Characterization of Gas Turbine
Blade Layers Using
Inclusive Dates: 07/03/00 - 11/03/00
Background - Advanced gas turbines are increasingly used for electrical power generation. The turbine blades are protected from high-temperature oxidation and corrosion by metallic coatings made of CoCrAlY or NiCoCrAlY materials, often with a top layer of aluminum. As the coating ages, aluminum is depleted from the coating structure to support a protective aluminum oxide layer on the outer surface. The remaining service life of these blades depends on the state of degradation of the coating and, in particular, the remaining aluminum content. A nondestructive evaluation (NDE) technique is needed to provide a measurement of the aluminum content or the thickness of the aluminum-rich layers in the coating to help determine remaining coating life.
The Electric Power Research Institute (EPRI) has initiated a program to develop a field-deployable NDE system for life assessment of blade coatings. As a first step in this program, EPRI performed a nonfunded, round-robin testing program to identify the most suitable NDE technique using well-characterized test samples. EPRI requested that SwRI be one of the test participants.
Approach - A previously developed SwRI model based on multifrequency eddy current testing (ECT) was further developed and validated. The model used ECT measurements to predict layer thicknesses associated with degradation of PWA-286 gas turbine blade coatings. This capability allowed SwRI to participate in the EPRI round-robin test and provide test results to EPRI to show potential of the Institute’s method for future EPRI funding. It also allowed SwRI to take advantage of the well-characterized EPRI test specimens to validate the model.
Accomplishments - The model was further
developed to allow determination of the thickness of the aluminum-rich
layer of the coating, as well as the thicknesses of the inner and outer depleted
layers. The model was tested and validated on a set of initial EPRI specimens in
which the coating thicknesses were made known to SwRI. Good agreement was
obtained between predicted and actual thicknesses, particularly the
layer, which is of most importance. Blind tests were then performed on
additional specimens. The specimens were later examined metallurgically by
EPRI, and layer thicknesses were provided to SwRI. The model predicted the layer
thicknesses reasonably well with an rms error of 18.5 micrometers.