SPLIT TENSILE STRENGTH OF CONCRETE
To determine the split tensile strength of concrete of given mix proportions.
Scope and Significance:
The tensile strength is one of the basic and important properties of the concrete. The
concrete is not usually expected to resist the direct tension because of its low tensile strength and
brittle nature. However, the determination of tensile strength of concrete is necessary to
determine the load at which the concrete members may crack. The cracking is a form of tension
Apart from the flexure test the other methods to determine the tensile strength of concrete
can be broadly classified as (a) direct methods, and (b) indirect methods. The direct method
suffers from a number of difficulties related to holding the specimen properly in the testing
machine without introducing stress concentration, and to the application of uniaxial tensile load
which is free from eccentricity to the specimen. As the concrete is weak in tension even a small
eccentricity of load will induce combined bending and axial force condition and the concrete
fails at the apparent tensile stress other than the tensile strength.
As there are many difficulties associated with the direct tension test, a number of indirect
methods have been developed to determine the tensile strength. In these tests in general a
compressive force is applied to a concrete specimen in such a way that the specimen fails due to
tensile stresses developed in the specimen. The tensile stress at which the failure occurs is
termed the tensile strength of concrete.
The splitting tests are well known indirect tests used for determining the tensile strength
of concrete sometimes referred to as split tensile strength of concrete. The test consists of
applying a compressive line load along the apposite generators of a concrete cylinder placed with
its axis horizontal between the compressive platens. Due to the compression loading a fairly
uniform tensile stress is developed over nearly 2/3 of the loaded diameter as obtained from an
elastic analysis. The magnitude of this tensile stress O
(acting in a direction perpendicular to
the line of action of applied loading) is given by the formula (IS : 5816-1970):
= 0.637 P/dl
The ratio of the split tensile strength to cylinder strength not only varies with the grade of
the concrete but is also dependent on the age of concrete. This ratio is found to decrease with
time after about a month. The air-cured concrete gives lower tensile strength than that given by
moist-cured concrete. The flexural strength as obtained by rupture test is found to be greater
than the split tensile strength. This test is becoming very popular because of the following
i) The test is simple to perform and gives more uniform results than that given
by other tests.
ii) The strength determined is closer to the actual tensile strength of concrete than
the modulus of rupture value.
iii) The same moulds and testing machine can be used for compression and
Similar to the splitting of the cylinder cubes can also be split either (a) along its middle
parallel to the edges by applying opposite compressive forces through 15 mm square bar
of sufficient length or (b) along one of its diagonal planes by applying compressive forces
along two opposite edges. In the side splitting of cubes the tensile strength is obtained
= 0.642 P/S
and in diagonal splitting it is determined from O
= 0.5187 PS
where P is the failure load and S is the side of the cube.
Compression testing machine weighing machine mixer, tamping roks
i) Take mix proportion as 1:2:4 with water cement ratio of 0.6. Take 21kg of
coarse aggregate, 10.5 kg of fine aggregate 5.25kg of cement and 3.l5 litres of
water. Mix them thoroughly until uniform colour is obtained. This material
will be sufficient for casting three cylinders of the size 150mm diameter X
300 mm length.
In mixing by hand cement and fine aggregate be first mixed dry to
uniform colour and then coarse aggregate is added and mixed until coarse
aggregate is uniformly distributed throughout the batch. Now the water shall
be added and the ingredients are mixed until resulting concrete is uniform in
colour. Mix at least for two minutes
ii) Pour concrete in moulds oiled with medium viscosity oil. Fill the cylinder
mould in four layers each of approximately 75 mm and ram each layer more
than 35 times with evenly distributed strokes.
iii) Remove the surplus concrete from the tope of the moulds with the help of the
iv) Cover the moulds with wet mats and put the identification mark after about 3
to 4 hours.
v) Remove the specimens from the mould after 24 hours and immerse them in
water for the final curing. The test are usually conducted at the age of 7-28
days. The time age shall be calculated from the time of addition of water to
the dry ingredients.
vi) Test at least three specimens for each age of test as follows
i. Draw diametrical lines on two ends of the specimen so that they are in the
same axial plane.
ii. Determine the diameter of specimen to the nearest 0.2 mm by averaging
the diameters of the specimen lying in the plane of premarked lines
measured near the ends and the middle of the specimen. The length of
specimen also shall be taken be nearest 0.2 mm by averaging the two
lengths measured in the plane containing pre marked lines.
iii. Centre one of the plywood strips along the centre of the lower platen.
