Dissemination of IT for the Promotion of Materials Science (DoITPoMS)


Tensile Testing - Practical Basics

The uniaxial tensile test is the most commonly-used mechanical testing procedure.  However, while it is simple in principle, there are several practical challenges, as well as a number of points to be noted when examining outcomes. 

Specimen Shape and Gripping

A central issue concerns the specimen shape.  The behaviour is monitored in a central section (the “gauge length”), in which a uniform stress is created.  The grips lie outside of this section, where the sample has a larger sectional area, so that stresses are lower.  If this is not done, then stress concentration effects near the grips are likely to result in premature deformation and failure in that area.  Several different geometries are possible.
A typical sample and gripping system are shown in the photo below.

Tensile test sample set up

Photo 1: Static image showing a typical tensile specimen and set of grips

Measurement of Load and Displacement

All testing systems have some sort of “loading train”, of which the sample forms a part.  This “train” can be relatively complex  -  for example, it might involve a rotating worm drive (screw thread) somewhere, with the force transmitted to a cross-head and thence via a gripping system to the sample and then to a base-plate of some sort.  It does, of course, need to be arranged that, apart from the sample, all of the components loaded in this way experience only elastic deformation.  The same force (load) is being transmitted along the complete length of the loading train.  Measurement of this load is thus fairly straightforward.  For example, a load cell can be located anywhere in the train, possibly just above the gripping system.  In some simple systems, such as hardness testers or creep rigs, a fixed load may be generated by a dead weight.

Measurement of the displacement (in the gauge length) is more of a challenge.  Sometimes, a measuring device is built into the set-up  -  for example, it could measure the amount of rotation of a worm drive.  In such cases, however, measured displacements include a contribution (elastic) from various elements of the loading train, and this could be quite significant.  It may therefore be important to apply a compliance calibration.  This involves subtracting from the measured displacement the contribution due to the compliance (inverse of stiffness) of the loading train.  This can be measured using a sample of known stiffness (ensuring that it remains elastic).

Displacement Measuring Devices

Several types of device can be used to measure displacement, including Linear Variable Displacement Transducers (LVDTs), eddy current gauges and scanning laser extensometers.  These have resolutions of the order of 1 µm.  More specialised (and accurate) devices include parallel plate capacitors and interferometric optical set-ups, although they often have more limited measurement ranges.

Alternatively, displacement can be measured directly on the gauge length, eliminating concerns about the system compliance. Devices of this type include clip-gauges (knife edges pushing lightly into the sample) and strain gauges (stuck on the sample with adhesive). The latter have good accuracy (±0.1% of the reading), but are limited in range (~1-2% strain). They are useful for measurement of the sample stiffness (Young's modulus), but not for plastic deformation. A versatile technique, useful for mapping strains over a surface, is Digital Image Correlation (DIC), in which the motion of features ("speckles") in optical images is followed automatically during deformation, with displacement resolutions typically of the order of a few μm.

In the video below, which also shows the development of the stress-strain curve, clip gauges are being used to measure the displacement, and hence the strain. It can be seen that, when the strain reaches a certain level (~25% in this case), the specimen starts to"neck". This is apparent in the video and it can also be seen that the onset of necking coincides, at least approximately, with a plateau (peak) in the nominal stress – nominal strain curve. This important phenomenon is examined in more detail on the next page.


Video 1: Tensile testing of annealed Cu sample (video and evolving nominal stress-strain plot)