Development of a Modal-Based Turbomachine Blade-Disk Attachment Inspection Technique

Abstract

Turbine blade failures are among the leading causes of steam turbine failure. Failure types typically include cracking, rubbing, blade fouling, and foreign object damage. There is currently a range of non-destructive testing methods used to detect damage at the blade-disk attachment zone, all of which involve disassembling of the blade from the disk for periodic inspection. Evidence indicate that a method to detect damage at the blade-disk attachment zone using a non-contact, non-destructive in-situ off-line modal-based structural health monitoring technique could be useful under some circumstances. Such a technique would have the advantage of eliminating the necessity to disassemble blades during inspection. This would result in significant cost savings. Also, defects associated with the disassembly and reassembly of blades would be avoided. Thus, the aim of this study was to develop a modal-based turbomachinery blade disk attachment inspection technique. Modal parameters were acquired from a robust experimental modal analysis of freely supported low-pressure steam turbine blade-disk segment assemblies. Artificial single-location cracks were intentionally introduced into the turbine blades by cutting a 1 mm thickness notch at three probable damage locations, namely, at the upper pinhole on the leading-edge pressure side, above the root at the base of the aerofoil on the leading-edge and on the trailing-edge. In this work, a finite element analysis of the bladed disk segment assemblies was carried out with and without damage. To validate the reliability of the numerical models, the numerical results were correlated with the measured values, the results of which showed a strong correlation. Finally, a parametric study was conducted in which various healthy and damaged blade-disk cases were systematically investigated. This was done to examine the sensitivity of the blade natural frequency to damage. The artificial damage above the root was found to cause the largest changes in natural frequency. These changes were even more pronounced for assemblies with two blades. Receiver operating characteristic curves were used to assess the discriminatory ability of the results. Each damage case was found to be unique and therefore identifiable from its corresponding healthy case.