Qantas A380 uncontained #2 engine failure

[jlaryh2]

Time Line

My first post, and perhaps my last when you gentlemen get thru with me.

Time Line

May I suggest that all the discussion about temperatures, pressures, speeds, vibrations, etc become totally irrelevant within mili-seconds of the disk rupture. It really only takes less than 'mili-seconds' (consider rotor speed/RPM) for the section of disk to excit the case of the engine, allowing pressures, temps, and vibes to depart the scene---and rotor speeds accellerate/decel depending on the rotor involved. No idea how the 'un-loaded' IPC rotor speed would react from ram air.

[Lonewolf_50]

May I suggest that all the discussion about temperatures, pressures, speeds, vibrations, etc become totally irrelevant within mili-seconds of the disk rupture.

Mostly agreed.

Would you also agree that it is those data up to the pont of failure that are of great interest in getting to root cause, as well as a possible means of using existing sensors to identify a performance change that allows the pilots to
throttle back,
or
secure

an engine changing from steady state into "troubled state" before it gets to "it just failed, oops there goes the turbine wheel" state?

From previous discussion, with different cueing, perhaps the "time for 30 seconds" protocol for engine performing in X state will in some cases change to "retard to idle" or "secure"

if the proper cueing can be figured out from sensors. Being able to adapt the checklists (and the ECAMS available) for the malfunction before it becomes an emergency or materiel failure allows the company to save an engine (to be repaired upon landing), rather than sending bits of itself through the airframe. It's also both safer, and more comforting, for the paying passengers.

It is possible that current sensors can't do that, or provide the granularity required.

[jlaryh2]

Enginne Instrumentation

Yep, I totally agree that the data 'up to the point of failure' is definitely valid data, and after the fact is is the best monday morning quarter back you can have. Of the three indications the crew had (or could have had) was oil temp, vibration, and rotor speed(?). While I haven't been able to read the FDR for timing of the indications, my experience is that none of these indications would have been of significant value to cause any experienced pilot to take immediate action to shut the engine down wothout some anlysis, and in this case he would not have been successful in preventing such a disturbing failure. Even had he been looking directly at the cockpit instruments for this particular engine they are not of sufficient accuracy/detail (such as later provided by the FDR printout) to tell the PIC to shut down the engine in the time span he was faced with.

[edmundronald]

failure diagnosis

It seems to me the ability to diagnose a failing engine is probably reduced by the closeness of a well-functioning engine to its failure regime. So if the engine were a bit overdesigned, it would probably give more warning of failure - and would also offer some more reserve engine power when other engines found themselves unwilling to supply such ...

Edmund

[bearfoil]

Fair Enough,

"Thirty second chronometer started at overtemp"? ECAM enunciated "Engine Fire", F/O reset chronometer back to zero. So more than thirty seconds??

[Lonewolf_50]

bear, as I am not familiar with the Emergency Procedure/Malfunction Procedure, I can only hope that a current 380 driver may be able to expand on what was discussed in Capt Evans interview.

Don't know enough to make sense of the procedures used.

[35YearPilot]

EGT Sensors

T44 is measured by 14 thermocouples (in parallel) in the LP turbine nozzle guide vanes. (GenFam 5-19)

Take this temperature, correct (trim) it with factors stored in the Data Entry Plug (DEP) on the side of the EEC, and you end up with a standardised reference value that is called "EGT".

For info:
Max EGT is 900 deg C (takeoff).
The HPT entry temperatures get up to 1600 deg C.

[barit1]

I've seen other engines with thermocouples similarly configured.

What puzzles me is that the parameter is labeled "EGT" - when it's plainly an inter-turbine temperature. CF6 and CFM's are similarly mis-identified in this respect. We don't call a "pump" a "noisemaker", do we?

[noip]

Bearfoil, Lonewolf,

A mis-understanding of what transpired with respect to the chrono. During certain ECAM procedures, a timer starts to guide the crew as a convenience when to proceed to the next step - eg discharge the Fire Extinguishing Agent.

