Search

A heavy-duty gas turbine engine at the heart of a seawater desalination plant on the Arabian Peninsula exhibited multiple impact damages on their compressor airfoils that were detected during an inspection. Many blades showed heavily damaged airfoils and some even lost fractured-off parts. One moving blade in row 8 lost 50 % of its airfoil. Most compressor blade tips from stages 4 to 15 were damaged and bent. It was reported from the site that turning gear operation had been omitted once after engine shut-down in the recent past, i. e., the still hot rotor had not been turned until cool-down, resulting in rotor bend. This is believed to have caused the tip rubbing and airfoil break-off during the next engine start.

The aim of this contribution is to review cases of repeated snap ring failures in large gas turbine engine service due to intergranular attack, or intergranular corrosion, IGC, and to shed some light on metallurgical aspects of the wet corrosion resistance of the subject high-alloy martensitic stainless chromium steel X39CrMo17-1, DIN 1.4122. It is important to note that only snap rings, a.k.a. retaining rings for bores or Seeger rings, were affected by those failures, namely fractures resulting from intergranular attack (IGA, Kornzerfall). It was determined that the metallurgical root cause of failure was sensitization, resulting from secondary chromium carbide precipitation at grain boundaries (GB), and the ensuing chromium depletion in narrow bands adjacent to the GB, which rendered the matrix there sensitive to selective corrosion (the chromium content in these narrow seams, measuring only a few 10 nm in width, drops below the corrosion resistance threshold of ca. 10.5 %). Snap rings, as the name suggests, require a certain “snappiness” to serve their purpose. It is because of this requirement that all snap rings that experienced failures in the field were heat treated per DIN EN 10088-2, providing for a low tempering temperature of max. 250 °C, leading to a relatively high hardness of ca. 580 HV 10, providing said “snappiness”. It turned out that the microstructure resulting from this heat treatment (H/T) is sensitized, rendering the material sensitive to IGA. To the knowledge of the authors, no other components made of this martensitic stainless chromium steel are given this H/T, but rather a high-temperature tempering of 650 °C–750 °C per DIN EN 10088-3, producing a lower hardness in the order of ca. 300 HV 10 and largely a lack of “snappiness”, the latter being not required by those non-snap ring applications. Metallurgically, it became clear that high-temperature tempering per DIN EN 10088-3 eliminates sensitization by “backfill” diffusion of chromium from the non-depleted matrix into formerly depleted regions next to GB. It was therefore decided in the authors’ organization to eliminate X39CrMo17-1 per DIN EN 10088-2 as a possible material selection. The metallurgical investigations described in this contribution corroborate and support this decision.

Transverse through-wall airfoil cracking was found in a row 3 twin turbine vane segment of a type used in heavy-duty gas turbine engines for power generation. A laboratory investigation determined liquation cracking (LC), a hot cracking mechanism, as the metallurgical cause of failure. As became apparent only after the investigation, the root cause of failure was a malfunctioning plasma gun used for thermal spraying the thermal barrier ceramic top coat onto the subject part.

A metallographic section from the crack area of an exhaust gas turbocharger was supplied by the customer. The shaft made of quenched and tempered low-alloy steel and the charger wheel made of a nickel-based superalloy were joined by laser welding. For design reasons, the root side of the weld contained a geometric notch, at which a high cycle fatigue crack initiated during the test facility operation of the turbocharger. At the time of examination, it went through almost the entire weld root. As a corrective action, it was recommended that the welded joint be replaced by a shrink fit.

Several high-pressure lift oil pump failures were reported from the field. These assemblies are used to inject lubrication oil into the sliding bearings of heavy rotating turbomachinery equipment, such as steam and gas turbine rotors, often weighing in excess of a hundred tons. This ensures that the rotor shaft “floats” on a lube oil film even at low rotational speeds, when hydro-dynamic lubrication conditions have not yet been fully established. A fractured spline shaft, which was the driven shaft of one of the failed pumps, was received from the client for the determination of the metallurgical root cause of failure. The subject spline shaft failed due to torsional overload. This cracked the hardened case and initiated high cycle fatigue (HCF) cracking as secondary damage. The main fatigue cracks were nucleated at overload fractures in the hardened case, in the fillet radii at the base of the teeth of the spline shaft. No evidence of any material defects that could have contributed to the failure or could even have been causative for it was found.