Place the specimen on the plywood strip and align it so that the lines
marked on the end of the specimen are vertical and centered over the
plywood strip. The second plywood strip is placed length wise on the
cylinder centred on the lines marked on the ends of the cylinder.
The assembly is positioned to ensure that lines marked on the end
of specimen are vertical and the projection of the plane passing through
these two lines interest the centre of the platen.
vii) Apply the load without shock and increase it continuously at the rate to
produce a split tensile stress of approximately 1.4 to 2.1 N/mm
/min, until no
greater load can be sustained. Record the maximum load applied to specimen
viii) Note the appearance of concrete and any unusual feature in the type of failure.
ix) Compute the split tensile strength of the specimen to the nearest 0.25 N/mm
Observations, Calculations & Results
Dia of the
Length of the
Breaking load (N)
Effect of height/diameter ratio on strength of cylinders
Standard cylinders are of height h equal to twice the diameter d, but sometimes specimens of
other proportions are encountered. This is particulary the case with cores cut from in situ
concrete: the diameter depends on the size of the core-cutting tool whereas the height of the core
varies with the thickness of the slab or member. If the core is too long, it can be trimmed to
the h/d ratio of 2 before testing but, with too short a core, it is necessary to estimate the
strength of the same concrete as if it had been determined on a specimen with h/d =2.
ASTM C 42-90 and BS 1881 : Part 120 : 1983 (the latter by implication) give correction
factors (Table(a)) but Murdock and Kesler found that the correction depends also on the
level of strength of the concrete
Table (a): Standard Correction Factors for Strength of Cylinders with Different Ratios of
Height to Diameter
Height to diameter ratio Strength correction
ASTM C 42-90
2.00 1.00 1.00
High strength concrete is less affected by the height/diameter ratio of the specimen, and
such a concrete is also less influenced by the shape of the specimen; the two factors should be
related as there is comparatively little difference between the strengths of a cube and of a
cylinder with hld=1.
The influence of strength on the conversion factor is of practical significance in the case of
low strength concrete, if cores with hid smaller than 2 are tested. Using ASTM C 42-90 and,
even more so, BS 1881 : Part 120:1983 factors, the strength that would be obtained with an
h/d ratio of 2 would be overestimated: yet, it is in the case of concrete of low strength, or
suspected of having too low a strength, that a correct estimate of strength is often particularly
Choice of the standard height/diameter ratio of 2 is suitable, not only because the end
effect is largely eliminated and a zone of uniaxial compression exists within the
specimen, but also because a slight departure from this ratio does not seriously affect the
measured value of strength. ASTM C 42-90 states that no correction is required for values
of h/d between 1.94 and 2.10.
The influence on strength of the ratio of height to the least lateral dimension applies also
in the case of prisms. Of course, if the end friction is eliminated, the effect of h/d on
strength disappears but this is very difficult to achieve in a routine test.
The end effect decreases more rapidly the more homogeneous the material; it is thus less
noticeable in mortars and probably also in lightweight aggregate concrete of,low or
moderate strength where a lower heterogeneity arises from the smaller difference between
the elastic moduli of the cement paste and the aggregate than is the case with normal weight
aggregate. It has been found that, with lightweight aggregate concrete, the value of the
ratio of strengths of a standard cylinder to a cylinder with a height—diameter ratio of 1 is
between 0.95 and
0.97. This has, however, not been confirmed in Russian tests on concrete
made with expanded clay aggregate where a ratio of about 0.77 was reported.
