Development of a Direct Tensile Test for Compacted Earth

Blocks Reinforced with Natural Fibers

A. Mesbah1; J. C. Morel2; P. Walker3; and Kh. Ghavami4

Abstract: Many thousands of years after its development, unburnt earth block masonry continues to offer an environmentally sustain-
able and cost-effective alternative to other conventional wall building materials, such as concrete blocks and clay fired bricks, in both
developing and developed countries. However, earth is a brittle building material with low tensile strength, and as a consequence tensile
cracks in response to external actions or restrained shrinkage are often observed. Ductility and tensile strength may be improved by adding
binders or natural fibers to the soil mixture. In this technical note, initial findings of an ongoing study of the tensile characteristics of
compacted earth blocks reinforced with natural fibers is reported. The pre- and postcracking characteristics of compacted earth blocks in
direct tension were determined using a test procedure developed for this work.
DOI: 10.1061/共ASCE兲0899-1561共2004兲16:1共95兲
CE Database subject headings: Clays; Fiber reinforced materials; Tensile strength; Building materials; Blocks.

Introduction years, to reduce shrinkage cracking and improve tensile strength.

In recent tests, natural fibers in compressed earth blocks have also
Earthen architecture has a continuous heritage dating back at least
been shown to reduce the size of shrinkage cracks and to improve
10,000 years 共Houben and Guillaud 1994兲. Compressed earth ma-
durability and postcracking tensile strength 共Ziegler et al. 1998兲.
sonry blocks, formed using moist soil compacted mechanically to
improve physical characteristics, are a relatively recent develop-
ment that have gained popularity over the past 50 or so years. In
this time, many projects using compressed earth blocks have been Objectives and Scope
successfully completed in both developing and developed coun-
tries. Benefits of using earth in this manner include improved The primary objective of the work presented in this technical note
strength and durability as compared with adobe, while maintain- is to study the tensile characteristics of compacted earth blocks
ing significantly lower embodied energy levels than alternative reinforced with randomly distributed sisal fibers. The note de-
materials. However, problems arise from the materials’ low ten- scribes the development of a direct tensile strength test for com-
sile strength, brittle behavior, and deterioration in the presence of pacted earth blocks. Initial experimental results for plain and sisal
water. fiber-reinforced blocks in tension are presented, together with rec-
Stabilization with a hydraulic binder, such as cement or lime, ommendations for wider applications of the technology and fur-
or a waterproofing agent, such as bitumen, can significantly im- ther research work.
prove strength and water resistance 共Houben and Guillaud 1994兲.
Typically, binders are added at between 4 and 10% of the soil dry
weight. However, use of these additives also significantly in- Materials
creases both material costs, an important deterrent to their greater
use in developing countries, and their environmental impact. As a
Soil
more sustainable alternative to cement and bitumen, natural fi-
bers, such as straw, have of course been used in adobe and other The natural soil used for this investigation was sourced from the
traditional forms of earthen construction for many thousands of area around Lyon in southern France, a region famous for its large
number of earthen buildings. Characteristics of the soil, including
1
Dept. Génie Civil et Bâtiment, URA CNRS 1652, ENTPE, Lyon, grading and standard Proctor test results, are outlined in Table 1.
France.
2
Dept. Génie Civil et Bâtiment, URA CNRS 1652, ENTPE, Lyon,
Sisal Fibers
France.
3
Dept. of Architecture and Civil Engineering, Univ. of Bath, BA2 Sisal fibers, sourced from the leaves of the sisal plant 共Agave
7AY, Bath, UK. family兲, were selected as reinforcement of the earth blocks. Sisal,
4
Dept. of Civil Engineering, PUC de Rio, Brasil. used for a wide variety of products including ropes and floor
Note. Associate Editor: Christopher K. Y. Leung. Discussion open coverings, has been used successfully in previous studies of com-
until July 1, 2004. Separate discussions must be submitted for individual
pressed earth blocks 共Olivier and El Gharbi 1995; Ghavami et al.
papers. To extend the closing date by one month, a written request must
be filed with the ASCE Managing Editor. The manuscript for this techni- 1999兲. The fibers used in this work were on average between 20
cal note was submitted for review and possible publication on November and 50 mm long, with an average cross-sectional area of 0.104
15, 2001; approved on January 27, 2003. This technical note is part of the mm2. Tensile behavior of the individual fibers is linear-elastic
Journal of Materials in Civil Engineering, Vol. 16, No. 1, February 1, until fracture. Mechanical characteristics, taken from a previous
2004. ©ASCE, ISSN 0899-1561/2004/1-95–98/$18.00. study 共Ghavami et al. 1999兲, are summarized in Table 2.

