Testing - Concrete Mix Design, Quality Control and Specification - page 136

machine may be little more than the difference in the cost of cement
requirement according to two different machines per month.
7.3.2 Testing machine technology
Obviously a correct result will not be obtained unless the stress is uniformly
distributed over the test specimen (and any deviation in this respect will
lead to a lower result).
An assumption is made that the faces of both the test specimen and the
testing machine platen are absolutely flat and that the load will be applied
concentrically. Quite small differences in planarity can make very large dif-
ferences in contact area and therefore in stress distribution. With cube spec-
imens this problem will worsen with older and higher strength specimens
because the older concrete (i.e., 28 day rather than 7 day) will be more
rigid, that is less subject to plastic distortion. With cylinders the problem
is different. Here the capping compound (e.g., where sulfur caps are used)
will flow equally at any age. The platen planarity may be slightly less criti-
cal but any plastic flow allows stress concentrations to develop unless the
original cylinder ends are very close to flat.
Spherical seatings are provided to allow one platen to rotate to compen-
sate for any tendency for the two opposite faces of the test specimens not to
be exactly parallel. This introduces its own problem in that, if the spherical
seating were effective during the whole test, any eccentricity at all would
lead to a bending moment in addition to an axial force, so reducing the
failure load. Therefore spherical seatings must be lubricated with a very
light machine oil specifically so that the oil will break down under pressure
and allow the seating to lock solid after an initial adjustment. Extreme
pressure lubricants, such as graphite grease, must be avoided, as they will
produce lower and more variable results. For cubes this is even more impor-
tant because, since the specimen is tested perpendicular to the direction of
casting (and therefore water gain or bleeding), its physical center may not be
its “center of resistance”, that is, if the cube is stronger at the bottom than
at the top, its center of resistance would be displaced toward the previously
bottom face when turned on its side for testing.
A further influence of the platen-specimen interface, again especially
with cubes, is that friction provides a lateral restraint to the Poisson's ratio
spreading effect and so increases the test strength. Day (inadvertently) dem-
onstrated this many years ago when he tested cubes coated with a wax curing
compound. The compound may have increased the actual concrete strength,
but it certainly caused a drastically reduced load at failure. The reason for
test cylinders to have a height to diameter ratio of 2 is to avoid this effect in
the central area where failure actually takes place. This is probably the main
reason for the difference between the test strength of cubes and cylinders
from the same concrete. It may also be the reason why this effect is reduced