Metallurgical Investigation and Failure Analysis in GTD-111 Stage 2 Buckets From an Industrial F-Class Gas Turbine

GE, one of the most important gas turbine OEM companies, developed F class technology in the 1980s, and its first F-technology unit entered commercial service on June 6 in 1990 at Virginia Electric & Power Company's Chesterfield site [1]. Since that time, the F-technology turbines have been improved via advanced material technology from Equiaxed (EA) structure to Directionally Solidified (DS) and further Single Crystal (SX) structure, improved cooling design and the use of advanced thermal barrier coatings (TBC's). GE marked its F-technology units as 7FA.0x, and currently 7F.03, 7F.04, and 7F.05 frames are dominant. The 7F technology is core to GE Gas Power's portfolio of products and services, and approximately 950 units have been installed to produce ∼175 GW of power [2].

7F.03 stage 2 buckets (S2Bs) creep performance has been weak since the beginning, and GE has issued several TILs (1334, 1858, 1863, 1859, and 1984) and updated the design from original version to a modified version, and then to the current compound fillet version [3]. In many instances, creep damage and cracking has been observed at the tip shroud suction side fillet location. Table 1 shows a comparison of the original model to OEM modifications (new designs) and non-OEM applications for blade material, cooling hole configuration, airfoil coatings, cutter teeth and z-notch hardened surfaces.

The investigated S2Bs in this study were of the compound fillet design and made of GTD-111 DS material. DS casting is the solidification that occurs from the farthest edge of the casting and progresses toward the sprue during which the temperature of the mold is precisely controlled to promote the formation of aligned stiff crystals as the molten metal cools [4]. The DS casting serves to reinforce the component in the direction of alignment and is expected to have improved creep performance as compared with EA structure.

Several different microstructural characterization tools were utilized to evaluate the failed stage 2 gas turbine bucket failures. High temperature mechanical testing was also conducted to compare the properties to typical GTD-111 base material (BM) and to compare creep damage initiation to that observed in the failed buckets.

Prior to sectioning, visual inspection was conducted and the gas turbine bucket samples were three dimensionally scanned using the Creaform HandySCAN 3DTM equipment and modeling software. The three-dimensional (3D) models were then compared to determine if there were any changes in overall geometry between the various buckets.

After identifying areas of interest, sectioning was conducted using wire electric discharge machining (EDM). Metallurgical samples were hot mounted in a conductive Bakelite material and polished using an automated polishing procedure to obtain a 0.04 μm surface finish. These samples were then used for optical and scanning electron microscopy (SEM) imaging.

A digital Keyence VR-3200 large area measurement system utilizing a white light emitting diode source was used to capture three dimensional fractography images at the failed fracture surface. A digital Keyence VHX-7000 optical microscope was used to capture general images of the observed damage in the failed bucket samples. The SEM was used to capture general images at higher magnifications of the fracture surface and polished samples showing damage. Several different detectors, such as the electron backscattered diffraction (EBSD) and energy dispersive spectroscopy (EDS) were used to evaluate grain structure and chemical composition, respectively.

μXRF compositional mapping was conducted with a Bruker M4 Tornado Plus. The larger chamber size (600 mm × 350 mm × 260 mm) provided a method to map a larger section of the gas turbine bucket to evaluate differences in composition. The experiment was performed under a vacuum of ∼1 mbar, a 50 kV voltage and dual rhodium X-rays were used to map the entire surface of the airfoil section in one of the buckets. The X-ray spot size of each measurement was ∼20 μm.

Chemical composition of the material was measured using inductively coupled plasma optical emission spectrometry (ICP-OES) for Al, B, Ca, Co, Cr, Cu, Fe, La, Mn, Mo, Nb, Ni, P, Si, Ta, Ti, V, W, and Zr. Inductively coupled plasma mass spectroscopy was used for As, Bi, Pb, Sb, and Sn. Combustion methods were used for C and S. Inert gas fusion was used for O and N.

