CONSTRUCTION MATERIALS

1. Influence of the testing procedures in the mechanical characterization
of adobe bricks
Dora Silveira, Humberto Varum ⇑
, Aníbal Costa
Civil Engineering Department, University of Aveiro, Campus Universitário de Santiago, 3810-193 Aveiro, Portugal
h i g h l i g h t s
" Mechanical characterization of adobe bricks collected from existing constructions.
" Cylindrical and cubic specimens subjected to simple compression tests.
" Bricks subjected to flexural tests and cylindrical specimens to splitting tests.
" Comparison of results obtained with different testing procedures.
" Rigorous characterization of the deformations suffered by adobe under compression.
a r t i c l e i n f o
Article history:
Received 9 August 2012
Received in revised form 23 October 2012
Accepted 22 November 2012
Available online 25 December 2012
Keywords:
Adobe brick
Mechanical characterization
Experimental testing
Testing procedures
Earth architecture and construction
Traditional construction
Rehabilitation
a b s t r a c t
A study of the influence of testing procedures in the mechanical characterization of adobe bricks tradi-
tionally used in Aveiro district, Portugal, was conducted. Cylindrical and cubic adobe specimens were
subjected to simple compression tests, adobe bricks to flexural tests, and cylindrical adobe specimens
to splitting tests. With these tests it was possible: (i) to collect more data to improve the knowledge
of the mechanical behavior of adobe; (ii) to determine correlations between results obtained with differ-
ent testing procedures; and (iii) to perform a rigorous characterization of the deformations suffered by
adobe under compression, by measuring deformations directly on test specimens.
Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction
1.1. Adobe construction in Aveiro
In Portugal, and particularly in Aveiro district, adobe was a
widely used construction material until the mid-20th century.
Adobes were made with coarse sand, generally with a moderate
or low silt–clay fraction, and with air-lime mortar with variable
hydraulicity [1], usually in a proportion of 25%–40%. Currently, as
confirmed by surveys recently conducted [2,3], there is still a very
significant percentage of adobe building in this region, many of
which are in use. A large number of these buildings present impor-
tant cultural and architectonic value.
Adobe buildings which are not adequately designed and
strengthened, generally present a very poor performance when
subjected to seismic actions (e.g., [4,5]). The majority of buildings
in Aveiro district are not properly strengthened, presenting various
structural and non-structural anomalies, and it is thus necessary to
adequately rehabilitate and strengthen these buildings. To support
rehabilitation and strengthening measures it is essential to create
knowledge about the mechanical behavior of the adobe masonry
traditionally used in the region. At the University of Aveiro, a
research group has been developing work with this objective [6].
1.2. Motivation and objectives
Part of the research work developed at University of Aveiro has
been focused on the study of the mechanical behavior in compres-
sion and tension of the adobes traditionally used in Aveiro district
[7]. The study of the mechanical behavior of adobe bricks is an
important step in the investigation of the mechanical behavior of
0950-0618/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.conbuildmat.2012.11.058
⇑ Corresponding author. Tel.: +351 234 370 938; fax: +351 234 370 094.
E-mail addresses: dora.silveira@ua.pt (D. Silveira), hvarum@ua.pt (H. Varum),
agc@ua.pt (A. Costa).
Construction and Building Materials 40 (2013) 719–728
Contents lists available at SciVerse ScienceDirect
Construction and Building Materials
journal homepage: www.elsevier.com/locate/conbuildmat

2. adobe masonry [8], although it should be noted that tests per-
formed on adobe specimens are only indicators of the quality of
adobe, and not of the quality of masonry [9].
Standards addressed to adobe construction (e.g., [9–11])
indicate that simple compression tests shall be conducted on
adobe bricks or cubic specimens. ‘The Australian earth building
handbook’ [12] allows the possibility of testing adobe bricks or
cylindrical specimens. In addition, most of these documents
recommend the conduction of flexural tests on adobe bricks.
In the research work developed at University of Aveiro, simple
compression tests and splitting tests (also designated as ‘diametral
compression tests’) have been conducted on cylindrical specimens
[7]. Given the characteristics of the available laboratory facilities,
the extraction of cylindrical specimens from adobe bricks is easier
than the extraction of cubic specimens – this is true if the material
does not present large coarse particles in its constitution, and pre-
sents good cohesion of aggregates. Additionally, the testing of
cylindrical specimens presents other relevant advantages. In sim-
ple compression tests conducted on cylinders with a height to
diameter ratio of approximately 2, the obtained strength is closer
to the unconfined compressive strength, when compared to the
strength obtained in the testing of cubes or bricks (according to
studies conducted on concrete specimens) [13]. The conduction
of simple compression tests on cylindrical specimens also allows
an easier and more accurate measurement of the deformations
experienced by specimens. Cylindrical specimens have greater
height than cubic specimens, which facilitates the application of
displacement transducers. In addition, when subjected to simple
compression, cylinders present a stress distribution which is closer
to uniaxial, when compared to the stress distribution in cubes [13].
According to studies conducted on concrete specimens, splitting
tests also present advantages when compared to flexural tests –
a splitting test is closer to a direct tensile test, and the obtained re-
sults are less variable than those obtained in flexural testing [14].
Many authors have conducted studies for the mechanical char-
acterization of adobes taken from existing constructions in differ-
ent parts of the World (e.g., [15–18]), and also of adobes usually
produced for the study of the effectiveness of different possible
compositions (e.g., [18–20]). Generally, compression tests are con-
ducted on adobe bricks or adobe cubic specimens (e.g., [15–20]),
without reference to the influence of confinement effect on the ob-
tained results, and normally tensile strength of adobe is evaluated
by the conduction of flexural tests (e.g., [15,16]). In general, how-
ever, the existing knowledge concerning the influence of the geom-
etry of specimens and of testing procedures on the mechanical
characterization of earth specimens is very limited. Nevertheless,
the importance of such study has been recognized and the first
steps to contribute to this knowledge have been taken (e.g., [8]).
