Uniaxial testing in compression is in many ways simpler and easier than in tension. There are no concerns about gripping and no possibility of necking or other localised plasticity. The sample is usually a simple cylinder or cuboid. However, there are some potential difficulties. One of these is the danger of (plastic) buckling, particularly if relatively large strains (>~10%) are to be created. In order to avoid this, the aspect ratio (height / diameter) must be kept relatively low - probably not much more than unity. Since a very large sectional area might lead to excessive load requirements, this often means that the height (gauge length) of the sample is limited. This in turn leads to relatively low displacements, placing a premium on measurement accuracy (with the points made about compliance calibration, when referring to tensile loading, applying equally here).
Effect of Friction between Sample and Platen
There are also concerns about the effect of friction. This is potentially important, since one outcome of friction is that the stress and strain fields become non-uniform - see the simulation below, so that the nominal stress-strain curve cannot be converted to a true version via use of the analytical equations (even if the value of the coefficient of friction, μ, is known).
In practice, it is common to apply a lubricant to the contact surfaces of the sample and to assume that any effect of friction will be small. However, the high contact pressure tends to force lubricant out of the region between platen and sample, so this assumption may not be valid.
Two extreme cases can be identified. The first, which is commonly assumed, is that there is unhindered sliding at the interface (μ = 0). The sectional area will remain uniform along the sample length during deformation (no barrelling) and there is no frictional work. The complementary limiting case is that of no sliding. This also involves no frictional work, but barrelling occurs from the start of the test. While the exact shape of the “barrel”, as a function of the applied load, will depend on the aspect ratio, it is clear that significant barrelling will invalidate the test - the stress now varies along the length of the sample and the relationship between the true stress-strain curve and the outcome will be a complex one.
In practice, there is likely to be at least some frictional sliding (with (μ > 0, so that energy is dissipated), but also some barrelling. The sliding is likely to occur over only part of the surface, since the interfacial shear stress rises with increasing distance from the loading axis (where it is zero). The outcomes of this can be explored using the simulation below, which is based on FEM modelling.
Simulation 7: FEM simulation of a compression test