This paper proposes a partial-destructive method for connecting adjacent orthogonal

masonry walls by means of aramid fibre reinforced polymer (AFRP) rods, in order to
improve the overall performance of masonry buildings under horizontal forces. The proposed
method is supported by an experimental campaign to assess the effectiveness of the
strengthening measure, and by an analytical study to develop equations suitable for design.
The experimental tests showed that the connection between adjacent masonry walls is
actually effective in increasing both their strength and stiffness. It is also shown that the
developed analytical equations satisfactorily predict the relevant design quantities.
Keywords: connection of adjacent walls; strengthening of URM; improvement of flexural behaviour;
out of plane overturning collapse; FRP strengthening.
1. INTRODUCTION
It is common to find in our cities old buildings with load bearing masonry walls that were designed
only to resist vertical loads and with no consideration of horizontal loads. This is usually reflected
in the arrangement of walls perpendicular to each other, which are not clamped along the edges.
This lack of connection may be found either in backbone (or main) walls that intersect with each
other, or in other walls considered as secondary, but made of stone that are given a structural
function, or even in walls that were independent in the original construction, because designed as
vertical load bearers or because partly independent and without a load-bearing function.
When seismically retrofitting these buildings, it may be convenient to take advantage of all existing
walls to provide resistance to horizontal forces by connecting them to each other at their intersections.
By doing so, two originally unconnected walls are made into a single one with a T-shaped cross-
section. The result is that the strength and the stiffness of each single wall are increased. This allows
avoiding insertion of additional walls or thickening of the existing ones, with all the obvious
difficulties related to these strategies. In the past, such strengthening measures were carried out by
STRENGTHENING OF MASONRY WALLS BY TRANSVERSE CONNECTION
THROUGH AFRP RODS: EXPERIMENTAL TESTS AND ANALYTICAL MODELS
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 61
Marco Vailati
1
, Giorgio Monti
2
Manuscript received on 17
th
June 2012, reviewed and accepted on 21
st

