You are on page 1of 24

International Journal of Architectural Heritage

ISSN: 1558-3058 (Print) 1558-3066 (Online) Journal homepage: https://www.tandfonline.com/loi/uarc20

Experimental and Numerical Response of Arch


Bridge Historic Masonry Under Eccentric Loading

Gianmarco de Felice & Stefano De Santis

To cite this article: Gianmarco de Felice & Stefano De Santis (2010) Experimental and Numerical
Response of Arch Bridge Historic Masonry Under Eccentric Loading, International Journal of
Architectural Heritage, 4:2, 115-137, DOI: 10.1080/15583050903093886

To link to this article: https://doi.org/10.1080/15583050903093886

Published online: 03 Feb 2010.

Submit your article to this journal

Article views: 311

View related articles

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=uarc20
International Journal of Architectural Heritage, 4: 115–137, 2010
Copyright  Taylor & Francis Group, LLC
ISSN: 1558-3058 print / 1558-3066 online
DOI: 10.1080/15583050903093886

EXPERIMENTAL AND NUMERICAL RESPONSE OF ARCH


BRIDGE HISTORIC MASONRY UNDER ECCENTRIC
LOADING

Gianmarco de Felice and Stefano De Santis


University Roma Tre, Rome, Italy

This study contributes to the knowledge of the compressive behavior of brickwork used in
railway bridges and shows how to incorporate this information in structural modeling. An
experimental investigation is carried out on brickwork specimens made with old clay bricks
and lime mortar to reproduce the original components and the arrangement of masonry arch
bridges built in Italy at the end of the nineteenth century. The specimens are subjected to
monotonic and cyclic displacement-controlled compression tests, under centered and eccentric
loading. Based on experimental results, a beam model with fiber cross-section is used to
describe the macroscopic behavior of brickwork, where the fiber constitutive relationship is
estimated according to the concentric tests. Eccentric tests are finally simulated and the
comparison between theoretical predictions and experimental results reveals the capability of
the model to reproduce the global force-displacement and bending moment-curvature experi-
mental behavior and its suitability for the structural analysis of masonry arch bridges.

KEY WORDS: masonry, historic brickwork, arch bridges, fiber beam model, eccentric
loading

1. INTRODUCTION
Masonry arch bridges, built extensively in Europe in the second half of the
nineteenth century, are today still an important element of the railway infrastructure
system. As a consequence of the age of the bridges and the increase in traffic load and
train speed, there is a definite need for assessing their effective load-carrying capacity;
actually, the classical assumption of infinite strength, currently used in the assessment
of arches, may be not adequate for brick masonry railway bridges, where high loading
is combined with weak material. Since stress concentration is to be expected under
travelling load, due to the eccentricity of the load resultant over the cross-section of
the vaults, a refined estimate of brick masonry strength and ductility seems to be one
of the important steps towards a refined assessment of the effective load-carrying
capacity.
The behavior of brick masonry subjected to compression has been widely
investigated over a long period of time regarding the behavior under uniform com-
pression (for example, among others, Al Shebani and Sinha 1999; Oliveira et al. 2006;
Roberts et al. 2006), for the response under combined compression and bending
moment (Drysdale and Hamid 1982; Ding 1997; Brencich et al. 2007; Cavalieri et al.
Received 31 January 2009; accepted 4 June 2009.
Address correspondence to Gianmarco de Felice, University Roma Tre, Department of Structures, via
Corrado Segre 6, 00146, Rome, Italy. E-mail: defelice@uniroma3.it

115
116 G. DE FELICE AND S. DE SANTIS

2005), or for specimens subjected to combined in-plane forces and moments (Braga
et al. 1996; Mojsilović and Marti 1997). However, most contributions deal with
contemporary rather than historic masonry. Apart from some studies on the behavior
of masonry under compression (Venu Madhava Rao et al. 1997; Aprile et al. 2001, de
Felice and Carbone 2006), or under overload phenomena (Anziani et al. 2000), with a
view toward strengthening techniques (Valluzzi et al. 2005), only a few papers have
been devoted to historic brick masonry, which however, may behave completely
differently from contemporary masonry, due to both brick and mortar properties
(Brencich and de Felice 2009).
Some data on the mechanical properties of bricks and mortar used in arch
bridges may be found in historic treatises such as in (Rondelet 1802; Curioni 1874;
Séjourné 1913; Gay 1924); more recent experimental results can be found in (Barbi et
al. 2002), where the behavior of historic bricks under compression is presented, in
(Sala et al. 2008), where the mechanical properties of lime mortar, enriched with
natural pozzolana, are derived from experimental tests, or in (Binda et al. 1988),
where the influence of mortar types on the compressive strength of masonry is
outlined.
The present study aims at investigating the structural properties of the brick-
work used in railway arch bridges built in Italy at the end of nineteenth century, when
subjected to cyclic compression. The starting point is the consideration that the
behavior of masonry under compression plays a non-negligible role in the load-
carrying capacity assessment of the bridge. In fact, as shown in (Brencich and De
Francesco 2004), depending on the material ductility, the behavior of the bridge close
to failure may vary widely: only when masonry exhibits an adequate ductility capacity,
does the classical four-hinge mechanism take place, otherwise, failure is reached with
crushing within the critical cross-section.
With this in mind, three bridges built between 1890 and 1894 along the railway
line between Rome and Viterbo have been surveyed and the masonry properties
assessed, such as to reproduce in the laboratory specimens having characteristics
similar to the brickwork used in the bridges. This might appear as a particular choice,
but it should be considered that most of the railway bridges in Italy were built in a
relatively short period of time (in the second half of the nineteenth century), using
almost everywhere the same materials with the same construction rules (Brencich and
Morbiducci 2007). This is the reason why bridges with similar overall dimension,
material properties and structural details, may be found, without distinction, in
different regions.
The original samples of bricks that were available were tested under cyclic
compression in order to determine their effective mechanical properties. Then,
fortunately, a stock of old bricks with the same dimensions as the original speci-
mens was found, that, once tested, displayed a very similar response. As regards
the mortar, the original type is a hydraulic mortar with lime and pozzolana, which
was reproduced in the laboratory, on the basis of a petrographic and micro-
chemical analysis, carried out on some fragments taken from the three bridges
under study.
Using the stock of old bricks (dimensions 27013055 mm3) and the mortar
reproduced in the laboratory, 19 brickwork specimens were built, according to the
same arrangement of the bricks detected in the vaults, as shown in Figure 1. The
specimens were tested under cyclic-centered compression (four specimens) and cyclic

