EC225 crash near Bergen, Norway April 2016
Join Date: Sep 2007
Location: USA
Posts: 601
Likes: 0
Received 0 Likes
on
0 Posts
In my post I stated, "...where the main gearbox TTO spiral bevel pinion suffered catastrophic plastic failure of the gear teeth from elevated temperatures...". This is a conclusionfrom theCAB investigation report A09A0016 sec. 1.12.3 . From the picture of the failed TTO spiral bevel pinion included in the report section linked, most notable is that all of the teeth are missing and it does appear to be the result of severe plastic deformation. However, there is also evidence of similar type damage on both of the pinion's bearing journals which would agree with your statement about the tapered roller bearings failing. Since both the gear teeth and bearing journal damage appears to be severe plastic deformation resulting from loss of mechanical strength due to the material being heated well above its tempering temperature, it seems logical to conclude a C64 pinion would have performed much better than a 9310 gear. At the tempering temperature used with C64 gears, a 9310 gear will have experienced a reduction in tensile strength of ~35%. The loss of drive to the tail rotor would seem to be the result of the TTO pinion losing all of its gear teeth. But I may be wrong.
Lonewolf_50-
ISF of gear tooth flanks definitely provides some additional scoring margin under marginal lubrication conditions such as that experienced during loss of lube operation. The reduced mean surface roughness provided by ISF improves the min hydrodynamic oil film thickness the gear contacts can safely operate at. In case you're interested, the technical term for the relationship between hydrodynamic oil film thickness and mean height of the mating gear surface asperities is called "lambda ratio".
Join Date: Feb 2009
Location: A nice place
Posts: 160
Likes: 0
Received 0 Likes
on
0 Posts
Ye gods you’ve got to get through a number of pages before you get to this topic.
Anyone privy to information beyond the AH solid wall of silence that SKF v Timken bearings has any relevance to good/bad 2nd stage planetary gears?
Anyone privy to information beyond the AH solid wall of silence that SKF v Timken bearings has any relevance to good/bad 2nd stage planetary gears?
I hope you are in no way implying this is from an Airbus product. The 206 part number on it confirms it is a Bell part. Do you have any first hand information on this, did you find it on the internet, do you know whether or not there were any chip lights, I seriously doubt they just found this at overhaul. It is of course possible but based on my experience with Bell light and mediums helicopters it is not likely this would have been undetected. This photo doesn't really even belong in this thread.
An element of the discussion is bearings in helicopter gearboxes, whether this particular bearing in #1535 came out of a AH or Bell or Robinson product, whether made by SKF (in this case) or somebody else doesn't matter. It does show the nature of damage that can occur in a failed bearing, and in that context I think shows relevance in the general discussion.
I think gullibell's post is relevant.
It points out that there are not a lot of huge design differences in most helicopter MGB's.
These components are invariably not made by the OEM.
There are plenty of MGB failures that are not catastrophic and there are some that but for some grace easily could have that do not get into the public domain.
Something really obscure happened here.
There are a large number of factors in this accident and some that may be difficult to trace unfortunately, for various reasons, which will make any fix possibly beyond economic as it seems.
It points out that there are not a lot of huge design differences in most helicopter MGB's.
These components are invariably not made by the OEM.
There are plenty of MGB failures that are not catastrophic and there are some that but for some grace easily could have that do not get into the public domain.
Something really obscure happened here.
There are a large number of factors in this accident and some that may be difficult to trace unfortunately, for various reasons, which will make any fix possibly beyond economic as it seems.
I think gullibell's post is relevant.
It points out that there are not a lot of huge design differences in most helicopter MGB's.
These components are invariably not made by the OEM.
There are plenty of MGB failures that are not catastrophic and there are some that but for some grace easily could have that do not get into the public domain.
Something really obscure happened here.
There are a large number of factors in this accident and some that may be difficult to trace unfortunately, for various reasons, which will make any fix possibly beyond economic as it seems.
It points out that there are not a lot of huge design differences in most helicopter MGB's.
These components are invariably not made by the OEM.
There are plenty of MGB failures that are not catastrophic and there are some that but for some grace easily could have that do not get into the public domain.
Something really obscure happened here.
There are a large number of factors in this accident and some that may be difficult to trace unfortunately, for various reasons, which will make any fix possibly beyond economic as it seems.
Is it really true that ' something obscure happened here'?
Au contraire, it seems obvious to me that the gearbox design has insufficient margin, based on the record.
That is a solvable problem, which AH tried to finesse after the first crash.
