The last stage blades of low-pressure steam turbine rotors are among the most highly-stressed components in modern power generating plants. The ongoing drive for increased efficiency has seen a proliferation in the number of designs incorporating larger last stage blades with curved axial entry fir tree roots (CAEFTR); the curvature of the root attachment allows more flexibility in the aerodynamic design of the aerofoil with improved inter-blade spacing. The tendency of CAEFTRs to suffer failure induced by stress corrosion cracking, high-cycle fatigue cracking or low-cycle high-strain fatigue is well documented and can be shown to be most likely to occur in the first two serrations of the blade root. Finite element analysis and actual failures confirm the regions under highest risk and have driven developments in ultrasonic phased array techniques to achieve detection of defects in these regions whilst in situ, in turn avoiding the huge costs associated in decommissioning and dismantling rotors to perform alternative NDE surface inspections. Due to the complexity of the root geometry, there are many difficulties in applying ultrasonic techniques due to limited scanning surfaces, inter-blade spacing and disorientation of the active ultrasound trajectory and the region under test. In this presentation the author will show that the application of novel phased array techniques and unique inspection design has led to increased sensitivity to smaller defects and comprehensive coverage of CAEFTR rotor designs in situ. It will also be shown how the application of these techniques has negated the need to upgrade both equipment and resources to the use of2D arrays, thereby reducing inspection costs significantly whilst achieving higher repeatability and sensitivity. Novel inspection design has led to the ability to reduce the number of scans required, enabled single scan encoded data recording over the whole blade root, increased sensitivity, improved detectability and increased coverage, whilst reducing inspection costs and time.
Read full abstract