I do not know what methods (torque, wrench arc, ...) are authorized in ATA100 pubs for Trent maintenance. And I won't speculate on whether a torque wrench was available/calibrated/properly used. Your question still stands.

Yep, Bolty is our man.
Hey Bolty, you need to add a Y at the end of your Sirname.
Bolty McBolty ;)

Some news on the repair plans for the aircarft.
No return to home base.

Qantas A380 engine failure | Repairs set to begin | $150 million bill

4-corners
 

I am more than only dissatified that this link was taken of this thread.
Four Corners - QF32
It is the first foto sequence that shows more than the ATSB fotos do. E.G. it shows where the stub oil pipe was mounted and it shows in that blow up of the recovered IP disk segment that there is decouloration of the bore section - blueish signature -
This leads to two conclusions:
a. there was indeed an oil fire that weakened the disk and
b. the stub pipe broke probably by fatique
What immediately raises the question what caused that fatique failure ??
If I am not completely off track fatique failure is caused by intermittend loads. We also know by those graphs published in the ATSB report that oil pressure was decreasing from normal levels but stayed well above critical amounts, as was the quantity of oil available.
This once again very clearly indicates the total failure of that oil tube happened at the time the engine disintegrated and only the oil spelled through the fissure had accumulated enough of an amount to start that fire.
It is by no means identified in that report what really is the cause of the selfdestruction of the engine and subsequently almost the loss of QF 32. But typically it mentioned that 19 engines had received a "new module" and just that one that blew apart didn´t.
But I think by now every one that has had a closer look to the sequence of events has to come to the conclusion that vibrations of an unexpected amount has killed that engine. Though outside leakage that also is testimony of the last incident of a QF airbius flying to London on 3 engines.
I believe it is time to release those unknowns of the engine behaviour. It may hurt RR but it will give a safer operation of that engine / aircraft combination a chance.
I hope that RR is far beyond our imagination of whatneeds to be done.
Jo

Rolls Royce under scrutiny.
 

Mar 28, 2011

"A Four Corners investigation into last year's mid-air explosion of a Qantas A380 jet engine raises serious concerns about quality controls at Rolls-Royce.

It reveals the engine-maker was aware of a faulty weld on an oil pipe in one its Trent 900 engines a year before the explosion happened.

Despite the initial warning, the company went on to manufacture another faulty pipe.

The 2009 crack was caused by a sub-surface void which had been created in the welding process.

Three more oil pipes were found that were considered by regulators to fall outside the safety margin for thickness."

Rolls-Royce under scrutiny after Qantas explosion - ABC News (Australian Broadcasting Corporation)

A new wing or a repair?
 

Some interesting updates about the fate of the Qantas A380 on this thread.

News on QF / Trent 900
 

I just found this report in "flightglobal"
Partial power loss on Qantas A380 caused by oil leak

I think of greatest importance is this sentence of the report
Quote:
The ATSB add that it is now conducting a "holistic investigation" into the oil leak and other reported cases of oil leaks in Trent 900 engines.

As I understand this announcement, ATSB has come to the conclusion that there has to be more behind these engine problems - QF 32 and others - than just oil feed pipes.
Jo

Nothing New here

All Engine models leak oil sometimes. Where it leaks is important. Internally is important as it can ignite.

The problems with this engine are under close watch to ensure that corrective action is effective.

"The Australian" newspaper writes today:
"Rolls-Royce was unable to tell how many more of its Airbus A380 engines might explode after last year's Qantas near-disaster near Singapore because of a lack of records."

see link:
Records missing on A380 engines | The Australian

vic cairns

From the article, I see the minimum value for wall thickness of this stub pipe is .5mm. Can that be correct?? It is also the minimum "crest" measurement for the splines at IP shaft rigid coupling. The AD specifies a range for splines of 0.5mm to 2.65mm, so we know the AD is correct, but 0.5mm for an oil feed pipe? I think that is about 0.045 inch?

change in fatigue location
 

The interim report text calls out a (to me very minor) change in the diagnosed fatigue crack location of interest. This version is posted in the new interim report:
http://i6.photobucket.com/albums/y23...May2011rev.jpg
while this is the one provided late last year--downsampled by me to be a better match to the new and to be more suitable for posting here:
http://i6.photobucket.com/albums/y23...-bore_2010.jpg
Here is a link to the full resolution version from last year:
full resolution 2010 picture
(depending on your browser and Photobucket, the high resolution one will display at less than full resolution when you click my link--the clue may be a plus sign in your mouse pointer--one more click will take you up to full res.

