You are on page 1of 12

NUMERICAL AND EXPERIMENTAL ANALYSIS OF THE OUT-OF-

PLANE CAPACITY OF UNREINFORCED MASONRY WALLS


Moritz LÖNHOFF 1, Hamid SADEGH-AZAR 2

ABSTRACT

Unreinforced masonry (URM) is still used for many buildings in Europe and all over the world. In many regions,
the verification of the load-bearing capacity of URM walls in an earthquake scenario is mandatory. In load cases
with horizontal forces, like earthquakes or explosions, the out-of-plane capacity of URM walls often gets crucial.
In currently used design methods, the maximum earthquake load is applied on quasi-static models, neglecting
many important influencing parameters. This results in inaccurate and often conservative estimations of the out-
of-plane capacity. To realistically determine the out-of-plane capacity, parameters like geometry, restraints,
vertical loads and dynamic effects need to be considered.
To investigate the influence of those parameters, two experimental tentative test series are carried out at the
University of Kaiserslautern. Therefore, heat insulating unreinforced masonry walls are tested on a shaking table,
loaded with real earthquake time histories. In the tests, the influence of vertical loading and vertical stiffness of
the upper boundary on the out-of-plane load bearing capacity is investigated. Results are compared with analytical
methods and numerical simulations. The comparison shows that the out-of-plane capacity of URM walls is higher
than estimated by simple analytical models.

Keywords: unreinforced masonry walls, out-of-plane behavior, nonlinear analysis, shaking table tests

1. INTRODUCTION

In 2005, the German earthquake design code (DIN 4149) was included in the technical building
regulations. This made earthquake design checks mandatory in many regions in Germany. As in many
other areas in the world, the load bearing capacity of loaded and unloaded unreinforced masonry (URM)
walls must now also be ensured under earthquake loads. The collapse of URM walls as well as falling
debris must be prevented for the design earthquake load. The stability transverse to the plane (out-of-
plane capacity) often gets decisive in this scenario.
With the planned implementation of DIN EN 1998 and the introduction of the new earthquake hazard
maps for Germany, the requirements for an earthquake-proof design are tightened once more, which
makes an economic design even more difficult. For buildings that are important for the protection of the
population (for example, hospitals, fire stations, power stations, schools and cultural institutions), higher
demands are required on earthquake resistance, which means that even higher earthquake loads need to
be considered.
The design models for the load-bearing capacity of masonry walls currently used in DIN EN 1996 and
KTA 2201.3 are quasi-static methods in which the PGA is used as an input variable. These models lead
to conservative estimations as many simplifications are used. More advanced displacement-based
models (Griffith) lead to more realistic results but still neglect important influencing parameters.
Investigations by Doherty et al. (2002) have already shown that the actual load-bearing capacity of URM
walls under earthquake loads is significantly higher than predicted in the simple models mentioned
above. In the quasi-static experimental out-of-plane tests carried out in the framework of the EU project

1
Dipl.-Ing., Institute of Structural Analysis and Dynamics, University of Kaiserslautern, Germany,
moritz.loenhoff@bauing.uni-kl.de
2
Prof. Dr.-Ing., Institute of Structural Analysis and Dynamics, University of Kaiserslautern, Germany,
hamid.sadegh-azar@bauing.uni-kl.de
INSYSME (Innovative Systems for Earthquake Resistant Masonry Enclosures in R. C.) considerable
load-bearing reserves have been identified (Butenweg et al. 2004). In addition, there are load-bearing
reserves resulting from the dynamic effects after cracking of the wall. Geometric instability (failure) of
the wall occurs only if the deformation of the wall exceeds the stability limit (Doherty et al. 2002). The
main reason for the failure of the wall are therefore geometrical properties, especially the size of the
maximum deflection. However, due to the rapid and frequent change of the excitation direction during
the earthquake, such large deformations only occur at high accelerations. In addition to the geometrical
failure, falling debris must also be avoided, which is strongly influenced by the type of bricks used.
However, disregarding time-dependent dynamic properties can lead to unnecessary and uneconomical
rehabilitation measures or retrofitting in the seismic assessment of existing masonry walls and to
uneconomical design of new buildings (Sadegh-Azar 2012).

