Engineering Structures 30 (2008) 2194–2205

www.elsevier.com/locate/engstruct

Stone masonry in historical buildings — Ways to increase their

resistance and durability
Emı́lia Juhásová a,∗ , Ramiro Sofronie b , Rogério Bairrão c
a Institute of Construction and Architecture, Slovak Academy of Sciences, Dúbravská cesta 9, 845 03 Bratislava, Slovakia
b Ecoland, University of Bucharest, Bd, Marasti 59, RO-71331 Bucharest, Romania
c Laboratório Nacional de Engenharia Civil, Av. Brasil, 101, 1700-066, Lisbon, Portugal

Available online 7 September 2007

Abstract

The paper is devoted to stone masonry structures in view of possible improvement of their dynamic-seismic resistance and durability. Many
existing structures and large models were investigated by the authors in the framework of national and international research projects. The
investigation of a large stone masonry model, tested on the shaking table, had as main purpose the evaluation of the behaviour of new and
rehabilitated masonry structures reinforced with polymer grids produced in the European Union. The test plan consisted of the following parts:
tests of individual material properties; seismic tests of a masonry model reinforced with polymer grids inserted horizontally in prescribed bed
joints; seismic tests of a masonry model reinforced on parts of a wall surface by vertical polymer grids bonded with fibre-reinforced plaster.
The paper presents the selected representative results and synthesis of obtained data. The polymer grids and fibre mortars enhance the seismic
resistance of stone masonry structures. They contribute also to the durability of historical and other stone masonry structures, especially in cases
of vertical polymer grids combined with fibre lime–cement plaster.
c 2007 Elsevier Ltd. All rights reserved.

Keywords: Seismic vulnerability; Retrofitting; Upgrading; Fibre mortars; Polymer grids; Stone masonry; Seismic tests; Seismic resistance

1. Introduction The action effects need to be taken into account in developing

Historical buildings and monuments create a part of
cultural heritage and therefore, countries, unions, national and
international bodies should pay attention to their maintenance
and preservation to protect them for future generations. A limit
state method, as a general worldwide method for the design of
new structures and the assessment and rehabilitation of existing
ones, has not been developed for failures due to the material
deterioration to the extent that it has for failures due to actions
as earthquakes, settlements, wind, snow, etc., combined with
gravity loads. Both for new and old structures it is important
to identify an action-effect process starting from the structure’s
environment. Then, the mechanisms are active to transfer the
environment into environmental actions on structural materials.
∗ Corresponding address: Institute of Construction and Architecture, Slovak
Academy of Sciences, Earthquake Engineering Laboratory, Dubravska cesta 9,
845 03 Bratislava, Slovakia. Tel.: +421 2 59309216; fax: +421 2 54773548.
E-mail addresses: juhasova@savba.sk (E. Juhásová), ecoland@ecoland.ro
(R. Sofronie), bairrao@lnec.pt (R. Bairrão).
methods for the prediction of working life, and developing tools
for its prolongation [1–8].
New international/European standards define shorter design
working life in comparison with those ones in national stan-
dards. The indicative design working life in EN 1990: 2002 [9]
presents lower figures in comparison with national standards,
e.g. STN 73 0031: 1989 [26], as shown in Table 1. The structure
shall be designed such that deterioration over its design working
life does not impair the performance of the structure below that
intended. Environmental effects and anticipated level of mainte-
nance should be taken into account. Although the indicative val-
ues of Table 1 will record the wide use, the customer, national
authority or the designer can decide for higher design work-
ing life, including necessary changes in respective safety factors
and other parameters dependent on the design working life.
The indicative design working life in Table 1 does not
correspond to actual demands on historical buildings, where
the recommended life should be as long as possible. Thus, the

c 2007 Elsevier Ltd. All rights reserved.

0141-0296/$ - see front matter
doi:10.1016/j.engstruct.2007.07.008
 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205 2195

Fig. 1. Deterioration of stones in lower part of defensive walls — the medieval castle. The detail is on the right.

Fig. 2. Example of the old stone masonry building under repair — left; ruins of an old stone masonry structure — right.