It would be valuable to have the possibility to test cylinders or
cubes, and to conduct flexural or splitting tests, depending on the
characteristics of adobes, and on the existing conditions for the
extraction and testing of specimens. For adobe, however, there
are no studies establishing correlations between the results ob-
tained with these different procedures. With the work developed
and presented in this article, in addition to contributing more
information to the mechanical characterization of adobes tradi-
tionally used in Aveiro district, it is intended to contribute with
an initial proposal for the correlation between results obtained
with the following different testing procedures: simple compres-
sion tests on cylindrical specimens vs. simple compression tests
on cubic specimens; and flexural tests vs. splitting tests. It is
important to note, however, that the strength of a specimen does
not only depend on the type of test conducted and shape of the
specimen, but also on its dimensions, and thus the conclusions
drawn in this study are applicable to the specific dimensions of
the tested specimens.
In the research work previously carried out at University of Ave-
iro [7], and given the available laboratory resources, the deforma-
tion of specimens considered in the analysis corresponded to the
relative displacement of the testing machine platens. However, a
more accurate measurement of the deformation should be per-
formed directly on specimens. In the present work, and with the
objective of contributing to a more rigorous study of the stress–
strain behavior curves, modulus of elasticity, and Poisson’s ratio
of the adobes characteristic of Aveiro district, displacement trans-
ducers were applied directly on the test specimens.
2. Selection, preparation and testing of specimens
2.1. Adobe bricks
A total of 30 adobe bricks were collected from three houses,
with 10 adobes collected per house, in different locations of Aveiro
district (Fig. 1): (i) house ‘H12’, located in the parish of Bunheiro, in
Murtosa municipality, which was being subjected to a reconstruc-
tion process; (ii) house ‘H13’, located in the parish of Monte, in
Murtosa municipality, which was being subjected to a demolition
process; and (iii) house ‘H20’, located in the parish of Cacia, in Ave-
iro municipality, which was also undergoing demolition.
These locations were selected since this study is part of a larger
investigation work, which focuses particularly on the adobe con-
struction in Murtosa and Aveiro municipalities [2,3]. In the build-
ings, the collected adobe bricks were protected by lime plasters,
and thus presented a good conservation state (Fig. 2). The mean
dimensions and specific weight of the collected adobe bricks, typ-
ical of the adobes traditionally used in Aveiro district, are, respec-
tively: 0.41 Â 0.28 Â 0.13 m3
, and 16.2 kN/m3
, for ‘H12’; 0.46 Â
0.32 Â 0.12 m3
, and 15.0 kN/m3
, for ‘H13’; 0.44 Â 0.24 Â 0.12 m3
,
and 15.1 kN/m3
, for ‘H20’.
2.2. Technical recommendations
The following documents were used as reference in the prepa-
ration of specimens and conduction of tests: ‘The Australian earth
building handbook’ [12], for simple compression and flexural test-
ing; and the RILEM technical recommendation ‘CPC 6 Tension by
splitting of concrete specimens’ [21], for splitting testing. The rec-
ommendations in these documents were taken as guidelines, and
were not strictly followed due to limitations of the available labo-
ratory facilities, and due to the fact that these documents refer to
materials for new constructions while the present study is directed
to materials collected from existing constructions. The referenced
recommendation for splitting testing of concrete [21] was adopted,
since there is no technical recommendation for the conduction of
splitting tests on cylindrical adobe specimens.
2.3. Preparation of specimens
Cylindrical specimens were extracted by rotary core drilling
from entire adobe bricks (for simple compression tests), and from
the halves of adobe bricks resultant from flexural tests (for split-
ting tests), with diameters ranging from 78 to 93 mm (Fig. 3a). This
variation in diameter is due to different erosion suffered by the
material during the drilling process. Specimens were cut with a
height to diameter ratio of approximately 2, where possible [21],
and never less than 1. For specimens with a height to diameter ra-
tio equal or inferior to 1.75, tested in simple compression, correc-
tion factors were applied in the calculation of compressive
strength. Given that there are no correction factors directed to
adobe specimens, factors for concrete were used [22]. Cubic spec-
imens were cut from the same adobe bricks from which cylindrical
720 D. Silveira et al. / Construction and Building Materials 40 (2013) 719–728

3. specimens (for simple compression tests) were extracted (Fig. 3b).
For flexural tests, the lateral faces of bricks were cut in order to ad-
just their width to the dimensions of the testing machine (Fig. 3c).
The number of specimens, per house and type of test, is indicated
in Table 1, and the mean dimensions are presented in Table 2. From
each adobe brick, the maximum possible number of specimens was
extracted, and it was attempted to maintain dimensions as
uniform as possible.
To facilitate the identification of the test specimens and the
analysis, specimens were labeled according to their provenience,
shape and type of test conducted. The following notation was
adopted:
Hi aj
clc
clt
cb
pr
8
>>><
>>>:
9
>>>=
>>>;
k;
where ‘H’ indicates the type of construction (in this case, house); ‘i’
is the index which represents the number of the construction from
which the adobe brick was collected; ‘a’ indicates the type of mate-
rial (in this case, adobe); ‘j’ is the index which represents the num-
ber of the adobe brick from which the test specimen was extracted;
‘clc’ corresponds to a cylindrical specimen, submitted to simple
compression testing; ‘clt’ corresponds to a cylindrical specimen,
submitted to splitting testing (tension); ‘cb’ corresponds to a cubic
specimen, submitted to simple compression testing; ‘pr’ corre-
sponds to a rectangular parallelepipedic specimen, submitted to
flexural testing; and ‘k’ is the index which represents the number
of the test specimen.
2.4. Testing
Specimens were tested in the laboratory, using an ‘ELE Multi-
plex 50-E’ testing machine. In simple compression tests, a 50 kN
load ring was used, and in flexural and splitting tests, a 10 kN load
ring was used. The test setups are presented in Fig. 4. In all simple
compression tests and, when necessary, in flexural and splitting
tests, a layer of mortar (constituted by thin sand and water) was
placed in the testing interface, for regularization.