2. STATE OF THE ART

The out-of-plane behavior of URM walls has been investigated in various studies (Kariotis et al. 1981,
Dawe and Seah 1989, Angel et al. 1994, Doherty and Griffith 2000, 2002+2004, Simsir et al. 2004,
Tondelli et al. 2016, Penner and Elwood 2016). In particular, the connection at the top of the wall was
identified as one of the most important boundary conditions. Experimental investigations of Dafnis et
al. (2000), Meisl et al. (2005) and Dazio (2008) were especially focused on the influence of the boundary
conditions on the out-of-plane behavior of URM walls. Dafnis conducted shaking table tests on walls
with varying connections at the top of the wall. He considered complete joints, partially filled joints and
gaps that may occur due to shrinking of the mortar. Meisl investigated the influence of different soils,
but also used two different joints, one with poor and one with good quality. In experimental studies by
Dazio, different significant factors on the out-of-plane capacity were investigated. Both the bearing
conditions at the upper and lower end of the wall as well as the axial load were varied. The walls were
tested on a shaking table using ground motions. The load was thereby increased until destruction of the
wall. The investigations led to the conclusion that the boundary conditions can have a larger influence
on the walls capacity then the slenderness.
In practical applications, the models of Paulay and Griffith as well as the models from KTA 2201.3 and
DIN EN 1996 are often used. In the Model of Paulay (Paulay and Priestley 1992), the wall is loaded
with a constant load, which is calculated from the maximum earthquake acceleration and the inertial
forces of the wall. A gaping joint is permitted and it is assumed that only compressive stress is
transmitted between the two rigid slabs (Fig. 1a).
The method of Griffith (Griffith et al. 2004), where the wall is also divided into two rigid slabs and is
deflected virtually by one at the center of the wall. With the principle of work and energy the equilibrium
conditions are determined and the maximum acceleration is calculated (Fig. 1b). Additionally a
deformation-based model is proposed, that uses an idealized trilinear force-deformation relationship.
The German standard DIN EN 1996 provides a very simplified analytical model for the design of
masonry, but not explicitly for earthquake design. Hinges on the bottom and top of the wall are assumed
in this model. It uses the flexural strength and elasticity theory for the calculations (Fig 1c). The German
Nuclear Safety Standards Commission (KTA) also provides an analytical model (KTA 2013). Herein
the arc effect of masonry is implied. A triangular stress distribution, which extends over three-quarters
of the cross-sectional width, is assumed in the wall center. This model can only be used if an axial load
is applied because otherwise no arching effect exists (Fig 1d).
None of the previously conducted research investigated the out-of-plane behavior of thermally insulated
bricks with relatively small webs, which are used in Germany in many new buildings recently.

2
a b c d

Figure 1. Analytical models: (a) Paulay; (b) Griffith; (c) DIN EN 1996; (d) KTA 2201.3

3. EXPERIMENTAL INVESTIGATIONS

In order to investigate the out-of-plane capacity of URM walls, the analytical methods have been
assessed and compared with numerical simulations in Lönhoff et al. (2017). The investigations lead to
the conclusion that the capacity estimated by the common methods from codes and literature are
inaccurate and in many cases conservative since decisive effects are neglected. One major parameter
not considered in the models is the influence of the dynamic time-dependent behavior of the wall. In the
numerical simulation, the rapid alternation of direction of the excitation leads to only slight deflections
in the center of the wall. This also limits the stresses in the mortar joints and the blocks. Additionally,
the wall extends in the axial direction due to the rotation of the blocks during rocking, which leads to an
increase of axial loading. This effect has an additional stabilizing effect on the wall and thus increases
the load-bearing capacity.
To verify the results of the numerical analysis as well as the load-bearing reserves determined from it,
a series of experimental dynamic tests is planned at the University of Kaiserslautern. Therefore,
unreinforced masonry walls with realistic dimensions will be loaded in the x- and z-direction with
earthquake time histories on the universities shaking table in order to determine the actual load-bearing
capacity.
In two tentative tests conducted so far, 2.5 m high masonry walls were tested in the newly developed
test rig on the shaking table. The loading was based on the time history of the El Centro earthquake and
corresponds to a very high and energetic earthquake loading with peak accelerations of more than 3 m/s²
(Fig. 2).