Table 1 potential energy. The strength of stone is higher than that

Comparison of indicative design working lives of bricks and that of mortar. All consequences of stress
Category EN 1990: STN Examples concentrations around any structural faults are supported only
of a 2002 73 0031 by mortar. Originally, stone masonry was made with lime
structure (years) (years) mortar. The thickness of bed layers was chosen according to the
1 10 15 Temporary structuresa dimensions of stone units. The ductile lime mortar contributes
2 10–25 30 Replaceable parts like gantry girders, bearings to reduce the effects of stress concentrations. However, it
3 15–30 30–50 Agricultural and similar structures has a low tensile resistance. Many buildings and structures
4 50 80–100 Buildings and other common structures of stone masonry, under their own weight or due to some
5 100 100–500 Monumental buildings, bridges, etc.
settlements, result in cracking. By replacing lime with cement
a Structures or their parts to be reused should not be considered as temporary. in mortars, the behaviour of stone masonry was not improved as
expected. Indeed, even if bearing capacity to gravitational loads
special strategy is needed both for maintenance, conservation, increased, the lack of ductility of cement mortars made it more
repair, rehabilitation, intervention measures, etc. This paper is vulnerable to any lateral static and dynamic loads including
addressed prevailingly to historical stone masonry structures those of earthquakes. While stones lasted for centuries and their
that represent a significant part of the world’s cultural heritage. deterioration was caused only due to weathering, water, wind
Any rehabilitation of these structures is conditioned by the and temperature impacts, nowadays increasing effects of air
exterior (sometimes also interior) appearance (Figs. 1 and 2). pollution, chemical and biological attacks speed up damage not
If the demand of bare stone masonry facade is the fundamental only to mortar but also directly to stones. The example in Fig. 1
condition, intervention measures are different from those that represents the case when the failure of stone masonry units
could be applied for facades covered by plaster. preceded that of mortar. If such places appear in lower parts
of either historical buildings or defensive walls, the essential
2. Stone masonry in historical structures problem of structure existence should be solved urgently.
For generations, stones, bricks and wood were basic 3. Simple masonry structure performance
construction materials in the majority of historical structures.
The stone masonry essentially differs from the clay brick Theory of strain and stresses in elastic and plastic bodies
masonry. Stone units are stiff and therefore unable to develop and media describes the conditions for the failure in general.
elastic deformations, like burned clay bricks, accumulating Properties of loading in space and time, and strength limits,
2196 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205
Fig. 3. Limestone used for static tests. The end of diagram shows an atypical start of failure due to local crushing of stone (force control mode of compression test).
Table 2
Properties of limestone samples
Test results (MPa)
Limestone cube compression tests — strength 67.31/68.36/19.32
mean/median/STD
Low strength limestone — standard compression 67/72/15
strength mean/median/STD
Limestone prism compression test — Young’s modulus 37 478
Low strength limestone — standard Young’s modulus 35 000
with the contribution of energy dissipation and redistribution
of carrying capability, control the ultimate resistance of a
structure.
The expected modes of a structure failure and related
analytical models should be estimated and described. In
general, a structural member can possess a number of
fundamentally different potential failure modes. The structural
behaviour can depend upon the effects of one or more
actions that will not be varied systematically. The response is
interconnected with the resistance of materials used, and both
can depend on time and duration effects, scale and size effects,
different environmental, loading and boundary conditions, and
material degradation effects [10–12,27].
Traditional materials in view of masonry remain popular and
frequently used also nowadays. Since loading is prevailingly
vertical and static, the system is able to sustain the effects
of self weight and imposed loads. Horizontal actions and
consequently shear forces and stresses affect both masonry
units and binding mortar. Cracks and failure start in the most
vulnerable parts either in stone units alone, in mortar or at
the interface between the stone and mortar. The properties of
stone masonry are investigated by laboratory and in situ tests,
as hammer tests, jacks, core sample tests, chemical analyses,
etc. The compression strength of healthy stone units is more or
less independent of direction. The shear strength and modulus
exhibit values generally in recommended standard ranges. The
values obtained on limestone samples are in Table 2 and in
Fig. 3. Compression tests were executed on cubes 20 × 20 ×
20 mm (6 samples) and prisms 20 × 20 × 40 mm (3 samples).
The mortar that is convenient for stone masonry should
allow plastic deformations. There exist a lot of mortars
with different properties. Some of them can be identified
and optimized for the wide range applications in historical
structures. Catalogues, material orientated standards, national
and international databases and producers’ offer are utilized
according the needs of a specific structure.
The European Standard EN 1996-1-1 2005 [10] refers to
the design of new masonry structures. It recommends that
masonry mortars for use in bed joint reinforced masonry should
not have a compression strength f m less than 2 MPa, and
for use in other types of reinforced masonry, not less than
4 MPa. Modulus of elasticity is variable, the recommended
value for mortars in masonry is E = 1000 f k , where f k
is characteristic compression strength. Shear modulus should
be 0.4E. Considering that in the case of horizontal loading
the mortar plays a dominant role, the reasonable value of
masonry modulus E m for the predicting analysis could be about
2000 MPa. The offer of different mortars is large; the codes,
technical specifications and databases give information on their