Comparison of strengths of cubes and cylinders
The restraining effect of the platens of the testing machine extends over the entire height
of a cube but leaves unaffected a part of a test cylinder. It is, therefore, to be expected that
the strengths of cubes and cylinders made from the same concrete differ from one another.
According to the expressions converting the strength of cores into the strength of equivalent
cubes in BS 1881: Part 120:1983, the strength of cylinder is equal to 0.8 of the strength of a
cube but, in reality, there is no simple relation between the strengths of the specimens of the two
shapes. The ratio of the strengths of the cylinder to the cube increases strongly with an
increase in strength and is nearly 1 at strengths of more than 100 MPa (or 14000 psi).
Some other factors, for example, the moisture condition of the specimen at the time of
testing, have also been found to affect the ratio of strengths of the two types of specimens.
Because European Standard ENV 206:1992 recognizes the use of both cylinders and
cubes, it includes a table of equivalence of strengths of the two types of compression
specimens up to 50 MPa (measured on cylinders). The values of the cylinder/cube strength
ratio are all around 0.8. The CEB—FIP Design Code gives a similar table of equivalence but,
above 50 M Pa, the cylinder/cube strength ratio rises progressively, reaching 0.89 when the
cylinder strength is 80 MPa. Neither of these tables should be used for purposes of conversion
of a measured strength of one type of specimen to the strength of the other type. For any
one construction project, a single type of compressive strength test specimen should be
It is difficult to say which type of specimen, cylinder or cube, is 'better' but, even in
countries where cubes are the standard Specimen, there seems to be a tendency, at least for
research purposes, to use cylinders rather than cubes and this has been recommended by
RILEM (Reunion Internationale des Laboratoires d'Essais et de Recherches sur les Materiaux et
les Constructions) — an international organization of testing laboratories. Cylinders are
believed to give a greater uniformity of results for nominally similar specimens because their
failure is less affected by the end restraint of the specimen; their strength is less influenced by
the properties of the coarse aggregate used in the mix; and the stress distribution on horizontal
planes in a cylinder is more uniform than on a specimen of square cross-section.
It may be recalled that cylinders are cast and tested in the same position. Whereas in a
cube the line of action of the load is at right angles to the axis of the cube as-cast. In
structural compression members, the situation is similar to that existing in a test cylinder,
and it has been suggested that, for this reason, tests on cylinders are more realistic. The
relation between the directions as-cast and as-tested has, however, been shown not to affect
appreciably the strength of cubes made with unsegregated and homogeneous concrete.
The stress distribution in any compression test is such that the test is only comparative
and offers no quantitative data on the strength of a structural member.
Tests for strength in tension
Although concrete is not normally designed to resist direct tension, the knowledge of tensile
strength is of value in estimating the load under which cracking will develop. The absence
of cracking is of considerable importance in maintaining the continuity of a concrete
structure and in many cases in the prevention of corrosion of reinforcement. Cracking
problems occur when diagonal tension arising from shearing stresses develops, but the
most frequent case of cracking is due to restrained shrinkage and temperature gradients. An
appreciation of the tensile strength of concrete helps in understanding the behaviour of
reinforced concrete even though the actual design calculations do not in many cases explicitly
take the tensile strength into account.
Strength in tension is of interest also in unreinforced concrete structures, such as dams,
under earthquake conditions. Other structures, such as highway and airfield pavements, are
designed on the basis of flexural strength, which involves strength in tension.
There are three types of test for strength in tension: direct tension test, flexure test, and
splitting tension test.
A direct application of a pure tension force, free from eccentricity, is very difficult.
Despite some success with the use of lazy-tong grips, it is difficult to avoid secondary
stresses such as those induced by grips or by embedded studs. A direct tension test, using
bonded end plates, is prescribed by the U.S. Bureau of Reclamation.
Influence on strength of moisture condition during test
The British as well as ASTM Standards require that all the test specimens be tested in a
'wet' or 'moist' condition. This condition has the advantage of being better reproducible than
a dry condition' which includes widely varying degrees of dryness.