The Australian handbook [12] recommends a testing rate of
1–5 mm/min for simple compression tests. The rate limits indicated
in this handbook, for flexural tests, and in RILEM CPC 6 [21], for
splitting tests, are directed to load-controlled devices, and the
available testing machine is strain-controlled; in addition, RILEM
CPC 6 [21] addresses the testing of concrete, which is a material
with higher strength capacity and higher stiffness than adobe.
The limits indicated in the Australian handbook [12] for simple
compression tests were thus considered in all the performed tests.
Load was applied without shock and increased continuously until
Fig. 1. Houses: (a) ‘H12’; (b) ‘H13’; and (c) ‘H20’.
Fig. 2. Adobe bricks collected from: (a) ‘H12’; (b) ‘H13’; and (c) ‘H20’.
Fig. 3. (a) Cylindrical; (b) cubic; and (c) rectangular parallelepipedic test specimens.
Table 1
Number of test specimens.
Test Number of test specimens
H12 H13 H20
Simple compression (cubes) 7 16 9
Simple compression (cylinders) 7a
15b
6c
Flexural (bricks) 7 5 4
Splitting (cylinders) 10 12 7
No. of specimens where adequate measurement of deformations was possible:
a
5.
b
5.
c
2.
D. Silveira et al. / Construction and Building Materials 40 (2013) 719–728 721

4. failure, with the moving head of the testing machine travelling at a
rate of 1.5 mm/min.
2.5. Measurement of deformations
In simple compression tests, the measurement of the deforma-
tion of specimens was conducted using Gefran ‘PZ12’ rectilinear
displacement transducers (50 mm model) (Fig. 4a). In each cylin-
drical specimen, two or three transducers were placed in the verti-
cal direction, equally spaced. Initially, in each cubic specimen, two
transducers were placed in the vertical direction, in two opposite
vertical faces, and one transducer was placed in the horizontal
direction in one of the vertical faces. The majority of the deforma-
tion values measured in cubes were, however, affected by different
error sources and, therefore, were excluded. The errors are possibly
associated to the small dimensions of the cubic specimens which
did not allow a representative deformation measurement. In addi-
tion, in cubes, the stress distribution deviates from the uniaxial dis-
tribution [13]. Only the measurements conducted in one of the
cubes were considered, and Poisson’s ratio was calculated for that
specimen (section 3.6). The number of cylindrical specimens where
adequate measurement of deformations was possible is indicated
in Table 1.
3. Presentation and discussion of results
3.1. Compressive strength
Compressive strength was obtained through simple compres-
sion testing of cylindrical and cubic specimens. The mean compres-
sive strength values obtained for each adobe, for the two types of
specimen, are presented in Fig. 5, with indication of mean values
and coefficients of variation per house under analysis (calculated
considering the results obtained for all the individual test
specimens).
The mean compressive strength, calculated per adobe under
analysis, for cylindrical specimens, varies between 0.23 MPa
(‘H20_a07’) and 1.02 MPa (‘H12_a03’), with a global mean value
of 0.58 MPa. For cubes, the mean compressive strength, calculated
per adobe, ranges from 0.28 MPa (‘H13_a10’) to 1.21 MPa
(‘H12_a06’), with a global mean value of 0.54 MPa. Results present
considerable variability, particularly the results obtained in the
testing of cylindrical specimens belonging to house ‘H13’.
The mean compressive strength values obtained by testing cyl-
inders for the buildings under analysis, are close to the lower mean
values obtained for other buildings in Aveiro district analyzed in
the study previously conducted [7].
3.2. Flexural and splitting tensile strengths
Flexural and splitting tensile strengths were obtained through
flexural testing of bricks and splitting testing of cylinders, respec-
tively. The mean strength values obtained for each adobe, with
the different testing procedures, are presented in Fig. 6, where
mean values and coefficients of variation per house under study,
calculated considering the results obtained for all the individual
test specimens, are also indicated. For adobe bricks where no split-
ting tensile strength value is indicated, it was not possible to ex-
tract intact cylindrical specimens from the adobe halves resultant
from flexural testing.
The mean flexural tensile strength, calculated per adobe, ranges
between 0.20 MPa (‘H13_a01’) and 1.03 MPa (‘H12_a08’), with a
global mean value of 0.56 MPa. The mean splitting tensile strength,
calculated per adobe, varies between 0.03 MPa (‘H13_a01’) and
0.28 MPa (‘H12_a02’), with a global mean value of 0.16 MPa. The
splitting strength values are close to the results obtained for other
Table 2
Mean dimensions of test specimens.
Specimen Mean dimensions (m)
H12 H13 H20
Cylinder (compression and splitting tests) H 0.16 0.16 0.16
D 0.09 0.09 0.09
H/D 1.8 1.8 1.8
Cube (compression test) H (=L = W) 0.11 0.10 0.10
Brick (flexural test) L 0.41 0.46 0.43
W 0.24 0.24 0.22
H 0.12 0.11 0.12
Notation: H – height; D – diameter; L – length; W – width.
Fig. 4. (a) Simple compression; (b) flexural; and (c) splitting tests.
722 D. Silveira et al. / Construction and Building Materials 40 (2013) 719–728

5. buildings located in Aveiro district, investigated in the preceding
study [7].
Results show high variability, especially the results obtained for
house ‘H13’. As referred in section 1.2, according to studies con-
ducted in concrete, results from flexural tests generally present
higher variability than results from splitting tests. In the present
study, this was verified for houses ‘H12’ and ‘H20’, but not for
‘H13’ which, for splitting tests, presents a very high coefficient of
variation. More tests would be necessary, comprising more build-
ings, to confirm if, for adobe specimens, flexural tests tend to pres-
ent higher variability of results.
