Because fatigue analysis is based on a proportional-load assumption, the load is applied in step format. Automatic time-stepping is disabled (AUTOTS,OFF), as it has no effect on convergence and accuracy in load-step cases.

It is often desirable to control output frequency to the results file (.rst) by suppressing selected element or nodal solutions (OUTRES). For a crack-initiation and -growth analysis, however, it is good practice to monitor and examine fracture calculations on all load steps and substeps during the entire crack-growth process (OUTRES,CINT,ALL). The fracture-calculation data set is generally small compared to element and nodal data sets and has only a minimal effect on output file size.

The simulation uses two load steps, the first to apply a trivial load, the second to apply the true load. This method completes the local mesh adjustment around the edge-crack front in the first load step so that results are independent of the original mesh. The recommended magnitude of the trivial load ranges from 1e-8 to 1e-4 of the magnitude of the true load. The effect of local mesh adjustment is discussed in Edge-Crack Growth.