0,4
0,3
0,2
acceleration [g]

0,1
0
-0,1 0 5 10 15 20 25 30
-0,2
-0,3
-0,4
time [s]

Figure 2. El Centro time history


3
The used bricks are thermally insulated blocks with a thickness of 0.24 m, a length of 0.248 m and a
height of 0.249 m. They have 10 and 15 mm thick webs and eight chambers filled with mineral wool
for the thermal insulation (Fig. 3). The stone strength class is eight. The used mortar is a thin-bed mortar.

Figure 3. First row of thermal insulated brick wall

The bottom and top of the wall is framed by U-profiles, which are needed for the transportation and
mounting of the wall as well as the transfer of the applied loads from the shaking table into the wall.
The U-profile at the bottom is filled with mortar. Therefore, the rotation of the lowest brick row is
mostly restrained. This does not correspond to the actual situation on site but is here used to have
clearly determined boundary conditions in order to calibrate the numerical models.
To ensure horizontal stiffness at the top of the test rig there are four bracings connected to the Pi-
profile at the top. Between this profile and the U-profile of the wall, springs can be used to provide
vertical stiffness as well as vertical loading of the wall. The horizontal load at the top is transferred
through a connection between Pi- and U-profile (Fig. 4).

Figure 4. Schematic sketch of the test rig and upper boundary

During the tests, displacements and accelerations of the wall are recorded. Accelerations are measured
at the bottom, center and top of the wall using 100mV/g sensors. Displacements are recorded in the
quarter points of the wall with laser distance sensors that are able to measure large deformations (Fig.
4+5).

4
First tentative test series

In the first tentative test with the 2.50 m masonry wall, the applied axial load was 0.9 kN. The springs
at the top boundary are installed in the center of the profile (see Fig. 4) leading to a vertical stiffness of
3050 kN/m. For the coupling of the Pi- and U-profile to transfer horizontal loads at the top, wooden
wedges are used in this test. With this method of coupling, a free rotation of the top of the wall is not
possible. The El Centro time history (Fig. 2) was applied in horizontal direction only. It was increased
by scaling up to an acceleration of 9 m/s² (3 x El Centro).

Figure 5. Experimental dynamic seismic testing on the shaking-table of the University of Kaiserslautern

The first two tests with 1x and 2x El Centro time history, corresponding to 3 m/s² and 6 m/s², led to no
measurable displacements of the wall. The excitation of 3 x the El Centro earthquake with accelerations
of 9 m/s² led to the cracking of the wall in the mortar joint at 7/10 height. Still, the displacements of the
wall were small with a maximum deformation at ¾ height of 1.2 cm (Fig. 6). Also, the opening of the
cracked joint was very small during the test. No further damage of the wall was observed in this test
series.

1,5 3/4-point
1/2-point
1
displacement [cm]

1/4-point
0,5

0
0 5 10 15 20 25 30
-0,5

-1

-1,5
time [s]

Figure 6. Displacements in the first tentative test series (3xEl Centro)


5
Second tentative test series

In the second test series, the same type of wall was used while the boundary conditions at the top of the
wall were changed. For the vertical stiffness, the same springs with a vertical stiffness of 3050 kN/m
were used. The horizontal load transmission at the top was this time done with ball pod bolts to allow a
rotation of the top of the wall (Fig. 7). Additionally, a gap of 3mm was added between top of the wall
and vertical springs to see if the rotation of the top is actually possible as well as to investigate the
influence of the vertical stiffness on the deformation and capacity of the wall. No vertical load was
applied. The load was again applied in horizontal direction only.

Figure 7. Rotation of the top boundary

With this configuration the excitation with a scale factor of the El Centro earthquake of 1 and 2 did not
lead to any damage in the wall and therefore very low deformation. A load of 3 x El Centro (9 m/s²) led
to the cracking of the mortar joint at 4/5 of the wall height. Since the rotation of the first row is restrained
due the used U-profile, the joint above the first brick row cracked so the bottom part of the now two-
slab system is free to rotate (Fig. 8).