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 117

a b c

Figure 1. Ronciglione railway arch bridge: a) view; b) detail of the vault; and c) brick arrangement in the vaults
(figure is provided in color online).

eccentric compression (15 specimens), with eccentricity ranging from 30 mm to 110


mm (eccentricity versus brick height ratio ranging from 0.15 to 0.37).
Finally, the parameters of a non-linear uniaxial constitutive law for masonry as
an equivalent homogeneous material were deduced from the centered tests; then, the
eccentric tests were simulated by means of a beam model with fiber cross-section
which proves to be adequate in representing the whole experimental behavior, in terms
of both load-displacement and bending moment-curvature diagrams.

2. MASONRY COMPONENTS
The experimental investigation was carried out using an MTS hydraulic load
frame with 500 kN load rating (Eden Prairie, MN, USA). In the loading branches, the
tests were made under displacement control to obtain the overall shape of the
stress–strain relationship, with velocity 0.01 mm/sec and acquisition frequency of
10 Hz. Whereas, in the unloading branches, force control with velocity vmin¼500 N/
sec was used, to avoid the complete unloading of the specimen.

2.1. Bricks
The original bricks used in the construction of the bridges of the railway line
built between 1890 and 1894 from Rome to Viterbo, measure 27013055 mm3. To
determine their mechanical properties, 20 cubic samples with side of 4550 mm were
extracted from six bricks and submitted to monotonic and cyclic compression tests.
The tests were performed in the direction normal to the face of the bed joints, which
were previously carefully leveled; no devices were employed to reduce the friction
between the specimen surface and the plates of the load frame.
For the purpose of reproducing masonry specimens to be tested in the laboratory, a
stock of old bricks with the same characteristics and dimensions as the original samples was
found; the bricks were produced at the beginning of the twentieth century in a clink situated
in Monterotondo, not far from the bridges. Their mechanical properties were determined
by testing 11 cubic specimens of approximately 48 mm side, extracted from six bricks.
The whole stress-strain curves are reported in Figures 2 and 3 for the original
and historic bricks, respectively: both brick stocks show very similar properties in

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


118 G. DE FELICE AND S. DE SANTIS

Figure 2. Experimental stress-strain curves of a) monotonic and b) cyclic compression tests on the original
brick cubic samples belonging to Rome-Viterbo railway arch bridges (figure is provided in color online).

Figure 3. Experimental stress-strain curves of a) monotonic and b) cyclic compression tests on historic
brick cubic samples used for building masonry specimens (figure is provided in color online).

terms of strength and post-peak behavior, with a slightly higher stiffness of old bricks
compared to the original ones. The average properties are shown in Table 1: the
average compressive strength is approximately 30 MPa while the average tensile
strength, estimated with three-point bending tests, is approximately 3.8 MPa. An
apparent stiffness increase is observed in the first branch of the response curve,
which derives from the bedding effect induced by the asperities and the unavoidable
non-perfect parallelism of the loaded surfaces; however a linear branch can be clearly
identified, from the end of the ‘‘bedding phase’’, until approximately 70% of the
maximum load; the elastic stiffness is estimated by interpolation of this linear branch,
giving an average value of approximately 3100 MPa and 3900 MPa for the original
and the historic bricks, respectively. Both stocks of bricks also behave similarly under
cyclic loads, showing their capability to sustain loading–unloading cycles in the

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 119

Table 1. Experimental properties of original and historic bricks

Compressive Deformation at Elastic Tensile


strength fb [MPa] peak stress e(fb) stiffness Eb [MPa] strength fbt [MPa]

Original brick samples (compression tests: 18 samples)


Mean value 29.85 11.110-3 3067 —
Standard deviation 7.33 3.210-3 731 —
Historic brick samples (compression tests: 11 samples; 3PBT: 7 samples)
Mean value 30.51 9.410-3 3918 3.79
Standard deviation 5.24 2.110-3 395 0.24

Figure 4. Failure pattern of a cubic brick sample at the end of the compression test (figure is provided in
color online).

softening branch, without appreciable stiffness degradation, in spite of the high level
of deformation and damage reached.
Figure 4 shows the evolution of damage during the compression test: the first
cracks occur at approximately 75%–85% of peak load, together with the expulsion of
fragments from the corners; at peak load, the cracks involve the whole height of the
specimen and then increase in size and number during the softening branch of the load
displacement curve, inducing a progressive lateral dilatation of the specimen; finally,
under very high strain, slices or scraps are discharged from one or more lateral faces.