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / JANUARY/FEBRUARY 2004 / 95

Table 1. Soil Characteristics
Characteristic Value
Gravel content 共2–10 mm兲 3%
Sand content 共0.06 –2.0 mm兲 43%
Silt content 共0.002–0.06 mm兲 42%
Clay content 共⬍0.002 mm兲 12%
Liquid limit 33.3%
Plastic limit 23.3%
Plasticity index 10.0%
Methylene blue result 1.27
Optimum moisture content 16.2%
Maximum dry density 2,260 kg/m3

Composite Material
Sisal fibers were added to the natural soil during wet mixing. A
total mass of fibers equivalent to 0.5% of initial soil dry weight Fig. 1. Test specimen
was used throughout this study 共irrespective of fiber length兲. This
quantity of fibers is sufficiently small that methodologies used on
raw soil, such as the standard Proctor compaction test, are also are not at present sufficiently well understood to allow reliable
applicable for the fiber-reinforced soil 共Mesbah et al. 1999; Morel replication of tapered specimens.
et al. 2000兲. During preparation of the material, the soil moisture Therefore, given the difficulties outlined in testing cylinders
content is initially brought up to its optimum value, established by and manufacturing representative nonprismatic specimens, it was
Proctor testing to achieve maximum dry density, in the mixer. The decided to test standard blocks produced by a press. To overcome
sisal fibers are then added slowly by hand and mixing is contin- problems of anchorage failure, the block was sawed along a sec-
ued to ensure a uniform distribution of the fibers throughout the tion at midheight to create a weakened cross section, where the
soil matrix; care is required to avoid aggregation of the fibers crack is expected to form 共Fig. 1兲. During load application, move-
within the matrix. ment of the ram was measured to provide displacement of the
crack. Using the simplifying assumption that the tensile stress is
uniformly distributed along the crack plane at failure, the applied
Development of Direct Tensile Test tensile stress is just given by the applied tensile force divided by
the cross-sectional area along the crack plane.
It is well established that direct tensile strength testing of materi- There are, of course, problems with the preceding approach for
als such as rock, concrete, masonry, and soil are difficult to per- measuring precracking behavior. By cutting the block, the tensile
form due to problems associated with specimen anchorage. Con- stress is no longer uniform throughout the specimen, introducing
sequently, indirect alternatives, such as the bending or modulus of significant stress concentrations that are likely to propagate early
rupture test and the Brazilian test, are more commonly used cracking. Secondly, the process of saw-cutting blocks may intro-
共Atassi 1997兲. The Brazilian test has been used previously to test duce microcracking along the test plane, further reducing the ten-
compacted earth blocks 共Olivier et al. 1997兲, though for the more sile strength of the earth section. Opposite ends of the test block
ductile fiber-reinforced earth blocks it is not considered suitable, are unlikely to be perfectly parallel and may also be distorted as
as the test conditions are heterogeneous and postcracking load is the moist earth shrinks when the blocks are sun-dried. Conse-
not readily controlled 共Rocco et al. 1999兲. quently, significant levels of tensile stress may be introduced
Initially, a direct tensile test, similar to that developed by along the test plane as the blocks are fixed into the testing appa-
Atassi for fiber-reinforced concrete, was used 共Atassi 1997兲. Each ratus. To overcome this problem, blocks were carefully, not ran-
end of the cylindrical earth specimen was glued directly onto domly, selected for the tensile strength testing. Despite these
opposing platens of the tensile test apparatus. However, this test problems for precracking response, the adopted test procedure is
was often unsuccessful, as the glued connection developed insuf- considered valid for measuring postcracking behavior of fiber-
ficient strength to allow tensile failure of the earth specimen. Ta- reinforced specimens, the main objective of this investigation.
pered specimens might be used to overcome this problem 共Ziegler
et al. 1998兲; however, it is very difficult to compress material in a
tapered form that also remains representative of compacted earth Experimental Results
blocks produced using a given block press. The structure of the
finished block is dependent on friction along the mold sides, com- The adopted block size was 100⫻140⫻295 mm. The width and
pactive effort, and speed 共static or dynamic兲. These parameters depth of the cut notch were 6 and 25 mm, respectively. This notch
size was determined to ensure tensile failure along that cross sec-
tion while also ensuring minimum intervention and disturbance to
Table 2. Typical Mechanical Properties of Sisal Fibers 关from the specimens. Notch shape was dictated by the available method
Ghavami et al. 共1999兲兴 of cutting, using a diamond saw blade, selected to minimize the
Young modulus Unit dry weight risk of damage. After cutting and prior to testing, each specimen
Strain failure 共N/mm2兲 共kN/m3兲 was dried in a ventilated oven at 35°C until a constant mass was
attained. The rate of tensile displacement during testing was con-
0.056 11,800 9.4
stant throughout at 0.002 mm/s.

