Detection of Defects in Thermal Barrier Coatings by Thermography Analyses

Size: px

Start display at page:

Download "Detection of Defects in Thermal Barrier Coatings by Thermography Analyses"

Dayna Wilkins
5 years ago
Views:

1 Materials Transactions, Vol. 44, No. 9 (23) pp to 185 #23 The Japan Institute of Metals Detection of Defects in Thermal Barrier Coatings by Thermography Analyses Hua-nan Liu 1, Michiru Sakamoto 1; *, Kazushi Kishi 1, Kazuhisa Shobu 1, Tatsuo Tabaru 1, Hiroshi Tateyama 1 and Yoshio Akimune 2 1 Institute for Structural and Engineering Materials, National Institute of Advanced Industrial Science and Technology, Tosu , Japan 2 Smart Structure Research Center, National Institute of Advanced Industrial Science and Technology, Tsukuba , Japan Thermographic imaging under steady-state heat flow was used to nondestructively detect the defects in thermal barrier coatings (TBCs). The finite element method (FEM) analyses and the experimental thermography observations were performed using artificially grooved ZrO 2 plates and the indentation-tested TBC-metal specimens. The FEM results show that: (1) Defects (nonuniformity or internal cracks) of the TBC can be effectively detected by the thermographic imaging method; (2) The apparent thermal images would be far greater than the real sizes of the defects, but the half-height sizes of the temperature profile were found to give good estimates for the latter; and (3) The higher value of heat flow would contribute to the detections. Besides these results, the influence of the defect situations (morphology, size and position) on the thermal images was also predicted by the FEM analysis. Although, due to some difficulties in preparing the test specimens, the quantitative comparison between each FEM result and that of the actual measurement was not performed, however, the experimental results of the grooved ZrO 2 plates were found to be in good agreement with the FEM predictions. For the thermographic experiments of the indentation-tested specimens, both the internal-cracks and the thickness-nonuniformity of the TBCs were successfully observed, and the smallest cracks detectable had a diameter far less than 1 mm. (Received May 19, 23; Accepted July 24, 23) Keywords: thermal barrier coating, infrared thermography analysis, nondestructive detecting 1. Introduction Thermal barrier coatings (TBCs) have been widely used in a variety of gas turbine and diesel engine applications. Since there are critical demands to improve the overall system efficiency, investigations on TBC systems are being extensively conducted so as to prolong their lifetime and/or to increase their service temperature. In this regard, the most important issue is the long-term degradation mechanism of TBC, which has been mostly studied by correlation to the non-uniform growth of a thermally grown oxide (TGO, which is mainly Al 2 O 3 ) between TBC and the bond coating (BC, which is NiCrAlY-base, Pt 3 Al-base, or else). However, TBC systems inherently contain a certain amount of macroand micro-defects, which cannot be eliminated completely. These defects are apparently of critical importance for shortterm failure of the TBC system, so that nondestructive evaluation on the TBC has been studied extensively to date. We expect, further, that defects may play a very important role even for the long-term degradation of the TBC. A possible failure mechanism of TBC systems would be that defects result in a nonhomogeneous heat flow through the TBC and cause a nonhomogeneous temperature distribution at the TBC-BC interface (where TGO grows). This, in turn, results in the nonhomogeneous growth of TGO and eventually provides fracture origins. At any rate, the nondestructive assessment of internal defects is apparently important and indispensable for both the improvement and the application of the TBC systems. A thermal imaging technique for nondestructive defectdetection has been proposed, 1,2) but only limited reports are available in the open literature. Concerning the application of this technique to TBC systems, Ito et al. 3) studied the thermal *Corresponding author, michiru-sakamoto@aist.go.jp properties of a plasma sprayed ZrO 2 coating (25 mm in thickness) by laser-flash heating. With the aid of infrared thermography analyses, they successfully detected the penny-shaped interfacial cracks with diameters over 3 mm, but failed in those of diameters less than 1 mm. With the view of investigating the spatial nonuniformity of the thermal conductance of the TBC and its effect on the long-term degradation behaviors, we studied the thermal imaging detecting method under steady-state heat flow and found that this method is very effective for detecting even a smallsize defect. This paper reports the results of both the FEM calculations and experimental thermal analyses. 2. FEM Calculations We conducted the FEM calculations to simulate the influence of the experimental condition (testing temperature) and the defect situations (morphology, size and position) on the temperature distribution of the TBC surface. The software, MSC Visual Nastran for Windows 22, was used in this study. Two kinds of defects models, as shown in Fig. 1, were employed. These models represent the thickness-nonuniformity and the internal-crack, respectively, of the TBC film, both of which were assumed to be a groove with a rectangular cross-section. The upper face (back face) of the TBC was the interface of the metal-tbc system, the temperature of which was assumed to be constant everywhere, while the underside surface (detecting face) was assumed being cooled by the atmosphere (refer Fig. 3). The TBC material employed was the 8%Y 2 O 3 ZrO 2 with properties listed in Table 1. The input boundary conditions were the temperature of the back surface and the heat flow from the detecting surface. Since it is difficult to make a reasonable assumption for the value of the heat flow from the TBC surface to the