Can they solve it while still making a return? After the second crash, I don't think so. AH has to eat the dirt and fix the gearbox, not for the 225s sake, but to be an acceptable supplier going forward.
I recognize that AH depends on its suppliers, so that the fault is not solely that of AH, but AH blessed the final product.
Still, AH can and will surely have the suppliers eat part of the cost
Join Date: Sep 2007
Location: USA
Posts: 601
Likes: 0
Received 0 Likes
on
0 Posts
It is indeed true that the bearings used in most helicopter gearboxes are manufactured by a vendor like SKF or Timken. But the bearings are made to a detailed specification supplied by the OEM.
The particular type of planet gear bearing shown in the example above is quite common with helicopter gearbox epicyclic stages. But one problem with these epicyclic stages that use more than 3 planet gears is their capability to evenly share loads between the planet gears. With the example of the EC225 output stage that used 8 planet gears, the load mismatch between the planet gears/bearings could be up to +/-25%. And the fatigue analysis for these planet gears/bearings would need to take this into account.
The particular type of planet gear bearing shown in the example above is quite common with helicopter gearbox epicyclic stages. But one problem with these epicyclic stages that use more than 3 planet gears is their capability to evenly share loads between the planet gears. With the example of the EC225 output stage that used 8 planet gears, the load mismatch between the planet gears/bearings could be up to +/-25%. And the fatigue analysis for these planet gears/bearings would need to take this into account.
Join Date: Jun 2016
Location: Aberdeen
Posts: 90
Likes: 0
Received 0 Likes
on
0 Posts
It is indeed true that the bearings used in most helicopter gearboxes are manufactured by a vendor like SKF or Timken. But the bearings are made to a detailed specification supplied by the OEM.
The particular type of planet gear bearing shown in the example above is quite common with helicopter gearbox epicyclic stages. But one problem with these epicyclic stages that use more than 3 planet gears is their capability to evenly share loads between the planet gears. With the example of the EC225 output stage that used 8 planet gears, the load mismatch between the planet gears/bearings could be up to +/-25%. And the fatigue analysis for these planet gears/bearings would need to take this into account.
The particular type of planet gear bearing shown in the example above is quite common with helicopter gearbox epicyclic stages. But one problem with these epicyclic stages that use more than 3 planet gears is their capability to evenly share loads between the planet gears. With the example of the EC225 output stage that used 8 planet gears, the load mismatch between the planet gears/bearings could be up to +/-25%. And the fatigue analysis for these planet gears/bearings would need to take this into account.
In any single revolution, ovality of the sun gear or of the ring gear may affect the sharing of load over the set of planet gears. Could there be a numerical phenomenon whereby the number of gear teeth, the number of planets and the initial assembly position of the gears with respect to their individual maximum ovality axes could combine at the same position and repeat with more frequency (and reduced fatigue life) than say a similar module with 9 planets and corresponding differences in teeth numbers? It may sound a bit far fetched but it might explain how a gearbox could pass all the inspection checks at major overhaul yet fail relatively soon afterwards if that one gear repeatedly takes a higher share of loading than the others.
It shouldn’t be as difficult as cracking the Enigma code machine (another cluster of cogs) to work out if this can happen.
I still think the primary ‘cause’ will be found to be a manufacturing process producing an irregularity in the polycrystalline material but the design margins that should prevent this leading to failure do not appear to have been wide enough to compensate for manufacturing or operating margins.
Join Date: Feb 2009
Location: A nice place
Posts: 160
Likes: 0
Received 0 Likes
on
0 Posts
Would that +/-25% be for a brand new matched set or a mixture (after an overhaul) of new and ‘on-condition’ gears that may have used up some of their wear and fatigue lives?
In any single revolution, ovality of the sun gear or of the ring gear may affect the sharing of load over the set of planet gears. Could there be a numerical phenomenon whereby the number of gear teeth, the number of planets and the initial assembly position of the gears with respect to their individual maximum ovality axes could combine at the same position and repeat with more frequency (and reduced fatigue life) than say a similar module with 9 planets and corresponding differences in teeth numbers? It may sound a bit far fetched but it might explain how a gearbox could pass all the inspection checks at major overhaul yet fail relatively soon afterwards if that one gear repeatedly takes a higher share of loading than the others.
It shouldn’t be as difficult as cracking the Enigma code machine (another cluster of cogs) to work out if this can happen.
I still think the primary ‘cause’ will be found to be a manufacturing process producing an irregularity in the polycrystalline material but the design margins that should prevent this leading to failure do not appear to have been wide enough to compensate for manufacturing or operating margins.