High temperature creep testing was carried out in accordance with ASTM E139 [5]. Specimens were sectioned from the dovetail region of the buckets and tested at two different test conditions. The first at 900 °C and 240 MPa and the second at 850 °C and 300 MPa. Samples were machined with an overall length of 50 mm, gauge length of 12.7 mm, thread diameter of 6 mm and gauge diameter of 3.2 mm. Temperature was maintained to ±0.1 °C using type S platinum/rhodium thermocouple wire.

A total of three different stage 2 buckets from a GE 7F.03 unit were evaluated after they were removed from service due to cracking in the Z-notch area in the shroud. The buckets were on their 3rd interval, with the first two intervals operating a typical 24,000-hour cycle with standard rejuvenation heat treatment repairs between each cycle. Numerous buckets were also blended, and weld repaired during the second interval cycle. During the third interval the buckets were removed after 7,348 h and 31 starts due to the same repeated cracking issue in the pressure side Z-notch region of the shroud. Figure 1 shows the three different buckets and the corresponding 3D scans for each. No major differences were observed in the overall geometry of each blade, but some dimensional differences were noted at the pressure side Z-notch relief area, where cracking was observed (see Fig. 2). The scan data was used to measure the radius and thickness at each Z-notch location in the components along with visual observations, which are all shown in Table 2. Two of the buckets showed clear signs of cracking, but one of the buckets did not show any surface indications.

The three different evaluated buckets were represented by two different casting houses, denoted by the cast number from Table 2. Thus, the chemical composition of P68-B10 and P4-B5 was measured to determine any differences in major and minor alloying elements, as well as trace elements. The detailed 28 element composition for each cast heat was compared to the typical values for GTD-111 [6] and is shown in Table 3. All of the major alloying elements were in-line with typical values for GTD-111. However, the Zr values for the K2DP heat (for the P4-B5 bucket) were not detected and this is a controlled minor alloying element for GTD-111 material. Zr is typically added in trace amounts for grain boundary strengthening [7]. Additionally, the S and O values were also higher in the K2DP heat, which may also have detrimental effects on creep rupture behavior [8,9]. Literature knowledge, visual observations of the cracked buckets and the elemental composition results shown here suggest the K2DP heat of material may have higher metallurgical risk compared to the K2DH heat.

To further evaluate the extent of cracking and determination of the potential damage mechanism, each sample was sectioned, and various techniques were used to evaluate the damage. Figure 2 shows a representative blade and the initial cut taken below the shroud for each of the evaluated components. The top down (looking above) and from underneath (looking underside) views are shown with important nomenclature explaining various feature of the bucket, such as the rail, shroud, airfoil, shank, and dovetail regions. The pressure side (PS) and suction side (SS) are also denoted, showing Z-notch relief areas for both the PS and SS of the bucket.

Initial visual observations in the P68-B10 bucket showed clear cracking indications at the PS Z-notch radius, as shown in Fig. 3. The crack was ∼2.5 mm in length and the location was further sectioned at several locations for more detailed evaluations. Figure 4 shows the various locations that were sectioned. A single cut across the airfoil from the PS to the SS was made for μXRF compositional evaluations and to determine if any additional damage existed in this area, which has been shown to be a common area of failure in some designs [3]. The Z-notch area was further sectioned using electric discharge machining (EDM) to provide a section for fractography and for polishing to evaluate damage beyond the crack tip.

The fracture face of the PS Z-notch radius was 3D imaged and evaluated using standard SEM imaging. The surface topography map showing the Z-notch tip and a SEM secondary electron image of the fracture face are shown in Fig. 5. The 3D scan clearly showed the fracture ‘ridges’ were parallel to the crack and the SEM image showed the crack was intergranular in nature and occurred at either grain boundaries or dendritic regions. There were also no signs of fatigue related damage, such as beach marks or striations.

The second section near the Z-notch area was mounted and polished to reveal features beyond the crack tip. This section was further evaluated using standard SEM backscattered imaging and EBSD methods. Figure 6 shows the SEM images and the location where EBSD was conducted, which verifies damage (voids) was following a low angle grain boundary ahead of the crack. Oxidation was also observed at the crack opening as well as agglomerates of various carbides at the grain boundary. There were no signs of damage in the matrix, and the overall microstructure showed a typical bimodal distribution of γ′ precipitates.

