An analytical model for simulation of heat flow in plasma-sprayed thermal barrier coatings

Numerical (finite difference) and analytical models have been developed for the simulation of heat flow through plasma-sprayed coatings, allowing the effective thermal conductivity to be predicted as a function of microstructural parameters. The structure is assumed to be composed of lamellar materi...

Full description

Bibliographic Details
Main Authors: Golosnoy, I.O (Author), Tsipas, S.A (Author), Clyne, T.W (Author)
Format: Article
Language:English
Published: 2005-06.
Subjects:
Online Access:Get fulltext
Description
Summary:Numerical (finite difference) and analytical models have been developed for the simulation of heat flow through plasma-sprayed coatings, allowing the effective thermal conductivity to be predicted as a function of microstructural parameters. The structure is assumed to be composed of lamellar material (splats), separated by (thin) pores, within which there are areas of contact (bridges). The analytical model is based on dividing the material into two regimes, within which the heat flow occurs either by unidirectional serial flow through lamellae and pores or by being funneled through the regions of the lamellae above and below the bridges. The validity of this model is demonstrated by a comparison of the predictions obtained from it and those obtained from the numerical model. The effects of pore geometry on conductive and radiative heat transfer within the coating have been investigated over a range of temperatures and gas pressures. It is shown that the main factor controlling the conductivity is the intersplat bridge area. Comparisons are also presented with experimental conductivity data, for cases in which some attempt has been made to characterize the key microstructural features. The study is oriented toward thermal barrier coatings, based on zirconiayttria top coats. It is noted that the effect of microstructural sintering, which tends to occur in these coatings under service conditions, can be predicted using this model.