In any single revolution, ovality of the sun gear or of the ring gear may affect the sharing of load over the set of planet gears. Could there be a numerical phenomenon whereby the number of gear teeth, the number of planets and the initial assembly position of the gears with respect to their individual maximum ovality axes could combine at the same position and repeat with more frequency (and reduced fatigue life) than say a similar module with 9 planets and corresponding differences in teeth numbers? It may sound a bit far fetched but it might explain how a gearbox could pass all the inspection checks at major overhaul yet fail relatively soon afterwards if that one gear repeatedly takes a higher share of loading than the others.
It shouldn’t be as difficult as cracking the Enigma code machine (another cluster of cogs) to work out if this can happen.
I still think the primary ‘cause’ will be found to be a manufacturing process producing an irregularity in the polycrystalline material but the design margins that should prevent this leading to failure do not appear to have been wide enough to compensate for manufacturing or operating margins.
Anyone got information on AH SKF/Timken good/bad belief?
Spalling
Agree that the Bell pinion photo is relevant, but it may have helped initially if it explained that it was a different product.
Spalling is a quite normal "wearout" mode in gear or bearing designs. It is actually a form of metal fatigue itself at just below the surface of contact due to contact stress. It can initiate at asperities, or even surface damage from dents, handling, or debris going through the system. It progresses from smaller pits into larger flakes. Often a small pit, or beginning spall, will not produce a significant chip enough to be indicated, although they will usually stick to the magnetic detector commonly found on most aerospace transmissions. The progressive nature of the spall produces more chips, which will accumulate, and bridge the gap on the detector. My 50 cents guess was that the one in the photo was detected in such manner.
Spalls will progressively worsen in the direction of the contact, which also means that examination can sometimes pinpoint the initiating location, such as a dent.
Issue with the subject mishap pinion is that the raceway spall produced high stress concentration enough to produce a secondary fatigue made, which became catastrophic. Probably was not ever predicted.
If someone were so inclined, they could compare the rim thickness of the two examples, as a ratio of tooth size. Suspect the Bell version is inherently thicker, thus more stiff, thus less relative bending stress. Thus, less likely to experience a secondary rim fracture fatigue. Now of course that is 20,000 foot view, without any real calculation of stress due to load. There should be a stress level that the 225 pinion should be able to tolerate, even with surface spalls present. What makes this questionable now is that the local stress is probably higher than predicted due to unknown reasons, and it may be that by the time the questions are answered, it may point out that the gearbox design is over-rated for power capability.
Spalling is a quite normal "wearout" mode in gear or bearing designs. It is actually a form of metal fatigue itself at just below the surface of contact due to contact stress. It can initiate at asperities, or even surface damage from dents, handling, or debris going through the system. It progresses from smaller pits into larger flakes. Often a small pit, or beginning spall, will not produce a significant chip enough to be indicated, although they will usually stick to the magnetic detector commonly found on most aerospace transmissions. The progressive nature of the spall produces more chips, which will accumulate, and bridge the gap on the detector. My 50 cents guess was that the one in the photo was detected in such manner.
Spalls will progressively worsen in the direction of the contact, which also means that examination can sometimes pinpoint the initiating location, such as a dent.
Issue with the subject mishap pinion is that the raceway spall produced high stress concentration enough to produce a secondary fatigue made, which became catastrophic. Probably was not ever predicted.
If someone were so inclined, they could compare the rim thickness of the two examples, as a ratio of tooth size. Suspect the Bell version is inherently thicker, thus more stiff, thus less relative bending stress. Thus, less likely to experience a secondary rim fracture fatigue. Now of course that is 20,000 foot view, without any real calculation of stress due to load. There should be a stress level that the 225 pinion should be able to tolerate, even with surface spalls present. What makes this questionable now is that the local stress is probably higher than predicted due to unknown reasons, and it may be that by the time the questions are answered, it may point out that the gearbox design is over-rated for power capability.
Join Date: Sep 2007
Location: USA
Posts: 601
Likes: 0
Received 0 Likes
on
0 Posts
Would that +/-25% be for a brand new matched set or a mixture (after an overhaul) of new and ‘on-condition’ gears that may have used up some of their wear and fatigue lives?
In any single revolution, ovality of the sun gear or of the ring gear may affect the sharing of load over the set of planet gears. Could there be a numerical phenomenon whereby the number of gear teeth, the number of planets and the initial assembly position of the gears with respect to their individual maximum ovality axes could combine at the same position and repeat with more frequency (and reduced fatigue life) than say a similar module with 9 planets and corresponding differences in teeth numbers? It may sound a bit far fetched but it might explain how a gearbox could pass all the inspection checks at major overhaul yet fail relatively soon afterwards if that one gear repeatedly takes a higher share of loading than the others.