strength, adhesiveness, bending and shear resistance, and other
parameters. The appropriate new mortars are developed to
withstand the static and dynamic load effects.
The idea of improving mortar properties is in the use
of isotactic polypropylene fibres of type Dimapos as an
additive for mortars and plasters. The result is an increase in
ductility and tension/shear resistance. Supplementary studies
and experiments confirm the reasonable stiffness and strength
of such fibre lime–cement mortars. The mortar samples of
different composition and age were tested under compression
(cubes 20 × 20 × 20 mm), under bending (prisms 20 × 20 ×
80 mm) and under direct tension (prisms 20 × 20 × 80 mm),
with main results as shown in Table 3 [16–20]. LC — is
an abbreviation for lime–cement mortar (75% sand, 18.75%
lime and 6.25% cement); FLC — is an abbreviation for fibre
lime–cement mortar (77.8% sand, 9.65% lime, 12.5% cement
and 0.05% fibre).
The alternative reinforcing of masonry structural members
with polymer grids shows great potential. This work involves
the analysis of how the polymer grids contribute to the dynamic
performance of a structure, when they are appropriately inserted
into the horizontal bed mortar layers between stone blocks.
The blocks can be regular or irregular in shape. If further
the masonry, either plain or reinforced, is wrapped with
polymer grids and plastered, then it becomes a composite
 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205
Table 3
Properties of mortars and polymer grid
Test results
LC mortar compression tests (28 days–132 samples) — strength mean/median
FLC mortar for plaster (prism tension tests–6 samples) — Young’s modulus
FLC mortar for plaster (cube compr. tests–36 sampl.) — compression strength
FLC mortar for plaster (prism tension tests–6 samples) — tension strength
Polymer grid SS 20 – 1 layer for 240 mm thickness of wall — tension force at strain = 3%
with higher integrity and better behaviour under seismic or
other dynamic actions. Laboratory tests on wallets, masonry
walls, and shaking table tests, together with numerical analyses
validated this innovative method [21,23–25].
Polymer grids as reinforcement in bed layers prevent the
diagonal cracks and increase the shear-tension resistance of
mortar layer, as the mortar ductility is better controlled. It
is similar concept like in early times when wood or several
rows of clay brick masonry were inserted in stone masonry.
Actually, e.g. Byzantine and Ottoman stone masonry buildings
included horizontal layers of clay bricks in order to account for
settlements, that was resulting to increased ductility.
The rehabilitation of existing historical or other structures
calls for the complex strategic approach, utilizing range and
limits of stiffness and strength, both of original materials and
new composites prospective for use.
4. Limestone masonry model tested on the shaking table
A large, nearly 30 tons heavy, 3D asymmetrical model of
a structure made of natural stone masonry, reinforced with
polymer grids, was built and tested on a shaking table in two
phases: In the first phase the masonry received polymer grid
reinforcement only in horizontal bed layers. After a series
of tests involving seismic loading, and when a reasonable
damage level was reached, the model was repaired by externally
bonding the polymer grid with special screws and a fibre
lime–cement plaster. Then it was tested again in the second
phase with increasing seismic loading. The test campaign ran
on the LNEC Lisbon 3D shaking table. An alternative of
the model, without any reinforcement and representing the
original historical stone masonry structure, was not built and
tested because of available financial sources. However, such
research has been executed on large brick masonry models,
with the further retrofitting only by vertical polymer grids and
fibre lime–cement plaster [16,18,19,24,25]. The results can be
extrapolated also for stone masonry structures.
The stone masonry model was designed as an asymmetrical
structure with two windows on the east side, a door on the
south side, one window on the west side, a solid wall on the
north side and an indent in the north-west corner. No floor slab
was provided. The lintels above the opening were replaced by
application of a single strip of polymer grid in the mortar layer.
The model was built in full scale 1:1, dimensions in plan were
4010 × 3580 mm, its height was 3600 mm. The thickness of
the walls was 240 mm. The dimensions of the model in plan are
given in Fig. 4.
2197
(MPa)
0.8045/0.80
1125
10–14
1.6–1.9
2.5 kN (ultimate 6 kN)
Fig. 4. Dimensions of the stone masonry model in plan.
The original stone masonry model was not plastered. The
weak lime–cement mortar had a compression strength of
0.85–1.0 MPa. The limestone used has mean compression
strength of 67 MPa and Young’s modulus of 37 GPa, higher
than that of concrete. These characteristics correspond to the
low strength limestone, as shown in Table 2.
The measurement points on the model were chosen to
monitor the essential absolute accelerations and chosen relative
displacements. The scheme of the model with the position
of the measurement points is in Fig. 5. Design and drawings
were elaborated in University of Bucharest, after consulting
and agreement of design with ICA SAS Bratislava and LNEC
Lisbon. The model was built and tested in the laboratories of
LNEC Lisbon (Fig. 6).
Before starting the tests, preliminary calculations of the
natural frequencies were carried out in ICA SAS Bratislava and
LNEC Lisbon, applying a number of finite element calculation
codes and different degrees of FE mesh refinement. Results
of calculations showed the space effect in natural modes with
significant out-off-plane deformation of walls. The apparent
modulus of elasticity used in these calculations was E m =
2100 MPa, and the Poisson’s ratio ν as equal to 0.1. The
estimate of E m was based on experience with other masonry
structures, previously analysed and tested. Fig. 7 illustrates
shapes of the first five natural vibration modes. Calculated
and experimental values of respective natural frequencies are
given in Table 4 for the original model and, in Table 5 for the
repaired one. ρ is the considered unit mass, d is the thickness
2198 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205