2 1846 H. Liu et al. Nonuniform model Metal TBC d Nonuniform location.3 T=6 C d=.1;.3;.5 T=6 C T=6 C Crack model Metal TBC Fig. 1 Table 1 δ ω d Crack.3 Model sketch for the typical defects of the metal-tbc systems. Properties of the materials used in the thermal calculations. Material (W/mK) C p (J/kgK) (kg/m 3 ) " 8%Y 2 O 3 ZrO SUS Air ( C) d=.5mm d=.3mm d=.1mm Width.5h h Heat Flow, q/kwm Surface, T/ C (d) (e) Fig. 3 FEM results for the nonuniform-model with different defect widths. Model. Meshes & Boundary conditions. Temperature distribution on the detecting surface. (d) Definition of the half-height width of the temperature profile. (e) Calculated half-height width of the temperature profile. Fig. 2 Value of heat flow from the ZrO 2 surface as a function of the temperature. atmosphere, we made a simple measurement of this property using an 8%Y 2 O 3 ZrO 2 plate (.5 mm in thickness). These results are shown in Fig. 2, which showed that the heat flow value was roughly a linear relationship with temperature. With this curve, the heat flow values of the TBC surface at different surface temperatures were deduced. Figure 3 shows the calculated surface temperature of the nonuniform model at different defect widths, which showed that the subsurface nonuniformity would result in a temperature anomaly on the TBC surface (in this case, the anomaly location was relatively hotter than the surroundings). Under the input boundary conditions, the width of the anomaly location was far bigger than that of the defect, indicating that the thermal image would be far greater than the actual size of the defect. With an increase in the defect width, the temperature difference (between the anomaly location and the surroundings) obviously increased (Fig. 3), showing that the larger-size defects are relatively easier to be detected. By using the width at half height of the temperature profile to represent the width of the image (Fig. 3), the size of the defects and their images were compared and these results are shown in Fig. 3(d). The smaller the defect size, the greater the size ratio of the image to defect. In these calculations, it had been assumed that the temperature was constant at the TBC- BC interface even at the mound. This would be less reasonable for smaller defects. Nevertheless, the half-height width of the temperature profile would be substantially greater than the actual size of the nonuniform-type defect. Figure 4 shows the testing temperature (or, the heat flow) dependence of the temperature profile. These profiles were very similar in shape except for their magnitude. An increase