In the research work previously carried out [7], compressive
and tensile strengths of adobes traditionally used in Aveiro district
were compared to strengths obtained by other authors for adobes
from different parts of the world, and were compared to the
strength limits indicated in earth construction standards, and thus
this type of comparison will not be repeated here.
3.3. Stress–strain behavior curves
The stress–strain behavior curves resulting from the simple
compression tests conducted on cylindrical specimens, plotted un-
til peak stress, are presented in Fig. 7.
Three different proposals of compression stress–strain behavior
laws, expressed by exponential functions and calibrated with the
obtained results were determined. The following notation was
used: ‘rc’ – compressive stress (MPa); ‘ec’ – compressive strain
(‰); ‘fc’ – compressive strength (or peak stress) (MPa); ‘e2
3
fc
’ – strain
produced by the stress corresponding to two-thirds of compressive
strength.
The first proposed curve presents peak stress, strain at peak
stress, and secant modulus of elasticity at one-third of peak stress,
equal to the respective mean values obtained in the testing of the
cylindrical adobe specimens (Fig. 7). The equation of this curve is
given by: rcðecÞ ¼ À0:668 Ã 3:61 Ã 10À11
ec
þ 0:668. Considering
compressive strength as a variable, it is given by:
rcðecÞ ¼ Àfcð1:05 Ã 10À7
Þ
ec
fc þ fc. The initial section of this curve rep-
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
a03 a04 a05 a06 a03 a07 a08 a09 a10 a06 a07 a08 a09 a10
Compressivestrength(MPa)
Adobe unit
Cylinders Cubes Cylinders (mean strength) Cubes (mean strength)
H12
CV (cylinders): 24%
CV (cubes): 31%
H13
CV (cylinders): 47%
CV (cubes): 29%
H20
CV (cylinders): 28%
CV (cubes): 16%
Fig. 5. Mean compressive strength per adobe brick, obtained in the testing of cylindrical and cubic specimens.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
a01 a02 a07 a08 a09 a10 a11 a01 a02 a04 a05 a06 a01 a02 a03 a04
Flexural/splittingtensilestrength(MPa)
Adobe unit
Flexural strength Splitting strength
Mean flexural strength Mean splitting strength
H12
CV (flexural): 35%
CV (splitting): 30%
H13
CV (flexural): 51%
CV (splitting): 73%
H20
CV (flexural): 24%
CV (splitting): 21%
Fig. 6. Mean flexural and splitting tensile strengths, per adobe brick.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Stress(MPa)
Strain (%º)
Experimental curves Theoretical curve 1
Theoretical curve 2 Theoretical curve 3
Fig. 7. Stress–strain behavior curves.
D. Silveira et al. / Construction and Building Materials 40 (2013) 719–728 723

6. resents well the experimental results; however, the final section
presents a very sudden decrease in stiffness, with a slope of
approximately zero, which is not representative of the final branch
of the results obtained in the conducted tests.
The second proposed curve presents peak stress, strain at peak
stress, and secant modulus of elasticity at 80% of peak stress, equal
to the respective mean values obtained in the conducted tests
(Fig. 7). The equation of this curve is given by: rcðecÞ ¼
À0:668 Ã 1:12 Ã 10À6
ec
þ 0:668. Considering compressive strength
as a variable, it is given by: rcðecÞ ¼ Àfcð1:05 Ã 10À4
Þ
ec
fc þ fc. The final
section of this curve represents well the experimental results;
however, this curve presents an initial stiffness significantly
inferior to the mean initial stiffness obtained in the experimental
tests.
The third curve (Fig. 7) is constituted by two branches described
by two exponential functions. The first function, defined for stres-
ses in the range [0;2/3fc], presents secant modulus of elasticity at
one-third of peak stress and at two-thirds of peak stress, equal to
the respective mean values obtained in the testing of the
cylindrical adobe specimens. The second function, defined for
stresses in the range ]2/3fc;fc], presents peak stress and strain at
peak stress, equal to the respective mean values obtained in the
conducted tests. At the transition point, corresponding to two-
thirds of peak stress, the two functions present the same slope.
The equation of this curve is given by:
rcðecÞ ¼
À0:498 Ã 5:10 Ã 10À16
ec
þ 0:498; ec 6 e2
3
fc
À0:385 Ã 3:79 Ã 10À4
ec
þ 0:678; ec e2
3
fc
8
:
;
with e2
3
fc
¼ 6:41 Ã 10À2
‰. Considering compressive strength as a
variable, it is given by:
rcðecÞ ¼
À0:744 à fc à ð5:99 à 10À11
Þ
ec
fc þ 0:744 Ã fc; ec 6 e2
3
fc
À0:576 à fc à ð5:17 à 10À3
Þ
ec
fc þ 1:01 Ã fc; ec e2
3
fc
8
:
;
with e2
3
fc
¼ 9:60 Ã 10À2
à fc‰. This curve constitutes a good represen-
tation of the experimental results, throughout all the range of val-
ues measured in the tests.
It is important to note that, to determine the stress–strain
behavior laws considering compressive strength as a variable, the
secant stiffness values obtained for reference points (one-third,
two-thirds and 80% of peak stress, and peak stress) were consid-
ered constant for different levels of compressive strength.
3.4. Strain at peak stress
The mean strain at peak stress, per adobe under analysis,
derived from the stress–strain behavior curves obtained in the
simple compression tests conducted on cylindrical specimens, is
presented in Fig. 8, with indication of mean values and coefficients
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
a03 a04 a05 a06 a03 a07 a09 a09 a10
Strainatpeakstress(‰)
Adobe unit
Mean strain
H12
CV: 53%
H13
CV: 37%
H20
CV: 15%
Fig. 8. Mean strain at peak stress per adobe brick, obtained in the compression testing of cylindrical specimens.
0
5
10
15
20
25
30
a03 a04 a05 a06 a03 a07 a09 a08 a09 a10
Modulusofelasticity(x103
MPa)
Adobe unit
E Mean E Mean E
H13
CV (E): 45%
CV (Epeak): 33%
H20
CV (E): 36%
CV (Epeak): 35%
Epeak peak
H12
CV (E): 28%
CV (Epeak): 24%
Fig. 9. Mean modulus of elasticity per adobe brick, obtained in the compression testing of cylindrical specimens.
