Gas turbine owner/operators that do not send key personnel to user-group meetings may be taking unnecessary financial and safety risks. There is, perhaps, nowhere better to learn about issues you should be aware of. Consider the recent liberation events associated with third-stage buckets (S3B) in 7FA gas turbines. This was an agenda item of great interest to attendees at the 2016 conference of the 7F Users Group, May 9-13, at the Rosen Shingle Creek Hotel in Orlando.

The group received an alert on S3B failures from a user at the 2015 meeting and was reminded of it only a few weeks before this year’s conference by a post on the 7F Forum describing another such failure. Yet another third-stage liberation event was reported a week before the users would gather in Florida. In sum, the editors identified six 7FAs that have experienced S3B failures, and one 9FA.

The OEM’s proactive response to the latest two incidents included a preliminary explanatory message for posting on the 7F Forum prior to the meeting, a formal presentation on the issue right after the keynote on GE Day (Thursday, May 12), and a deep dive during the interactive hot-gas-path (HGP) breakout Thursday afternoon, which included participation by about half of the user attendees.

GE’s experts said they believe the underlying cause of the failures is casting defects that caused one blade in each of the affected machines to fracture. The liberated airfoil segment then tore up other S3Bs (Fig 1), sending metal fragments downstream as far as the heat-recovery steam generator (Fig 2). In one instance, the trail of metal terminated at the first row of tubes in the HRSG, with any remaining airborne shards entrained in the exhaust stream captured by the field of finned tubes without damage to pressure parts (Fig 3).

1. Just the upper third of one third-stage turbine bucket liberated and caused all this damage

2. Pieces of turbine buckets liberated, providing a pathway to the HRSG

3. The few remaining pieces of turbine buckets entrained in the gas-turbine exhaust stream were stopped by the first row of finned tubes in the HRSG, without damage to pressure parts

In the latest case, affected components have been sent by the OEM to a third-party for analysis, to help determine the root cause. This could take until yearend, or longer, depending on what investigators find. Believing casting quality is likely the underlying issue, as a first step GE reviewed x-rays for third-stage buckets made from three months before to three months after a failed bucket was cast. Task takes one experienced person about 24 hours to check x-rays for all 92 buckets in the third row. A positive result of this investigation was that several units were shut down for inspection.

GE participants in the 7F meeting appeared confident that S3B failures of the type experienced thus far were unlikely to occur after about 6000 hours of service (for repaired buckets). They also viewed S3B fractures as a low risk for operators, pointing to the company’s experience in repairing and inspecting more than 55,000 S3Bs over the years. The reliability of these airfoils was cited at 99.2% through three HGP inspections.

The OEM is keeping an open mind regarding cause as the investigation proceeds. Its plan is to follow up with the users in early summer with progress report and to update owner/operators perhaps as frequently as monthly thereafter.

Case histories

Key facts associated with each of the liberation events, as told to CCJ ONsite by the owner/operators of the affected engines, are summarized below:

No. 1, 7FA.03, 2016

S3B failed 3600 hours following a major inspection (six months into a planned four-year interval). Bucket row came from another of the company’s units following its refurbishment by the OEM in 2015 after 29,000 hours of service. The crack migrated inward from the trailing edge about 1.5 in. before failure. Note that S3B cracks typically are found between about 3 and 10 in. above the base of the 18-in. airfoil. Unit maintenance is under an LTSA and the OEM replaced the third stage with pre-owned buckets having about 2000 hours of service and no repairs. Project took 20 days.

No. 2, 7FA.04, 2016

Failure happened after 800 hours of baseload operation following a major inspection that included an AGP (advanced gas path) upgrade by the OEM with new third-stage buckets. This was the only incident involving new buckets and may be an “infant-mortality” issue—different from the other five failures profiled here. They occurred on buckets that had been repaired at least once, possibly pointing to the need for refinements to repair processes by all participants—third parties as well as the OEM.