3 Detection of Defects in Thermal Barrier Coatings by Thermography Analyses 1847 T=6; 7; 8 C d=.5 T=6 C d=.1;.3;.5 78( C) q T= 8 C q=-246kw/m 2 T= 7 C q=-28kw/m ( C) 52 d=.1mm d=.3mm d=.5mm dt=2 C T= 6 C T= 8 C T= 7 C T= 6 C C 7 C 8 C (d) Fig. 4 FEM results for the nonuniform-model at different heating temperature. Model and boundary conditions. Temperature distribution on the detecting surface. Relative temperatures of the detecting surface. (d) Calculated half-height width of the temperature profile. in the value of the heat flow increased the temperature difference of the defect s image (Fig. 4), but had no influence on the image shape (Fig. 4(d)). The FEM results for crack models (in this case, the cracks were of high thermal resistance) are shown in Figs The influences of both the defect width and the testing temperature on the thermal images were similar to those for the nonuniform-model cases (Figs. 5 and 6). However, the size ratios of the image to defect were relatively smaller, and the half-height widths of the images were much closer to the actual width of the defects (Figs. 5 and 6(d)) Figures 7 and 8 show the influence of the location and thickness of the defects on their thermal images. The closing to the TBC surface or the increase in the thickness of the defects increased the temperature difference of the images (Figs. 7 and 8), indicating that those thicker and/or near-surface defects were relatively easy to detect. In both cases, the variation in the image size was very limited (Figs. 7 and 8). The FEM results can be summarized as: (1) The apparent thermal images were larger than the actual size of the defects; (2) The high heat flow value contributes to the defectdetections; and (3) Thick cracks and/or near-surface cracks were easily detected. 3. Experimental Thermography Analyses.5.54 Fig. 5 FEM results for the crack-models with different defect widths. Model and boundary conditions. Temperature distribution on the detecting surface. Calculated half-height width of the temperature profile. A thermal analysis was performed on a simple ZrO 2 plate (2 mm 2 mm :5mm in size) with two artificial grooves (1. in width and.1 mm in depth) on one side of the surface. During the experiments, the grooved face was placed on a stainless steel base (SUS34) heated by an electric heater, while the upper face was observed using a thermographic imaging apparatus (JTG-63 Thermoviewer), as shown in Fig. 9. The thermal image of the ZrO 2 plate is shown in Fig. 9, in which the groove images were clearly observed. The apparent widths of the images were about 2 mm which were twice that of the grooves, and the measured temperature difference was about 2 degrees. For comparison, an FEM calculation was performed on this

4 1848 H. Liu et al ( C) T=6; 7; 8 C d=.5; δ=.1; w=.1 q T= 8 C q=-246kw/m 2 T= 7 C q=-28kw/m 2 T=6 C 54 58( C) 5 δ =.5;.1;.15 d=.5; w=.1 δ=.5mm δ=.1mm δ=.15mm 6 56 dt=8 C T= 6 C (d) T= 6 C T= 7 C T= 8 C C 7 C 8 C Fig. 6 FEM results for the crack-model at different heating temperature. Model and boundary conditions. Temperature distribution on the detecting surface. Relative temperatures of the detecting surface. (d) Calculated half-height width of the temperature profile. system. These results are given in Fig. 1 showing that the size of the groove images had been successfully predicted, but the calculated temperature difference (that was 13.2 degrees) was far greater than the measured one. This discrepancy was mostly attributed to the translucency of the ZrO 2 plate, because the infrared radiations not only increased the internal heat transfer of the plate (this decreased the temperature difference of thermal image), but also influenced the imaging apparatus (the camera also caught part of the inner infrared radiations). At any rate, it was reasonable to conclude that the FEM analyses gave correct predictions for the thermal images and sizes of the defects δ=.5mm δ=.1mm δ=.15mm Fig. 7 FEM results for the crack-model with different crack locations. Model and boundary conditions. Temperature distribution on the detecting surface. Calculated half-height width of the temperature profile. T=6 C 52 56( C) ω=.1;.5;.1 d=.5; δ=.1 ω=.1mm ω=.5mm ω=.1mm ω=.1mm ω=.5mm ω=.1mm Fig. 8 FEM results for the crack-model with different crack thickness. Model and boundary conditions. Temperature distribution on the detecting surface. Calculated half-height width of the temperature profile.

Detection of Defects in Thermal Barrier Coatings by Thermography Analyses 1849 3. ZrO 2 (2 2.5) Stainless steel Groove (1.

These specimens were prepared by plasma-coating the ZrO 2 (:5 mm thickness) on the Ni-base superalloy plate, and then indentation tested with the ball-shaped indenters with diameters of 1, 2, 4 and 6

It can be seen that some lateral cracks have been generated in the TBC beneath this indent (Fig. 11). By a simple testing method (as sketched in Fig.