Metallic heat shields, used in combustion chambers of heavy-duty gas turbine engines and made of nickel-based superalloys, were accidentally heat treated during fabrication, using wrong parameters. There were concerns about embrittlement. Nondestructive metallurgical material characterization using the replica technique verified embrittlement by secondary phases. In spite of this, it was recommended to the client to use the affected parts as is, this on the grounds that no grain growth was observed and hardness was not elevated. While the case study presented in this contribution may appear trivial at first sight, it interestingly shows how metallographic examinations may sometimes be performed non-destructively, without sacrificing any engine components. The inclined reader should also note the superb quality of the photomicrographs presented in this contribution. They were taken from replica foils, not actual metallographic sections.

In this contribution, the most interesting and educational failure cases are presented that the author came across during his over twenty years of laboratory practice as manager of the Materials Testing Laboratory of the Berlin Gas Turbine Plant of Siemens’ Power and Gas Division. The case studies are presented and categorised in accordance with VDI Guideline 3822, the German failure analyst’s guide to the subject of how to organise and run a root cause failure analysis. An effort was made to have each of the main four categories of failure causes represented, namely failures due to mechanical loading, corrosive failures, failures due to thermal loading, and tribological failures. Case studies include turbomachinery components that failed due to tensile overload, stress corrosion cracking, intergranular corrosion, hydrogen embrittlement, hot cracking, fretting, erosion, and galling. Affected components include valves, retaining rings, tubing and piping, burners, rotor disks, lifting lugs, and casings. Some of the presented cases were published in the new section “Failure Analysis” of Practical Metallography between October 2011 and the present time. Others were oral presentations at the Metallography conferences and at the annual failure analysis conferences “VDI Jahrestagung Schadensanalyse”, held during that time. The focus of discussion of the failure cases in this paper is the metallurgical evaluation of failure causes. This is the approach taken in many small and industrial laboratories. A holistic approach of a failure case, which includes calculation and simulation methods such as finite element analysis, and which also implies a knowledge of the service stresses intended by design as well as the actual loading situation of the failed part, is not the aim of this contribution.

Supposedly simple cases of failure are most often best suited to communicate the principles of component failure analysis in the field of materials engineering to a wide readership, especially to those peers in the specialist community who are just beginning to familiarize themselves with the subject. The present case of failure relates to components that failed as early as during the assembly, and more specifically, during the final assembly stage of combustion chamber components for heavy-duty gas turbine engines. Hence, they lost their functionality (in fact, the common definition of component failure). At tightening torques of the nuts opposite of the tapered heads as low as below 25 Nm, so-called twistoff bolts which, when welded into combustion chamber sheets, take on the function of stud bolts, sheared off. By way of exception, a materialographic failure analysis could show that the primary cause of the failure was not the component’s design, but the disregard of the drawing specifications during final assembly. However, on a secondary level, design deficiencies had to be mentioned, as untempered welded joints in martensitic chromium steels invariably act as metallurgical notches. If the respective part is subjected to dynamic loads, as is the case in virtually all turbo machinery, they are thus to be avoided.

In this contribution, a case study is presented describing the failure of a combustion chamber assembly in a non-OEM (Original Equipment Manufacturer) gas turbine engine used for power generation. It showed how even advanced fabrication methods, such as Electron Beam (EB) welding, could trigger fatigue fracture, even if there are no material defects, no weld imperfections, no fabrication flaws, and even if everything is within specified limits. As is so often the case in component failures, the fact that failures occur anyway, despite the absence of out-of-spec material properties, and even though there were no fabrication flaws, is attributable to the design; which is often not sturdy enough to withstand unexpected dynamic loading.

Short fillet welds used to fasten a large retainer ring to so-called dog bone seals failed in the turbine exhaust casing of a non-OEM heavy-duty gas turbine engine used for power generation. The subject fillet welds fractured due to high cycle fatigue loading. Neither weld imperfections nor any other material defects were found that could have contributed to the failure. It was concluded that an unfavorable design, specifying very short fillet welds for fastening the dog bone seal segments to the retainer ring, was the root cause of failure. In a purely static loading situation, this design would probably not have failed. However, in a dynamic loading scenario as is the case in any gas turbine engine exhaust, such a design is simply not sturdy enough.