The tests of stone masonry model consisted of phases and

Fig. 5. Position of accelerometers on the tested model. I1, I2, I3 and I4 indicate
points of applied hammer impacts.
of walls. Calculated values are those estimated before tests.
More detailed analytical models and a sensitivity study are in
progress.
Experimental natural frequencies were determined from
results of impact, sweep sine and ambient tests, using the
transfer functions and experimental modal analysis. The
“spring” effect of shaking table actuators on natural frequencies
was investigated by using “down” and “up” table positions
during impact tests. Some natural modes were not sufficiently
excited, due to the position of a corresponding impact point.
subphases.
Phase 0 — originally not prescribed. The original model
without any plaster and girders, stiffened only by polymer grids,
was transported from the place of construction to the shaking
table by crane; this led to an accident. One side of the support
failed, and the model fell on the floor, thus being subjected to a
stress state resembling the case of uneven settlement. The result
was the pattern of cracks illustrated in Figs. 8 and 9. They
appeared above the door on the south side and another shear
crack propagated from the window on the east side to the base
on the north side. Should it had been a model without polymer
grids in bed joints, the model would have been totally damaged.
Phase 1 — shaking table tests of the original model without
plaster. Here, through subphases the sequence of tests was the
following:
(a) Impact tests with acting force in chosen points I1 to
I4 (Fig. 5); During these tests the model was hit by a
wooden hammer successively near A7, A29, A9, A10. This
procedure was repeated with sufficient time breaks and the
table in parking position — down; and in operating position
— up. Response time histories were recorded with sampling
frequency f s = 1000 Hz.
(b) Series of harmonic sweep sine tests; During each sweep
sine test the model was excited in one direction x or y,
respectively, by sine seismic input of a constant amplitude
0.1 m s−2 or 0.2 m s−2 in a frequency range from 5 to 35 Hz.
The result was basic information about natural modes and
frequencies of the model connected to the shaking table
under excitation along a single direction. Used sampling
frequency f s = 400 Hz.

Fig. 6. One-storey stone masonry model before tests: (a) Original model — grids only in prescribed bed joints; (b) Repairing procedure; (c) Repaired model —
added vertical grids inside of plaster. The thickness of repaired walls increased to app. 32 cm, due to the rough surface of stone masonry walls.

Fig. 7. Calculated natural modes of vibration of the original stone masonry model.
 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205 2199

Table 4
Calculated and experimental values of natural frequencies — original model

Mode j 0 1 2 3 4 5

Calculated frequencies (Hz) 11.29 14.84 17.23 20.34 24.68

E m = 2100 MPa; ν = 0.1; ρ = 2 t/m3 ; d = 24 cm
Experimental frequencies (Hz) from tests: 10.20 11.71 16.00 18.76 23.30
Impact I1 – x down (A2-mid)
Impact I1 – x down (A3-corner) 10.83 11.59 – – 22.67
Impact I1 – x up (A2-mid) 8.31 12.22 15.36 – 22.04
Impact I1 – x up (A3-corner) 7.68 10.08 15.11 19.02 23.80
Impact I4 – y down (A5-mid) 10.45 11.96 16.62 – 22.17
Impact I4 – y down (A4-corner) 10.32 14.86 17.50 – 21.41
Impact I4 – y up (A5-mid) 8.69 11.84 14.99 18.77 22.54
Impact I4 – y up (A4-corner) 8.31 13.22 14.36 18.51 22.29
E imp mean,exp (MPa) 1121 1337 1582 1794 1861

Down — means parking position of the shaking table; up — means ready for operation position of the shaking table.

