There are three categories of graphite creep experi­ments. The first type was the restrained creep experiments. In these experiments a graphite specimen, usually dumbbell in shape, is restrained from shrinkage by a tube or split collar manufactured from graphite that shrinks less than the specimen of interest. In the case of anisotropic graphite, the tube or split collar could be manufactured with its longitudinal axis aligned with the more dimension­ally stable grain direction; the specimen would be manufactured with its axis perpendicular to this. These types of experiments are relatively easy to deploy but are difficult to assess as the load is not directly measured and a ‘creep law’ has to be assumed in the assessment of the results.88 A variation on these experiments was the graphite spring tests used in Calder Hall to define primary creep.86

A second important type of experiment was the ‘out-of-pile measurements’ technique.89,90 These experiments, importantly, give ‘real-time results.’ However, this type of experiment is difficult to install in a reactor and results are obtained only for one specimen.

The final type of experiment is the in-pile rig loading using a string of samples and usually taking advantage of the MTR flux gradient to obtain data on samples to various levels of fluence. There have been various designs of simple strings of specimens loaded either in tension or in compression. Tensile creep tests are vulnerable, that is, if one specimen fails, results for the whole string of specimens could be lost. However, there have been various rig designs aimed at overcoming this problem.