724 D. Silveira et al. / Construction and Building Materials 40 (2013) 719–728

7. of variation per house under study (calculated considering the re-
sults obtained for all the individual test specimens). Strain at peak
stress ranges between 0.24‰ (‘H20_a10’) and 1.27‰ (‘H12_a03’),
with a global mean value of 0.47‰. Results present significant var-
iability, especially the results obtained for house ‘H12’.
The obtained values of strain at peak stress are approximately
20 times inferior to the ones obtained in the study previously
conducted at University of Aveiro [7]. An explanation for the signif-
icantly inferior deformations verified in the present study is given
in section 3.5.
3.5. Modulus of elasticity
For cylindrical specimens, modulus of elasticity (‘E’) was calcu-
lated as a secant modulus at one-third of peak stress [23]. Secant
modulus of elasticity at peak stress (‘Epeak’) was also calculated.
The mean modulus of elasticity values, calculated per adobe brick,
and their global mean values and coefficients of variation,
corresponding to each house under study (calculated considering
the results obtained for all the individual test specimens), are pre-
sented in Fig. 9. In this figure, the secant modulus of elasticity at
peak stress for adobe ‘H20_a08’ is not presented, given that in
the testing of the cylindrical specimen extracted from this adobe
it was not possible to perform a correct measurement of the defor-
mation at peak stress.
The mean modulus of elasticity, calculated per adobe, ranges
from 7609 MPa (‘H12_a05’) to 25,000 MPa (‘H20_a09’), with a glo-
bal mean value of 13,214 MPa. The mean secant modulus of elas-
ticity at peak stress, calculated per adobe, varies from 803 MPa
(‘H12_a03’) to 3061 MPa (‘H13_a07’), with a global mean value of
1777 MPa. Results show significant variability, particularly the re-
sults obtained for house ‘H13’.
Stiffness degradation was calculated between one-third of peak
stress and peak stress. Secant stiffness presents a global mean deg-
radation of 85% (coefficient of variation: 10%). For the lowest
strength class of concrete (C8/10), secant stiffness degradation cal-
culated between one-third of peak stress and peak stress is 58%
[24], much lower than the obtained for adobe. It is interesting to
note that the higher value of secant stiffness degradation for adobe
is coherent with the tendency verified for concrete [24] – the lower
the concrete strength class, the higher the stiffness degradation.
The mean values of modulus of elasticity obtained for adobe by
other authors, the mean value obtained in the previous study con-
ducted in the University of Aveiro, and the mean value obtained in
this study, are presented in Table 3. The modulus of elasticity val-
ues obtained for adobe in the present study are much higher than
the values obtained by other authors [18–20,25] and in the re-
search work previously carried out at University of Aveiro [7],
and are closer to the lower typical values of modulus of elasticity
for concrete – which normally vary between 17 and 31 GPa [26]
– and the typical values of modulus of elasticity for ceramic clay
bricks – which usually range from 5 to 30 GPa [26]. This may be
partially due to the fact that adobes traditionally used in Aveiro in-
cluded the addition of a significant fraction of lime binder, which
contributes to an increased stiffness of the material, while adobes
in other parts of the world, are normally made without using a bin-
der. The main justification for these differences, however, must be
related to the method adopted in the measurement of deforma-
tions in compression tests. Due to the technical difficulties associ-
ated with performing the measurement of deformations directly
on test specimens, the measurement is often conducted on the load
application system and thus the additional deformation suffered
by the testing devices and especially in the interface with the
specimen, is included in the deformation recorded. This was the
methodology followed in the work of other authors, as well as in
the study previously conducted at University of Aveiro [7]. In the
present work, however, and given the improved available labora-
tory conditions, deformations were measured directly on the spec-
imens, and thus the additional deformation suffered by the system
and interfaces was not measured, obtaining therefore a more pre-
cise value for the modulus of elasticity.
3.6. Poisson’s ratio
For the determination of Poisson’s ratio, longitudinal and trans-
verse deformations of several of the cubes tested in simple com-
pression were measured. However, and as referred in section 2.5,
reliable measurements of transverse deformation were only ob-
tained for one of the tested cubes (‘H12_a04_cb05’), and thus Pois-
son’s ratio was calculated solely for that cube.
Given that the existing standards and technical recommenda-
tions directed to earth construction do not include indications for
the calculation of Poisson’s ratio, the following expression, adapted
from the ‘ASTM C469/C469M-10’ standard [27], was applied:
m = et2/e2. In this expression the following notation was used: ‘m’ –
Poisson’s ratio; ‘et2’ – transverse strain at mid-height of the speci-
men produced by the stress corresponding to 40% of peak stress; ‘e2
– longitudinal strain produced by the stress corresponding to 40%
of peak stress.
For the cube under analysis (‘H12_a04_cb05’), a Poisson’s ratio
of approximately 0.10 was obtained. This value is low when com-
pared with Poisson’s ratio values for other materials. For a large
part of materials, this ratio varies between 0.25 and 0.35 [28].
For concrete it is lower, and similar to several ceramic materials
[29], ranging approximately between 0.1 and 0.2 [28], with a value
of 0.18 being typically considered [29]. The Poisson’s ratio obtained
for the adobe cube under analysis thus coincides with the lower
limit of the range of values typically considered for concrete. This
value is the result of a single test, and it is only a first indication
of adobe Poisson’s ratio. Further studies on this subject are there-
fore necessary.
3.7. Correlations
3.7.1. Compressive strength of cubes and cylinders
The correlation between the compressive strengths obtained in
the testing of cylinders (‘fc,cyl’) and in the testing of cubes (‘fc,cub’)
was studied. For each adobe brick, the mean compressive strength
obtained in the testing of cylinders extracted from that adobe was
plotted against the mean compressive strength obtained in the
testing of cubes extracted from the same adobe, and the following
best-fit linear correlation was determined: fc,cyl = 0.94fc,cub (Fig. 10).
