Detection of Cracks in Turbomachinery Blades by Online Monitoring

Abstract

The presence of a crack in a blade can change the natural frequencies of that blade. It has long been a goal to detect blade cracks by assessing the change in a measured vibration frequency of the blade over time. It has been found that prior frequency assessment methods can be less accurate than is desirable to reliably detect the relatively small frequency changes that are typically associated with blade crack sizes of practical interest. This paper describes a method in which potential temporal changes in the frequencies of individual blades are assessed by periodically analyzing complete rows of blades using mistuning analysis techniques that treat the blade rows as coupled systems, in contrast to other techniques that consider each blade individually in turn. This method, while computationally complicated and challenging, has been found to be capable of detecting blade root cracks that are much smaller than those that can be detected using other techniques. Moreover, this method has been demonstrated to detect cracks that are much smaller than the critical size for mechanical separation of the blade from the rotor. This improved frequency assessment technique has been used to identify more than 30 blades with frequency changes that were considered to be potential indicators of blade cracks. Subsequent inspections verified indications in all of those blades. In addition to providing operational guidance, the frequency change data were used to infer the time periods during which crack growth had occurred.