Construction and Building Materials 25 (2011) 303–307

Contents lists available at ScienceDirect

Construction and Building Materials

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

The effect of scale on the structural behaviour of masonry under compression

A. Mohammed a,*, T.G. Hughes b,1, A. Mustapha c
a
Civil Engineering Programme, Abubakar Tafawa Balewa University Bauchi, PMB 0248, Bauchi, Bauchi State, Nigeria
b
Civil Engineering Department, Cardiff School of Engineering, Cardiff CF24 0YF, UK
c
Civil Engineering Department, University of Maiduguri, PMB 1069, Maiduguri 600001, Borno State, Nigeria

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

Article history: This paper discusses the effect of scale or size effect on the structural behaviour of masonry under com-
Received 3 September 2008 pression. Small scale modelling has been employed as a means to understanding masonry structural
Received in revised form 9 April 2010 behaviour over the years, because testing prototype masonry structures is both costly and difficult to
Accepted 7 June 2010
control in terms of instrumentation. An experimental programme involving tests at four scales namely
Available online 2 July 2010
prototype, half, fourth and sixth scale were undertaken under compression with a view to understanding
how compressive action affects masonry structural behaviour over the range of scales under consider-
Keywords:
ation. The results show that masonry compressive strength is increased as the scale is reduced but the
Masonry
Model scale
stiffness is not significantly affected.
Compressive strength Ó 2010 Elsevier Ltd. All rights reserved.
Stiffness
Size effect

1. Introduction
Small scale masonry model testing has been carried out for
many decades. Early researchers in this area include [1–3]. Tests
by these authors has established that it is possible to model ma-
sonry behaviour at reduced scales but not the strength and stiff-
ness, this has also been reported by Refs. [4,5]. Recent interest in
masonry modelling has arisen because of the need to assess and
maybe strengthen existing historic masonry structures like bridges
and buildings. For instance there are over 40,000 masonry arch
bridges in the UK. Most of these bridges are over 100 years old,
while some are as old as 500 years. Increasing traffic speeds and
weights have made assessing both the ultimate and serviceability
requirements of these bridges necessary.
Furthermore there is also the need to understand the structural
behaviour of masonry structures under extreme natural events like
windstorms, floods, earthquakes etc., since some of these events,
like flooding have become recurrent actions posing danger to thou-
sands of people inhabiting or working in masonry structures. For
example in January 2005, up to £250 m worth of damage was
caused by flooding due to very heavy rainfall in Carlisle, England
[6]. In Iran an earthquake in Bam killed about 35,000 people and
flattened about 90% percent of the city’s mainly masonry houses
[7]. Because of the issues associated with the cost implications of
* Corresponding author. Tel.: +234 80 35145747.
E-mail addresses: abba_gm@yahoo.com (A. Mohammed), Hughestg@cardiff.
ac.uk (T.G. Hughes).
1
Tel.: +44 2920 875716.
full size masonry tests, coupled with the danger of instrumentation
destruction at failure, repeatability and difficult boundary condi-
tions it has become increasingly necessary to carry out such tests
at reduced scales.
One aspect of this study is reported in this paper, looking at ma-
sonry behaviour under compressive loading at prototype, 1/2, 1/4
and 1/6 model scales. This paper primarily aims to compare ma-
sonry structural behaviour at prototype and small scales.
2. Experimental design
The aim of this research as it has been stated is to compare
experimentally brickwork structural behaviour at prototype and
model scales. However a reference scale was needed that provides
significant opportunities for modelling overall structural behaviour
of, for example arch bridges, whole buildings, retaining walls etc.,
as well as for undertaking the parametric study. Ref. [8] has sug-
gested twelfth scale as a limiting scale in small scale masonry
modelling while other authors [2,4,9] have reported reasonable
model to prototype scale correspondence using half, third, fourth
and sixth scale model masonry. An examination of the different
factors led to the choice of the sixth scale as the limiting scale
for this research, because it is most suitable for modelling whole
masonry structures and components in a controlled laboratory
environment. Two further scales of half and fourth scale were also
investigated for the first programme of tests in order to get a com-
plete picture of masonry behaviour across the range of scales. It
was intended that material tests would be first carried out to

