Such test assemblies contain defects to detect, size and evaluate. Generally, such defects are either very artificial (calibration holes, crotches) or induced by particular techniques to stimulate as much as possible real defects. They are called realistic defects.

The success of this simulation is often judged visually, after destructive examination. The defect is qualified as being "realistic" if it resembles, for the observer, a real defect.

Unfortunately, the defect should not be introduced to satisfy the eye of the observer, but well to resemble to a real defect for the particular NDT technique used. It is the NDT technique considered that must "declare" if the defect is a good simulation or not.

It is obvious that human view has very different perceptions than the ultrasonic "view", than the X-tray "view" and than the Eddy Currents "view". A defect will be realistic for these techniques if it produces, when detected by these techniques, a signal similar to the one of the natural defect it must resemble.

Very often, simulations, apparently very realistic for the human observer, are completely wrong for particular NDT techniques. Many qualifications are performed on these defects, which are unrealistic for NDT. Vice-versa, several artificial defects, which could hurt the human observer looking for realism, are good defects for specific NDT techniques and hence must be used for NDT systems performance evaluation and qualification trials.