Figure 8. Location of cracked joints

After cracking of the wall, another test with the 3 x El Centro was conducted. Here, larger deformations
and larger opening of the cracked joint (see Fig. 7) occurred. Still no geometrical failure of the wall was
observed. During the larger rotation of the blocks and the associated elongation of the wall, the gap
between wall and vertical springs was closed and the vertical stiffness was activated. This prevented the
top block from further rotation and therefore limited the horizontal deformation of the wall. This
prevented the wall from a geometrical failure.
The maximum displacement at ¾ height in the test with 3 x El Centro earthquake was about 2.8 cm (Fig.
9).

6
3 3/4-point
1/2-point
2
1/4-point
displacement [cm]

0
0 5 10 15 20 25 30
-1

-2

-3
time [s]

Figure 9. Displacements in the second tentative test series (3xEl Centro)

In a subsequent test of the wall with 3xEl Centro, the large deformations and the rocking of the wall led
to local failure of the bricks as well as a horizontal displacement of the two top rows. Due to the rotation
of the bricks during the rocking process, large forces are transferred to the outer web of the bricks. This
led to a shear failure of the connection between the outer and inner web. The horizontal shifting of the
top rows was 10-17mm (Fig. 10).

Figure 10. Failed outer web of the bricks

7
To investigate the influence of the destroyed joint, bricks and the shifted top rows further tests with 1x
and 2x El Centro were conducted with the same wall.
While in the first tests with 1x and 2x El Centro before cracking of the wall, there was no measurable
deformation of the wall, now observable displacements occurred. In the test with the 1x El Centro the
maximum deformation was 1.8 cm (Fig. 11). In the test with 2 x El Centro it was 3.7 cm, larger than the
displacements in the test with 3x El Centro before destruction of the bricks and shifting of the top rows
(Fig. 12).

1,5 3/4-point

1 1/2-point
1/4-point
displacement [cm]

0,5

0
0 5 10 15 20 25 30 35
-0,5

-1

-1,5

-2
time [s]

Figure 11. Displacements in the second tentative test series after cracking (1xEl Centro)

3 3/4-point
2 1/2-point
1/4-point
displacement [cm]

0
0 5 10 15 20 25 30 35
-1

-2

-3

-4
time [s]

Figure 12. Displacements in the second tentative test series after cracking (2xEl Centro)

A subsequent test with 3x El Centro led to an abortion of the experiment due to exceedance of the forces
acting on the shaking table. The ultimate out-of-plane capacity of this type of wall could therefore not
be determined yet.

8
4. ANALYTICAL AND NUMERICAL ANALYSES

The used walls are also analyzed with the analytical models from KTA, Paulay and Griffith described
in Chapter 2. As an example, for the calculation of the maximum attainable acceleration with methods
of Paulay equation (1) is used.

8 𝑊𝑊 𝑡𝑡 2 ⋅ 𝑃𝑃 + 𝑊𝑊
𝑞𝑞 = 2
∙ �𝑃𝑃 + � ∙ � − − ∆� (1)
ℎ 2 2 4 ⋅ 𝑓𝑓𝑚𝑚 ⋅ 𝑡𝑡

with the wall height h, the vertical load P, the dead weight of the wall W, the wall thickness t, the
compressive strength of the masonry fm and the displacement ∆ of the wall centre defined by Paulay
(Fig. 1a). For further information on the used models and calculations see Lönhoff et al. (2017-1).
Characteristic values are used in all cases. For the first test series with an axial load of 0.9 kN the models
lead to a capacity of 0.09 m/s² for KTA, 0.28 m/s² for Griffith (equilibrium) and Paulay and up to 0.51
m/s² for Griffith (displacement) (Fig. 13) assuming a severe damage of the joints (see Doherty et al.
2002).
The actual displacement of about 1 cm in the experiment with an acceleration of 0.9 m/s² shows that
there are quite some reserves left until a geometrical failure will occur.