Table 5
Calculated and experimental values of natural frequencies — repaired model

Mode j 0 1 2 3 4 5

Calculated frequencies (Hz) 13.33 16.15 18.79 19.78 27.53

E m = 2100 MPa; ν = 0.1; ρ = 2.4 t/m3 ; d = 32 cm
Experimental frequencies (Hz) from tests: 5.18 8.41 11. 5 13.24 – 24.54
Impact I1 – x up (A2-mid)
Impact I2 – x up (A4-corner) 5.18 7.03 10.7 – 19.70 –
Impact I4 – y up (A4-corner) – – 11.45 – – 24.46
Ambient – – – 13.55 17.62 20.89
E imp mean,exp (MPa) – 704 1013 1067 1869 1504

Fig. 8. Cracks in the original stone masonry model: (a) south side; (b) east side. Solid (blue) lines are cracks after the transport. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article.)

(c) Series of ambient (like white noise) excitation in one
direction x or y, respectively; Records and analysis results
were obtained by similar methods used in the sweep sine
tests. Sampling frequency f s = 400 Hz.
(d) Seismic tests with simultaneous earthquake excitation in
two directions (x, y); The earthquake input started from
low level of earthquake intensity and it was successively
increasing up to the significant appearance of cracks in
masonry. Earthquake like response time histories were
recorded with sampling frequency f s = 400 Hz. Maximum
seismic input acceleration (equivalent of PGA — peak
ground acceleration) reached values 0.36g and 0.34g,
for directions x and y, respectively. The related input
accelerograms along with their 5% response spectra are in
Fig. 14 — left.
During impact tests, only accelerations were recorded.
Remaining tests included also measurements of relative
displacements — “strains” above openings on east and south
sides of the model. The acceleration records of seismic response
in representative points on the model (Fig. 5) and input
motion of the shaking table allow the calculation of model
to shaking table relative deflection time histories. Deflections
give also information on out-of-plane vibration of walls. It
was confirmed that polymer grids in bed joints supported the
2200 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205

Fig. 9. Cracks on the most damaged parts of the walls: the original stone masonry model after shaking table tests. 0.02g (1) — means the first sweep sine test in x
direction.

Table 6
Key values of the seismic response of stone masonry model (original and repaired)

Exp. natural frequency Estimated damping Maximum input Maximum top rel. displ. Maximum top abs. Estimated max. base shear (kN)
(Hz) ratio acc. (g) (mm) acc. (g)
Stone masonry model 1:1, original, only grids above openings (m tot = 34.8 t), the seismic input based on Herceg Novi earthquake 2D x, y (polymer
grids in bed joints)
f 1x = 8.31 0.05 0.359 S-16.72; N-18.74 S-1.211; N-1.033 127.6
f 1y = 11.84 0.05 0.339 E-26.20; W-10.39 E-0.780; W-1.003 101.4
f 1θ z = 23.80 0.05 – –

Stone masonry model 1:1, repaired (m tot = app.42 t), the seismic input based on Herceg Novi earthquake 2D x, y (polymer grids in bed joints and
vertical polymer grids inside of fibre lime–cement plaster)
f 1x = 8.41 0.05 1.29 S-9.72; N-8.14 S-1.866; N-1.763 249.2
f 1y = 11.46 0.05 1.04 E-13.61; W-13.47 E-2.187; W-1.735 289.37
f 1θ z = 24.46 – – – – –

integrity, however, they gave the allowance for comparatively
high out-of-plane deflection amplitudes (Fig. 10 and Table 6).
In the majority of existing old structures, the wooden floor
systems are comparatively flexible. Wooden floors influence
the dynamic response and behaviour both in plane and out-
of-plane. This can also bring some stability problems [13–15].
The partial help is coming from polymer grids in bed joints, as
they support by friction the individual disconnected stones. The
substantial improvement can be reached by using of vertical
grids.
Phase 2 — shaking table tests of the repaired model with
vertical grids and fibre plaster. The sequence of tests was
slightly different from phase 1, and it was as follows:
(a) Impact tests of model with acting force in chosen points
of upper part of model near A7, A29, A9, A10; Response
time histories were recorded with sampling frequency f s =
2000 Hz:
(b) Series of ambient (like white noise) excitation in one
direction x or y, respectively; used sampling frequency
f s = 400 Hz.
 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205 2201

Fig. 10. Time histories of horizontal relative deflections — the original stone masonry model. Achieved maximum seismic input: PGA xS = 0.359g; PGA yE =
0.339g.

Fig. 11. Cracks on the most damaged parts of the walls: the repaired stone masonry model after shaking table tests.
2202 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205

Fig. 12. Cracks in the repaired stone masonry model: (a) south side; (b) east-north corner.

Fig. 13. Time histories of horizontal relative deflections — the repaired stone masonry model. Achieved maximum seismic input: PGA xS = 1.290g; PGA yE =
1.040g.