0950-0618/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.conbuildmat.2010.06.025
304 A. Mohammed et al. / Construction and Building Materials 25 (2011) 303–307
determine the properties of the constituent materials that make up
the masonry; the bricks and mortar e.g. compressive strength and
flexural strength etc. Knowing the individual material properties
will aid in understanding the composite behaviour of the masonry
assembly; by determining how properties of the components con-
tribute to overall brickwork behaviour under various conditions of
loading.
In order to fully understand the composite behaviour of the ma-
sonry specimens, consideration was made of how best to capture
the structural behaviour of a masonry structure, because in any
masonry structure, its various elements are under the influence
of a variety of actions. For instance a masonry wall could at any
point be under a vertical compressive load, out-of-plane and or
in-plane lateral forces, etc. Even though most of these forces act to-
gether, it is better to isolate each one to understand the mechanics
of its action. Therefore to understand the fundamental behaviour
of masonry, each of the actions was taking separately and consid-
ered for each of the chosen scales. Consequently in order to under-
stand the behaviour of masonry under some of the most common
actions to which it is subjected to, the following tests were chosen;
compressive strength, flexural strength, shear strength and bond
strength tests. Only the effects of scale on the structural behaviour
of masonry in compression is reported here since that is the pri-
mary aim of the paper, in depth discussion on the other effects
are discussed in [10].
2.1. Choice of bricks and mortar
Refs. [8,10] have reported some issues regarding the firing of re-
duced scale model clay bricks. These issues included the fired mod-
el bricks being irregular in shape as well as some bricks more burnt
than the others. Because of these problems a cutting procedure
developed and used successfully by Refs. [5,8] was employed for
the manufacture of the model clay bricks from a standard solid pro-
totype clay burnt brick of approximately 215  102.3  65 mm.
Cutting dimensions of the model bricks are shown in Table 1. The
table shows that the dimension for a ‘‘half” scale brick is 0.45 times
that of a prototype brick. This is because obtaining the ideal 0.45
scale factor will result in getting only one half scale brick from a
prototype brick, which implies a lot of wastage. Thus the 0.45 factor
was used which resulted in getting eight half scale bricks from a
prototype brick. The selection of the prototype brick was made
based on consideration of its suitability for cutting considering fac-
tors like lack of manufactured voids, strength, internal structure
(whether full of cracks or not) and ease of cutting. A mortar of des-
ignation (iii) equivalent to M4 was selected as a suitable mortar be-
cause of its medium compressive strength. Two types of sands were
used in making the mortars; normal builder’s sand for the proto-
type mortar and finer sand, Congleton HST 95 with an average grain
size of 130 lm for the model scale mortar. This sand was chosen be-
cause of the very small joints in the 1/6 scale (1.6 mm) and 1/4 scale
(2.5 mm). Further considerations on the experimental design are
provided in [10].
The compressive strength test was chosen because masonry is
loaded in compression in the majority of the situations under
which it is used. Therefore it is necessary to understand the
Table 1
Average dimensions of prototype and model bricks.
Scale Length (mm) Width (mm) Depth (mm)
Prototype 215.0 102.5 65.0
Half 96.8 46.1 29.3
Fourth 53.8 25.6 16.3
Sixth 35.8 17.1 10.8
mechanics of brickwork in compression for both prototypes and
models. In addition to the compressive strength, the stiffness of
the masonry under compression would also be investigated in or-
der to compare the stiffness behaviour of prototype and model
scale masonry.
3. Materials and methods
3.1. Materials
The bricks used in the research were solid burnt clay bricks and some of its
