Good high quality field failure data has often been described as the ultimate source of failure data. However, not all field failure studies are high quality. Some field studies simply do not have the needed information. Some field studies make unrealistic assumptions. The results can be quite different depending on methods and assumptions. Some methods produce optimistic results that can result in bad designs and unsafe processes.
This paper presents some common field failure analysis techniques, shows some of the limitations of the methods and describes important attributes of a good field failure data collection system.