2.2. Mortar
The original mortar was made with lime and pozzolana, as shown by petro-
graphic and microstratigraphic analyses and microchemical tests (Figure 5) per-
formed on samples taken from the railway bridges. The samples contain several grey
mortar fragments, with a low cohesion level, immersed in a large quantity of powder

a b c

Figure 5. Mortar sample: a) photograph; b) shiny section enlargement 40x; and c) 100x (figure is provided in
color online).

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


120 G. DE FELICE AND S. DE SANTIS

Figure 6. Failure pattern of mortar under compression: a) cubic specimen; b) prismatic specimen with ratio
H/W ¼ 2; and c) with ratio H/W ¼ 4 (figure is provided in color online).

coming from mortar disaggregation, and dark grey round pozzolana particles (prob-
ably coming from a volcanic deposit in the Sabatini Mountains, not far from Rome).
The bonding element is a light grey carbonated lime. The granulometry varies from
lime to fine gravel, between 0.01 and 4.50 mm; the distribution of the aggregate within
the matrix is homogeneous and the grains are not oriented; the original ratio between
binder and aggregate is approximately 1:4.
A mortar with similar characteristics was then prepared in the laboratory to be
used with the previously described bricks for building masonry specimens. To estimate
the mortar properties, 11 cubic mortar samples with 40 mm sides and eight prismatic
samples, having 4040 mm2 base dimensions, with slenderness ratio height to width
H/W¼2 and 4, were tested under compression (Figure 6). The average mechanical
properties and their standard deviation are listed in Table 2, where the elastic mortar
stiffness was estimated as described in the previous section. The results show that
greater slenderness gives a lower compressive strength, which derives from the

Table 2. Experimental results of mortar samples under compression tests

Compressive Deformation at Elastic


strength fm [MPa] peak stress e(fm) stiffness Em [MPa]

Cubic specimens: 5 samples


Mean value 1.62 10.610-3 228.7
Standard deviation 0.21 3.810-3 61.8
Prismatic specimens (slenderness ratio H/W¼ 2): 4 samples
Mean value 1.46 7.810-3 267.2
Standard deviation 0.23 4.310-3 138.2
Prismatic specimens (slenderness ratio H/W ¼ 4): 4 samples
Mean value 1.27 5.810-3 329.2
Standard deviation 0.23 0.710-3 156.6

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 121

non-negligible confining effect of the loading plates. The tensile strength of the
mortar was estimated through three-point bending tests on five prismatic speci-
mens having dimensions 4040160 mm3, and an average value of 0.44 MPa was
found. The experiments were performed in conformity with the European
Standard EN 1015-11 and UNI EN 12390-6.

3. BRICKWORK UNDER UNIFORM COMPRESSION


In all, 19 brickwork specimens were made using the bricks and the mortar pre-
viously described and tested, after being aged for 2 years. Aiming at assessing the influence
of the specimen size on the experimental response, and because of the limited availability
of historic bricks, specimens with different height and width were built, as shown in Figure
7. All the tests were performed on the basis of Standard UNI EN 772-1 using the MTS
hydraulic load frame described in the previous section; the general experimental plan on
masonry samples is shown in Table 3. Four specimens were tested under cyclic-centered
axial load; these measure 140140310 mm3, and consist of five layers of bricks; two
specimens (S-CG type) have a head mortar joint at the second and fourth brick layer,
while the other two specimens (S-C type) do not. Both horizontal and head joints were
10712mm thick. A spherical articulation was used to ensure a uniform stress distribution
between specimen and loading plates. The global displacements were acquired by means
of four linear potentiometers (stroke 50mm, resistance 4.0 k, and independent linearity
0.15%), while local brick vertical and horizontal strains were acquired by means of two
resistive strain gauges (l0¼30mm), that were connected to the bricks by means of an acrylic
glue, applied on an epoxy-based resin film that ensures the regularity of the contact
surface. Table 4 lists the main properties resulting from the tests.

Figure 7. Schematic illustration of masonry specimen types and experimental set-up for centered and
eccentric compression tests.

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


122 G. DE FELICE AND S. DE SANTIS

Table 3. Experimental plan

Test Specimen type Eccentricity e0 (mm) Number of Specimens

Centred cyclic compression S-C 0 2


S-CG 0 2
Eccentric cyclic compression S-C 30 2
S-CG 30 2
S-E5 40 2
60 2
80 1
100 2
S-E9 70 2
110 2

Table 4. Experimental results of masonry prisms under cyclic axial load

Initial Average Softening


Compressive Deformation at stiffness unloading-reloading branch
Specimen strength fM [MPa] peak stress e(fM) EM [MPa] stiffness [MPa] slope [MPa]

S-C1 7.43 18.110-3 662.1 3191.3 -335.5


S-C3 7.96 17.110-3 668.3 2978.5 -351.6
S-CG1 4.37 11.210-3 726.5 2497.0 -127.1
S-CG3 4.12 13.110-3 585.0 2236.2 -162.8

As regards the cyclic behavior, it appears that the stiffness in the re-loading
branches is higher than in the first loading branch. This phenomenon depends on the
deformation of horizontal mortar joints where some compaction arises and inelastic
strains occur, depending on the maximum stress sustained during load history. As a
consequence, no variation in stiffness is detected in subsequent loading–unloading
cycles, while, once the load goes beyond the unloading value, the behavior reverts to
the skeleton curve of the monotonic test.
Vertical cracks (Figure 8) appear at approximately 80% of maximum load in a
brick tier and then rapidly propagate into the other tiers of the specimen. After peak
load, the cracks increase in size and at the very end of the test, the external slices are
ejected (Figure 9). The graphs in Figure 10 show the stress-strain curves of the speci-
mens without (S-C type) and with (S-CG type) head mortar joints.
A significant reduction in strength is found in the specimens with head mortar
joint, from approximately 7.6 MPa to approximately 4.2 MPa as shown in Table 4.
The crack pattern shows that, in S-CG specimens, the head joint behaves as a
preferential weakness plane that leads to a lower failure load (Figure 9). This reduc-
tion in strength is accompanied by a slight reduction of the softening branch slope,
while almost no variation in the stiffness of the elastic branch is recorded.