It shouldn’t be as difficult as cracking the Enigma code machine (another cluster of cogs) to work out if this can happen.
I still think the primary ‘cause’ will be found to be a manufacturing process producing an irregularity in the polycrystalline material but the design margins that should prevent this leading to failure do not appear to have been wide enough to compensate for manufacturing or operating margins.
In any single revolution, ovality of the sun gear or of the ring gear may affect the sharing of load over the set of planet gears. Could there be a numerical phenomenon whereby the number of gear teeth, the number of planets and the initial assembly position of the gears with respect to their individual maximum ovality axes could combine at the same position and repeat with more frequency (and reduced fatigue life) than say a similar module with 9 planets and corresponding differences in teeth numbers? It may sound a bit far fetched but it might explain how a gearbox could pass all the inspection checks at major overhaul yet fail relatively soon afterwards if that one gear repeatedly takes a higher share of loading than the others.
It shouldn’t be as difficult as cracking the Enigma code machine (another cluster of cogs) to work out if this can happen.
I still think the primary ‘cause’ will be found to be a manufacturing process producing an irregularity in the polycrystalline material but the design margins that should prevent this leading to failure do not appear to have been wide enough to compensate for manufacturing or operating margins.
Here is a good article on the issue of planet gear load sharing analysis if you're interested in reading more about it.
Join Date: Jun 2016
Location: Aberdeen
Posts: 90
Likes: 0
Received 0 Likes
on
0 Posts
AGMA 6123-B06 table 8 provides planet gear mesh design load factor recommendations based on numbers of planets, gear accuracy grade, application, mounting arrangement/flexibility, pitch line velocity, etc. The EC225 output stage would seem to fall under "application level 3" which recommends using a mesh load design factor of 1.30 for the "heaviest loaded planet". These are load factors used for design/analysis of the gear system, are based on accumulated industry experience, and tend to be conservative. However, obtaining even load sharing when you have a large number of planets (ie >4) and a very lightweight, flexible supporting structure, can be extremely difficult in practice.
Here is a good article on the issue of planet gear load sharing analysis if you're interested in reading more about it.
Here is a good article on the issue of planet gear load sharing analysis if you're interested in reading more about it.
Just compare it against the industrial planet carrier on the cover photo of the article (pdf download version).
One significant drawback of making the pins and carrier 'flexible' is that it puts a greater share of planet load on the upper bearing race. That effect on the raceways could easily outweigh any benefit to the gears of having carrier/pin flexibility especially as the bearing load distribution becomes more unequal as torque increases.
Doubtless this will all have been factored into the original fatigue design(?). It should be evident in higher occurrence of spalling in the upper races than lower ones.
Last edited by Concentric; 16th Aug 2016 at 03:19. Reason: Reference pdf version.
Join Date: Jun 2016
Location: Stuttgart
Posts: 40
Likes: 0
Received 0 Likes
on
0 Posts
Is there any other reason the upper stage is significantly
wider than the lower
There must as well be a reason related to fatigue
that they don´t stiffen the conical housing any more.
but anyways, if a stiff conical housing would prevent MRA detatch from AC
but in its lifetime using 2-3 gearboxes more, it would be worth it.
wider than the lower
There must as well be a reason related to fatigue
that they don´t stiffen the conical housing any more.
but anyways, if a stiff conical housing would prevent MRA detatch from AC
but in its lifetime using 2-3 gearboxes more, it would be worth it.
TS,
Power is Torque times Speed. At high speeds, in stages near the engine, lower torque is experienced so single stage reductions and smaller gear sets are adequate. As you get nearer rotor speed, torques become very large, epicyclics make more sense as way of spreading load across many components, and facewidth typically increases as torque increases.
Power is Torque times Speed. At high speeds, in stages near the engine, lower torque is experienced so single stage reductions and smaller gear sets are adequate. As you get nearer rotor speed, torques become very large, epicyclics make more sense as way of spreading load across many components, and facewidth typically increases as torque increases.
Join Date: Sep 2007
Location: USA
Posts: 601
Likes: 0
Received 0 Likes
on
0 Posts
Interesting article, thank you. The 225 planet carrier certainly looks like it has low enough stiffness to share the load well over 8 planets.
Just compare it against the industrial planet carrier on the cover photo of the article (pdf download version).
