Vibration is probably the most common type of repeated cyclic stress, and fatigue failures are among the most common service failures.  Fatigue strength is therefore an important property to be considered in designing any part which will be exposed to vibration or any type of frequent, intermittent loadings.

Testing specimens at different stress levels (S) and measuring the number of cycles-to-failure (N) produces an S-N curve.  Thant S-N curve allows designers to make a direct estimate of the expected life of the part interm of stress - a basic design parameter.  By locating on the graph the number of cycles similar to that expected during the service life of the part, designers can quantify the appropriate design stress.

There are two factors to remember when working with S-N curves.  First, the validity of the values for a particular application depends on the similarity of the test conditions to those of the actual use environment.  Second, a safety factor must be included in design calculations to compensate for possible flaws in the part  - such as voids, foreign material embedded in the plastic, or surface scratches which act as stress concentrators.

Changing any variable that affects heat transfer can cause a shift in the S-N curve.  Such variables include the frequency of loading cycles, specimen thickness, and ambient temperature.  Hysteretic heating (caused by constant loading-unloading) can be important factor in fatigue failure and the designer, in order to minimise heating and its  effects, should:

• Reduce stress by adding ribbing, or by redidtributing the load .
• Avoid stress concentrators such as sharp corners or abrupt changes in cross sectional geometry.
• Reduce the frequency of loading cycles if possible.
• Avoid unnecessarily thick walls which reduce heat transfer and thermal dissipation.