5 Detection of Defects in Thermal Barrier Coatings by Thermography Analyses ZrO 2 (2 2.5) Stainless steel Groove (1..1) Thermography analyses were also performed on the indentation-tested TBC specimens. These specimens were prepared by plasma-coating the ZrO 2 (:5 mm thickness) on the Ni-base superalloy plate, and then indentation tested with the ball-shaped indenters with diameters of 1, 2, 4 and 6 mm under the applied load of 1 N. Figure 11 shows the view of the specimen s surface and the morphology of the cross-section of a 1 mm-indent location. It can be seen that some lateral cracks have been generated in the TBC beneath this indent (Fig. 11). By a simple testing method (as sketched in Fig. 12), the cracks under the indents were successfully observed by infrared thermography as a representative thermal image shown in Fig. 12. It must be noted that there were surface dimples at the indents, however, these dimples would not contribute to the cold spots of the image (instead, they would give hot spots), i.e., these cold spots would have been a result of internal defects. Since the diameter of the images were 1 2 mm, the diameter of the 2 Indentation location Fig. 9 Thermal analysis of a ZrO 2 plate heated by a stainless steel plate (3 mm thickness). Testing method. Thermal image. 3 C 3 C 7 Superalloy-(8%Y 2 O 3 -ZrO 2 )Specimen Indentation-tested mm(3-metal &.5-ZrO 2 ) 3.5 Air(1..1) SUS34 ZrO 2 Resin Heat treated at 115 C for 1h Indentation C q=-69kw/m mm Fig. 1 FEM results for the grooved ZrO 2 plate using the boundary conditions described in Fig. 9. Temperature distribution on the cross section. Temperature distribution on the testing surface. Crack 8%Y 2 O 3 -ZrO 2 Bond coat Superalloy.3mm Fig. 11 Photograph of the indentation-tested TBC specimen. Coated surface. Cross section at the indentation location.

6 185 H. Liu et al. Thermocouple Specimen Heater Indentation location CCD Indentation location Nonuniform part (Heattreated at 115 C for 1h) Fig. 13 Infrared photography of the indentation-tested TBC specimen, which shows the resulting nonuniform temperature-anomaly. detectable defect would be less than 1 mm. As depicted in Fig. 13, a large-scale hot temperature-anomaly was also observed on a specimen, and it would be reasonable to attribute this anomaly to the nonuniform thickness of the TBC film. 4. Summary (Heattreated at 115 C for 2h) Fig. 12 Thermal analysis of indentation-tested TBC specimen. Testing method. Representative thermal image. FEM calculations and thermograpy analyses were performed on the artificially grooved ZrO 2 plate and the indentation-tested metal-zro 2 specimens in order to develop a nondestructive defect-detecting method for TBC systems. The FEM analyses gave the following results: (1) The apparent thermal image of the defect would be much larger than the actual size of the defect, but the half-height width of the temperature profile gives a good estimate for the latter; (2) A high heat flow value gives a greater temperature difference in the defect s image, and is favorable for the defect detection, and (3) Thick cracks and those near the specimens surface produced a larger temperature difference, and are easy to detect. The experimental thermography analyses were successfully applied to the defect-detection of the TBC systems. Under steady-state heat flow conditions, cracks with diameters much less than 1 mm were detectable. Acknowledgements The authors express their appreciation for support of this work by the New Energy and Industrial Technology Development Organization of Japan (NEDO). Appreciation is also expressed to Professor Y. Kagawa and Associate Professor S. J. Zhu, University of Tokyo, for their valuable comments during the course of the work, and to Mr. M. Kawamura and Mr. A. Shinmi, Kawasaki Heavy Industries, Ltd., for providing the testing specimens that were used in this investigation. REFERENCES 1) V. P. Vavilov, V. V. Shiryaev and E. Grinzato: Insight 4 (1989) ) V. P. Vavilov, E. Grinzato, P. G. Bison, S. Marinetti and C. Bressan: Materials Evaluation 54 (1996) ) Y. Ito, M. Saito, H. Kashiwaya, M. Oishi and T. Kaneko: J. of Ceramic Society of Japan 97 (1989)

Effects of TGO Roughness on Indentation Response of Thermal Barrier Coatings

Effects of TGO Roughness on Indentation Response of Thermal Barrier Coatings Copyright 2010 Tech Science Press CMC, vol.17, no.1, pp.41-57, 2010 Effects of Roughness on Indentation Response of Thermal Barrier Coatings Taotao Hu 1 and Shengping Shen 1,2 Abstract: In this paper,