I'm not claiming this to be important--but it is new.

Bear,

.5mm is a little less than .02". Even thinner than you thought.

archae86

It is extremely important. The smooth face is not indicative of either wear or fatigue, imo. For all we know that could be a keyway to accept a "c" ring, coupling m/f lines. It's tempting, but the "pipe" was done to exhaustion, so suffice that I still believe the "misalignment" is wear from excess vibration twixt male and female, not a blown end bore assignment. To add (edit), the length of the arrow indicating fatigue could be the limit of the metal as built. In other words, the smooth area defines the "former" position of the missing metal. That is oddball, and if I didn't know better, I would say that interpretation is misleading.......not to mention that the gouge at the end of the pipe mimics the limit of vibration. Note the "swirl" marks on the face of the injury? One might almost calculate the frequency of the damaging vibration from those telltales....

chuchu

Smokin' Eddie, that ain't too pretty much thick enough for oil, innit ?

bear

Bear,
The "new" area they now point out as the fatigue crack almost looks to me as a brittle failure (no ductility) rather than a fatigue failure, in other words, it just snapped due to lack of support of the rest of the structure that was progressively failing. I would have a tendency to agree with you relative to vibration.

TD

Hi TD...

This picture has always shown to me a coupling I have known as a "Quik-Couple". Male/Female, a sliding sleeve to align and retain the two mates. The "Misaligned counter bore" I have assumed is NOT a machining fault, but the result of a loose connection where the male, the counter-bored part is vibrating in the walls of the recipient.

A "counter bore" is the attaching piece, "counter" meaning mate. Its "misalignment" responsible for the sloughing off of wall material.

The striae on the walls show a vibration that likely blew off the large pieces, and in so doing, opened up the "cone" of play that caused the brittle fracture you speak of ?

The investigation seems to view it as a manufacture fault, maybe they have other info that corroborates that view. If it is faulty to start with, then surely misalignment would result in a poor connection, which would then lead to more vibration of the connection, then leading to failure as you state. Or maybe cause and effect as you state. Chicken - egg.

Would be nice to have pictures of the whole pipe assembly - I still feel we're only looking at half (or less) of the picture.

ATSB May 18 Update
 

Is there new information contained in this ATSB Interim Factual Update on 18 th May? Interested in expert comments.

Thanks




The Australian Transport Safety Bureau is investigating an occurrence involving a Qantas A380 aircraft that experienced an uncontained engine failure over Batam Island, Indonesia on 4 November 2010. The aircraft landed safely in Singapore having returned with the aircraft's No 2 engine shut down. There were no injuries.
The investigation team has inspected the damaged engine and components and determined the sequence of events that led to the failure of the engine disc.
The investigation is also examining the airframe and systems damage that resulted from the engine disc burst to understand its effect on those systems and the impact on flight safety. That includes their effect on the aircraft’s handling and performance and on crew workload. A flight simulator program was used to conduct a number of tests in a certified A380 flight simulator. Analysis of the flight simulation test data is ongoing.
The investigation is continuing.

The investigation team has developed into a large multi-agency group consisting of the Australian Transport Safety Bureau (ATSB) as the lead investigation agency with assistance from the French Bureau d’Enquêtes et d’Analyses (BEA), the Air Accident Investigation Branch of United Kingdom (UK AAIB), the Air Accident Investigation Bureau of Singapore (AAIB Singapore), the National Transport Safety Committee of Indonesia (NTSC), and advisors to the various investigation bodies from Rolls-Royce, Airbus, SAFRAN Sagem, Honeywell (USA and UK), Aerolec UK and Singapore Aero Engine Services Private Limited.

Since the on-site phase of the investigation, the European Aviation Safety Agency (EASA) has facilitated a meeting between EASA, Rolls-Royce, Airbus and the Civil Aviation Safety Authority of Australia (CASA) with the ATSB, BEA and UK AAIB attending as observers. That meeting was to establish if the Rolls-Royce Trent 900 engine and the Airbus A380 airframe met the design certification requirements for the engine and airframe in light of the significant damage that resulted from this event. The engine manufacturer - Rolls-Royce - and the airframe manufacturer - Airbus - presented technical data and findings to that group. EASA and CASA agreed that, based on the information supplied, the airframe and engine meet the certification requirements. However, further investigation into the aircraft’s structure and systems and engine behaviour is continuing to fully understand this event and establish if there are safety issues that need to be addressed.