(c) Seismic tests with simultaneous earthquake excitation in
two directions (x, y); The earthquake input started from
low level of earthquake intensity, and it was successively
increasing up to the limit of shaking table. Used sampling
frequency f s = 400 Hz. Maximum input acceleration
achieved values 1.29g and 1.04g for directions x and y,
respectively. The corresponding input accelerograms along
with their 5% response spectra are in Fig. 14 — right.
(d) Repeated series of ambient (like white noise) excitation in
one direction x or y, respectively; used sampling frequency
f s = 400 Hz.
The pattern of cracks indicates that the damage of repaired
model was accumulated in the lower parts of walls, near the
base plate (Figs. 11 and 12).
The seismic response of the repaired model was different
than that of the original model. Fig. 13 shows the respective
horizontal relative deflections of the repaired model against
the shaking table. The out-off-plane vibration of walls nearly
disappeared (Table 6).
In Table 7 presented vector extremes Acc xy, Vel xy mean
maximum values of space vectors of seismic inputs verified in
time domain. Critical frequencies Crit f A were calculated under
assumption that maximum acceleration and maximum velocity
are bounded nearly as in harmonic vibration. Critical ranges of
frequency Crit f B were determined on the basis of calculated
acceleration response spectra and represent the range of natural
frequencies at which maximum local stresses in the structure
can be expected (Fig. 14).
 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205 2203

Fig. 14. Time histories and response spectra of actual extreme seismic inputs: left — the last test of the original model; right — the last test of the repaired model.

Table 7
Seismic input extremes — achieved during the tests of original and repaired stone masonry model

Required max acc. Acc x (g) Acc y (g) Acc z (g) Acc x y (g) Vel x (mm/s) Vel y (mm/s) Vel z (mm/s) Vel x y (mm/s)
Original model f s = 400 Hz; f p = 0.8 Hz;
0.25g 0.359g 0.338g – 0.359g 158.853 181.383 – 192.652
Crit f A (Hz) 3.528 2.908 2.912
Crit f B (Hz) 2–5 2–5

Repaired model f s = 400 Hz; f p = 0.8 Hz;

0.9g-R 1.289g 1.036g – 1.292g 414.649 509.277 – 542.666
Crit f A (Hz) 4.854 3.176 3.717
Crit f B (Hz) 2–5 2–5