August 2012 as per publication policies
of NED University Journal of Research.
ABSTRACT
1
Post-Doctoral Researcher, Department of Structural and Geotechnical Engineering, Sapienza University of Rome, Italy. Ph. (+39) 06 49919254,
Fax. (+39) 06 3221449, Email: marco.vailati@uniroma1.it.
2
Full Professor, Department of Structural and Geotechnical Engineering, Sapienza University of Rome, Italy. Ph. (+39) 06 49919197,
Fax. (+39) 06 3221449, Email: giorgio.monti@uniroma1.it.
M. Vailati and G. Monti
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 62
Marco Vailati is a Post Doctoral Researcher in the Department of Structural Engineering and
Geotechnics at Sapienza University of Rome, Italy where he received his Masters in Innovation
in Design, Rehabilitation and Control of Structures: Assessment and Retrofitting in Seismic
Areas and PhD in Structural Engineering, respectively, in 2004 and 2011. His research Interests
include earthquake risk assessment of building structures, and strengthening of masonry walls
and reinforced concrete structures with advanced techniques.
drilling holes and inserting metal bars in them, which were then partially filled with mortar injections.
However, metal bars are often subjected to corrosion if not properly injected.
Therefore, a technique to connect two originally unconnected walls into a T-shaped layout has been
studied and tested, which makes use of thin rods of aramid fibre reinforced polymer (AFRP) inserted
in small holes drilled in the flange wall. These rods are then anchored to the lateral surfaces of the
web wall through spread fibres, which are then glued onto the web wall surface, thus providing
efficient anchorage. Since the connection has to restrain the vertical sliding between the connected
walls, the rods are placed at 45, so as to provide tension components in both sliding verses.
In this way, the strengthening technique maximizes the capacity of the existing structural walls with
minimum invasivity.
2. TESTING SET-UP
The tests described below have been carried out at the Department of Structural and Geotechnical
Engineering of the Faculty of Architecture of the Sapienza University of Rome, Italy.
2.1 Geometry and Boundary Conditions
The configuration of the testing rig with the walls is shown in Figure 1.
The boundary conditions are: F
V
= o
0
.
t
.
l; u
v1
= u
v2
= 0; u
h
= imposed; F
h
= measured.
where o
0
is the average stress in the web wall cross-section; t and l are the web thickness and length,
respectively; u
v1
and u
v2
are the vertical imposed displacements at the two flanges; u
h
is the horizontal
imposed displacement; F
h
is the horizontal measured force. Therefore, it can be understood that all
tests were performed under displacement control, in order to follow any possible degrading branch
in the wall response.
Figure 2 shows a typical wall configuration, along with a detail of the connection with the anchorage.
The measuring equipment is constituted by five strain gauges, numbered from 1 to 5; the first four
measure the vertical relative displacements between web and flange, while the last measures the
horizontal absolute displacement.
Giorgio Monti is a Full Professor at Sapienza University of Rome, Italy. His research interests
include modelling, analysis and assessment of reinforced concrete and masonry structures
under seismic excitation, structural health monitoring, strengthening techniques with innovative
materials (FRP), strategies for the preservation of historical towns, and reliability analysis of
structures and infrastructures in seismic zones. He is an active member of national and
international committees for the development of seismic design codes.
Figure 1. Configuration of testing rig and walls.
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 63
2.2 Materials
Clay bricks with premixed mortar were used to build the walls, while AFRP rods were used to
connect them. The walls mean mechanical properties were: compressive strength (f
d
) = 6.00 MPa
(0.87 ksi); shear strength (t
d
) = 1.38 MPa (0.2 ksi); Youngs modulus (E) = 2,700 MPa (392 ksi);
shear modulus (G) = 900 MPa (131 ksi).
The AFRP rod properties were: tensile strength (f
yd,c
) = 1400 MPa (203 ksi); Youngs modulus (E
d,c
)
= 60,000 MPa (8.7x10
6
ksi)
The total tensile strength of the connection system was: (a) configuration 1 R
T,1
= 32 kN (7 kips);
(b) configuration 2 R
T,2
= 16 kN (3.6 kips).
In configuration 1, the spread of the terminal anchors lays along the bar axis, while in configuration
2 it is arranged at 90 (Figure 3).
2.3 Connections between web and flange walls
The connection between web and flange walls is constructed in few simple steps (Figure 4). The
details of these phases are as under
Phase a: drilling holes from the outer face of the flanges, tangent to the web wall faces
Phase b: inserting the rods through the flanges and fixing them in the hole with mortar
Phases c and d: gluing the spread ends to the surface of both walls
Phases e and f: applying aramid sheets with vinylester resin to cover and strengthen the anchorages
The AFRP rods were of 5.5 mm (0.22 in.) diameter, while the holes crossing through the flange
were of 7 mm (0.28 in.) diameter.
3. TEST RESULTS
The walls specimens consist of one web wall and two flange walls at its ends. They are instrumented
as shown in Figure 2.
Figure 2. Displacement transducers on a masonry wall specimen. At top right the anchorage
detail of the rods.
Note: All dimensions are in mm; 25.4 mm = 1 in.
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 64
M. Vailati and G. Monti
Figure 4. Construction phases of the connections between flanges and web walls.
Figure 3. Detail of AFRP rod anchorage: the two reference configurations (above), the test
configuration (below).
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 65
With reference to Figure 2, transducers 1, 2 and 3, 4 are used to measure the relative displacement
between the web wall and flanges, while transducer 5 measures the horizontal displacement between