important mechanical properties are given in Table 2. Normal building sand was
used for making the mortar in the prototype tests while HST 95 Congleton sand,
a very fine sand with an average grain size of 120 lm was used for making the mor-
tar in the model tests, the gradings of the two sands is shown in Fig. 1 in relation to
the grading limits set by BS13139, it shows that the normal building sand is within
the grading limits and the model sand is within the sixth scale grading limit (set by
simply diving the sieve sizes for the main grading limits by a factor of 6). Some of
the important material properties of the mortars are summarised in Table 3. Other
constituents used for making the mortar are ordinary Portland cement (OPC) and
hydrated lime.
3.2. Construction procedure
The bricks were first pre-wetted by totally immersing in a water tank for 20 min
before laying them on their sides in a horizontal position as is usually done in pre-
fabricated masonry panels. Pre-wetting of the units was necessary so as to condi-
tion the suction properties of the units in order to achieve a good bond between
the units and mortar bed. The horizontal method of construction was employed
for both the prototype and models, in order to achieve a repeatable and controllable
way of making the specimens since the traditional laying method is amenable to
significant workmanship variations that could mask the structural behaviour of
brickwork. Another benefit of this method of construction is cancelling of the differ-
ential compaction of the mortar beds during curing because of the different weights
of the units across the four scales. Otherwise some form of simulating gravity stres-
ses would have to be resorted to; for example by undertaking the tests in a centri-
fuge, which could bring about complexities into testing programme. The bricks
were separated by wooden spacers cut to the desired mortar bed joint thickness
of 10 mm in the prototype tests, while tile spacers of 5 mm, 2.5 mm and 1.6 mm
were used for the 1/2, 1/4 and 1/6 model scales, respectively. Mortar was then
placed into the bed spaces defined by the spacers whilst the units were laid on their
sides in specially made moulds placed on top of a vibrating table. The whole assem-
bly was vibrated gradually as the mortar was placed to fill up the bed spaces de-
fined by the spacers used. This method of construction ensured uniformity in the
bed joints of the specimens. A picture of the triplet specimens in the four scales
is shown in Fig. 2.
3.3. Test procedure
Three brick high triplet specimens were tested in the in load controlled testing
machine at a loading rate of 0.18 kN/s, 0.04 kN/s and 0.02 kN/s in the prototype, 1/2
and 1/4 scales, respectively. In the 1/6 scale the test was undertaken at a displace-
ment control of 0.006 mm/s, this was because testing in load control was not
achievable in that particular testing machine which was most suitable for testing
the 1/6 scale specimens. The chosen displacement rate was arrived at by first con-
ducting trial tests to determine the time taken to achieve failure in the 1/6 scale
tests, and then adjusting the rate of displacement in order to achieve failure in a
comparable time to the load controlled tests.
Linear variable displacement transducers (LVDTs) were attached to both faces
of the specimens in the prototype, 1/2 and 1/4 scale specimens for the determina-
tion of their stiffness properties, while specially made model masonry clip gauges
(MMCGs) were used in the 1/6 scales because of the small size of the specimens.
Details of the clip gauges are given elsewhere [5]. The stiffness of the masonry
was calculated as the secant modulus at a third of the maximum stress reached.
Table 2
Mechanical properties of prototype and model bricks.
Test Prototype Half Fourth Sixth
Compressive strength (N/mm2) 29.2 30.6 41.9 47.4
COV (%) 14.3 8.4 9.7 32.7
Modulus of elasticity (N/mm2) 11,500 – – –
COV (%) 31 – – –
Poisson’s ratio 0.06 – – –
COV (%) 48.1 – – –
 A. Mohammed et al. / Construction and Building Materials 25 (2011) 303–307 305