So, if the ECAM message that was timing down disappears only to re-appear, then the ECAM timer will be reset.

hth

n

[barit1]

edmundronald:

It seems to me the ability to diagnose a failing engine is probably reduced by the closeness of a well-functioning engine to its failure regime. So if the engine were a bit overdesigned, it would probably give more warning of failure - and would also offer some more reserve engine power when other engines found themselves unwilling to supply such ...

All well and good, except that overdesigned means overweight, and profit margins make an overdesigned product uncompetitive. Gerhard Neumann, longtime head of a major engine maker, was notorious for telling his engineers "If it works right on the first test, it's too heavy!"

And "reserve engine power" means excess assymetric thrust, and that means reevaluating the whole aircraft to be sure the pilot can keep it flying straight.

[Fargoo]

I've seen other engines with thermocouples similarly configured.

What puzzles me is that the parameter is labeled "EGT" - when it's plainly an inter-turbine temperature. CF6 and CFM's are similarly mis-identified in this respect. We don't call a "pump" a "noisemaker", do we?

Similarly called TGT on some engines, just semantics isn't it? EGT used as a familiar acronym to suggest engine temperature.

[bearfoil]

barit1

I see it a little differently. RR has been tweaking this design (kudos) for a while, and certainly not adding weight. To the Contrary. Pushing a design to its Thrust-Weight-Service limits is a design goal. Overbuilt is overweight. Overdesigned is a realization just this much past a functional limit. One can get too cute (not here), in asking just a bit too much of a powerplant. Margins are trimmed, weight being one. Weight can mean strength, or durability, or waste; the engineer, test will decide when to stop.

Just a different viewpoint, I don't mean to impute an application of this opinion here.
It is when things start to occur that are previously eliminated from consideration start to happen, that design issues are raised, IMO.

Relative to this occurrence, I see an attempt to suggest a little mystery, and a pinch of poor work by someone "else". This failure, I think is pretty straightforward, after all.

bear

[Turbine D]

barit1

In a way it is a misnomer because the EGT reading doesn't come from the exit end of the LPT like you would expect. I explained in a previous post why the thermocouples are located where they are in modern engines, e.g., inside the stage 1 LPT hollow vanes.
However it is an important engine diagnostic feature and one that is paid close attention to during engine design. The object in design is to design the greatest EGT margin possible, actual operation EGT verses maximum EGT at takeoff. The longer the margin can be maintained, the longer the engine can stay on wing and the healthier the rotor tip clearances and absence of blade tip wear is. Also, it can indicate the accumulation and build up of foreign material on the airfoil surfaces that reduces aerodynamic efficiencies. Water washing has become popular to restore EGT margins these days.

Turbine D

[barit1]

bearfoil & Turbine D - Having been around aircraft engines 55 years (and very close to turbines 44 years) I'm well aware of the semantic issues. Every reference I've found for the "EGT" abbreviation - regardless of the shop it came from - says it's EXHAUST gas temp. That's obfuscation, because I think P&W was the last company to actually measure it at the (core) exhaust.

Thermodynamically, and from a response-time standpoint, the LPT inlet is the right place to place the chromel-alumel thermocouples that (I believe) are standard on most engines.

And "overdesigned" vs "overweight" - in my experience they're nearly synonymous IN MOST CASES. The modern engine shop cannot afford either, although it's certainly nice to build some growth capability into a new design. The CF6 did this rather well, starting at 40k and growing to over 60k while improving just about every operating measure. After its early troubles, customers came to regard it as a desirable powerplant.

[lomapaseo]

Somewhere in all this I completely forgot why the EGT is important or not in all this discussion about the IPT disk burst.

After the fact it only confirms some of the seconday events of a dieing engine.

Is there some other theory behind this discussion?

[Old Engineer]

EGT-- Exhaust Gas Temperature:

If this EGT were actually that, meaning measured at the exhaust, it would have a significance in measuring the maximum theoretical thermal efficiency of the engine. That is, it sets a limit of efficiency in burning the fuel which the actual efficiency of the engine cannot possibly exceed.

