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

DoITPoMS Teaching & Learning Packages Creep Deformation of Metals Uniaxial Creep Testing - Practical Basics

# Uniaxial Creep Testing - Practical Basics

Many of the practical issues outlined in the TLP on Mechanical Testing (for Plasticity) apply equally to Creep Testing.  For tensile testing, the gauge length must have a smaller sectional area than the region that is gripped, such that the latter undergoes only elastic deformation.  This is not the case for compressive testing, where the sample usually has a uniform section along its length, and is located between hard platens.  For Creep Testing, however, there are additional challenges.  For example, it is commonly carried out at high temperature, so a furnace (with good thermal stability) is needed, and all of the sample must be held at the selected temperature.  Also, the load must be sustained for long periods  -  perhaps just a few hours, but in some cases periods of many weeks or even many months might be needed.  Such conditions bring slightly different challenges from those of conventional (stress-strain) testing.

## Nominal or True Levels of Stress and Strain?

As for conventional testing, there is the issue of true or nominal versions of both stresses and strains.  However, there is a difference with Creep Testing.  For Stress-Strain Testing, the applied load is being ramped up continuously, so the issue reduces to that of converting any particular load level to a stress in the sample.  Provided the stress is uniform throughout the sample, a simple equation can be used to convert an applied load to a true stress.  A similar operation allows an extension (nominal strain) to be converted to a true strain.   There is a rather more fundamental problem with Creep Testing.  As noted in the Introduction page of this TLP, creep is characterised by a series of strain-time curves for different levels of (true) stress.  Under fixed load conditions, the true stress changes continuously during the test, so these curves will be unreliable (for tests in which relatively large strains are generated).