the triplet compressive strength across the four scales, it is seen

Cummlative % passing

100
from the fitted trend line in the figure that generally there is an in-
80 crease in the compressive strength as the scale is reduced as re-
60 ported by various authors. There is also evidence of strength
40 anisotropy, because it is seen that the masonry strength in the half
20 and prototype scales are somewhat similar on one hand and in the
0 fourth and sixth scales on the other. This supports findings by Ref.
0.01 0.1 1 10 [10] on the possible anisotropy of masonry unit strength as well as
Sieve size (mm) other authors [12,13]. This behaviour could be due to the manufac-
Fine Limit Coarse Limit 1/6 FineLimit 1/6 Coarse Limit turing process of extruded clay units in which plastic clay is forced
HST 95 HST 60 Sharp sand through a die, thereby possibly bringing about anisotropy in the
unit properties because the extrusion process. The orientation of
Fig. 1. Grading curves for prototype and model sands within the BS limits.
the model units in relation how there were cut from the prototype
is shown in Fig. 4. The figure shows that the fourth and sixth scale
Table 3 models have similar orientation while the half scale model has the
Properties of prototype and model mortar (COV in brackets). same orientation to the prototype. This reinforces the earlier prep-
HST95iii MP
osition that there is a possible effect of anisotropy on the unit
strength because the similarities in the orientation of the units cor-
Mortar designation iii iii
respond with the similarities observed in the unit strength.
Vol. proportions 1:1:6 1:1:6
W/c ratio 1.8 1.55 In order to reduce the effect of the unit strength, the masonry
Water retentivity (%) 86 84 strength in Fig. 3 is re-plotted with the masonry strength norma-
Consistence retentivity (%) 55 55 lised with respect to their respective strength as shown in Fig. 5.
Comp. str. (N/mm2) 4.7(8.5) 5.0(1.8)
This figure shows a reduced rate of increase of masonry strength
Flexural str. (N/mm2) 2.3(5.3) 1.8(11.8)
Modulus of elasticity (N/mm2) 6500(8.7) 6300(17.8) as the scale is reduced. This suggests there is still some additional
Poisson’s ratio 0.10 0.12 factor that is responsible for the high increase in strength, apart
from the effect of unit strength alone.
Studies of the fracture of brittle materials (mostly concrete)
have shown that smaller sized specimens have higher strengths
4. Results and discussion than larger specimens [14]. According to the Griffith theory of brit-
tle fracture, the less the surface area of a material the stronger it is,
4.1. Compressive strength since there is less probability of flaws occurring [15]. Therefore this
size effect seen in testing brittle materials may also be one of the
Typical failure of the specimens across the four scales is by ten- reasons for the strength differences. But curiously there is little
sile splitting cracks in the axial direction. The cracks passed scale effect seen in the half scale results. This suggests a possible
through the unit and mortar bed on both the faces and ends of interplay of two factors at the smallest scales; firstly a possible
the specimens. Failure is sometimes occasioned by spalling off of anisotropy with respect to strength due to the manufacturing pro-
the units prior to failure. The observed failure mechanisms in the cess of clay brick and secondly the size effect phenomena.
four scales were all similar and agree with the known failure mech- The thinness of the mortar joint at the smallest scales; that is
anisms for masonry in compression. fourth and sixth scales, could also contribute to the higher masonry
Most small scale masonry model tests to date have shown that strength in the smallest scales. Because as stated by many authors
the models are stronger than the prototypes, as seen in [4]. But cur- the tensile stresses that cause failure in the brick are due to the
rent masonry theory assumes that if the mechanical properties of stresses induced by the mortar joint. Again using the Griffith con-
the brick and mortar are similar then there should not be a signif- cept that smaller sized elements are stronger than larger ones, it
icant strength difference between models and prototype [4,11]. would be expected that a thinner joint should be able to withstand
From Table 4, which shows a summary of the triplet compres- higher stresses than thicker ones; consequently the brick would be
sive strength test result and Fig. 3, which shows the variation of able to carry higher tensile forces in masonry with a thinner joint.

Fig. 2. Triplet specimens in the four scales.

306 A. Mohammed et al. / Construction and Building Materials 25 (2011) 303–307

Table 4
Summary of triplet masonry compression test results in the four scales.

Scale Mortar type Bed thickness (mm) Compressive COV (%) Stiffness (N/mm2) COV (%) Mortar cube
strength (N/mm2) strength (N/mm2)
Prototype MP-iii 10.0 9.2 10.8 5500 22.6 4.6
Half M95-iii 5.0 11.0 9.1 5200 26.9 4.9
00
Fourth 2.5 23.0 9.1 5400 6.6 4.7
00
Sixth 1.6 20.3 11.6 6000 20.1 4.7
Compressive Strength N/mm 2