3.1. Head Joint Effects


As it is known (Hilsdorf 1969), the failure of brick masonry under compression is
driven by tensile horizontal stress, which develops in the bricks, as a consequence of
the mismatch of transverse deformation between the brick and the mortar layers. In
the present case, this phenomenon is amplified due to the low strength and stiffness of

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 123

Figure 8. Crack pattern and corresponding stress-strain graph of a centered compression test on S-CG
masonry specimen (figure is provided in color online).

Figure 9. Failure pattern of masonry under centered compression: a) S-C specimens; b) S-CG specimens
(figure is provided in color online).

the mortar when compared with the brick; as a result of confinement of the mortar
within the bed joints, the strength of brickwork is much higher than that of mortar.
When head joints are present, while the average tensile stress in the brick remains
almost unchanged, a concentration of stress occurs and the maximum tensile stress
increases significantly along the vertical plane of the head joints. Now, since the failure of
the brick is brittle in traction, the strength is controlled by the maximum, rather than the
average, stress; which explains the great reduction in strength observed in experiments.
Conversely, the overall stiffness is mainly controlled by the average stress and therefore
no significant variation occurs between the two specimen types, as observed in the tests.
A finite element analysis with a three-dimensional elastic model, reproducing the
two brickwork patterns, provides a first explanation of the behavior observed: the stress
concentration resulting from the presence of the head joint is such that the same value of

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


124 G. DE FELICE AND S. DE SANTIS

a b

Figure 10. Experimental and numerical stress-strain curves of cyclic axial compression tests: a) S-C speci-
mens; b) S-CG specimens (figure is provided in color online).

maximum tensile stress is reached in the S-CG specimen for a vertical load approximately
40% lower than in the S-C specimen. The horizontal stress field in the two specimens,
when the maximum experimental load is applied (142 KN for S-C specimens and 76 KN
for S-CG specimens), is shown in Figure 11: both analyses give rise to almost the same
value of the maximum horizontal tensile stress in the bricks. The analyses are performed
considering the average experimental elastic modulus Eb¼3900MPa for the bricks and the
equivalent modulus Em¼60MPa for the mortar, much lower than the experimental value,
to account for the inelastic deformation which occurs in the test; in particular, Em is
defined by requiring the vertical displacement of the model to be the same as the one

Figure 11. FE analysis results: contour plot of horizontal tensile stress at failure. a) S-C specimens; b) S-CG
specimens.

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 125

measured in the test under maximum load. As regards the Poisson ratios, the values
 b¼0.15 and  m¼0.35, are assumed, for brick and mortar, respectively, where the latter is
much higher than the elastic Poisson ratio, so as to take into account that, when plastic
strains develop, the mortar behaves almost as an uncompressible medium.
For a full explanation of the experimental behavior, a refined non-linear model
would be required, that takes into account the inelastic deformation occurring in the
mortar bed joints and the crack forming in the head joint before collapse, since both
these effects strongly increase the tensile stress in the bricks. The recourse to such a
refined model would be suitable for small specimens, but not for structures such as
bridges. In the latter case, a simpler model should be defined in which masonry is
considered as an equivalent homogeneous material (Lourenco 1998), and the effect of
the head joint is included in the overall macroscopic properties of such material.

3.2. Uniaxial Behavior Under Compression


For the purpose of setting up a macroscopic model, the Kent and Park (Kent and
Park 1971) relationship is adopted for representing the uniaxial behavior of masonry
under compression. Originally formulated for confined concrete, the Kent and Park
model consists of an ascending branch represented by a second-degree parabolic curve,
a linear descending softening branch and a final constant branch. The model is defined
by four parameters, namely the peak stress s0, the corresponding strain e0, the ultimate
strain eu, that is reached at the ultimate stress level su. Denoting by s~ ¼ s=s0 , and
~e ¼ e=e0 the normalized stress and strain respectively, the three branches thus become:

s s ¼ 1 þ ðs~u 1
~ ¼ 2~e  ~e2 ; for 0  ~e  1;~ Þð~e1Þ
1 ; for 1  ~e  ; s
~¼s
~u ; for ~e   (1)

where  ¼ eu =e0 and s ~u ¼ su =s0 .


The appropriate parameters defining the Kent and Park model are estimated on
the basis of the experimental results of the centered compression tests: the peak
coordinates are deduced for the average experimental compressive strength:
s0 ¼ f exp exp
M and e0 ¼ eðf M Þ; the final constant branch has su ¼ 0, while the ultimate
strain eu is set to reproduce the experimental softening branch slope (Table 4).
The model, as previously defined, agrees reasonably well with the experimental
curves, as shown in Figure 10. Contrary to contemporary brickwork, where the behavior
is almost linear up to maximum stress, and the softening branch is much shorter and with a
steeper slope (Brencich and Gambarotta 2005; Brencich and de Felice 2009), the historic
brickwork considered here displays the capability of sustaining load well beyond the strain
e0, with a smooth transition between the ascending elastic and the descending softening
branches, which is probably due to the very weak mortar, and to the non-negligible ductility
of the bricks; both these aspects are encompassed by the Kent and Park model.