Occasionally, a test specimen may not be in a wet condition, and it is of interest to
consider what are the consequences of such departure from the standard. It should be
emphasized that only the condition immediately prior to the test is considered, it being
assumed that usual curing has been applied in all cases.
As far as compressive strength specimens are
concerned, testing in a dry condition
leads to a higher strength. It has been suggested that drying shrinkage at the surface
induces a biaxial compression on the core of the specimen, thus increasing its strength in the
third direction, that is, in the direction of the applied load. However, tests have shown that
well-cured mortar prisms and concrete cores, when completely dried, had a higher
compressive strength than when tested wet. These specimens were not subject to differential
shrinkage so that there was no biaxial stress system induced. The behaviour of the specimens,
described above, accords also with the suggestion that the loss of strength due to wetting
of a compression test specimen is caused by the dilation of the cement gel by adsorbed
water: the forces of cohesion of the solid particles are then decreased. Conversely, when on
drying the wedge-action of water ceases, an apparent increase in strength of the specimen is
The effects of water are not merely superficial as dipping the specimens in water has
much less influence on strength than soaking. Soaking concrete in benzene or paraffin,
known not to be adsorbed by the cement gel, has no influence on strength. Re-soaking
oven-dried specimens in water reduces their strength to the value of continuously wet-cured
specimens, provided they have hydrated to the same degree. The variation in strength due to
drying appears thus to be a reversible phenomenon.
The quantitative influence of drying varies: with 34 M Pa (5000 psi) concrete, an increase
in compressive strength up to 10 per cent has been reported on thorough drying, but if the
drying period is less than 6 hours, the increase is generally less than 5 per cent. Other tests
have shown the increase in strength, in consequence of 48-hour wetting prior to test, to
be between 9 and 21 per cent.
Beam specimens tested in flexure exhibit behaviour opposite to that of compression
test specimens: a beam which has been allowed to dry before testing has a lower modulus of
rupture than a similar specimen tested in a wet condition. This difference is due to the
tensile stresses induced by restrained shrinkage prior to the application of the load which induces
tension in the extreme fibre. The magnitude of the apparent loss of strength depends on the rate
at which moisture evaporates from the surface of the specimen. It should be emphasized
that this effect is distinct from the influence of curing on strength.
If, however, the test specimen is small and drying takes place very slowly, so that internal
stresses can be redistributed and alleviated by creep, an increase in strength is observed. This
was found in tests on concrete beams, and also on mortar briquettes. Conversely, wetting
a completely dry specimen prior to testing reduces its strength; interpretation of this
phenomenon is controversial)
The strength of cylinders tested in splitting tension is not affected by the moisture
condition because failure occurs in a plane remote from the surface subjected to wetting or
The temperature of the specimen at the time of testing (as distinct from the curing
temperature) affects the strength, a higher temperature leading to a lower indicated strength,
both in the case of compression and of flexure specimens.
Influence of size of specimen on strength
The size of test specimens for strength testing is prescribed in the relevant standards, but
occasionally more than one size is permitted. Moreover, from time to time arguments in
favour of use of smaller specimens are advanced. This point out their advantages: smaller
specimens are easier to handle and are less likely to be accidentally damaged; the moulds
are cheaper; a lower capacity testing machine is needed; and less concrete is used, which
in the laboratory means less storage and curing space, and also a smaller quantity of
aggregate to be processed. On the other hand, the size of the test specimen may affect the
resulting strength and also the variability of test results. For these reasons, it is important to
consider in detail the influence of the size of specimen on strength test results.
Concrete composed of elements of variable strength is reasonable to assume that the larger
the volume of the concrete subjected to stress the more likely it is to contain an element of a
given extreme (low) strength. As a result, the measured strength of a specimen decreases with an
increase in its size, and so does the variability in strength of geometrically similar specimens.