The ATSB, UK AAIB and Rolls-Royce have inspected the damaged engine and components
and determined the sequence of events that led to the failure of the engine disc. As a result of those findings, Rolls-Royce published a series of non-modification service bulletins (NMSB) with various amendments and a Service Bulletin (SB) to manage the continued serviceability of the Rolls Royce Trent 900 engine and EASA issued two Emergency Airworthiness Directives and one Airworthiness Directive. Qantas Airways initially placed operational restrictions on the Australian A380 fleet in conjunction with CASA. Commencing on 16 January 2011, those restrictions were progressively removed, as the fleet’s continued airworthiness was established.
The investigation has found that the intermediate pressure (IP) turbine disc failed as a result of an overspeed condition, liberating sections of the IP turbine disc that then penetrated the engine case and wing structure. The disc failure was initiated by a manufacturing defect in an oil feed pipe that resulted in a wall thickness reduction in an area that is machined to receive a coarse filter. That section of the oil feed pipe sustained a fatigue crack during engine operations that led to an internal engine oil fire that weakened the IP turbine disc. In turn, a circumferential fracture was induced around the disc, allowing it to separate from the IP turbine shaft. The unrestrained disc accelerated to critical burst speed. This led to the No 2 engine failure and subsequent significant penetration damage to the airframe structure and systems.
Since the ATSB’s Preliminary Report was issued, analysis of the oil feed pipe fracture surface has progressed and the investigation team has a better understanding of the failure mechanism. Technical reviews to date of the available evidence have established that the location of the fatigue cracking that was depicted in Figure 9 of the Preliminary Report1 is not the area of interest. The area of fatigue cracking and misaligned counter bore is now understood to be as shown in Figure 1 below.

Figure 1: Updated location of fatigue cracking

At the time of the accident, there were three oil feed pipe modification standards in the IP turbine module case (module 51) of Rolls-Royce Trent 900 engines. Oil feed pipe modules were manufactured to those modification standards as follows:

1. FW48020 standard modules, between October 2005 and May 2008 (42 units).
2. FW59326 standard modules, between July 2007 and March 2009 (67 units).
3. FW64481 standard modules, March 2009 and January 2011 (97 units).

　

As a result of this accident, Roll-Royce instigated the removal of Rolls-Royce Trent 900 engines from service with the following module 51 standards:

1. all FW48020 standard modules, which included the accident engine
2. 10 FW59326 standard modules
3. one FW64481standard module.

　
A lack of measurement records for the FW48020 standard oil feed pipes meant that Rolls-Royce was unable to establish whether those oil feed pipes had been manufactured to specification. A subsequent risk assessment by Rolls-Royce determined that there should be a fleet wide removal of the FW48020 standard engines from service.
The measurement records for a number of FW59326 standard engines were also not available. An on-wing measurement technique identified seven of those modules with an oil feed pipe wall thicknesses of less than the Rolls-Royce stipulated minimum acceptance limit of 0.5 mm. Those engines were removed from service. In addition, three other FW59326 standard engines
1 See www.atsb.gov.au/media/2888854/ao-2010-089%20preliminary%20report.pdf
- 2 - - 3 -
were removed from service after an evaluation of their manufacturing measurement records.
2 In Airbus products, the relationship between a flight crew order (or control input) and the aircraft response is termed a ‘flight control law’. The main objective of the normal control law is to provide instinctive and comfortable handling characteristics and comfort to those on board
Measurement records were available for all FW64481 standard module 51’s. However, the oil feed pipe wall thickness on one FW64481 standard module was found to be less than the minimum acceptable limit and that engine was removed from service.
The oil feed pipe wall thickness for all remaining Trent 900 engines in operation was found by Rolls-Royce to either meet or exceed the minimum acceptable manufacturing limit.
The ATSB, in conjunction with the UK AAIB and Rolls-Royce, is examining the circumstances and missed opportunities with the potential to have detected the reduced wall thickness and offset counter bore of the oil feed pipe prior to, during and after the manufacture of the module 51 assemblies. The ATSB is also reviewing the quality audits undertaken of, and the quality assurance system affecting, the module 51 design and manufacturing process and their effectiveness in detecting deficiencies in that process.