foundation and wall head.
The global effects of this strengthening method are illustrated in Figure 5-7. Figure 5 shows the
comparison at the first cycle between unstrengthened and strengthened wall. In the strengthened
wall, the improved collaboration between the orthogonal walls gives rise to an increase in the initial
stiffness. Note that, in the strengthened wall, the stiffness is increased approximately by 40%.
Figure 5. Comparison between strengthened and unstrengthened wall at the first cycle: increase
of stiffness.
Figure 6. Horizontal force vs. relative displacement between web and flanges walls.
Figure 7. Comparison between strengthened and unstrengthened wall at ultimate: increase
of flexural capacity. At top left: detail of crushing of masonry.
Note: 25.4 mm = 1 in.; 4.448 kN = 1 kips
Note: 25.4 mm = 1 in.; 4.448 kN = 1 kips
Note: 25.4 mm = 1 in.; 4.448 kN = 1 kips
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 66
M. Vailati and G. Monti
For low horizontal forces, the two curves are practically coincident; however, when passing 30%
of the unstrengthened wall failure load, the connections modify the wall system response. This delay
could eventually be avoided, if need, by pretensioning the rods. In fact, a truss-like mechanism can
be activated only when the rods are sufficiently stressed in tension.
In Figure 6, the difference in terms of relative displacement between the web and flange walls can
be observed: for the unstrengthened configuration it is 0.31 mm (0.01 in.), while in the other one
it is 0.02 mm (7.8x10
-4
in.). The smaller displacement in the strengthened case is due to the presence
of the rods.
Initially, the contacting surfaces slip relatively one to each other, however, when the rods start being
pulled, they provide their contribution by imposing compatibility of the displacements of both
surfaces. After this stage, the wall system shows a significant increase of strength and ductility, as
shown in Figure 7.
In Figure 7, the curves were obtained by inverting the load path at yield displacement (curve 1),
at 50% of ultimate displacement (curve 2), at ultimate displacement (curve 3). Note that each reversal
point is marked by a circle.
As observed from the tests performed, the strengthening effectiveness strongly depends on the
correct application of the rods. In order to exploit the material mechanical properties, the rods are
inserted at 45. As a matter of fact, for a plane stress state in shear, the maximum tensile force
is inclined at 45 with respect to the vertical, as shown in Figure 8.
Two effects can be observed when reaching the ultimate limit state
1) Detachment from the wall of the aramid sheet that covers the rods anchor;
2) Loss of the flange wall verticality
In Figure 9b, a tension failure of a rod is shown, with clear signs of breakage in its cross-section.
Figure 8. Shear transfer mechanism between web wall and flange.
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 67
Figure 9. Collapse by flexural mechanism of wall: (a) collapse of masonry with expulsion of
material to external direction; (b) traction failure of a rod.
4. COMPARISON WITH ANALYTICAL MODEL
4.1 Unconnected Walls
The stiffness obtained from the experimental results is compared to the analytical model proposed
by Tomazevic [3]
(1)
Figure 10 shows the comparison between experimental and predicted stiffness.
It can be seen in Figure 10 that Eq. (1) accurately predicts the elastic experimental stiffness, while
a 50% reduction gives a good estimate of the ultimate displacement.
Figure 10. Comparison between experimental and predicted stiffness and peak strength at
first cycle.
G
.
A
W
K
I
G h
l E
2
2h 1 1
+
o
=
.
. . . .
q r o p
Note: 25.4 mm = 1 in.; 4.448 kN = 1 kips
(a) (b)
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 68
M. Vailati and G. Monti
The wall capacity is assessed by considering two possible failure mechanisms [1, 2] that may
develop, i.e., flexural and diagonal shear, respectively, as follows
(2)
(3)
The prevailing collapse mechanism is the one with lower capacity.
Table 1 contains the parameters used to calculate the stiffness (Eq. (1)) and the capacities (Eqs. (2)
and (3)).
For the case at hand, the capacities are: f
yF
= 64.6 kN; f
yV
= 390.8 kN. Therefore, it is recognized
that failure is of the flexural type.
The experimental test on the unstrengthened walls exactly shows this behaviour, for a horizontal
force equal to 70 kN (15.7 kips), very close to the analytically predicted value. Figure 11 shows
the crack pattern, typical of a flexural failure mechanism (note that the flange walls, though present,
are not connected).
4.2 Connected Walls
The connection of the web wall to the flange walls improves the performance of the overall system,
by increasing its bending capacity thanks to the change in shape of the base cross-section, which is
h
mm
(ft.)
1850
(6)
l
mm
(ft.)
1550
(5)
t
mm
(in.)
120
(5)
Aw
mm
(in.)
1.86x10
5
(7x10
3
)
o
0.83
G
MPa
(ksi)
900
(130)
E
MPa
(ksi)
2700
(392)
f
m,d
MPa
(ksi)
6.00
(0.87)
t
0,d
MPa
(ksi)
1.38
(0.20)
o
0
MPa
(ksi)
0.97
(0.14)
Table 1 - Parameters used to evaluate the wall capacity
Figure 11. Crushing of masonry in web wall: left and right side, near foundations.
F
2
=
q r
o
f
y
H
0
p
2
1
1
0 0
0.85 f
d
.
t o o
.
.
V
f
y
t
b
=
1 . . .
.
1 5
d
.
.
1 5
d
1
0
+
o
t
w
t
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 69
now T-shaped, with a significant contribution from the compressed flange. The shearing capacity,
however, remains unaffected and equal to that of the web wall, since the flanges do not provide any
contribution to it.
The bending capacity of the T-shaped cross section is obtained by writing the equilibrium equation
of the web and flange system, considered as fully connected. Figure 12 shows a comparison of the
two configurations for adjacent and connected walls.
The equilibrium equation for the T-shaped wall is given in Eq. (4)
V
.
H = N
p
.