4. BRICKWORK UNDER AXIAL LOAD AND BENDING MOMENT


Three different specimen types were tested under cyclic eccentric axial load
(Figure 7 and Table 3). The first type (made in 4 samples denoted as S-C and S-CG)
is the same as for centered compression. It measures 140140310 mm3 and is made
of five tiers, each consisting of half-brick units. The second type, denoted as S-E5
(7 samples), measures 280140310 mm3 and is made of five tiers consisting of a brick

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


126 G. DE FELICE AND S. DE SANTIS

unit alternating with two half-brick units; a head mortar joint is present in the second
and in the fourth layers. Finally, the third type, denoted as S-E9 (4 samples), measures
420140550 mm3 and is made of nine tiers of bricks as shown in Figure 7. All the
specimens aim at reproducing the typical arch brick arrangement (Figure 1).
The eccentric axial load was applied by means of two steel bars with diameter
ؼ20mm and two steel HEA140 I-beams, stiffened with vertical flanges, in contact
with the bases of the masonry prism in order to apply compression with the desired
eccentricity, avoiding local stress concentrations. Displacement and strain data were
measured using linear potentiometers (electrical stroke 5 mm, 12.5 mm, 25 mm,
50 mm; resistance 0.44 k, independent linearity 0.25%) and resistive strain
gauges (l0¼30mm), as shown in Figure 12.
As shown in Figure 13, the damage pattern develops as follows: the first crack
appears in the compressed face at approximately 70% of peak load; then, once
maximum load is reached, the cracks spread vertically crossing the mortar joints,
until they affect the whole specimen height, while small cracks also form in the lateral

a b

Figure 12. Testing set-up for eccentric compression tests. a) S-C and S-CG specimens; b) S-E5 specimens;
c) S-E9 specimens.

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 127

Figure 13. Crack pattern of an eccentric compression tests on S-E9 masonry specimen and corresponding
load-displacement curve.

faces. At the same time, one or more horizontal mortar joints start to open in the
tensile side. With increasing deformation, a localization of cracks develops, with
crushing of one or two bricks on the compression side, and opening of the correspond-
ing horizontal bed joint on the tensile side (Figures 13–14). As in the case of centered
compression, the cyclic tests show the capability of brickwork to sustain loading-
unloading cycles, also in the softening branch.
The tests are displacement-controlled; the relative displacement and rotation are
derived from the transducers at plate corners; the load F is directly provided by the
MTS loading frame, while the moment M is computed taking into account the second
order effects, which derive from the rotation of the steel plates (Figure 15) according to
the following expression:
M ¼ F  e ¼ F  ðe0 þ e00 Þ (2)
In Equation (2) e0 is the initial eccentricity, which is kept constant during the test,
and e00 ¼  hp is the second-order eccentricity, where  ¼ ð1 þ 2 Þ=2 is the average
plate rotation and hp the plate height. The value of  is derived directly from the
potentiometers at plate edges as  ¼ ðL1  L2 Þ=2B, where Li is the relative
displacement of the i-transducer and B is the specimen width (i.e., the distance between
the transducers) as shown in Figure 15. In the strict sense, the bending moment given
by Equation (2) is acting at the top (or at the bottom) of the specimen, since at the
middle height, a further increase in eccentricity is expected, which derives from the
specimen deformation; however, this latter is approximately one order of magnitude
lower than e00 , and therefore can be neglected.
The results of the tests are shown in Figures 16–18 for the various specimens and
eccentricities. For each test the variation of effective eccentricity (increasing from the
initial value e0 ) is plotted, together with the load-displacement F- and moment-
curvature M-w diagrams, where  ¼ (L1 þ L2)/2 þ   e, and w ¼ /L.

4.1. Maximum Load Estimate


The experimental tests show that, for high eccentricities (load-resultant out
of the third medium), the maximum load no longer depends on the specimen width
B, but only on the distance between the load resultant and the compressive edge of

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


128 G. DE FELICE AND S. DE SANTIS

a b c

d e f

Figure 14. Failure pattern on S-E5 (a-c) and S-E9 (d-f) masonry specimens under eccentric axial load (figure
is provided in color online).

the specimen u ¼ B=2  e. In fact, by comparing the results of two specimens


(SC-G and S-E5) with different width B, by parity of the initial distance from
the compressive edge u0 ¼ B=2  e0 (u0 ¼ 40mm), a similar value of the maximum
load F0 is obtained (Table 5). A similar result is also observed for S-E5 and S-E9
specimens for u0 ¼100mm. In figure 19a the experimental results for all specimens
that include head joints are plotted in the F–u plane, showing an almost linear
dependence of the maximum load to the distance from the compressive edge; S-C
type specimens are not compared because of the absence of head joints, which lead
to a different average compressive strength. According to figure 19a, for high
eccentricities, a rough estimate of the load capacity can be obtained simply by
F0 2f exp
M ud, considering a uniform stress distribution (stress block) equal to the
masonry compressive strength f exp
M over a section 2ud (where d is the specimen
depth), as happens for perfectly plastic materials. The depth of the stress-block
y*¼2u is plotted in figure 19b against the neutral axis depth yc (i.e., the distance

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 129

Figure 15. Schematic illustration of brickwork specimen and fibre beam model in the initial and deformed
configurations.

between neutral axis and compressive edge). The experimental results provide an
average value of 0.73 for the ratio y*/yc, which is slightly lower than the value 0.80
currently used for reinforced concrete sections, due to the different inelastic strain
capacity of the materials.