30 8000

24
6000

Stiffness, N/mm2
18
4000
12

6 2000

0
0 0.2 0.4 0.6 0.8 1 1.2 0
0 0.2 0.4 0.6 0.8 1 1.2
Scale
Scale
Fig. 3. Triplet compressive strength in the four scales.
Fig. 6. Stiffness of masonry triplets in the four scales.

masonry could show a stronger compressive strength due to the

thinness of their joints alone. However this may be applicable to
only the very small scales like the sixth and fourth scales in this
study, because as we see from the results, the half scale strength
is similar to the prototype strength.
From Table 4, it can be seen that the mortar cube strengths for
the tests are similar to each other. The highest cube strength of
4.9 N/mm2 was in the half scale test. This is marginally higher than
the mortar cube strength in the prototype test by about 6% and
fourth and sixth scale tests by about 4%. Therefore since the mortar
strengths are comparable to each other in the four scales under
consideration, the difference in masonry strengths seen could be
principally due to the unit properties and the size of the joints at
the smallest scales.

4.2. Stiffness

The variation of masonry stiffness across the four scales are

Fig. 4. Orientation of model units in a prototype brick. shown in Fig. 5 and Table 4, they show that all the stiffness results
are similar; the half and fourth scales stiffness were 6% and 2%
more than the prototype stiffness, while the sixth scale stiffness
0.6 was 9% more than the prototype stiffness. Even though we have
seen substantial variations with regards to their strength proper-
Normalised Strength

