S. Sridharan, E. Jordan, M. Gell, University of Connecticut, Storrs, CT

Summary:

Remaining life assessment of thermal barrier coated (TBC) service components, following field service, is essential to determine whether the parts can be reliably used until the next planned inspection, to prevent unplanned shut downs and minimal part utilization. Premature spallation failure of TBC’s during service, resulting in the exposure of turbine components to hot gases and high temperatures has increased the need for a non-destructive technique to detect early damage and assess the remaining life. In this context, a measurement approach, based on an experimental study of the oxide stress evolution as a function of thermal cycling has been developed for detecting TBC damage and assessing remaining life. Stress measurements have been made on thermally cycled TBCs, subjected to multi-temperature and multiple hold time furnace cyclic tests using photo-stimulated luminescence piezospectroscopy (PLPS). Monotonic trends observed in the measured stress in the thermally grown oxide layer (TGO) versus cycles data obtained on this TBC system, cycled at a single temperature, has shown great promise in being able to determine TBC remaining life and has provided remaining life assessments to within ± 7 % accuracy, by using simple regression and neural network techniques. These results compare well with the current industrial life prediction models that only provide a 2 X factor of TBC life predictions. Stress data obtained from over seventy TBC specimens has offered us further scope to extend this remaining life prediction assessment, for the first time, to multiple temperatures and hold times. Besides the non-destructive stress measurements, TGO thicknesses and the initial bond coat surface geometry have also been measured non-destructively – TGO thickness from impedance measurements and the initial bond coat surface geometry using surface interferometry and these three parameters have been used to develop a simple and accurate TBC life prediction method.