If the combustion of the fuel takes place at 1600-C as reported here, and the rejection takes place at 850-C in cruise (it doesn't as this is the LPT inlet limit, so the rejection is at some lower temperature), then the efficiency is found by first adding 273-C to each number to convert to degrees K (Kelvin) above absolute zero. Thus 1873-K and 923-K are used in the following formula:

eff = (combustion temp, minus exhaust temp) divided by combustion temp.

For the HPT and IPT stages combined, this is 50.7 percent in this example.

To this, there is the added percent efficiency gained in the LP stage. The formula for this is:

+eff = (delta temp at LP stage) divided by combustion temp.

Theoretical efficiency is a measure against which to evaluate attained actual efficiency. No point in striving to exceed it, although theoretical efficiency can be increased by raising the combustion temp or by lowering the exhaust temp.

Of course, for running the engine, the LP inlet temperature is more useful.

Edit: When the LPT inlet temp rises, that is because gas temp in the IPT is rising. We are not told whether the alarm limit is because 900-C has been reached in service for an excess of 5 minutes, or whether 920-C has been exceeded. I would think the latter. We know that 950-C was reached, which I would think indicates a rise of 30-C in somewhere between 49 and, I think, 35 seconds. The post (p92) of engine timeline events does not include all the alarms to FD. The interview with Sr Check Capt indicates that the engine shutdown was commenced from the FD following the FO's personal timing of 30 seconds from the initiation of the first overtemp warning. The explosion (SCC's choice of words) of #2 occurred some very little time after the first 30 seconds as measured by the FO.

We are not told whether the 920-C not-to-be-exceeded limit is more significant to the IPT or the LPT. But given the appearance in picture of the LPT, and the absence of the IPT, I think we can infer that this temp is more significant in regard to the IPT, or to events in the IPT. Obviously the temp on the leading edge of the IPT blade disk (1 stage) will be higher than that exiting that disk. But the IPT entry temp appears impractical to measure directly.

Note that the interview puts to rest (IMO) any question of whether on the FD the crew was waiting through the reset of the timer to begin again the 30 seconds from a delayed start. The crew did not delay their actions in response to the initial alarm beyond the procedural 30 seconds, as I read that interview. And they were very alert in doing so. That also suggests that the indicated procedure needs to be revisited. Among other things, obviously.

OE

[barit1]

Old engineer: Absolutely right, except that measuring the combustor exit temperature is a b***h and except for very limited experimental tests, that temperature must be computed based on fuel/air ratio.

Measuring "EGT" (heh...) at the LPT inlet has at least two advantages:
1) Faster response time, more accurate during transients such as engine starting; and
2) It measures core (HP system) efficiency more directly - and the core is where the bulk of deterioration occurs.

lomapaseo: Absolutely right in the present case, although it serves as a confirming parameter in establishing the timeline of the failure. But it's been a red herring (probably my own fault) in this thread.

[Turbine D]

barit1 & Bearfoil

Well now that the "EGT" has been pretty well wrung out, I have been thinking about something in the Trent 900 engine. The most unique feature of this engine is the counterrotation, e.g., The HPC/HPT spool rotating in the opposite direction from the IPC/IPT spool. This is the first high bypass fan engine to contain this feature. The second and third will be the Trent 1000 and the GEnx engines. The difference between the Trent engines and the GEnx is this; In the GEnx, the fan, fan booster and LPT rotate in the opposite direction from the core engine, e.g., compressor and HP turbine spool. For all three engines, this is relatively new technology to save weight and improve efficiencies.

Referring to Rolls Royce's presentation sent to Qantas on 11/18/2010, Rolls says "the Trent 900's HP/IP support structure would be subject to high severity if the engines were operated to 540 psi at P30 on more than 75 flight cycles." In the Trent 900 there are 4 support structures which are commonly referred to as "frames". Of the 4, one supports the the rear of the IP compressor and the forward of the HP compressor and one supports the rear of the HP compressor and the HPT rotor and the IPT rotor. Contained in or supported by the frames are the bearings.