In a simple compression test, the specimen expands laterally as
the applied stress increases. Friction between the specimen and
the testing platens, however, hinders lateral expansion, increasing
the apparent strength of the material. This confinement effect de-
creases as the aspect ratio (‘ka’ – ratio between height and thick-
ness of the specimen) increases. The Australian handbook [12],
and ‘NZS 4298:1998 Materials and workmanship for earth build-
ings’ [10] present an aspect correction factor to calculate the
unconfined compressive strength of adobe bricks. The values of
this correction factor are similar to those derived for small walls
of fired clay brick masonry [8,30]. According to these documents,
the expected correlation between the strength of a rectangular par-
allelepipedic specimen with Ka = 2 and a cubic specimen (Ka = 1) is
approximately equal to 0.88. For concrete, a ratio of 0.80 between
the strength of a cylindrical specimen with Ka = 2 and a cubic spec-
imen (ka = 1) is generally considered [13], although research indi-
cates that this ratio depends on various factors, especially on the
strength level [31].
It was, therefore, expected to obtain a correlation between the
strength of cylindrical specimens and the strength of cubic
D. Silveira et al. / Construction and Building Materials 40 (2013) 719–728 725

8. specimens close to the referred ratios. The obtained correlation is
equal to 0.94. This close proximity between the results obtained
for cubes and cylinders suggests that the use of the regularization
mortar at the bases of specimens contributes to minimize the con-
finement effect produced by the testing platens.
3.7.2. Flexural tensile strength and splitting tensile strength
The correlation between the strengths obtained by splitting
testing of cylinders (‘ft,split’) and by flexural testing of adobe bricks
(‘ft,flex’) was studied. For each adobe brick, the mean strength
obtained in the splitting testing of cylinders extracted from the
adobe halves resultant of flexural testing was plotted against the
strength obtained in the flexural testing of the original adobe brick,
and the following best-fit linear correlation was determined:
ft,split = 0.30ft,flex (Fig. 11).
Flexural tensile strength is typically greater than splitting ten-
sile strength. In flexural tests, failure is controlled by the strength
of the material at the tension surface of the specimen, while in
splitting tests failure may start at any point in the tension diamet-
rical plane [14]. Taking into account the size effect principle, split-
ting tensile strength is thus expected to be lower than flexural
tensile strength [14]. Another possible reason for larger flexural
strength results can be related to the use of Hooke’s Law in the cal-
culation of strength values, when the material does not actually
behave according to this law [32]. Finally, the ratio between the
span and height of a brick tested in flexion is generally not large
enough for it to behave as a perfect beam, which also contributes
to larger strength values.
In several studies on the relation between splitting and flexural
tensile strengths carried out for concrete, the ratio obtained ranges
between 0.39 and 0.91, with a mean value between 0.60 and 0.70
[32,33]. This relation depends, among other factors, on the dimen-
sions of the test specimens and on the strength level of the
Table 3
Modulus of elasticity of adobe specimens, obtained in different studies.
Study Test specimen Measurement of
deformations
Modulus of
elasticity
(MPa)
Location Composition Condition Geometry
Present study Aveiro,
Portugal
Coarse sand (with moderate/low silt–
clay fraction), and air lime mortar
(usually 25%–40%)
Collected from existing
constructions; in good
conservation state
Cylinder: H % 0.15–
0.18 m D % 0.08–
0.09 m
Performed directly on test
specimens
13,214
Previous study [7] Measurement of the
relative displacement of
testing platens
225
Studies by other
authors
Catanzaro,
Italy [18]
Gravel clay, with a proportion of clay
ranging from 16% up to 40%
Collected from existing
constructions; in poor
conservation state
Cube:
0.05 Â 0.05 Â 0.05 m3
Measurement of the
relative displacement of
testing platens
15–87
Ancona,
Italy [19]
Clayey soil, straw and coarse sand, in
variable proportions
New (produced for the
study)
Parallelepiped:
0.15 Â 0.23 Â 0.13 m3
Measurement of the
relative displacement of
testing platens
98–211
Yazd, Iran
[20]
Clayey soil New (produced for the
study)
Parallelepiped:
0.19 Â 0.19 Â 0.05 m3
Not indicated %85a
Mexico
[25]
Not indicated (representative of the
‘‘average adobe’’)
Not indicated Not indicated Not indicated 1471
a
Estimated from the stress–strain curve presented in [20].
0.4
0.6
0.8
1.0
1.2
fc,cyl = 0.94 fc,cub
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Comp.strengthofcylinders(MPa)
Comp. strength of cubes (MPa)
fc,cyl = 0.94 fc,cub
Fig. 10. Correlation between the compressive strengths obtained in the testing of
cylinders and cubes.
0.0
0.1
0.2
0.3
0.0 0.2 0.4 0.6 0.8 1.0
Splittingtensilestrength(MPa)
Flexural tensile strength (MPa)
ft,split = 0.30 ft,flex
Fig. 11. Correlation between splitting tensile strength and flexural tensile strength.
0
5
10
15
20
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Modulusofelasticity(x103
MPa)
Compressive strength (cylinders) (MPa)
E = 13927 fc
Fig. 12. Correlation between modulus of elasticity and compressive strength.
726 D. Silveira et al. / Construction and Building Materials 40 (2013) 719–728

9. material [32]. The greater the strength, the greater the expected
value of this ratio [32,34]. In the present study, a ratio of 0.30
was obtained. This ratio is lower than those determined for con-
crete, which is consistent with the demonstrated tendency, as
adobe strength values are much lower than concrete strength
values.
3.7.3. Modulus of elasticity and compressive strength
The correlation between the modulus of elasticity and compres-
sive strength obtained in the testing of cylinders was studied. For
each cylinder, the respective modulus of elasticity was plotted
against the compressive strength, and the following best-fit linear
correlation was determined: E = 13927 fc (Fig. 12). The presented
data points are considerably scattered and thus more tests are
necessary to strengthen the validity of this correlation.
