Project Spotlight
The turbine end of the LP section of the turbine experienced a blade failure (see missing blade in Figure 1). Upon inspection, it was observed that several more blades had cracks occurring about ¼ from the blade root (see Figure 2). Furthermore, the cracks tended to occur on blades associated with the free edge of a shroud. The turbine had operated successfully for many years prior to the failures.
EDI performed a modal analysis of the turbine wheels at both ends of the rotor. Both wheels were investigated since only the turbine end wheel had failed. It was hoped that comparisons of the dynamical characteristics of the turbine end wheel and generator end wheel would show a difference that could have led to failure.
Because of the high modal density of this type of structure, a strain gage installed at the approximate location of the failure was used for a reference. This deemphasized blade modes not associated with the failure. A 1 gram accelerometer was attached at the various points on all blades. Impact response data from these transducers was used to obtain the modal parameters (see Figure 3).
The data showed several responses, however a group response near 260 Hz exhibited the proper shape to have caused the experienced failures as shown in this animation (500 Kb). It was still unclear; however, what caused the turbine to fail after so many years of successful operation.
No significant differences between the turbine-end wheel and generator-end wheel blade dynamics were observed. However, metallurgical analyses of a blade indicated that the blades were experiencing corrosion pitting. Such pitting would cause stress-intensification in the blades.
It is possible that a difference in the forcing function (e.g., the torsional mode shape or steam pulsation) could have been responsible for the failures occurring only at the turbine-end wheel. However, the client elected not to pursue the additional testing and analysis required to investigate these hypotheses. Instead, the client elected to simply re-blade the wheels. Note that because of the suspected mode-shape of the failures, the middle two or three blades in each group could have been salvaged. It is unknown whether the client elected to reuse the middle blades or if all blades were replaced.