A first-stage turbine vane was received in the laboratory directly from fabrication, prior to its use in engine service. The part had not yet been covered with its customary coating system that protects these parts against hot corrosion. A first visual inspection revealed multiple cracks on the airfoil’s hot gas path side, fairly centered in the part. After cutting the part open, it soon became apparent that the cracking was even more severe inside, suggesting crack initiation from that cooled side. Fractography allowed to determine liquation cracking as the metallurgical failure mechanism. Since the part was received immediately after pre-heat before plasma coating, that process step was concluded to have caused the cracking.

Helical tension springs used in switch mechanisms of medium voltage switches failed prematurely in test rigs. The switch mechanism assemblies did not reach the required number of cycles to failure. The springs failed by low cycle fatigue fracture. They were phosphate coated for corrosion protection but not shot peened. The material used for the springs is a usual patented spring steel. Chemical composition, mechanical properties, and microstructure of the failed springs and a reference spring alike were all unremarkable. The subject springs were assembled and rig tested at three different locations but failed only at two of them, not in the third. It was concluded that misalignment from assembly was the root cause of failure.

This short article presents a simple case of failure, from a materials engineering point of view. One may therefore ask what makes it worth publishing. The inclined reader, though, will notice the somewhat surprising nature of the macroscopic damage pattern. The type of defect reported in the context of the first examination of the component, which had to be performed non-destructively, was therefore wrong. This is an example of the risk of misinterpretation in cases where the customer dictates how the examination is carried out that all those working in the field of damage analysis are aware of. It also demonstrates that using the scanning electron microscope is not a luxury, not even in the manufacturing control, as some damage patterns cannot be elucidated without using this device that has revolutionized the component failure analysis.

One ring of a gland seal from a gas compressor was found fractured. The failure was detected because water leaked into the leaking gas system. The supplier apparently never had issues with this assembly. The field record is unremarkable, according to the manufacturer. However, the incident natural gas compressor is used in intermittent service only, while the assembly was originally designed for more or less permanent operation, according to the supplier of the compressor.

A novel tip of a fuel gasification burner, designed-to-SLM (Selective Laser Melting) and made by SLM, was introduced by the client. This component failed prematurely by cracking at the cooled tip. The metallurgical cause of the failure was overheating due to blocked cooling passages, probably as a result of contaminated cooling water, leading to corrosion in the non-stainless piping systems upstream of the burner. Corrosion products, mainly iron oxides, spalled off said non-stainless components, causing blockages in the intricate cooling cavities of the AM-built burner tip. This lead to excessive thermal overload, reaching the melting temperature of the alloy at the burner tip. As a result, the subject burner tip failed by TMF cracking. Neither manufacturing nor design flaws were identified in the course of this failure investigation. It is speculated that the subject component will perform well if cooled properly. The root cause of the failure is the operation of the cooling water system of the subject burner that allowed loose corrosion products to form and to enter the burner tip.

In this contribution, some aspects of fractography are discussed, arguably the failure analysts' most potent analytical tool in metallurgical failure analyses. The characterisation of fracture surfaces is indispensable when it comes to getting to the bottom of things in regards to component failures. When machine parts or entire assemblies fail by fracture of individual members, the fracture surface contains the data that tells the failure analyst about the history of the failed component, and hopefully will also shed some light on the exact failure mode that eventually lead to fracture. That makes fractography, the art of reading fracture surfaces, so valuable. This applies to all material classes, i. e. metals, ceramics, and plastics alike. This paper, however, focuses on failures of metallic components only. When there is no fracture surface in a failed part, one is produced by forcing open cracks or other imperfections that might be suitable for this purpose. Once this is achieved, the failure analyst will open his or her toolset of macro- and microfractographic analysis equipment. It is not exaggerated to say that the scanning electron microscope (SEM) revolutionised the art of microfractography from the early 1960 s on. In this paper, a number of examples for this are given.

Serial fabrication of novel selective laser melted (SLM) heavy-duty gas turbine burner parts was established. This is an additive manufacturing (AM) process. Thermocouples (T/C) are manually tungsten inert gas (TIG) brazed to these components. After fabrication, they exhibited severe cracking within the brazed T/C joints and had to be scrapped. A laboratory order for a destructive metallographic investigation was placed by the client with the aim of determining the metallurgical cause of the cracking. The crack path is interdendritic. The crack propagated within the braze metal only. The crack morphology is consistent with solidification cracking (SC), a hot cracking mechanism. No evidence of liquid metal embrittlement, or LME, was found.

