Predicting Stresses in Thermal-Barrier Coatings

John H. Glenn Research Center, Cleveland, Ohio

A methodology for predicting stresses and the resultant cracking in plasma-sprayed thermal-barrier coatings (TBCs) has been developed. The methodology is built around a computer code that implements a finite-element model that simulates the evolution of stresses, strains, and related phenomena in a TBC. The economic and technological value of the methodology lies in its potential to provide a more systematic basis for designing reliable and durable TBCs for advanced gas turbine engines by reducing the amount of time-consuming empirical testing needed to assess alternative TBC designs.

A TBC typically comprises a ceramic top coat deposited on an alloy bond coat that has been deposited on a superalloy substrate that the TBC is intended to protect. Together, the TBC and the substrate constitute a complex material system. Finite-element modeling is necessary to predict the complex, interactive thermomechanical and chemical behaviors of the component material layers.

This Small Part of the Finite-Element Model shows the geometric relationships of the various layers and the simulation of surface roughness by use of a sinusoidal waviness.

The present finite-element model is capable of making such predictions over a range of conditions to which TBCs are exposed during operation of engines and during thermal cycling in burner test rigs. Originally developed within the framework of the Lawrence Livermore National Laboratory code NIKE2D, the model has recently been implemented into a new framework using the commercial FEA code ABAQUS™, a product of Hibbitt, Karlsson & Sorensen, Inc. Variable phenomena that are represented in the model include transient thermal behavior, multiple thermal cycles, and oxidation of the bond coat. The initiation of fractures and the propagation of cracks are represented by a stress-based crack-initiation criterion and a fracture-mechanics submodel. The creep behavior of each constituent material is represented by a temperature-dependent power-law creep submodel.

Oxidation of the bond coat is represented by use of provisions within the codes that allow for the transformation of constituent materials. The bond-coat oxide starts at the bond-coat/top-coat interface and grows into the thickness of the bond coat. In the model, on the basis of empirical data on the rate of growth of an oxide layer on a bond-coat material, bond-coat-alloy finite elements are replaced by bond-coat-oxide finite elements at fixed intervals of time, starting at the bond-coat/top-coat interface.

The model can be used to assess the effects of numerous material, process, or geometric variables on the stress behavior within a TBC. Five principal variables originally studied and characterized during model development were oxidation, bond-coat creep, top-coat creep, bond-coat thermal expansion, and interfacial roughness. The outputs of the model include stresses as functions of time, location, and direction. The model has been applied to a burner-rig specimen, which is a 25.4-mm-diameter rod of Waspalloy (or equivalent superalloy) with a side TBC composed of 0.13-mm-thick NiCrAlY bond coat and a 0.25-mm-thick top coat comprised of a mixture of zirconia with 8 weight percent yttria. To simulate the effect of surface roughness of a typical TBC, the radial coordinate of the bond-coat/top-coat interface was made to vary as a sinusoidal function of the circumferential coordinate, with a peak-to-valley amplitude of 10 µm (see figure).

The numerical results of the application have been interpreted as signifying that oxidation of the bond coat exerts a strong effect on stresses in the ceramic layer and that stresses induced by oxidation are influenced by other factors, including bond-coat creep, top-coat creep and bond-coat roughness. It was also concluded that the progression of cracking is a result of the combined action of creep, oxidation, and thermal cycling. An accurate description of the entire process requires a model including these factors. It was further concluded that as complex as the model is, it is still too simple to provide a complete description of the failure of a TBC, and that for greater accuracy, it would be necessary to account for such other factors as (1) sintering phase changes in the oxide, bond coat, and ceramic layers and (2) cracking and changes in composition in the ceramic layer.

This work was done by Andrew Freborg and B. Lynn Ferguson of Deformation Control Technology, Inc., for Glenn Research Center. For further information, access the Technical Support Package (TSP) free on-line at www.nasatech.com/tsp under the Materials category.

Inquiries concerning rights for the commercial use of this invention should be addressed to