e
p
+

N
a
.

e
a
(4)
Rearranging Eq. (4) and introducing non-dimensional notation, the following capacity equation is
obtained
(5)
where M
U,T
is the flexural capacity of strengthened T-shaped wall; M
U,I
is the capacity of the
(unstrengthened) I-shaped wall and is given in Eq. (6); q, t and are non-dimensional factors (Eq.
(7)). Note that the first factor is of mechanical nature whereas the others are related to geometry.
Figure 12. Notation of variables and stress distribution before failure: isolated wall (left);
connected wall (right).
(6)
(7)
Applying Eq. (5) to the specific case of the test walls, characterized by the following parameters:
t= 0.077, = 4.58, q= 0.23, it is possible to assess the magnitude of the force that triggers the
bending failure mechanism (106 kN (23.8 kips)). The result is practically identical to that obtained
in the test, of 110 kN (24.7 kips).
In order to calculate the capacity of the strengthened system, in this case it is also necessary to duly
account for the contribution of the compressed flange. As it is known, Eq. (1) takes into account
both bending (K
B
) and shear (K
S
) stiffness of the walls (Eq. (8)).
(8)
M
U,T
M
U,I
=
.
(1+ t . )
t
2
.
(1 )
q
+ r q
M
U,I
=
1
2
to
0
(1 q)
r q
2
o
0
q = = =
0.85
.
f
d
; t
t
a
1
p
1
a
t
p
;
K
I
=

K
B
+ K
S
=
12EJ
h
3
1.2h
+
GA
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 70
M. Vailati and G. Monti
As mentioned above, the presence of the flange only modifies the flexural response of the wall,
leaving the shear behaviour almost unaffected.
It will, therefore, be sufficient to express the second moment of area for a T-shaped section in place
of that for the I-shaped section, in order to predict the stiffness of the strengthened system. The
second moment of area for the T-shaped section is given as
By re-defining Eq. (8) we obtain
(10)
In Eq. (10), o assumes its usual meaning. Note that the second moment of area of the T-shaped
section significantly contributes to the flange width, i.e., the part of the wall at the end involved in
the flexural resisting mechanism.
Although there is a dependency on the thickness ratios of the connected walls this type of problem
requires specific study, which can only be addressed in qualitative terms here. Considering this, it
may be helpful to address the problem according to Tomazevic [3], so that the flange width is defined
as in Figure 13.
The proposed approach must be considered in the context of a more general application of the
method in real cases, for buildings made of bricks.
In the specific case of the walls tested, the problem is not so significant, since all flanges lengths
were entirely involved in the resisting mechanism.
4.3 Design of the connections
Based on the above discussion, a design equation is proposed here for an easy application of the
method. The equation is capable of correlating the sliding force between the faces of the two
connected walls with the number of rods, which is essential for a correct design of the connection
system.
By looking at the distribution of contact stresses in Figure 12, the sliding force between the web
and flange walls can be evaluated. Since the vertical load is only applied on the web wall, the stress
at the flange base is the reaction to the combined compressive force and bending moment generated
by the horizontal force. The sliding force at the interface is thus given by
Figure 13. Definition of the geometry of the flange in the connected walls.
(11) F
s
= (0.85
.
f
d
- o
0
)
.
t
a
1
a
K
T
=
GA
w
GA
w
12EJ
T
1.2h
2
1
h
+ q r o
. .
(9)
[l
a

.
(l
p
+ t
a
)
2
- l
2
p

.
(l
a
- t
p
)]
2
- 4
.
l
a
.
l
p
.
(l
p
+ t
a
)
.
(l
a
- t
p
)
.
[(l
p
+ t
a
) - l
p
]
2
12
.
[l
a