Figure 16. Experimental results of eccentric (e0 ¼30mm) axial load tests on S-C (up) and S-CG (down)
masonry specimens (figure is provided in color online).

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


130 G. DE FELICE AND S. DE SANTIS

Figure 17. Experimental results of eccentric axial load tests on S-E5 masonry specimens for different
eccentricities (e0 ¼40mm, 60mm, 80mm, 100mm) (figure is provided in color online).

4.2. Plane Section Assumption


The main purpose of this section is to establish whether the plane section
assumption is adequate for representing the behavior of masonry under eccentric
compression. Clearly, the experimental set-up of the tests, equipped with the steel
plates, by which the load is applied, ensure that the plane section is enforced at the

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 131

Figure 18. Experimental results of eccentric axial load tests on S-E9 masonry specimens for eccentricities
e0 ¼70 mm and 110 mm (figure is provided in color online).

Table 5. Experimental results of masonry specimens under eccentric axial load

Maximum Corresponding Maximum Corresponding


Specimen load displacement bending moment curvature
type Name e0 (mm) u0 (mm) F0 (KN) (F0) (mm) M0 (KNm) w(M0) (m-1)

S-C S-C2 30 40 41.2 2.41 1.51 0.18


S-C4 30 40 60.4 3.67 2.26 0.18
S-CG S-CG2 30 40 36.2 3.25 1.72 0.43
S-CG4 30 40 37.7 4.07 1.62 0.33
S-E5 S-E5-1 40 100 121.1 8.61 5.48 0.12
S-E5-2 40 100 141.1 5.22 6.21 0.09
S-E5-3 60 80 104.6 6.41 6.92 0.14
S-E5-7 60 80 107.3 5.04 7.04 0.12
S-E5-4 80 60 76.7 5.89 7.07 0.30
S-E5-5 100 40 34.5 4.62 3.85 0.27
S-E5-6 100 40 32.6 3.80 3.79 0.38
S-E9 S-E9-1 70 140 186.2 6.22 13.13 0.025
S-E9-4 70 140 201.2 8.07 14.36 0.035
S-E9-2 110 100 137.1 7.89 15.37 0.054
S-E9-3 110 100 139.9 8.26 15.67 0.052

boundary of the specimen. To check the strain distribution at a local level within the
specimen, three sets of potentiometers have been fixed to the bricks by means of
threaded steel bars with diameter ؼ5mm in the front and back sides of the specimen
(Figure 12). The deformation of the cross-section during the test is therefore estimated

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


132 G. DE FELICE AND S. DE SANTIS

Figure 19. a) maximum load (F0) vs. distance to the compressive edge (u); b) stress block depth (y*) vs.
neutral axis depth (yc).

(a) (b)

Figure 20. Section deformation of brickwork under eccentric loading at different load steps: a) e0 ¼70mm;
b) e0 ¼110mm.

as the relative displacement between the couples of points equipped with instruments,
divided by the distance between them. The results are plotted in Figure 20 for two
different eccentricities (e0 ¼ 70 mm and e0 ¼ 110 mm) and compared with the average
strain of the specimen, as detected by the global potentiometers at its edges. It is shown
that the cross-section remains almost linear, and rotates around the neutral axis, as the
deformation increases up to the value 0 ¼ ðF0 Þ, corresponding to maximum load.
Then, a slight deviation from the plane behavior is displayed and the neutral axis
moves towards the compression side. Clearly, at this stage, the mortar joints have
started to open and several cracks to develop within the specimen, leading to strain
concentration and deviation from a linear distribution. The comparison between
average and local strain shows that the former is higher, up to attainment of the
maximum load, probably because it is influenced by the crushing occurring within the

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 133

mortar joints in contact with the steel plates, and by detachment of the plates from the
specimen at the tensile edge. At higher deformation, the local strain increases until it
exceeds the global strain. Finally, the wide cracks developing at the end of the test,
disturb the instrumental readings, which become unreliable.
In conclusion, according to the experimental results, it seems that, from an
engineering point of view, brick masonry under eccentric compression behaves
according to the plane section assumption, also in the non-linear range, at least until
a state of heavy damage is reached.

4.3. Failure Criterion


For the purpose of defining the strength domain of brickwork, the normalized
plane M/M0 - F/F0 is used, where F0¼f expM B d, and M0¼ F0 B/4, where B and d are the
specimen width and depth, respectively (Figure 21) and f exp
M is the average experimental
compressive strength. The normalization permits a comparison of the results of the
different specimen types in the same graph, as shown in Figure 21, where the experi-
mental data are plotted together with those presented in (Brencich and de Felice 2009).
The numerical estimates are obtained according to the plane section assumption, by
integrating, over the cross-section of the specimen, the Kent and Park relationship
previously defined. The three curves shown in Figure 21 are obtained for compressive
strength equal to 0:9  1:1f exp
M , thus taking into account the intrinsic uncertainty in its
estimate (Brencich et al. 2007). The Figure 21 shows that the axial force-bending

Figure 21. Strength domains of brickwork under eccentric loading according to the Kent&Park model and
corresponding experimental data.

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


134 G. DE FELICE AND S. DE SANTIS

moment interaction can be predicted with a good approximation. Anyway, some


experimental points lie out of the region delimited by the numerical prediction; this
could be ascribed to some discrepancies between the effective and the expected
eccentricities.