Original, measured quantities were accelerations. Corre-
sponding velocities were calculated using the numerical step in-
tegration procedure. The integration was applied together with
the Butterworth high-pass filter. The sampling frequency f s was
given from measurements, the cut-off frequency f p was chosen
to be sufficiently low in comparison with the frequency con-
tent of records. The selection of filter was verified by Matlab
Toolbox procedures.
It can be seen from Table 7 that differences between
the unidirectional seismic input maximum and the space
vector seismic input maximum are slightly less than those
recommended in standards. This is valid for all tests and
is partially influenced by the used seismic input, mass and
dimensions of the model, and the shaking table properties.
5. Limits of the dynamic resistance of original and repaired
stone masonry
Each presented and tested model configuration has its
dynamic resistance limits. The same concerns existing old
stone masonry structures. The development of cracks and
partial damage decrease values of natural frequencies and
consequently influence the seismic response time history.
The analytical estimate of the resistance can be based
either on the secant modulus approach or on the ideal
elastic–plastic nonlinear model [15,22]. The allowable space
for a dynamic resistance capacity is thus determined by the
natural frequencies and damping variation and the strength
and displacement limits. However, the tested models reached
their ultimate state through the limit strength and integrity. It
should be understood that the integrity of masonry structure
and its components does not mean the monolithic state without
any crack. The cracks are allowed but limited. The main
achievement of the presented retrofitting (also upgrading)
procedure is in the redistribution of stresses and in the
prevention of collapse.
The response of a structure is a product of global and
local stress–strain states resulting from the interaction of
natural modes/frequencies and seismic input properties in
time and space [6,17,20]. It can happen that the response
to unidirectional seismic excitation is higher than that one
to multicomponent excitation, and/or one mode response can
overrun multimodal response. The more detailed verification is
needed, if the structure is susceptible to a sudden initiation of
local and/or global damage.
The relation of maximum seismic response and maximum
seismic input can but need not be influenced by varying
natural frequencies. The changes of natural frequencies and
2204 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205
Fig. 15. Application of polymer grids combined with fibre lime–cement mortar/plaster in stone masonry structures.
total mass can be less important, and the main player is the
integrity of a structure. Therefore, the response of repaired
model exhibited smaller increments of absolute accelerations
and relative deflections in comparison with the response of
original model (Fig. 13 and Table 6).
Table 6 gives the review of achieved seismic response
extremes during the original and repaired model tests. Global
limits, usually presented by the total base shear force versus
top relative displacement relationship, are not fully compatible
with local stresses of a system. Especially, this must be
taken into account when analysing the original stone model
behaviour. The acceleration response ratio (maximum response
acceleration to PGA) recorded variation only about two or
fewer (Fig. 14). This phenomenon suggests the effect of large
masses and dimensions of tested stone masonry models.
The acceptable integrity was observed already during the
original model tests. After repair, the critical regions have
changed the position. The dissipation of energy has run
near the base mat corners and only slight cracks appeared
throughout the model. In general, the very good integrity
and the excellent increase of the seismic resistance capacity
underline the behaviour of the repaired model.
6. Discussion on intervention measures
The results have shown remarkably progressive information.
The overall behaviour of reinforced stone masonry model
has discovered several important features. Originally fragile
composition of stone units and mortar after introduction of
horizontal grids has changed the behaviour into nearly “flexible
one”. The seismic response of the original model included
contribution of large deformation of walls in out-of-plane
direction, resembling the behaviour of shells. Even, what is
important, the integrity of walls near openings has remained in
acceptable limits and no collapse appeared. The achieved PGA
was 0.36g. The second phase of tests presented nearly “stiff
box” behaviour of the repaired model and a very high seismic
resistance. The achieved PGA was more than 1g. The observed
mechanisms of dynamic response have opened the door for
practical applications in stone masonry structures, either new
or repaired.
The practical use of the retrofitting procedure is intercon-
nected with the requirements on the service properties and the
appearance of a structure (Fig. 15). Thus, for new/old structures
there exist options:
(a) only fibre lime–cement mortar — new and partially old
structures;
(b) horizontal polymer grids + fibre lime–cement mortar —
new and partially old structures;
(c) vertical polymer grids + fibre lime–cement plaster — any
new and old structures;
(d) both horizontal and vertical polymer grids + fibre
lime–cement mortar/plaster — new structures.
The horizontal grids in chosen bed joints of old stone
masonry can be considered for the repair of ruins and upper
parts of defensive walls. However, in the other parts it means
using the rather tedious technique of lateral inserting of grids,
combined with the mortar injection.
7. Conclusions
In this paper few aspects of material properties and the
seismic response estimate of stone masonry structures are
depicted. The composite system of a plain stone masonry
structure presents the nonlinear dynamic response, which is
accompanied with the damping increase and reversible or
irreversible changes of stiffness. The local and total failure can
appear suddenly as the result of extensive vibration and a low
resistance capacity. The vulnerability can be mitigated by the
proper choice of upgrading materials and their combination. In
