Structural hardware testing can be a form of comparative testing during the development phase.  That means that the test conditions do not always have to be an exact simulation of service conditions as long as the variables considered are tested the same way.  However, it is strongly recommended that service conditions be approximated as closely as possible.  How closely the test conditions have to resemble service conditions depends upon the predictability of the effect of a change in conditions.

The following guidelines are applicable to structural hardware testing for damage tolerance.  First, the specimen should contain the design and manufacturing details that are the subject of the investigation.  The load should be properly distributed at the point of interest.  Second, if the purpose is to validate a piece of structure for damage tolerance, then load sharing, load interaction, and load transfer among different members should be simulated or otherwise accounted for.  Type of loading (bending, tension) should be as in service, or be such that the stress distribution at the critical location is as in the actual structure.  Special care should be taken that no undesired bending is introduced due to load eccentricities.  This requires intelligent grip design.  It may also require some special structure to distribute the loads properly from these areas into the specimen.  Third, the nominal stress at the critical location should be as in service.  Experiments should be performed on a flight-by-flight basis with landing loads included.  A reasonable number of stress levels should be used.  The stress sequence within a flight should be representative of service usage (see Section 5) or arranged in lo-hi-lo sequence.  Block loading should not generally be applied.