Optimised wavelet-based local tomography has been found to be a useful non-destructive evaluation tool for studying the microstructure in thermal barrier coatings. Two-dimensional wavelet-based local tomography produced high-resolution images of regions inside the investigated samples. The investigated samples were cur from two thermal shock tested jet engine burner cans. The inside of the burner cans was coated with a thick thermal barrier coating consisting of a top coating made of partially stabilised zirconia (PSZ) and a bond coating (BC) of NiCoCrAlY. The coatings were manufactured with plasma spraying with two different spraying parameters. This resulted in samples with and without a segmented crack network in the as-sprayed top coating.
The obtained wavelet-based local CT-imagcs of the investigated samples from the thermal shock tested burner cans reveal cracks and pores in the microstructure. The obtained pixel-sizes in the local CT-images were 4.0 μm and 3.1 μm respectively. Small or no improvements in resolution arc made when the pixel-size is in the vicintiy of the X-ray focal spot size, in this case 5 μm. A traditional CT-investigation, collecting global data that covers the whole cross-section of the sample, would have resulted in 2.6 and 6.8 times bigger pixel-sizes respectively. A CT-image reconstructed from global data would therefore not resolve all features seen in the wavelet-based local CT-images.
To obtain CT-images with a high image quality a mathematical optimisation procedure is used to find the optimal equipment settings for collecting the CT-data. The geometry-sensitivity/noise, density-sensitivity/noise and mass attenuation-sensitivity/noise ratios are maximised using the weighted minimax method. The sensitivity/noise ratios are calculated using a mathematical model of the X-ray imaging system. The image quality is improved if the sample completely shades the detector and optimised equipment settings for this case are used. The sensitivity/noise ratios in the local CT-investigation of one of the samples are increased with at least 20% compared to if air gaps would have been present.
A procedure to suppress ring artefacts in the reconstructed CT-images is presented. The procedure is found to give good results.