0,60

0,50
acceleration [g]

0,40

0,30

0,20

0,10

0,00
Paulay Griffith (eq) Griffith KTA
(disp)
1. series: 0.9 kN axial load
2. series: no axial load

Figure 13. Out-of-plane-capacity from analytical methods for the first and second test series

For the second test series without any axial loading, the model of Paulay and Griffith (equilibrium) gives
a capacity of 0.19 m/s², KTA leads to zero capacity since the model is based on the arching effect which
does not exist without any loading. Griffith (displacement) gives a capacity of 0.38 m/s² (Fig. 13). The
applied acceleration in the test was also 0.9 m/s², leading to a deformation of about 3-4 cm. Though,
there are also reserves available until reaching the ultimate capacity of the wall.
In none of the models the vertical stiffness of the top boundary, which has a big influence on the capacity,
is considered. Also the gap between wall and springs which was used in the second test series is not
considered by the models. This is one key parameter leading to imprecise and conservative predictions
of the out-of-plane capacity.

The experimental tests were also analyzed using the developed numerical 2D model. Therefore, the wall
was modeled with 10 slab elements representing the 10 brick rows. The mortar joint was modeled using
nonlinear spring element neglecting any tension stiffness. The vertical spring at the top was considered
9
using linear spring elements and the applied axial force was considered by an implied displacement. The
time history was simultaneously applied at the bottom and top of the wall. For more details on the
numerical model see Lönhoff et al. (2017).

1,5 experimental
1 numerical
displacement [cm]

0,5

0
0 5 10 15 20 25 30
-0,5

-1

-1,5

-2
time [s]

Figure 14. Comparison of experimental and numerical deformation of the first test series/3xEl Centro

The deformation of the numerical model is in good accordance with experimental data when large
displacements occur. The smaller deformations of the wall are not present in the numerical model,
instead, the deformation is zero (Fig. 14). The low horizontal deformations of 3 mm in the experiment
correspond to a gap opening of only 0.5 mm, which occurs due to the imperfections of wall, bricks and
joints that are not represented by the model. Another reason for the differences may be that in the
simulation the average of the measured bottom and top acceleration was used. As shown by Beyer and
Lucca (2015) different accelerations at bottom and top can lead to an increase of the deformation.

5. CONCLUSION AND OUTLOOK

In the present paper, experimental, analytical and numerical investigations on the out-of-plane capacity
of unreinforced masonry walls have been presented. Two tentative test series with heat insulating brick
walls were conducted on a shaking-table. The first test series was carried out with a small axial load and
a vertical stiffness at the top boundary. In the second test series, a gap between top of the wall and top
boundary was introduced without any axial load. The tests showed that the influence of the gap/vertical
stiffness of the top boundary as well as the axial load of the wall have an influence on the out-of-plane
behavior of the wall. Also the condition of the joints has a large influence on the deformation of the
wall. In the analytical analysis, typical methods from codes and literature have been used for a
comparison. Since many important effects are neglected in the models, the results show inaccurate and
conservative estimations of the out-of-plane-behavior and capacity.
The numerical simulation of the experimental test shows a good accordance of the large deformations
in case of the loaded wall. The small deformations of the wall are not present in the numerical model.
In the further course of the project, numerous tests are planned in order to investigate the influence of
various parameters on the out-of-plane capacity. First, different axial loads and different vertical
stiffness (with and without gaps) of the upper support will be investigated. In addition, different block
and mortar types and various wall dimensions are to be investigated. In ancillary tests, the ultimate
capacity of the wall will be determined by increasing the loading until geometrical failure of the URM
walls, also investigating the influence of the vertical component of the earthquake excitation on the walls
capacity.
After experimental investigation and identification of the most important influencing parameters, the
numerical model will be further developed and adapted so all those parameters can be represented by

10
the model. With this model, further parameter studies will be conducted to evaluate the influence of
several parameters like bricks, mortar, geometry, vertical loading, vertical stiffness as well as the
influence of the used earthquake record.
In a final step, an engineering model that takes all important effects into account will be created to
estimate and evaluate the out-of-plane capacity of URM walls.

6. ACKNOWLEDGMENT

The ongoing investigations presented here are supported by the Arbeitsgemeinschaft Mauerziegel in the
Bundesverband der Deutschen Ziegelindustrie e.V. and the research field HiPerCon of the University
of Kaiserslautern.