.
(l
p
+ t
a
)- l
p
.
(l
a
- t
p
)]
J
T
=
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 71
The number of rods to be inserted to ensure that the connection resists until bending failure is then
given by
(12)
where R
T,1-2
is the connection tensile strength, according to one of the possible configurations, as
indicated in section 3. The total number of the rods to be applied is therefore equal to 4n, because
they are applied on both faces of the web wall and because they are inserted at 45.
5. CONCLUSIONS
The comparison of the performance of connected walls with that of adjacent walls (based on the
experiments carried out) indicated that a strength increase of
R
= 1.57 and displacement increase
of
d
= 3.2 was obtained. As a result, there is an increase in the capacity of about 60%, while ductility
increases by a factor of 3. Note that, beyond affecting the main resisting mechanisms, the proposed
strengthening technique provides both local and global ductility as this application can reverse the
hierarchy between failure mechanisms (shear and flexure), thus favouring the latter, which is more
ductile. Finally, it was ascertained that the developed equations are accurate in predicting both the
capacity and the stiffness of the strengthened wall system.
ACKNOWLEDGMENTS
The authors wish to thank the SACEN Company of Naples, Italy, for providing and applying the
NAILTEX AFRP rods.
NOTATION
fm = Mean compression strength
t
0
= Mean shear strength
K
e
= Stiffness of masonry wall
A
w
= Shear area of I-section
h = effective height of wall
l = Length of wall
o = Factor dependent of boundary condition (0.83 or 3.33 in case of cantilever)
t = Thickness of wall
o
0
= Mean normal tension on the total section area
H
0
= Point along the wall where the moment change sign
b = Factor depending of wall slenderness, and it can be take 1 Ob=h/l O1.5
f
yF
= Flexural capacity of wall
f
yV
= Diagonal shear capacity of wall
N
P
= Resultant of normal force on web section
N
a
= Resultant of normal force on flange section
e
p
= Eccentricity of Np
e
a
= Eccentricity of Na
K
I
= total stiffness of wall (I-shaped section)
J
T
= moment of inertia of T-shaped section
K
T
= total stiffness wall (T-shaped section)
F
S
= Sliding force at the interface web/flange
n = Number of rods
REFERENCES
[1] Ministry of Infrastructure and Transportation of Italy. Istruzioni per lapplicazione delle nuove
norme tecniche per le costruzioni di cui al decreto ministeriale 14 gennaio 2008. Supplemento
ordinario n. 27 alla Gazzetta Ufficiale, 2009.
[2] Ministry of Infrastructure and Transportation of Italy. Nuove Norme Tecniche per Le Costruzioni.
Gazzetta Ufficiale della Repubblica Italiana 2008.
[3] Tomazevic M. Earthquake resistant design of masonry buildings. Series on Innovation in
Structures and Construction., England, Imperial College Press London, 1981. p. 109-158.
F
s
n = round
2
.
w

2
.
R
T,1-2
o p
NED UNIVERSITY JOURNAL OF RESEARCH, THEMATIC ISSUE ON EARTHQUAKES, 2012 72
M. Vailati and G. Monti
[4] Menegotto M, Monti G, Salvini S, Vailati M. Improvement of Transverse Connection of
Masonry Walls through AFRP Bars. In: Ye L, Yue Q, Peng F, Editors. Proceeding of 5th
International Conference on FRP Composites in Civil Engineering. Beijing, China: 2010.
p. 947-950.
[5] Anthoine A, Magonette G, Magenes G. Shear-compression Testing and Analysis of Brick
Masonry Walls. In: Proceedings of 10th European Conference on Earthquake Engineering.
Rotterdam, Netherlands: 1995.
[6] Magenes G, Calvi GM. Cyclic Behaviour of Brick Masonry Walls. In: Proceedings of 10th
World Conference on Earthquake Engineering. Rotterdam, Netherlands: 1992.
[7] Magenes G, Calvi GM. In-plane Seismic Response of Brick Masonry Walls. Earthq Eng Struc
Dyn 1997;26(11):1091-1112.
[8] Turnek V, Cacovic F. Some Experimental Results on the Strength of Brick Masonry Walls.
In: Proceedings of the 2nd International Brick Masonry Conference, Stoke-on-Trent, England:
1971. p. 149-156.
[9] Turnsek V, Sheppard P. The Shear and Flexural Resistance of Masonry Walls. In: Procedure
of the International Research Conference on Earthquake Engineering. Skopje, Macedonia:
1980. p. 517-573.
[10] Turco V, Secondin S, Morbin A, Valluzzi M.R, Modena C. Flexural and Shear Strengthening
of Un-reinforced Masonry with FRP Bars. Comp Sci Tech 2006;66(2):289-296.
[11] Tumialan G, Micelli F, Nanni A. Strengthening of Masonry Structures with FRP Composites.
In: Chang PC, Editor. Proceedings of the 2001 Structural Congress and Exposition. Washington
DC, USA: 2001. p. 1-8.