96 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / JANUARY/FEBRUARY 2004

 After initial cracking, the composite behavior continues to dis-
sipate energy for increasing crack displacements. The energy dis-
sipated after the crack opening was around 300 J/m2, whereas the
energy dissipated for unreinforced earth blocks varied from 25 to
50 J/m2. Typically the fibers did not fail in tension, but specimen
failure occurred due to pullout of the fibers from the earth matrix.
In summary, the parameters influencing the tensile strength of the
composite material are:
• Tensile strength of the plain earth matrix;
• Bond strength between the fibers and earth matrix;
• Fiber 共anchorage兲 length; and
• Fiber tensile strength 共not exceeded experimentally兲.
Fig. 2 presents the test results with two different fiber lengths,
20 and 50 mm. The behavior is similar, with tensile ductility
improved in both cases. But the greater fiber length 共50 mm兲
increases the residual postcracking stress level. As observed ear-
lier, the 20-mm-long fibers did not break during the test, suggest-
ing that their embedded length was insufficient to develop full
Fig. 2. Tensile behavior of fiber-reinforced blocks tensile capacity. With longer fibers the pullout behavior is im-
proved. Though for the same fiber content 共0.5% of dry weight兲
there are of course fewer 50 mm fibers in the matrix, with a
longer fiber length there is also a greater probability for any given
fiber to intersect the crack plane.
Tests on Plain Earth Blocks
Though the tensile test described herein has primarily been devel- Conclusions and Prospects
oped to study postcracking behavior of fiber-reinforced earth
blocks, a limited series of tests were undertaken to assess the This technical note sets out a proposed direct test method for
behavior of similar plain blocks under the test conditions. Brittle assessing the tensile strength of fiber-reinforced compressed earth
tensile failure of the plain blocks occurred at a ram extension of blocks. The primary advantage of the proposed methodology is
0.8 mm and a peak tensile stress between 0.18 and 0.23 MPa. that direct tensile testing is undertaken on actual full-size com-
However, due to the problems outlined previously with specimen pressed blocks, rather than replicate specimens where variations
preparation, these values are not considered a true representation in the method of manufacture are likely to significantly alter be-
of block tensile strength. As expected, once cracking occurs, the havior.
tensile stress reduces to zero with no further meaningful displace- By using the direct tensile test, it was possible to quantify the
ment. tensile reinforcing effects of randomly distributed sisal fibers in
earth blocks. Benefits of natural fiber reinforcement include both
Earth Reinforced with Sisal Fibers improved ductility in tension in comparison with plain earth
blocks and the inhibition of tensile crack propagation after initial
Results for sisal-reinforced blocks are outlined in Fig. 2. As for formation. Prior to cracking, the fibers would appear to have no
the plain earth blocks, there is a reduction in applied stress once noticeable effect on material behavior.
the peak value has been attained on cracking of the earth. How- Following these initial tests, further work is ongoing, first to
ever, due to the presence of the fibers, a residual tensile stress further optimize design of the tensile testing setup, including the
level between 0.10 and 0.12 MPa is maintained after cracking. At influence of specimen size and preparation on test results, fol-
the peak stress levels, tensile displacement was not sufficient to lowed by an evaluation of optimum fiber length and mix propor-
mobilize the maximum tensile stress in the fibers. In one test, the tions on tensile behavior. This will be followed by tests on ma-
tensile stress increased with displacement and then decreased sonry wall prism tests and analytical models of behavior. The
steadily to zero 共Fig. 2兲. tensile test developed is also suitable for other materials, such as
Prior to cracking, the fibers have not significantly modified earthen mortars and plasters. Indeed, there are further benefits
block behavior, as the elastic stiffness of the fiber content is low from using natural fibers in mortars and plasters, including inhi-
and the relative displacement between the fibers and the earth bition of shrinkage cracking and improved fiber distribution due
matrix is insufficient to mobilize frictional resistance. On forma- to the higher moisture content at mixing.
tion of the tension crack, further overall displacement of the
specimen is concentrated into the elastic extension of the sisal
fibers across the opening and the bond slippage of the fiber an- Acknowledgment
chorage in either side of the opening.
As the tensile load is applied after cracking a second, lower, The writers wish to acknowledge Odile Roque, Mouchine Maar-
peak stress was recorded for both fiber lengths 共Fig. 2兲, corre- oufi, Gilles Chanvillard, and Firas Atassi for their help in the
sponding to the combined maximum bond strength of the embed- experiments and for their advice.
ded fibers. The peak occurs at 1 mm extension for the 20 mm
fibers and at 2.6 mm extension for the 50 mm fibers. This result References
has previously been observed 共Morel et al. 2000兲 for natural fi-
bers used to reinforce sand. If fibers are to contribute significantly Atassi, M. F. 共1997兲. ‘‘Etude du comportement des bétons de fibres mét-
to tensile behavior prior to cracking, they require greater stiffness. alliques: Influence de la géométrie des fibres, du dosage et de