A number of M8 fuel oil pre-mix (FOPM) burner nozzles in a heavy-duty gas turbine engine were found loosened after some 32,000 operating hours. They were removed and tended to crack in the shaft upon disassembly. It was determined on site that they were fractured at the fillet radius below the head, at the thread's end. A metallurgical root cause investigation was ordered to determine the cause of the cracking. Selected dimensional checks, chemical analyses, microstructural examinations, and evaluations of fuel oil analyses conducted by external sources were performed to achieve the required results. Fractographic studies were not possible in this case due to the corrosion products covering the fracture surfaces. The FOPM burner nozzles fractured due to hydrogen-induced stress corrosion cracking (HISCC). No evidence of pre-existing cracks was found. There are no signs of over-torqueing nor of any other damage caused by manufacturing or assembly; hence this failure was caused by engine service and is not fabrication related.

The slotted center guide bolt of the center guide feature of the lower part of the outer shell of an annular combustion chamber was found fractured in a heavy-duty gas turbine engine used for power generation, after approximately 5.500 operating hours. The incident was a one-off event and not a recurring incident. No similar events were reported from the fleet; hence the failure was not considered a field issue. The metallurgical root cause investigation that was ordered to determine the failure mechanism revealed that the incident center guide bolt failed by fretting fatigue cracking, a high cycle fatigue (HCF) phenomenon.

During a minor inspection of a heavy-duty gas turbine engine, the hexagonal head of a bolt used for fixing a metal plate in the air intake was found on the floor of the air intake casing [1, 2]. The respective washer was not found. It is believed to have been ingested into the engine, having caused moderate compressor damage. The inspection was started after 83.500 operating hours. Only one fractured bolt was found. The subject bolt fractured due to stress corrosion cracking (SCC), induced by pitting corrosion and possibly assisted by crevice corrosion.

Leaf chains are flexible lifting devices which are often used as a link between crane hook and the load to be lifted. Actually, they are not designed for diagonal pulls, but for clean tensile stress. In practice, however, slight diagonal pull situations are often inevitable. A certain degree of security against such unscheduled load cases, is therefore generally expected from these lifting devices, which means that the production engineer assumes that leaf chains behave “tenderly”. In the present failure analysis, this was also the case until the leaf chain manufacturer changed the chain link supplier. The chain links this supplier delivered were no longer quenched and tempered but case hardened instead, which actually makes them more wear-resistant but also less tolerable towards the kind of bending loads inevitably occurring with diagonal pulls. As a result the leaf chain fractured and the load dropped during the lifting of a 6.5 t casing component. Since all occupational safety rules of the engineering company were followed, there were no further consequential damages except for some damages to the casing component.

A leakage occurred in the area of the weld of a vent nozzle in the pressure pulsation dampener of a natural gas compressor. It was caused by a crack adjacent to the weld on the side of the pipe.

In the course of a service life extension after 100 000 operating hours, the rotor of a heavy-duty gas turbine was inspected. During the visual inspection, some compressor wheel disks, particularly in the last stage, showed conspicuous features and flawed material. The only causal damage mechanism worth considering is erosion: only very little deformation was found in the metallographic sections' microstructure immediately underneath the flawed material areas (wear or deformation induced damages would have lead to more deformation) and, what is more, the microstructure showed no evidence of corrosion. In accordance with VDI guideline 3822, the exact designation of the damage mechanism is “Flow wear in the form of sliding jet wear in the two-phase mixture gaseous/solid”.

A particular engine user experienced the repeated fracture of new fixing bolts in combustion chambers upon first tensioning. The heads of the bolts fractured from their shafts upon torque-controlled tensioning. Two affected screws were received from site by the original equipment manufacturer's (OEM) laboratory to perform a metallurgical failure investigation and to determine the metallurgical root cause of failure. From the findings of the failure investigation described in this paper it is concluded that the metallurgical cause of failure in both cases was quench cracking, i. e. the imperfections leading to failure were induced upon manufacture and pre-existed in the bolts, i. e. they did not develop in service. These pre-existing cracks eventually caused failure by torsional overload upon first tensioning in assembly.

A number of perforated-plate fuel oil strainers, called “witch hat” filters for their peculiar shape, were found fractured. Detailed metallurgical failure examinations revealed deflagration within the filter body, i.?e. a mild explosion, as the cause of fracture by tensile overload. No evidence of fatigue was found on the fracture surfaces. The microstructure of the subject failed “witch hat” filters was unremarkable, except for the presence of slip lines immediately adjacent to the fracture surfaces. This is consistent with and evidence for plastic deformation in the area of the fracture. While the metallurgical or physical cause of failure was determined to be tensile overload, where the ligaments between the filter holes fractured, the root cause of failure was found to be abuse in engine service, that caused deflagration of fuel oil vapours within the filter body.

