behaviour of ancient masonry
specimens tested on diagonal compression. However, they are in- shear walls, PhD thesis,. Portugal: University of Minho; 2005.
[28] Hamid AA, Drysdale RG. Behavior of brick masonry under combined shear and
side the range of values obtained by other researchers for similar compression loading. In: Proc 2nd Canadian Masonry conference; 1980. p. 57–
masonry specimens. 64.
1380 J. Milosevic et al. / Construction and Building Materials 47 (2013) 1372–1380

[29] Atkinson RH, Amadei BP, Saeb S, Sture S. Response of masonry bed joints in [34] Binda L, Fontana A, Mirabella G. Mechanical behavior and stress distribution in
direct shear. J Struct Eng 1989;115:2277–96. multiple-leaf stone walls. In: Proceedings of 10th international brick block
[30] Riddington JR, Ghazali MZ. Hypothesis for shear failure in masonry joints. Proc masonry conference, Calgary, Canada; 1994. p. 51–9.
Instn Civil Eng 1990:89–102. [35] Roberti GM, Binda L, Cardani G. Numerical modeling of shear bond tests on
[31] Gabor A, Bennani A, Jacquelin A, Lebon F. Modelling approaches of the in-plane small brick–masonry assemblages. In: Computer methods in structural
shear behavior of unreinforced and FRP strengthened masonry panels. J masonry – 4, Florence, Italy; 1997.p. 145–52.
Compos Struct 2006;74:277–88. [36] RILEM TC: 76-LUM Diagonal tensile strength tests of small wall specimens. In
[32] Amadio C, Rajgelj S. Shear behavior of brick–mortar joints. Mason Int RILEM, Recommendations for the Testing and Use of Constructions Materials,
1991;5:19–22. London: E&FN SPON ;1994. p. 488–89.
[33] Magenes G, Calvi GM. Cyclic behaviour of brick masonry walls. In: Proc of the [37] Chiostrini S, Galano L, Vignoli A. On the determination of strength of ancient
10th World Conference on Earthquake Engineering, Madrid; 1992. p. 3517– masonry walls via experimental tests. In: Proceedings of the Twelfth World
522. Conference on Earthquake Engineering (CD-ROM), Auckland, New Zealand;
2000. paper no 2564.