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / JANUARY/FEBRUARY 2004 / 97

 l’anisotropie de leur répartition.’’ PhD thesis, INSA, Lyon, France 共in fibres subjected to shear.’’ Masonry International, 13共2兲, 54 – 62.
French兲. Olivier, M., and El Gharbi, Z. 共1995兲. ‘‘Sisal fibre reinforced soil block
Ghavami, K., Toledo Filho, R. D., and Perrazo Barbosa, N. 共1999兲. ‘‘Be- masonry.’’ Proc., 4th Int. Masonry Conf., The British Masonry Soci-
haviour of composite soil reinforced with natural fibres.’’ Cem. Concr. ety, London, 55–58.
Compos., 21, 39– 48. Olivier, M., Mesbah, A., El Gharbi, Z., and Morel, J. C. 共1997兲. ‘‘Mode
Houben, H., and Guillaud, H. 共1994兲. Earth construction—a comprehen- opératoire pour la réalisation d’essais de résistance sur blocs de terre
sive guide, Intermediate Technology, London. comprimée.’’ Mater. Constr. (Paris), 30, 515–517 共in French兲.
Mesbah, A., Morel, J. C., and Olivier, M. 共1999兲. ‘‘Comportement des Rocco, C., Guinea, G. V., Planas, J., and Elices, M. 共1999兲. ‘‘Size effect
sols fins argileux pendant un essai de compactage statique: Détermi- and boundary conditions in the Brazilian test: Experimental verifica-
nation des paramètres pertinents.’’ Mater. Constr. (Paris), 32, 687– tion.’’ Mater. Constr. (Paris), 32, 210–217.
694 共in French兲. Ziegler, S., Leshchinsky, D., Ling, H. L., and Perry, E. B. 共1998兲. ‘‘Effect
Morel, J. C., Ghavami, K., and Mesbah, A. 共2000兲. ‘‘Theoretical and of short polymeric fibres on crack development in clays.’’ Soils
experimental analysis of composite soil blocks reinforced with sisal Found., 38共1兲, 247–253.

98 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / JANUARY/FEBRUARY 2004