The ATSB, in conjunction with Airbus, BEA and AAIB UK is also examining the airframe and systems damage that resulted from the engine disc burst to understand the effect on those systems and the impact on flight safety. That includes their effect on the aircraft’s handling and performance and on crew workload.
As part of the investigation, a flight simulator program was developed by Airbus from data that was obtained from the aircraft’s digital flight data recorder and cockpit voice recorder, from the aircraft’s fuel quantity management system, and from pilot interviews conducted by the investigation team. An A380 test pilot and group of experienced A380 flight crews from Airbus, the BEA and the ATSB conducted a number of tests using that simulator program in a certified A380 flight simulator at the Airbus facility in Toulouse, France. Those tests sought to establish the aircraft’s handling capabilities with the simulated damaged fuel transfer system, damaged flight controls and lift augmentation devices, and damaged electrics and electronic systems having effect. Various speeds and flight profiles were examined that simulated the workload that was experienced by the crew during the event. The simulation found that the aircraft had operated in ‘normal control law’2, in which, regardless of a flight crew’s input, computers prevent the exceedance of a predefined safe flight envelope. If there are certain types or combinations of failures within the flight control system or its components, the control law automatically changes to a different configuration level: alternate law or direct law.


The investigation is continuing and will include:

1. the testing and analysis of the black-coloured soot residue that was found in the left wing internal (No 2) fuel tank.
2. additional analysis of the flight simulation test data
3. the examination of the airframe and systems damage that resulted from the engine disc failure
4. the ongoing review of the quality control and quality assurance system affecting the Trent 900 module 51 design and manufacturing process
5. analysis of the flight crew work load
6. the review of the aircraft’s maintenance, including engine workshop visits.

　
At the time of this update, the aircraft remained in Singapore, where repair schemes were being developed by Airbus and relevant components were being manufactured to facilitate that repair.
The gathering and compilation of the large amount of complex factual information is anticipated to be concluded by the end of July 2011. The analysis of that information and development and review of the investigation, including by directly involved parties in accordance with international protocols, is anticipated for completion by May 2012.

ASTB May 18 Update
 

eagle farm,

As I read this interim update, there really is nothing new except for the correction made as to the area of fatigue in the photo of the fractured stub pipe. The report does give the focus of the ongoing investigation activities, most of which is focused on the aircraft, not the engine. The only engine activity is that concerning manufacturing and quality control related to the stub pipe failure at Rolls Royce.

So there isn't much more to say until the final report comes out, sometime in the future. The recommendations contained in the final report ought to be interesting.

TD

TurbineD

A final note on the updated "Stub Pipe" pic. The BEA merely seem to have chosen an alternate way to show the area of fatigue cracking. In the original, the line incorporates the actual area. In the follow up, it uses "limit line" id. Different way to show the same thing. ("Here is where the fatigue stops", etc.).

bear

Bear,

You could be right, I just never thought of it that way.

TD

TD

Did you ever look at that A330 (Singapore) "double bang" emer. return?

It seemed a bit similar to Qantas UF, then disappeared?

bear

While the corrective action relative to the oil fire (faulty pipe manufacture, etc.) is entirely appropriate, I hope that R-R doesn't stop there.

The fact that the T900 engine cannot survive a IP shaft disconnect is really disturbing to the turbomachinery folks I see often. I give you the RB211/744 failure QF74, 31 Aug 2010, KSFO - the shaft broke, the IPT oversped, but it shed its blades before reaching burst speed. An uncontained failure (escaped turbine blade fragments), but with considerably less airframe damage.

A couple of aggravating circumstances here

It may not have been just a pure overspeed failure. Given the coloration of the disk there may have been abnormally hot temperatures at the bore weakening it further than the shed speed of the blades.

The uniqueness of the IP module area may have prevented the massive tangling between blade airfoils and stator vane airfoils.

nevertheless the first priority corrective action should be against the massive oil leak that caused the shaft attachment failure in the first place.

Yes, a sump fire is a nasty situation, but there appears to be a six-second interval between the shaft separation (evidenced by N2 IPC spooldown) and the IPT burst. Given the gas stream energy still available from N3, that unloaded IPT disc must have reached a very high speed.

I perceive a lack of "what if" planning. The detached disk drifted aft until it contacted some static structure (mid frame) and established what designers call a "false bearing", which had low enough friction to permit the disc to continue accelerating. There were no static nozzle guide vanes to clash with turbine blades, thus destroying the driving torque; the blades remained attached as the disk kept accelerating; and the next part to fail was THE DISK ITSELF.