the analysis, the apparent masonry modulus of elasticity in the
 E. Juhásová et al. / Engineering Structures 30 (2008) 2194–2205
range of E m = 1000–2000 MPa and Poisson’s ratio ν = 0.1
seem to be appropriate for the assessment of initial natural
frequencies and modes of vibration. The tools, which are
used in the nonlinear response analysis, should be reasonably
selected to estimate the stone masonry structure resistance
capacity. If ductility is supposed to be beneficial, the cracks can
be accepted, provided that neither loss of structure integrity nor
instability can appear. The proper detailing and the appropriate
measures for increasing the total resistance capacity are key
design and execution activities in the construction industry.
The described technology of retrofitting underlines the
integrity role in the enhanced dynamic resistance capacity of
stone and other masonry structures. The fibre lime–cement
plaster even participates on the durability increasing. It
is proved the high efficiency of polymer grids and fibre
lime–cement mortars at acceptable costs. This supports using of
similar interventions in existing structures including historical
ones. Mentioned measures have wider field of applications.
Not only safety aspects related to natural earthquake disasters
or terrorist hazards, but also protection against undesirable
transport effects and settlements can be reasonably and
conveniently solved by the presented intervention procedures.
Acknowledgements
The research was mainly funded by European Commission
in the framework of ECOLEADER Project HPRI-CT-2002-
00182 and partially by VEGA Projects 2/4078/24 and
2/7114/27. The financial support is gratefully acknowledged.
References
[1] Abrams DP. Performance-based engineering concepts for unreinforced
masonry building structures. Progress in Structural Engineering and
Materials 2001;1:48–56.
[2] Aničić D, Fajfar P, Petrović B, Szavits-Nossan A, Tomažević M.
Earthquake engineering. Beograd: Gradevinska knjiga; 1990.
[3] Bairrao R, Bursi O, Carydis P, Magonette G, Mouzakis H, Tirelli D,
et al. Benchmark testing and performance comparison of shaking tables
and reaction walls. In: Proceedings of the 13th world conference on
earthquake engineering. Vancouver: UBC; 2004. Paper 441.
[4] Bairrao R, Falcao Silva MJ, Juhásová E, Campos Costa A, Coelho
E. Shaking table tests of an asymmetrical limestone building. In:
Proceedings of the 1st European conference on earthquake engineering
and seismology. 2006. Paper 636.
[5] Coburn A, Spence R. Earthquake protection. London: John Wiley & Sons;
2002.
[6] Coelho E, Campos Costa A. Seismic testing and vulnerability assessment.
In: Proceedings of the Portuguese–US workshop on seismic behaviour of
masonry structures. 2004.
[7] Coias e Silva V. Technical feasibility of the national programme for
reducing the seismic vulnerability of the building stock. In: Reducing the
2205
seismic vulnerability of the building stock, NSE Lisbon, April 1, 2001.
Lisbon: LNEC; 2001. p. 15–56.
[8] Colombo A, Negro P. Advanced materials for strengthening, rehabilita-
tion and repair of structures. experimental activities in ELSA. In: Pro-
ceedings of the workshop mitigation of seismic risk support to recently
affected European countries. Ispra: JRC EC; 2000. 5/1–7.
[9] EN 1990: 2002 Eurocode. Basis of structural design. Brussels: CEN.
[10] EN 1996-1-1: 2005 Eurocode 6. Design of masonry structures. Part 1-
1: Common rules for reinforced and unreinforced masonry structures.
Brussels: CEN.
[11] EN 1998-1: 2004 Eurocode 8. Design of structures for earthquake
resistance. Part 1: General rules, seismic actions and rules for buildings.
Brussels: CEN.
[12] EN 1998-3: 2005 Eurocode 8. Design of structures for earthquake
resistance. Part 3: Assessment and retrofitting of buildings. Brussels:
CEN.
[13] Griffith MC, Magenes G, Melis G, Picchi L. Evaluation of out-off-plane
stability of unreinforced masonry walls subjected to seismic excitation.
Journal of Earthquake Engineering 2003;7(1):141–69.
[14] Javed M, Naeem Khan K, Penna A, Magenes G. Behaviour of masonry
structures during the Kashmir 2005 earthquake. In: Proceedings of the 1st
European conference on earthquake engineering and seismology. 2006.
Paper 1077.
[15] Juhásová E. Seismic effects on structures. Amsterdam: Elsevier; 1991.
[16] Juhásová E. Experimental investigation of technical interventions to
reduce seismic vulnerability of old existing buildings. Bratislava: ICA
SAS; 1996.
[17] Juhásová E, Coelho E, Bairrao R, Costa A, Virsta A, Sofronie R.
Enhancing the performances of natural stone masonry. In: Proceedings
of IASS 2005 conference. vol. 2. Cluj-Napoca; 2005. p. 959–66.
[18] Juhásová E, Hurák M, Zembaty Z. Assessment of seismic resistance of
masonry structures including boundary conditions. Soil Dynamics and
Earthquake Engineering 2002;22(9–12):1193–7.
[19] Juhásová E, Pezzoli P, Da Rin EM, Sofronie R, Zembaty Z. Resistance
of brick building model with arches before and after retrofitting.
In: Proceedings of the 11th European conference on earthquake
engineering. Rotterdam: Balkema; 1998. Paper 180.
[20] Juhásová E, Sofronie R, Juhás M. The key role of structure integrity and
new materials. In: Proceedings of the conference building for European
future, vol. 1. Brussels: ECCREDI; 2004. p. 527–44.
[21] Krstevska L. Development and application of non-Linear micro models
for evaluation of seismic behaviour of rc frames infilled with plain and
reinforced masonry. University of Skopje reports. Skopje: IZIIS; 2002.
[22] Paulay T, Priestley MJN. Seismic design of reinforced concrete and
masonry buildings. New York: John Wiley & Sons; 1992.
[23] Severn R, Juhásová E, Franchioni G, Popa G, Sofronie R. Mitigation of
seismic risk by composing of masonry structures. In: Proceedings of the
workshop mitigation of seismic risk support to recently affected European
countries. Ispra: JRC EC; 2000. 22/1–7.
[24] Sofronie R. Performances in seismic strengthening of masonry.
In: Proceedings of the 13th world conference on earthquake engineering.
Vancouver: UBC; 2004. Paper 182.
[25] Sofronie R, Juhásová E, Greening P. Seismic strengthening of masonry.
SECED Newsletter 2003;16(4):6–7.
[26] STN 73 0031: 1989 Civil Engineering Structures and Foundations. Basic
rules for analysis. Prague: VUNM.
[27] ISO DIS 13823 General Principles on the Design of Structures for
Durability. Draft 2006. Geneva: ISO.