The simulation of random service loads by cyclic loads is discussed. It is noted that their similarity can at present be shown on an empirical basis only. A few test programs are reviewed; a clear picture has not yet been obtained.

The results of program-fatigue tests on notched light alloy specimens and structures show that ∑N¯n values are on the average larger than 1; and macrocrack propagation in sheet material is slower than predicted by the Palmgren-Miner rule. In addition to the amount of microcracking, the second damage parameter of residual stress at the tip of the crack is of great importance. Both parameters allow a qualitative understanding of the deviations from the Palmgren-Miner rule.

The rule is a useful tool to make rough life estimates in the early design stage, but the major difficulty is to obtain relevant and accurate S-N data. The rule is found to be unreliable in judging whether a certain type of service loading would contribute substantially to the damage induced by other loads.

Full-scale testing is considered to be indispensable to establish the fatigue performance of a new aircraft design. With present-day experimental facilities, a simulation of the anticipated load-time history in service is advised for such a test. Since program-fatigue testing may be useful for design studies, recommendations are made for planning the load sequence in this type of test.