7. REFERENCES

Angel R, Abrams D, Shapiro D, Uzarski J, Webster M (1994). Behavior of Rein-forced Concrete Frames with
Masonry Infills. University of Illinois at Urbana-Champaign, Civil Engineering Studies, Structural Research
Series 589.
Beyer K, Lucca F (2015). Einfluss der Randbedingungen auf das verhalten aus der Ebene von Mauerwerkswänden.
14. D-A-CH-Tagung: Erdbebeningenieurwesen und Baudynamik. pp. 139-146.
Butenweg C, Meyer U, Fehling E (2014). EU-Projekt INSYSME: Innovative Techniken für erdbebensichere
Ausfachungswände aus Ziegelmauerwerk in Stahlbetonrahmentrag-werken Mauerwerk, 18(2): 78–81.
Dafnis A, Kolsch H, Reimerdes H.-G (2002). Arching in Masonry Walls Subjected to Earthquake Motions.
Journal of Structural Engineering.
Dawe J, Seah C K (1989). Out-of-plane resistance of concrete masonry infilled panels. Can. J. Civ. Eng. 16: 854-
864
Dazio A (2008). The Effect of the Boundary Conditions on the Out-Of-Plane Behavior of Un-reinforced Masonry
Walls. 14WCEE.
DIN EN 1996-1-1 (2013). Eurocode 6: Bemessung und Konstruktion von Mauerwerksbauten – Teil 1-1:
Allgemeine Regeln für bewehrtes und unbewehrtes Mauerwerk. Deutsche Fassung EN 1996-1-1:2005+A1:2012.
Doherty K, Rodolico K T, Lam N, Wilson J, Griffith M O (2000). The Modeling of Earthquake induced Collapse
of Unreinforced Masonry Walls Combining Force and Dis-placement Principals. 12WCEE.
Doherty K, Griffith M C, Lam N, Wilson J (2002). Displacement-based seismic analysis for out-of-plane bending
of unreinforced masonry wall. Earthquake Engng Struct. Dyn. 31: 833-850.
Griffith M C, Lam N, Wilson J (2004). Experimental Investigation of Unreinforced Brick Masonry Walls in
Flexure. Journal of Structural Engineering. 130(3).
Kariotis et al. (1981). Methodology for mitigation of seismic hazards in existing unreinforced masonry buildings:
wall testing, out-of-plane. Technical Report ABK-TR-04, ABK, A Joint Venture. USA.
KTA 2201 (2013). Auslegung von Kernkraftwerken gegen seismische Einwirkungen, Teil 3: Bauliche Anlagen.
Meisl et al (2005). Shake table tests on the out-of-plane response of unreinforced masonry. 10th Canadian
Masonry Symposium.
Lönhoff M, Dobrowolski C, Sadegh-Azar H (2017-1). Analyse des out-of-plane-Verhaltens von unbewehrten
Mauerwerkswänden. D-A-CH-Tagung: Erdbebeningenieurwesen und Baudynamik 2017. 17: 419-427. Weimar.
Lönhoff M, Dobrowolski C, Sadegh-Azar H (2017-2). Analysis of the out-of-plane capacity of unreinforced
masonry infill walls. Procedia Engineering. 199: 693-698.
Paulay T, Priestley M J N (1992). Seismic Design of Reinforced Concrete and Masonry Buildings, Wiley & Sons,
USA.
Penner O, Elwood K J (2016). Out-of-Plane Dynamic Stability of Unreinforced Masonry Walls in One-Way
Bending: Shake Table Testing. Earthquake Spectra. 32(3): 1675–1697.

11
Simsir C C, Aschheim M A, Abrams D P (2004). Out-Of-Plane Dynamic Response of Unreinforced Masonry
Bearing Walls Attached to Flexible Diaphragms. 13WCEE.
Sadegh-Azar H, Hartmann H-G, Wörndle P (2012). Sicherheitsreserven und Konservativitäten in der
normgerechten Erdbebenauslegung von Stahlbetonbauwerken. Beton- und Stahlbetonbau. Volume 6.
Tondelli M, Beyer K, DeJong M (2016). Influence of Boundary Conditions on the Out-of-Plane Response of Brick
Masonry Walls in Buildings with RC Slabs. Earthquake Engng Struct. Dyn. 45: 1337-1356.

12

You might also like