Cold-drawn and cold-bent seamless Alloy 617 pipes were received by a gas turbine engine manufacturer for final assembly. They exhibited multiple cracking on the surface. The extrados of bending zones were most affected by this condition. It was determined that the metallurgical root cause of the problem was inadequate heat treatment by the supplier of the semi-finished product. The cracking upon cold bending the seamless pipes was caused by embrittlement of the pipes’ surface due to nitridation and a further drop in formability as a result of grin boundary embrittlement by secondary carbide precipitates.

Transcription of human papillomavirus type 18 (HPV18) DNA in the human cervical carcinoma cell lines HeLa, C4‐1 and SW756 was studied by nucleotide sequence analysis of HPV18‐positive cDNA clones isolated from a HeLa, C4‐1 and SW756 cDNA library, respectively, and the cDNA sequences were used to predict the potential encoded proteins. The cDNA clones from all three cell lines were found to be derived from virus‐cell fusion transcripts in which 3′‐terminal host cell sequences (different for each cell line) were spliced to 5′‐terminal exon sequences from the HPV18 E6‐E7‐E1 region. Three different types of cDNA clones can be distinguished according to the splicing patterns observed in the 5′ terminal HPV18 sequences. They carry as potential protein‐coding regions the HPV18 specific open reading frames E6 and E6* (generated by splicing and identical with E6 up to the E6* splice junction), E7 and E1 (only in HeLa). Translation of specific cellular genes from the chimeric viral‐cellular transcripts seems to be unlikely. The mapping of the 5′‐ends of the virus‐cell fusion transcripts indicates that transcription is initiated at a viral promoter. The similar patterns of HPV18 transcription in the three different cervical carcinoma cell lines suggest a functional role of HPV18 early genes for the malignant phenotype of these cells.

Over the past decade various clinical trials have used monoclonal antibodies as therapeutic agents against solid tumours. No consistent pattern of response or improved survival has yet emerged although antigenic heterogeneity and insufficient accessibility of cells in advanced tumours have been offered as explanations for these failures. We designed a study in which a monoclonal antibody was used to target minimal residual disease in an early stage of tumour cell dissemination in patients with colorectal cancer. Only patients in Dukes' stage C who had undergone curative surgery and were free of manifest residual tumour were admitted. 189 patients with colorectal cancer of stage Dukes' C were randomly assigned to an observation regimen or to postoperative treatment with 500 mg of 17-1A antibody, followed by four 100 mg infusions each month. A balance of risk factors in the two groups was achieved by dynamic randomisation procedure. After a median follow-up of 5 years, antibody treatment reduced the overall death rate by 30% (Cox's proportional hazard, p=0·04, log-rank p=0·05) and decreased the recurrence rate by 27% (p=0·03, p=0·05). The effect of antibody was most pronounced in patients who had distant metastasis as first sign of a relapse (p=0·0014, p=0·002), an effect that was not seen for local relapses (p=0·74, p=0·67). Toxic effects of 17-1A antibody were infrequent, consisting mainly of mild constitutional and gastrointestinal symptoms. During 371 infusions four anaphylactic reactions were seen, all controllable by intravenous steroids and none necessitated admission to hospital. Adjuvant therapy with 17-1A antibody extends life and prolongs remission in patients with colorectal cancer of Dukes' stage C.

The ZFXgene on the human X chromosome is structurally similar to the ZFYgene, which may constitute the sex-determining signal on the human Y chromosome. ZFYand ZFXdiverged from a common ancestral gene, as evidenced by similarities in their intron/exon organization and exon DNA sequences. The carboxy-terminal exons of ZFYand ZFXboth encode 13 zinc fingers; 383 of 393 amino acid residues are identical, and there are no insertions or deletions. Thus, the ZFY and ZFX proteins may bind to the same nucleic acid sequences. ZFYand ZFXare transcribed in a wide variety of XY and (in the case of ZFX) XX cell lines. Transcription analysis of human-rodent hybrid cell lines containing “inactive” human X chromosomes suggests that ZFXescapes X inactivation. This result contradicts the “dosage/X-inactivation” model, which postulated that sex is determined by the total amount of functionally interchangeable ZFY and ZFX proteins.