4. Conclusions
In Table 4 it is presented a summary of the results. Compressive
strength obtained by testing cylindrical specimens and tensile
strength obtained in splitting tests are close to the strength values
obtained in the study previously carried out at University of Aveiro
[7], and thus conform to what was initially expected. High values
of flexural tensile strength were obtained, close to the compressive
strength values of cubic and cylindrical specimens. This stresses
the fact that flexural testing of adobe bricks can overestimate ten-
sile strength.
The values of modulus of elasticity are much higher than those
obtained in the previous study [7] and by other authors [18–20,25].
One explanation to these higher values is the fact that deforma-
tions were directly measured on the specimens, and thus the addi-
tional deformation suffered by the testing system was not
considered.
The obtained results vary greatly, particularly for house ‘H13’.
High variability of results was expected because, at the time, pro-
duction control was inexistent, and since, even for the same adobe
construction, composition of adobes and manufacturing proce-
dures could vary significantly. Other authors testing adobe speci-
mens report similar variability of results, particularly when
adobes are taken from existing constructions (e.g., [15,18,19]).
The obtained correlations are summarized in Table 5. The com-
pressive strength obtained by testing cylinders is very close to the
strength obtained by testing cubes, as attested by the respective
correlation. One possible explanation is that the use of regulariza-
tion mortar between the specimens and the testing platens con-
tributed to the reduction of the confinement effect. The study of
the correlation between compressive and tensile strengths has
not been conducted here, as it was already performed and pre-
sented previously [7].
Stress–strain behavior laws were also determined, as approxi-
mate representations of the stress-strain curves obtained in simple
compression tests conducted on cylindrical specimens. Knowledge
of the stress–strain behavior laws of adobe is important, because
these curves express essential information about the properties
and mechanical behavior of adobe [28], are useful to support the
numerical modeling of the behavior of this material, and can assist
the validation of the results of experimental tests conducted by
other authors.
Overall, the work presented contributes to the enrichment of
the knowledge concerning the mechanical properties of the adobes
traditionally used in Aveiro district. This knowledge is important to
support the numerical modeling of adobe masonry and other fu-
ture studies, and to support the rehabilitation and strengthening
of existing adobe buildings. This work also contributes with an ini-
tial proposal for the correlation between results obtained with dif-
ferent testing procedures, allowing the possibility of selecting
between procedures, according to the characteristics of adobes
and existing conditions for the extraction and testing of specimens.
It is important to note that the obtained results and conclusions are
only directly applicable to the traditional lime adobe of Aveiro re-
gion; however, the testing procedures, types of analyses conducted
and general conclusions are transferrable to the study of other
types of adobes.
Finally, in the conduction of tests, and as previously noted in
[7], the lack of comprehensive European standards directed to
earth construction stands out. There is a need for standardized pro-
cedures adapted to earth construction, and particularly to adobe
construction, addressing not only new constructions, but also the
existing ones.
Role of the funding source
The research work presented is part of doctoral studies funded
by a scholarship provided by ‘FCT – Fundação para a Ciência e a
Tecnologia’, with the reference ‘SFRH/BD/39012/2007’.
Acknowledgments
The authors express their acknowledgments to: Eng. Daniel
Bastos from Câmara Municipal of Murtosa, the staff of the Civil
Engineering Laboratory of University of Aveiro, Eng. António
Figueiredo, Eng. José Carvalho, and Câmara Municipal of Aveiro
for the collaboration in the studies conducted; and the owners of
adobe buildings who allowed the collection of samples.
References
[1] Santiago L. The Gandaresa House of Aveiro district. Contribution to its
rehabilitation as cultural patrimony dissertation. Évora: University of Évora;
2007 [Portuguese].
[2] Silveira D, Varum H, Costa A, Lima E. Survey and characterization of the adobe
built park in Aveiro city. In: Proceedings of sixth Seminário de Arquitectura de
Terra em Portugal and nineth Seminário Ibero-Americano de Construção e
Arquitectura com Terra [CD-ROM]; 2010 February 20–23; Coimbra, Portugal;
2010. [Portuguese].
[3] Silva S, Varum H, Bastos D, Silveira D. Earthen architecture – investigation and
characterization of adobe buildings in Murtosa municipality. In: Maria F,
Mariana C, Filipe J, editors. Terra em seminário 2010. Argumentum (Lisbon);
2010. p. 236–9 [Portuguese].
[4] Mahdi T. Behavior of adobe buildings in the 2003 Bam earthquake. In:
Proceedings of Sismo Adobe 2005 Conference May 16–19. Lima, Peru; 2005
[CD-ROM].
[5] Elnashai AS, Gencturk B, Kwon OS, Al-Qadi IL, Hashash Y, Roesler JR, et al. The
Maule (Chile) earthquake of February 27, 2010. Consequence assessment and
Table 4
Summary of the obtained results.
Compressive
strength (MPa)
Tensile strength
(MPa)
Strain
at peak
stress
(‰)
Modulus
of
elasticity
(MPa)
Poisson’s
ratio
Cubes Cylinders Flexural Splitting
Min. 0.28 0.23 0.20 0.03 0.24 7609 –
Max. 1.21 1.02 1.03 0.28 1.27 25,000 –
Mean 0.54 0.58 0.56 0.16 0.47 13,214 0.10a
a
Corresponds to the measurement of deformations in one cubic specimen
(‘H12_a04_cb05’).
Table 5
Summary of the obtained correlations.
Compressive
strength
Tensile strength Modulus of elasticity vs. compressive
strength
Cylinders vs.
cubes
Flexural vs.
splitting
fc,cyl = 0.94 fc,cub ft,flex = 0.30 ft,split E = 13927 fc
D. Silveira et al. / Construction and Building Materials 40 (2013) 719–728 727

10. case studies. Urbana (IL): Mid-America Earthquake Center; 2010 [Report no.:
10-04].