4.4. Overall Behavior


As the cross-section remains substantially plane, at least until the curvature is
not too high, a fiber beam model (Spacone et al. 1996) may be used to simulate the
eccentric tests, where the uniaxial fiber behavior under compression follows the
previously defined Kent and Park stress-strain relationship, while no resistance is
assumed in traction. The model is drawn in Figure 15, where the central beam
represents the specimen and includes the rigid plates by means of which the load is
applied, so as to take due account of the second order effects developing during the
test. Moreover, in order to reproduce the experimental conditions, the model includes
two elasto-plastic non-tensile resisting elements, between the load plates and the
specimen, simulating the detachment of the plate at the tensile edge and the crushing
of the mortar at the compressed edge. Their calibration is carried out according to the
displacement data acquired by the 10mm transducers, set up for the purpose, on S-E9
tests (Figure 12). The analyses were carried out through the software OpenSEES in
displacement control.
The model, as it is defined, is shown to be capable of estimating the second order
eccentricity, which takes place due to the rotation of the steel I-beams as shown in
Figures 16–18. The accuracy in estimating the whole force-displacement and bending
moment-curvature behavior of brickwork under eccentric loading is shown in Table 6
and in Figures 16–18, where the numerical estimate is plotted against the experimental
results. Second order effects, represented by the plots of the eccentricity e00 are
considered in the model by means of a co-rotational geometrical transformation: at
each step the stiffness matrix of any element in its local reference system is updated
taking into account the end nodes position and rotation in the current deformed
configuration. The comparison reveals a good agreement between the model and the
experiments for specimen types S-C, S-CG, S-E5, both in the ascending and the
softening branches, with a slight overestimate of the latter. A higher discrepancy

Table 6. Comparison between experimental and numerical results

Experimental maximum Numerical Numerical/


bending moment M0exp (KNm) maximum Experimental
bending moment ratio M0num /
Specimen e0 (mm) u0 (mm) Specimen 1 Specimen 2 M0num (KNm) M0exp (average)

S-C 30 40 1.51 2.26 2.13 þ13%


S-CG 30 40 1.72 1.62 1.71 þ5%
S-E5 40 100 5.48 6.21 5.53 -5%
60 80 6.92 7.04 6.79 -3%
80 60 7.07 – 6.82 -4%
100 40 3.85 3.79 3.54 -8%
S-E9 70 140 13.13 14.36 14.12 þ3%
110 100 15.37 15.67 14.31 -8%

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 135

appears for the S-E9 specimen type, with eccentricity e0 ¼ 110mm (Figure 18), where
the numerical predictions overestimate the force and bending moment capacity and
are not able to follow the rapidly descending softening branch. Probably, one of the
reasons for this mismatch lies in the lack of fit between the numerical prediction of the
second order eccentricity e00 and the corresponding experimental value; this discre-
pancy is due to the difficulty in modeling the mortar layers in contact with the loading
plates, where severe crushing damage takes place during the tests. As shown in
figure 18, an even slight under-estimation of e00 , due to the very steep bending
moment-curvature softening branch, induces a strong over-estimation of the corre-
sponding force and moment predictions. It should be considered also that, in the
softening phase, the plane section assumption is no longer ensured (Figure 20) and
therefore a beam model is less reliable.

5. CONCLUSIONS
The results of the experimental investigation presented in this paper provide a
broad view of the mechanical behavior of masonry made with historic bricks and lime
mortar, when subjected to centered and eccentric compression. With respect to con-
temporary brickwork, a significant non-linear pre-peak branch and a subsequent long
softening branch are displayed, together with the capability of sustaining loading-
unloading cycles, without appreciable stiffness degradation. The tests under uniform
compression show the great influence of the head mortar joints, which act as planes of
weakness that greatly reduce the overall strength.
When combined axial load and bending moment is applied, the experimental
results show that, from an engineering viewpoint, the plane section assumption can be
considered to be valid at least until a state of heavy damage is reached. Therefore, the
mechanical response of masonry is evaluated by making use of a fiber beam model, for
which the Kent and Park stress-strain relationship is used to represent the non-linear
behavior of masonry under compression. The numerical predictions are finally com-
pared to experimental outcomes, showing to what extent the model is suitable for
representing the behavior of brickwork under eccentric loading.
On the basis of the results obtained by the eccentric compression tests, it
clearly appears that failure condition occurring with a hinge mechanism develops
at the same time as the material crushing in compression takes place. The pro-
posed modeling approach, which makes use of fiber beam elements, proves ade-
quate in following the non-linear response and the interaction between both
failure modes, thanks to a refined representation of the material behavior also in
the softening phase, unlike the classical elasto-plastic or elasto-brittle models.
Thus this approach would seem to be a promising tool for the structural analysis
of masonry pillars and arches, that may be used in the assessment of masonry arch
bridges (de Felice 2009).

ACKNOWLEDGEMENTS
This research was carried out with the financial support of the Reluis Project
‘‘Evaluation and reduction of seismic vulnerability of existing masonry buildings’’.