Standard optical imaging and μXRF compositional mapping was conducted in the section cut transverse to the airfoil, which including one of the cooling holes and the rail from the shroud region. Figure 7 shows the optical image of this cross section and the corresponding μXRF maps for primary alloying elements in GTD-111. The dark field optical image captured using LED ring lighting and compositional maps shown confirm that the casting was directionally solidified along the airfoil section. The μXRF maps also showed elemental segregation of Cr and Ti between the dendritic and interdendritic regions. There were no signs of creep damage at either the PS or SS airfoil radius.

The P81-B65 and P4-B5 buckets were also sectioned at the PS Z-notch region and polished to show microstructural features at the radius area. Figure 8 shows the optical images for each of the polished sections. There were no signs of damage in the P4-B5 bucket, consistent with visual observations. Damage was observed in the P81-B65 bucket and additional higher magnification optical images show cracking at the repair weld and the radius region. The weld defect also showed some porosity but did not appear to have any influence on the cracking observed at the Z-notch radius. The damage at the radius was consistent with the findings in the P68-B10 bucket shown previously, which was manufactured from the same casting house (K2DP).

Additional high-resolution images were captured using the SEM for the damage observed in the P81-B65 component. Several images showing the damage at different length scales are shown in Fig. 9. Again, damage was consistently found at the grain boundaries ahead of the crack tip.

The findings shown here suggest that the primary damage mechanisms were creep and bolstered by the metallurgical findings that show void cavitation damage at the grain boundaries and fractography showing fracture along boundaries.

Creep is accelerated by higher temperatures and applied stress. The Z-notch region is clearly a high stress concentration region due to the radius and the operational profile of the component (rotating centrifugal forces) results in a high stress region. To provide additional evidence that creep was the damage mechanism, sections were taken from the shroud region as well as the midspan of the airfoil in the P68-B10 bucket to compare overall γ′ precipitate size and distribution in the two areas. During service exposure at elevated temperature, the γ′ precipitate structure will grow and evolve [10–12] and this can be used to determine which regions operated at higher temperatures. Figure 10 shows the two different regions that were evaluated using high magnification SEM imaging. The overall γ′ structure was evaluated using image thresholding to determine the size (equivalent diameter) of each γ′ particle. The results showed that the secondary γ′ precipitates grew to a much larger average size (∼313 nm) in the shroud region (near the PS Z-notch) compared to the midspan of the airfoil, which showed an average secondary γ′ size of ∼102 nm. Figure 10 also shows the overall distribution (relative frequency) of γ′ size in each of the measured regions as a histogram. This suggests that the shroud portion of the component operated at a higher overall temperature.

The coating in the P68-B10 component was also evaluated using SEM-EDS techniques to determine the bond coat and TBC used. The EDS composition maps for Cr, Al, Y, Zr, and O are shown for the evaluated coating in Fig. 11. Individual locations at four different areas were quantified to show differences in composition between the BM, the region between the BM and bond coat, the bond coat, and the TBC. The results for each region are shown in Table 4. This confirmed that the bond coat was NiCoCrAlY (GT 33 InPlus), and the TBC was yttria stabilized zirconia oxide, which is consistent with expected OEM modifications. In this sample, a crack was observed between the TBC and bond coat layer. However, the use of a conductive hot-pressed mount was determined to be the cause for cracking, as multiple cold mounted samples showed no signs of cracking in the coating.