No. 3, 7FA.03, 2015

Buckets repaired at 48,000 hours by the OEM, failure occurred in less than 3000 hours following a return to service. Damage to exhaust casing, flex seals, and piping was experienced.

No. 4, 7FA.03, 2013

Rotor was replaced in the affected machine in 2003; the replacement rotor came bladed with new second- and third-stage buckets. Those buckets were removed for refurbishment by a third-party repair shop in 2009, after 48,000 hours of service. Following refurbishment, the set of buckets was warehoused until installed in a different unit in 2013. The upper one-third of one bucket liberated during a vibration event after operating for less than about 30 hours.

No. 5, 7FA.03, 2013

Buckets repaired at 48,000 hours by the OEM, failure occurred in less than 3000 hours following a return to service. Damage to exhaust casing, flex seals, and piping was experienced.

No. 6, 7FA.03, 2013

Buckets repaired at 48,000 hours by the OEM, failure occurred in less than 3000 hours following a return to service. Damage to exhaust casing, flex seals, and piping was experienced.

Rejuvenation

One independent metallurgist/repair expert told the editors he agreed with the OEM’s assessment that a casting flaw is the likely underlying cause of the failures on repaired buckets. He said a scratch—such as that incurred during the removal, handling, or reinstallation of buckets—or casting flaw creates a stress riser and a starting point for the failure to proceed. HIP or other heat treatments, designed to “heal” certain fully internal flaws can exacerbate some defects, in effect “triggering” the flaw.

High-cycle fatigue is generally thought to be the mechanism causing rapid propagation of any crack that may develop shortly after a return to service. The incident thumbnails above show all failures occurred, in round numbers, within about six months and in less than 3500 operating hours following restart after repair.

A second metallurgist/repair expert cautioned against jumping to conclusions as to failure causes without proper metallurgical analyses of the failed buckets, including the identification of any flaws that may have contributed to the problem. Until this is done, he said, it really is difficult to properly identify the role that any one factor, such as HIP, might have had on the failure.

The expert offered two possible scenarios of how damage might occur as a result of the repair process. The first is where the casting contains one or more near-surface defects—such as those caused by shrinkage during the casting process. In this case, there would be a thin membrane of material between the flaw and the surface that can be effectively “burst” by the pressure of the HIP gas.

The second is where there are tight oxide-filled cracks at the surface. The oxide prevents detection by the fluorescent penetrant used to verify the buckets are crack-free. The exposure to HIP, or vacuum heat treatment, can reduce oxides and allow detection of defects by penetrant.

In both scenarios, the flaws already exist and are merely exposed to NDE by the HIP process.

Inspection

For an understanding of inspection processes best suited to warn of possible impending issues in turbine blades after repair or in service, CCJ turned to Advanced Turbine Support LLC. In a telephone interview with Mike Hoogsteden, director of field services, and inspection experts Dustin Irlbeck and Brett Fuller, here’s what the editors learned:

Ultrasound, radiographic, eddy current, and penetrant inspection techniques are the alternatives. The OEM relies on fluorescent penetrant inspection (FPI) in its shop repair process and recommends it for field checks. FPI has the benefits of being inexpensive and unlikely to produce a false indication. However, it will not recognize a tight crack or lurking problems just below the surface of the airfoil.

Radiography is a “full-volume” inspection method, used primarily in the shop to check bucket internals. It relies on a density difference to identify a fault and a crack might not be revealed. Plus, a radiographic inspection is extremely time-consuming, taking perhaps four or five times as long as eddy current (EC).

Advanced Turbine Support suggests EC for 7FA S3B inspections. It is faster than both ultrasound and radiography and can “see” surface cracks as well as anomalies near the surface of the airfoil. The company has more than 18 months of successful experience inspecting last-row turbine blades in 501F and 501G machines. The protocol developed for this work has been used recently to inspect last-row turbine buckets in 7FAs.