So at least two opportunities to survive the sump fire were lost IN THE DESIGN PROCESS.

Edit: On second thought, I wish to change the word "survive" in the last sentence to "mitigate".

She's on her way back home...

:)

--Bill

pardon me for asking this..

but.. I really need to know what action(s) have been taken in light of this engine failure?

From what I know, the problems only affects those A380 with RRoyce Engines correct?

I'm due to travel with Singapore Airlines soon.. transatlantic flight to LA, USA..
I believe all SQ's A380's are fitted with RRoyce engines..

This kinda make me wary..

I have the option of flying with another airline.. they use B777-300ER for route to LA,USA..

I am rather confused now... go with SQ a380 or shall i just go with B777-300ER :)

any comments appreciated! Thx

Kucing

I wouldn't worry if I were you. This particular problem will have been examined to the Nth degree.

Turbine disc failures are very rare and have happened to most (if not all) large engines, not just Rolls Royce. It is even more unlikely to happen again now after all the work that has been done.

Just take the most convenient flight and enjoy it!

To be quite clear about QF32 and similar failures:

In ANY multispool engine, IF the high-pressure spool keeps running after a shaft separation of (one of) the lower-pressure spools, that low-pressure turbine will be driven to an overspeed condition.

Thereupon, one of perhaps three things will occur:

1) The turbine blades will, by design or accident, suffer a blade root failure. This creates medium-energy shrapnel, but also removes the driving torque from the free disc, so it can coast to a stop without further failure. eg QF74, SFO 744, RB211 engines.

2) If 1) doesn't happen, the free disc assembly will be driven aft where the blades will contact static parts such as a downstream nozzle guide vane ring; this will beat up the airfoils to the point that the disc stops accelerating. I have seen CF6 failures of this type.

3) if 2) doesn't happen, then the disc may continue accelerating to the point that it bursts. I give you QF32, Trent 900 - and also perhaps the T1000 test bench failure a few weeks before QF74.

Turbomachinery designers must take shaft failure into account in the design process. Some may do a better job than others.

barit1,

Good post!:ok:

I completely agree with what you said, it's all in the design...

Regards,

TD

Than there is the problem of too much oil leakage and resulting fire overheating the disk and it's attachment shaft to the point of burst before it even has a chance to move aft.

Thus the fix is to minimize a short term oil leak of this magnitude.

Obviously a lesson for all manufacturers in their quality control efforts.

lomapaseo:Yeah, that will address this particular failure sequence (overheated disc flange).

But decades of experience says there are potentially other causes for shaft separation, and the fundamental issue is the lack of any provision in the Trent to destroy the runaway turbine airfoils prior to disc burst.

I will grant that the six-second interval between N2 compressor spooldown and N2 turbine burst might yield promise for a software fix >> namely, fuel cutoff if sensed N2 is too far out of whack.

Barit1

I doubt that the regulators will agree with that

Do any of the manufacturers altually demonstrate a shaft separation at takeoof conditions, or do they just point at some design feature that might have worked in the past?

As in all accidents there are lessons to be learned by all

lomapaseo,

What you do is understand this is a possibility, and you design so that if the power drive arm would fracture releasing the turbine wheel from the shaft, it would not reach an overspeed condition where the disc would come apart. Although this was done on previous Trent engines, it wasn't done on this engine. It wasn't a lesson learned, it was a lesson forgotten, simple as that. You don't have to demonstrate it, you just use experience and good common sense to prevent it.

TD

So if you've never seen it in a model series it's good common sense

but once it happens it's an oops

Good design practice requires a failure modes / criticality analysis (FMECA), and a good analysis will consider historical known failure types. Not just those required by regulators, but ones that (for example) could taint the reputation of the product or the company.

For example, a modern propeller design incorporated features that lost sight of lessons learned over 6 or 7 decades. As a result, a regional airliner augured in 20 years ago. Neither the designers nor the FAA had retained the corporate wisdom to prevent the design error. :ouch:

barit1

The LP shaft by observation incurred no visible injury from the burst of the IPT. The HP shaft, nested withn the Drive Arm "Bell" is not seen. The third shaft of three, The Ishaft, terminates in a cylinder, the "Bell". Would you have any comment as to the vulnerability of this "Bell" to harmonic vibrations produced by the nested pair, and possible "whip" of this assembly, due wear of splines more forward of the IPT case? Considering the vibratory result can be multiplied through the length of the IP/HP shaft pair as it progresses aft, and in a super heated environment, would mechanical disruption play an important part in the burst (not to diminish the effects of "oil fire").