0.5 ties their deformation properties are very similar. This is further
illustrated in Fig. 7, a combined plot of stress/strain curves for
0.4
the four scales. It is seen from Fig. 6 that the slope of their elastic
0.3 region are markedly similar.
However results from other researchers [2,4,5] show a different
0.2
trend. Their results show that reduced scale masonry models are
0.1 softer than prototype by a factor approximately equal to the scale
factor. This may be due to the manner in which the models were
0
constructed in this experimental programme which is different
0 0.2 0.4 0.6 0.8 1 1.2
to the way the other researchers made their specimens, that is,
Scale
they all used conventional brick laying while the bricks were laid
Fig. 5. Normalised triplet compressive strength across the four scales. on their sides in this research. Because in conventional brick laying
the light weight of model bricks subject the mortar bed to a lower
bedding stress than the heavier prototype bricks, as a result the
This has been corroborated by Ref. [16] who found that prototype bed joints in the model scales are compacted to a lower degree
masonry with thin layer joints (1.3 mm) were 20% stronger in com- during curing than the bed joints in the prototype. Consequently
pression than those with 12 mm joints. This suggests that model the bed joints in scaled masonry models are less compacted which
 A. Mohammed et al. / Construction and Building Materials 25 (2011) 303–307
25
Compressive Stress, N/mm2
20
15
10
5
0
0 0.005 0.01 0.015
Average Axial Strain
Fig. 7. Stress/strain curves for selected triplet tests across the four scales.
could lead to a softer masonry response under compression. Ref.
[4] have suggested this as a possible reason for the softness of
small masonry models and it has also been reported by Ref. [17].
By comparing it to the stiffness of its constituents elements, it is
seen that the average stiffness of the masonry in the four scales;
5500 N/mm2 in this case is about half of the brick stiffness;
11,500 N/mm2 and roughly similar to the mortar stiffness
(6300 N/mm2 for prototype mortar and 6500 N/mm2 for bench-
mark mortar). This shows the influence of mortar in determining
the stiffness properties of masonry.
5. Conclusions
The results of the masonry tests at different scales has shown
that the strength of masonry triplet in compression was higher
than the prototype in the fourth and sixth model scales but the
similar to the prototype in the half scale. The same pattern was
also repeated in the tests of the unit strengths, therefore indicating
the strong influence of the unit in determining the masonry prop-
erties. There is evidence of anisotropy of strength in clay brick ma-
sonry possibly due to the manufacturing process of extruded clay
units which therefore makes the direction of loading on a cut mod-
el brick important. It can then be suggested that the increase in
model unit strength, could be attributed to two factors, namely;
strength anisotropy and size effect in quasi-brittle materials like
clay bricks.
It was found that triplet stiffnesses in the four scales were iden-
tical to each other and no scale effect was observed. The prototype
masonry and model stiffness were in good agreement with the pro-
totype and model mortar stiffness respectively. This may be attrib-
uted to the way the specimens were constructed, which has
effectively cancelled any differential compaction of the joints due
to the different weight of the unit bricks in the various scales. The
good agreement in the stiffness of the four scales may be evidence
of the importance of the mortar bed in determining the stiffness
307
properties of masonry. This finding is very significant bearing in
mind that model tests by other researchers showed a much softer
model response to the prototype under axial compression.
The findings here are important for testing the properties and
understanding the structural behaviour of prefabricated masonry
Prototype panels under compression. Because the results show that the stiff-
Half ness of prototype and model scale masonry are similar for masonry
Fourth
Sixth constructed while the units are laid on their sides as in prefabri-
cated masonry panels. Therefore using the conventional formulae
for calculating masonry stiffness from the unit and mortar strength
0.02 0.025
stiffness might lead to over estimating the masonry stiffness for
such prefabricated masonry panels.
References
[1] Vogt H. Considerations and investigations on the basic principles of model
tests for brickwork and masonry structures. Building Research Station,
Garston, Report No.: Library communication No. 932; 1956 [Translated from
German].
[2] Hendry AW, Murthy CK. Comparative tests on 1/3 and 1/6 scale model
brickwork piers and walls. In: Proceedings of the British Ceramic Society, No.
4; 1965. p. 45–66.
[3] Sinha BP, Hendry AW. The effect of brickwork bond on the load bearing
capacity of model brickwalls. In: Proceedings of the British Ceramic Society;
1968. p. 55–67.
[4] Egermann R, Cook DA, Anzani A. An investigation into the behaviour of scale
model brick walls. In: Proceedings of the 9th international brick/block
masonry conference, Berlin; 1991. p. 628–35.
[5] Hughes TG, Kitching N. Small Scale testing of masonry. Proceedings of the12th
international brick block masonry conference, 25–28 June, Madrid, Spain;
2000. p. 893–902.
[6] Kerr R. Flood aftermath still being felt; 2006. <http://news.bbc.co.uk/1/hi/
england/cumbria/4583364.stm>.
[7] Caglar A. Earthquake research groups seek lessons from Bam earthquake;
2004. <http://www.parstimes.com/news/archive/2004/washfile/
quake_construction_codes.html>.
[8] Taunton PR. Centrifuge modelling of soil/masonry structure interaction
[PhD]. UK: Cardiff University; 1997.
[9] Murthy CK, Hendry AW. Model experiments in load bearing brickwork.
Building Sci 1966;1:289–98.
[10] Mohammed A. Experimental comparison of brickwork behaviour at prototype
and model scales [PhD]. Cardiff: Cardiff University; 2006.
[11] Hilsdorf HK. Investigation into the failure mechanism of brick masonry loaded
in axial compression. In: Jonson FB, editor. Designing, engineering and
constructing with masonry products; 1969. p. 34–41.
[12] Jessop EL, Shrive NG, England GL. Elastic and creep properties of masonry. In:
Proceedings of the North American masonry conference, Colorado; 1978. p.
12-1–12-7.
[13] Shrive NG, Jessop EL. Anisotropy in extruded clay units and its effect on
masonry behaviour. In: Proceedings of the 2nd Canadian masonry symposium,
Ottawa; 1980. p. 39–50.
[14] Sabnis GM, Aroni S. Size effect in material systems – the state of the art. In:
Proceedings of the Southampton 1969 civil engineering materials conference,
Southamptom; 1971. p. 131–42.
[15] Moss JB. Properties of engineering materials. London: Butterworth; 1971.
[16] Porto Fd, Garbin E, Modena C, Valluzzi MR. Failure modes for in plane loaded
masonry walls made with thin layer mortar. In: 10th Canadian masonry
symposium, Banff, Canada; 2005. p. 694–703.
[17] Mohammed A, Hughes TG. Comparison of prototype and 1/6th model scale
behaviour under compressive loading. In: 10th Canadian masonry symposium,
June 8–12, Banff, Canada; 2005. p. 454–64.