Because the influence of size on strength depends on the standard deviation of strength. It
follows that the size effects are smaller the greater the homogeneity of the concrete. Thus, the
size effect in lightweight aggregate concrete should be smaller, but this has not been confirmed
with any degree of certainty, although there is some support for this suggestion in the
In the case of tests on the strength of concrete, we are interested in the averages of extremes as a
function of the size of the specimen. Average values of samples chosen at random tend to
have a normal distribution, so that the assumption of this type of distribution, when
average values of samples are used, does not introduce serious error, and has the advantage
of simplifying the computations. In some practical cases, a skewness of distribution has
been observed; this may not be due to any 'natural' properties of concrete but to the
rejection of poor quali t y concret e on the si te so that such concret e never reaches
the testing stage.
Size effects in tensile strength tests
Direct tension tests on cylinders of concretes with compressive strengths between 35
and 128 M Pa (5000 and 18 500 psi) were performed by Rossi et al.
They confirmed the
decrease in tensile strength and also in variability of test results with an increase in size: the
decrease in strength is larger the lower the strength of concrete. The coefficient of variation
also decreases with an increase in size of the specimen, but there is no apparent effect of the
strength of concrete on this relation. Rossi et a1.
explain this influence of strength in terms of
the heterogeneity of the mix components. Specifically, the size effect is a function of the ratio of
the specimen size to the maximum size of aggregate and of the difference in strength between the
aggregate particles and the surrounding mortar. This difference is small in very high
strength concrete and also in lightweight aggregate concrete.
Splitting tension tests on 150 mm diameter by 300 mm high (6 by 12 in.) cylinders and
100 mm diameter by 200 mm high (4 by 8 in.) cylinders have given an average ratio of the
strength of the former to the latter of 0.87; the average splitting tension strength of the
larger cylinders was 2.9 MPa (415 psi). The standard deviation for the larger cylinders was
0.18 M Pa (26 psi) and, for the smaller, 0.27 MPa (39 psi). The coefficients of variation
were, respectively, 6.2 and 8.2 per cent. It is worth observing that the coefficient of variation of
the splitting tension strength of 150 by 300 mm cylinders had nearly the same value as the
coefficient of variation of the modulus of rupture determined on beams with a 150 by 150
mm (6 by 6 in.) cross-section made of the same concrete.
The influence of the cylinder size on splitting tension strength was confirmed by Batant et
al. on the basis both of their own tests on mortar discs and also on the basis of tests on
concrete cylinders performed by Hasegawa et al. In both these series of tests, the size effect
disappears in large-size specimens
This topic is discussed in the next section.Cement compacts
have also been found to show the size effect when tested in splitting tension. The same applies in
the case of the ring test.
Size effects in compressive strength tests
It is interesting to note that the size effect disappears beyond a certain size so that a further
increase in the size of a member does not lead to a decrease in strength, both in
compression and in splitting tension. According to the U.S. Bureau of Reclamation, the
strength curve becomes parallel to the size axis at a diameter of 457 mm (18 in.), i.e. cylinders
of 457 mm (18 in.), 610 mm (24 in.), and 914 mm (36 in.) diameter all have the same
strength. The same investigation indicates that the decrease in strength with an increase in
size of the specimen is less pronounced in lean mixes than in rich ones. For instance, the
strength of 457 mm (18 in.) and 610 mm (24 in.) cylinders relative to 152 mm (6 in.) cylinders
is 85 per cent for rich mixes but 93 per cent for lean (167 kg/m3 (282 lb/yd
These experimental data are of importance in refuting a speculation that, if the size effect
is extrapolated to very large structures, a dangerously low strength might be expected. Evidently
this is not so because local failure is not tantamount to collapse.
The various test results on the size effect are of interest because size effects have been
ascribed to a variety of causes: the wall effect; the ratio of the specimen size to the maximum
aggregate size; the internal stresses caused by the difference in temperature and humidity
between the surface and the interior of the specimen; the tangential stress at the contact
surface between the platen of the testing machine and the specimen due to friction or
bending of the platen; and the difference in the effectiveness of curing. In this connection,
Day and Hague showed that the relation between the strength of 150 by 300 mm (6 by 12
in.) and 75 by 150 mm (3 by 6 in.) cylinders is not affected by the method of curing.