N speeds are not available at the time (per design) to effect a fuel cut. Instead, when N2/N1 values are rejected, the ECM recieives an NCD re: N3, and a value is providied that is 'theorized' by the Engine computer. (as I understand it). This allows the HP to soldier on, at least for six seconds?

regards

Lyman:

By use of the term "shaft failure" I mean the functional disconnect of the IPC from the IPT, regardless of the specific location of the break. The IPT drive arm is one component of the shafting. You can theorize all day long about vibration, shaft harmonics, etc. but these are merely possible logical routes to the ultimate shaft disconnect.

Good design practice assumes that the shaft will, despite all your efforts, suffer a failure someday; and it's the designer's job to mitigate the failure.

And yes, I know that R-R N2 sensing is at the compressor end, so there's no hard data from the IPT itself if a disconnect happens. But the observed N2 spooldown is non-characteristic when the other rotors are still up to speed, and should be interpreted as the signature of a shaft failure, IMHO.

I am trying to build on what is (was) known at the time. The AD warned specifically of aftward axial transit of a shaft, causing damage to the a/c, and those on the ground (sic). The wear, specific to the AD, was located at the splines forward of the IPT (at the other end of its shaft), and reporting of the initilal exam of the engne contents stated "Rigid Couling Failure". Rigid coupling failure was reported as the cause, not the result. The "Burst" was an artifact of the failure of the Bell, (Drive Arm), not the cause of the uncontained failure. Now it's possible to chiicken/egg this, but I am aware of the Rigid couple demise as the cause of the problem, and the couple disintegrated due, What? I don't think the couple would fail from heat, nor vibration alone. I also sense that the consensus here is that the wheel failed "first". i think that is wrong, from a read of the report. It is also inconsistent from the previous failure, (Miami), where the wheel failed from migration, not oil fire. imo.


For instance, couldn't the N2 spool down have been caused by the contact with the web? The actual friction and resistance slowing the shaft and puddling the "false bearing"? This while the shaft/turbine maintained its integrity, until the rigid coupling lost the Wheel?

your thoughts?

Lyman,

We have been through this many times. This engine had been inspected and passed the spline wear requirements before being placed back in service. The rigid coupling of the IP shaft, located in the compressor section of the engine, did not fail. The IP shaft did not move rearward. The initiation of the failure event started with an oil leak in a compartment just forward of the IP Turbine disk. A fire started as this area is hot. The fire caused the IP turbine disk to begin to heat above temperature capability of the disk alloy and rotational stresses being encountered at the time. Now to understand what happened next, you have to understand the metallurgy of superalloys used in disks.

The best disk superalloys have an operating temperature of between 1200℉ and 1400℉. These alloys have good ductility and creep rupture capabilities within operating temperature ranges. If the temperature is exceeded, creep and in the case of a spinning disk, outward growth (stretching) will occur fairly rapidly. As the stages of creep progress, the final stage progresses to failure very rapidly. So lets apply this to the IP disk in question.

The IP disk is attached to the turbine end of the IP shaft by a series of circumferential bolts which secure the power drive arm of the IP disk to the shaft. The designers determine the thickness of the power drive arm, the web thickness of the disk and the disk bore mass based on anticipated stress levels and temperatures during engine operations throughout the flight envelope plus a safety margin to preclude disk burst. So what happens when an oil fire develops in the compartment just forward but adjacent to the IP disk? The disk begins to overheat from the bore to the disk web with the overheating commencing from the forward side of the disk. As the temperature begins to exceed the superalloy capability, the disk begins to stretch somewhat unevenly from front to rear, but radially. But, the power drive arm is firmly attached to the end of the shaft. As the stress limits begin to be exceeded, the power drive arm fails, releasing the disk and the disk is free to rotate with no control over rotational speed. As the rotational speed increases with no impediment to slow it, it bursts. That, I believe, is what happened on this engine.

Regards,

TD

While I don't have details at hand, Lyman may be referring to the RB211 failure (QF74, SFO 8/2010) in which turbine shaft splines were wiped.