Frames are probably the second most difficult components to design next to HPT air-cooled blades. However, there is a 60 year history accumulated in frame design and design analysis to assist the designer. But there isn't a long history on designing frames were bearings contained in the frame or supported by the frame are rotating in the opposite direction at different speeds. Are there some unique design issues in play here?

I think the most critical of these two frames is the one supporting the rear HPC spool, the HPT rotor and IPT rotor. The biggest concern on this frame would be the stress and thermal gradients experienced during an engine cycle resulting in generation of LCF. Additionally, vibrations or slight unbalance could lead to HCF in certain areas if not accounted for properly. Contained within these frames are the oil feed tubes and return tubes which supply oil to the bearing and from the sumps.

I think Rolls Royce is telling Qantas (and maybe their other customers) a problem has been detected and we need to limit compressor outlet pressure, therefore, fuel flow and the generated maximum thrust to prevent failure of the support structure (frame) until a final modification can be inserted into your engines.

I am not dismissing the spline wear as a non-issue here, it could be contributing in some way.

Turbine D

[barit1]

Turbine D: Sorry, I can't add anything to your frame thesis. But I can't conjure up any special requirements in my mind.

[Old Engineer]

Turbine D wrote:

I am not dismissing the spline wear as a non-issue here, it could be contributing in some way.

My thoughts are the moment are that the spline wear may be being used as the canary in the mine, if in this case it was not the actual cause of the event.

Someone (saying this may help) posted a calendar-format brochure "Modern Gas Turbine" by RR in 2000. I can't find the reference, and because it was like a tri-fold publicity handout I didn't then realize its usefulness. Later I noticed it had individual cabinet drawings of the individual 8 modules of the RB211-type turbine, at different scales to suit the printer's layout (only an inconvenience).

What I noticed is that the LPT module #8 is furtherest aft and carries both reaction bearing for the LP shaft and the anchorage for the drag link that transfers the engine thrust (by compression of the link) to the pylon. The reaction bearing is on the axial centerline of the engine, but the drag link reaction is offset above it at the top outer perimeter of the case. The drag link imparts the entire 72000 lb thrust of the engine to the airframe. This arrangement will cause a bending moment in the turbine case that is most severe at TO or in climb thrust, as the case may be, once per cycle. Not always max, of course.

It is possible this bending is either elastic with rebound, or partly plastic due to high temp of the turbine case, with ever increasing permanent curvature of the case. That the inspection cycles of the splines are ever fewer cycles with increasing damage, a non-linear rate of damage is suggested. This leads me to suspect the latter-- ever increasing permanent curvature.

Such curvature can be relatively small, and yet cause unanticipated wear if the spline/gear teeth are precisely made-- which likely they are. I will explain why in a follow-on post. Vibration will also result and increase from the splines from this cause.

LCF and HCF:

That had not occurred to me, in regard to the spline wear. But of course if thermal stresses are the cause of LCF in this event, it would not be unusual for those to be the greatest stresses in certain structures, or here certain parts of the structure-- certain frames, as you say. It is so normal in structural design to think of the load stresses only (meaning non-thermal stresses) that the thermal stresses can be either overlooked, or just considered only for the major load-bearing elements of the particular structure. And fatigue is (or was in my day) so poorly understood, that after 50 years our understanding still continues to increase, usually the hard way.

Mostly a good measure is a structural layout that accommodates thermal expansion. A turbine has axial symmetry and is ideally unconstrained in longitudinal expansion. But the temperature rises are not uniform in all parts, particularly in sudden transient high loading, which can be awkward. The axial symmetry is not without small deviations, and the longitudinal expansion is eccentricly restrained by the thrust link. It is possible that bearing alignments in the frames are being degraded, with effects on the splines.

That doesn't mean, as you say, that the splines were the proximate cause in this event. However, I still don't see any ironclad proof that an oil fire preceded the breakup. Nothing to disprove it either. As bearfoil said (IIRC) conveniently nothing about oil fire can be contested either way, based on the data released.

OE