[6] Varum H, Figueiredo A, Silveira D, Martins T, Costa A. Outputs from the
research developed at the University of Aveiro regarding the mechanical
characterization of existing adobe constructions in Portugal. Inf Constr
2011;63:127–42.
[7] Silveira D, Varum H, Costa A, Martins T, Pereira H, Almeida J. Mechanical
properties of adobe bricks in ancient constructions. Constr Build Mater
2012;28:36–44.
[8] Morel JC, Pkla A, Walker P. Compressive strength testing of compressed earth
blocks. Constr Build Mater 2007;21:303–9.
[9] Reglamento nacional de edificaciones. Norma técnica de edificación NTE E.080
Adobe. Lima: SENCICO; 2006 [Spanish].
[10] NZS 4298:1998 Materials and workmanship for earth buildings. Wellington:
Standards New Zealand; 1998.
[11] 14.7.4: 2009 New Mexico earthen building materials code – New Mexico
administrative code. New Mexico, USA; 2009.
[12] Walker P. The Australian earth building handbook, HB 195–
2002. Sydney: Standards Australia; 2002.
[13] Domone PLJ. Strength and failure of concrete. In: Illston JM, editor. Constr
mater: their nature and behaviour. London: EFN Spon; 1994. p. 155–68.
[14] Ozyildirim C, Carino NJ. Concrete strength testing. In: Lamond JF, Pielert JH,
editors. Significance of tests and properties of concrete concrete-making
materials. West Conshohocken: ASTM International; 2006.
[15] Liberatore D, Spera G, Mucciarelli M, Gallipoli MR, Santarsiero D, Tancredi C,
et al. Typological and experimental investigation on the adobe buildings of
Aliano (Basilicata, Italy). In: Lourenço PB, Roca P, Modena C, Agrawal S, editors.
Proceedings of the fifth international conference on structural analysis of
historical constructions; 2006 November 6–8; New Delhi,
India. India: Macmillan India; 2006. p. 851–8.
[16] Rivera JC, Muñoz EE. Structural characterization of materials of earth
structural systems: adobe. Revista Internacional de Desastres Naturales,
Accidentes e Infraestructura Civil. 2005;5(2):135–48 [Spanish].
[17] Baglioni E, Fratini F, Rovero L. The materials utilised in the earthen buildings
sited in the Drâa Valley (Morocco): mineralogical and mechanical
characteristics. In: Proceedings of sixth Seminário de Arquitectura de Terra
em Portugal and nineth Seminário Ibero-Americano de Construção e
Arquitectura com Terra. 2010 February 20–23; Coimbra, Portugal; 2010 [CD-
ROM].
[18] Fratini F, Pecchioni E, Rovero L, Tonietti U. The earth in the architecture of the
historical centre of Lamezia Terme (Italy): characterization for restoration.
Appl Clay Sci 2011;53:509–16.
[19] Quagliarini E, Lenci S. The influence of natural stabilizers and natural fibres on
the mechanical properties of ancient Roman adobe bricks. J Cult Herit
2010;11:309–14.
[20] Eslami A, Ronagh HR, Mahini SS, Morshed R. Experimental investigation and
nonlinear FE analysis of historical masonry buildings. Constr Build Mater
2012;35:251–60.
[21] RILEM. CPC 6 tension by splitting of concrete specimens, 1975. In: RILEM
technical recommendations for the testing and use of construction materials,
London: EFN Spon; 1994. p. 21–2.
[22] ASTM C42/C42M–12 Standard test method for obtaining and testing drilled
cores and sawed beams of concrete. West Conshohocken: ASTM International;
2012.
[23] NP EN 1052–1: 2002 Methods of test for masonry – Part 1: Determination
of compressive strength. Brussels: European Committee for Standard-
ization (CEN), Caparica: Instituto Português da Qualidade (IPQ); 2002
[Portuguese].
[24] EN 1992-1-1:2004 Eurocode 2: Design of concrete structures – Part 1-1:
General rules and rules for buildings. Brussels: European Committee for
Standardization (CEN); 2004.
[25] Gavrilovic P, Sendova V, Ginell WS, Tolles L. Behaviour of adobe structures
during shaking table tests and earthquakes. In: Proceedings of the 11th
European conference on earthquake engineering. 1998 September 6–11;
Rotterdam, Netherlands.
[26] Cobb F. Structural engineer’s pocket book. Oxford: Elsevier Butterworth-
Heinemann; 2004 [CD-ROM].
[27] ASTM C469/C469M–10 Standard test method for static modulus of elasticity
and Poisson’s ratio of concrete in compression. West Conshohocken: ASTM
International; 2010.
[28] Gere JM, Goodno BJ. Mechanics of materials. Brief ed. Andover: Cengage
Learning; 2011.
[29] Weiss J. Elastic properties, creep and relaxation. In: Lamond JF, Pielert JH,
editors. Significance of tests and properties of concrete concrete-making
materials. West Conshohocken: ASTM International; 2006. p. 194–206.
[30] Krefeld WJ. Effect of shape of specimen on the apparent compressive strength
of brick masonry. In: Proceedings of the American society of materials,
Philadelphia, USA; 1938. p. 363–9.
[31] L’Hermite R. Current ideas on concrete technology. Paris: Documentation
technique du bâtiment et des travaux publics. 1955 [French].
[32] Melis LM, Meyer AH, Fowler DW. An evaluation of tensile strength testing.
Austin (TX): Center for Transportation Research, University of Texas at Austin;
1985 [Report No.: 432–1F].
[33] Popovics S. Relations between various strengths of concrete. Highway
Research Board; 1967 [Record No.: 210].
[34] Malhotra VM, Zoldners NG. Comparison of ring tensile strength of concrete
with compressive, flexural and splitting tensile strengths. J Mater
1967;12:160–99.
728 D. Silveira et al. / Construction and Building Materials 40 (2013) 719–728