To determine to variability of performance in the two different heats (from different casting houses), high temperature creep testing was conducted in material from the P68-B10 and P4-B5 components, which represented the K2DP and K2DH heats of material, respectively. Samples were removed from the shank area in the transverse orientation in relation to the directionally solidified airfoil structure. The reason for testing in this orientation was due to the observed damage at the grain boundary in each of the cracked buckets. Tests were conducted at 850 °C and 300 MPa as well as 900 °C and 240 MPa. Table 5 shows the creep testing results for each of the tested heats. Figure 12 shows a comparison of the data to typical behavior for GTD-111 material [13,14]. Results showed that each heat performed very closely to one another, suggesting no major differences in overall creep performance. The results also indicate that performance matched closely to typical GTD-111 behavior, but some marked decrease was notable at the higher stress and lower temperature test condition. These results also indicate metallurgical risk may have not played a prominent role in overall performance. However, the authors note that since the microstructures from the blade roots (where samples were taken) have seen the lowest service exposure temperature, longer-term tests or pretest thermal aging may be useful in further elucidating similarities or differences in creep behavior and careful inspection of the limited data shows while rupture lives were similar for the two casts there were small differences in creep resistance and failure ductility.

The overall operational profile for the entire 7FA gas turbine fleet at Duke Energy was evaluated to determine differences compared to other 7FA units. The overall fleet consists of 12 peaker units to meet occasional high demand energy needs and 11 base loaded units for standard operation. The fleet also consists of a mix between 7F.02 (2 units), 7F.03 (13 units), and 7F.04 (8 units) gas turbines. The average exhaust temperature was compared for each unit and is shown in Fig. 13.

For 7F units in the peaker operation mode, the units shall have more starts due to frequent start up and shut down, while the service hours are low, and creep damage is generally not considered as a primary risk. For the same 7F frame, high exhaust temperatures generally indicate high service temperatures in the turbines for similar workload. Among the 13 7F.03 units, only 3 units have run base loaded operation mode. The failed units and its sister unit ran hotter than the comparison 7F.03 base load unit. As compared with 12 peaker units (2 7F.02 and 10 7F.03), the failed unit ran hotter than 11 of them. As compared with 8 7F.04 units that have an advanced bucket design, the failed unit ran hotter than 4 of them.

Additionally, the percentage of operating hours exceeding exhaust gas temperatures of 635 °C were also shown as labels above each bar chart. Here, it is clearly shown that the failed base loaded unit operated at increased temperatures and time compared to other 7F.03 units. Several of the 7F.04 base loaded units have operated at even higher exhaust temperatures for longer periods of time, but no failures have been observed to date. This suggests that key design changes in the 7F.03 to the 7F.04 stage 2 buckets were beneficial. The detailed failure investigation and operational data analysis has led to new insights and the base loaded 7F.03 units are expected to be upgraded to the 7F.04 design to prevent reliability issues in future operation.

The extensive failure investigation of stage 2 buckets fabricated from GTD-111 directionally solidified material in a 7F.03 gas turbine unit revealed creep as the damage mechanism. Detailed metallographic evaluations showed that creep damage was dominate along the DS grain boundaries with microscopic creep voids forming ahead of the crack tip leading to the macroscopic cracking at the Z-notch region. Fractography also confirmed that fracture surface “ridges” were parallel to the crack, and the crack was intergranular in nature occurring at grain boundaries and dendritic regions. The manual blending resulting in variable geometry during the multiple repair/rejuvenation cycles did not appear to be important in determining which blades cracked. Although important differences were noted in chemical composition, metallurgical risk was not conclusively determined to be a significant factor based on limited creep rupture testing results. A more detailed evaluation of turbine exhaust temperature across the 7FA fleet also showed the base loaded unit with the failed S2Bs operated at a higher temperature and longer times compared to peaker units with the same 7F.03 S2B design. This along with detailed metallurgical findings further supports creep was the damage mechanism and not fatigue or other processes. Future plans are to upgrade the 7F.03 units to the 7F.04 design in all base loaded units to prevent future failures.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

ASTM =

American Society of Testing and Materials

EA =

equiaxed structure

EBSD =

electron backscattered diffraction

EDS =

electric discharge machining

EDS =

energy dispersive spectroscopy

DS =

directionally solidified structure

SEM =

scanning electron microscope

TBC =

thermal barrier coating

PS =

pressure side

SS =

suction side

XRF =

X-ray fluorescence