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


136 G. DE FELICE AND S. DE SANTIS

REFERENCES
Al Shebani, M., and S. Sinha. 1999. Stress-Strain characteristics of brick masonry under
uniaxial cyclic loading. Journal of Structural Engineering ASCE 125(6):600–604.
Anziani, A., L. Binda, and G. Mirabella Roberti. 2000. The effect of heavy persistent actions
into the behaviour of ancient masonry. Materials and Structures 33:251–261.
Aprile, A., A. Benedetti, and F. Grassucci. 2001. Assessment of cracking and collapse for old
brick masonry columns. Journal of Structural Engineering ASCE 127(12):1427–1435.
Barbi, L., S. Briccoli Bati, and G. Ranocchiai. 2002. Mechanical properties of ancient bricks:
Statistical analysis of data. In Proceedings of the 7th International Seminar on Structural
Masonry for Developing Countries, ed. F. Santos et al. Belo Horizonte, Brazil.
Binda, L., A. Fontana, and G. Frigerio. 1988. Mechanical behaviour of brick masonries derived
from unit and mortar characteristics. In Proceedings of the 8th International Brick/Block
Masonry Conference. Dublin, Ireland: Elsevier Applied Science Publisher LTD.
Braga, F., D. Liberatore, and E. Mancinelli. 1996. Numerical simulation of the experimental
behaviour of masonry panels and walls under horizontal loads. In Proceedings of the 11th
World Conference on Earthquake Engineering. Oxford, UK: Pergamon.
Brencich, A., and L. Gambarotta. 2005. Mechanical response of solid clay brickwork under
eccentric loading. Part I: Unreinforced masonry. Materials and Structures 38:257–266.
Brencich, A., and G. de Felice. 2009. Brickwork under eccentric compression: Experimental
results and macroscopic models. Construction and Building Materials 23:1935–1946.
Brencich, A., and U. De Francesco. 2004. Assessment of multi-span masonry arch bridges. Part
I: A simplified approach. Part II: Examples and applications, Journal of Bridge Engineering
ASCE 9:582–598.
Brencich, A., L. Gambarotta, and E. Sterpi. 2007. Stochastic distribution of compressive strength:
Effects on the load carrying capacity of masonry arches. In Proceedings of the ARCH’07, 5th
International Conference on Arch Bridges. Madeira, Portugal: Multicomp, Lda Publishers.
Brencich, A., and R. Morbiducci. 2007. Masonry arches: Historical rules and modern
mechanics. International Journal of Architectural Heritage 1:1–24.
Cavalieri, L., A. Failla, L. La Mendola, and M. Papia, M. 2005. Experimental and analytical
response of masonry elements under eccentric vertical loads. Engineering Structures
27:1175–1184.
Curioni, G. 1874. L’arte del fabbricare, ossia corso complete di istituzioni teorico-pratiche.
Torino, Italy: Negro.
de Felice, G., and I. Carbone. 2006. Experimental investigation on historic brickwork subjected
to eccentric axial loads. In Proceedings of the SAHC’06, 5th International Conference on
Structural Analysis of Historical Constructions. New Delhi, India: Macmillan India Ltd.
de Felice, G. 2009. Assessment of the load-carrying capacity of multi-span masonry arch bridges
using fibre beam elements. Engineering Structures 31(8):1634–1647.
Ding, D. 1997. Studies on brick masonry under compression. Materials and Structures
30:247–252.
Drysdale, R. G., and A. A. Hamid. 1982. Effect of eccentricity on the compressive strength of
brickwork. Journal of the British Ceramic Society 30:102–108.
Gay, C. 1924. Ponts en Maconnerie. In Encyclopedie de Ge´nie Civil et des Travaux publics. Paris,
France: Baillière et fils.
Hilsdorf, H. K. 1969. Investigation into the failure mechanism of brick masonry loaded in axial
compression. In Designing engineering and construction with masonry products, ed., F. B.
Johnson, 34–41. Houston, TX: Texas Gulf Publishing Co.
Kent, D. C., and R. Park. 1971. Flexural members with confined concrete. Journal of the
Structural Division ASCE 97(7):1969–1990.
Lourenço, P. B. 1998. Experimental and numerical issues in the modelling of the mechanical
behavior of masonry. In Structural Analysis of Historical Constructions II CIMNE, eds., P.
Roca, J. L. González, E. Oñate, P. B. Lourenço, 57–91.

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137


HISTORIC MASONRY UNDER ECCENTRIC LOADING 137

Mojsilović, N., and P. Marti. 1997. Strength of masonry subjected to combined loads. ACI
Structural Journal 94(6):633–642.
Oliveira, D. V., P. B. Lourenco, and P. Roca. 2006. Cyclic behavior of stone and brick masonry
under uniaxial compressive loading. Materials and Structures 39(2):219–227.
Roberts, T. M., T. G. Hughes, V. R. Dandamundi, and B. Bell. 2006. Quasi-static and high cycle
fatigue strength of brick masonry. Construction and Building Materials 20:603–614.
Rondelet, G. 1802. Traite´ theorique et pratique de l’Art de Bâtir. Paris, France: Didot Frères, Fils
et Cie.
Sala, E., I. Giustina, and G. A. Plizzari. 2008. Lime mortar with natural pozzolana: Historical
issues and mechanical behavior. In Proceedings of SAHC’06. 6th International Conference on
Analysis of Historical Constructions. Bath, UK: Macmillan India Ltd.
Séjourné, P. 1913. Grandes voutes. Bourges, France: Imprimerie Veuve Tardy.
Spacone, E., F. C. Filippou, and F. F. Taucer. 1996. Fiber beam-column model for non-linear
analysis of R/C frames. Part 1: Formulation. Earthquake Engineering and Structural
Dynamics 25:711–725.
Valluzzi, M. R., L. Binda, and C. Modena. 2005. Mechanical behavior of historic masonry
structures strengthened by bed joints structural repointing. Construction and Building
Materials 19:63–73.
Venu Madhava Rao, K., B. V. Venkatarama Reddy, and K. S. Jagadish. 1997. Strength
characteristics of stone masonry. Materials and Structures, 30:233–237.

INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 4(2): 115–137

You might also like