One significant drawback of making the pins and carrier 'flexible' is that it puts a greater share of planet load on the upper bearing race. That effect on the raceways could easily outweigh any benefit to the gears of having carrier/pin flexibility especially as the bearing load distribution becomes more unequal as torque increases.
Doubtless this will all have been factored into the original fatigue design(?). It should be evident in higher occurrence of spalling in the upper races than lower ones.
Just compare it against the industrial planet carrier on the cover photo of the article (pdf download version).
One significant drawback of making the pins and carrier 'flexible' is that it puts a greater share of planet load on the upper bearing race. That effect on the raceways could easily outweigh any benefit to the gears of having carrier/pin flexibility especially as the bearing load distribution becomes more unequal as torque increases.
Doubtless this will all have been factored into the original fatigue design(?). It should be evident in higher occurrence of spalling in the upper races than lower ones.
Load sharing is always a problem with any type of gear drive where torque is split into several parallel paths. The most recent example being problems with the torque-split arrangement used on the new CH-53K main rotor gearbox. The CH-53K MRGB uses a set of torsionally flexible quill shafts with spline teeth at each end that provide a very precise vernier-type angular index adjustment at assembly. In theory, this adjustment feature should compensate for any manufacturing tolerance errors and produce the required load distribution between the parallel gear paths. In practice, it proved more difficult than Sikorsky's gearbox designers anticipated.
Load distribution between epicyclic planet gears has also been a chronic problem with large wind turbine transmissions. Here is an interesting tech paper from Timken that describes their flexible planet pin concept.
Join Date: Jun 2016
Location: Stuttgart
Posts: 40
Likes: 0
Received 0 Likes
on
0 Posts
TS,
Power is Torque times Speed. At high speeds, in stages near the engine, lower torque is experienced so single stage reductions and smaller gear sets are adequate. As you get nearer rotor speed, torques become very large, epicyclics make more sense as way of spreading load across many components, and facewidth typically increases as torque increases.
Power is Torque times Speed. At high speeds, in stages near the engine, lower torque is experienced so single stage reductions and smaller gear sets are adequate. As you get nearer rotor speed, torques become very large, epicyclics make more sense as way of spreading load across many components, and facewidth typically increases as torque increases.
and onto the planetary gear. So to even transfer the load on to the gears there is a flex in the system.
I am sure that the higher loading distributed through the gears are accounted for in the design, and thus the gear tooth width.
The gears does not only handle the input torque as you suggest.
My point is that this allowable flex in the system. While probably increasing the individual gear life, is a great risk that the rotor will detach from AC if the gearbox breakdown.
the explanation confirms my previous point why not having 10 suspension bars and ultimately prevent the rotor share-off if the gearbox breaks down.
It could as well be a coincidence and the reason for only having
3 susp bars is only to save money and they thought it would be enough.
Join Date: Jun 2016
Location: Aberdeen
Posts: 90
Likes: 0
Received 0 Likes
on
0 Posts
The 1st stage carrier does look to be floating. As jimf671 points out this carries less torque. As far as I am aware, no 1st stage planet has catastrophically failed in flight although AH have found cracks in at least one removed from service according to the G-REDL report.
Also the suspension bars locating the upper mast bearing are unlikely to have contributed in any way to the fatigue crack in the planet gear that propagated to failure. Three bars is determinate and an ideal way of locating something with predictable distribution of loads. On this aircraft they failed as a consequence of gearbox breakup, they were not a cause of it. Once the lower location is lost, no number of bars at the upper bearing will prevent the mast pivoting. The bars can take axial loads only (i.e. in their own axis), normally tensile.
Last edited by Concentric; 17th Aug 2016 at 11:50. Reason: better image attached
In a drawing provided on the thread about the 2012 UK ditchings, one can clearly see a substantial roller bearing mounted in the lower web of the conical housing that supports side forces from the second stage planet carrier and the splined mast inside it.
225-coupeBTP.jpg Photo by FBav | Photobucket
225-coupeBTP.jpg Photo by FBav | Photobucket
Join Date: Jun 2016
Location: Aberdeen
Posts: 90
Likes: 0
Received 0 Likes
on
0 Posts
This may be a little off topic but since the 2012 UK ditchings were mentioned and we have a couple of good gearbox cutaway drawings at hand, it may be of interest to note the bearing arrangement on the bevel gear shaft. One at the bottom and two (one thrust; one radial roller?) above the bevel gear. What would have been the load distribution after the shaft broke half-way up?
I think it may have been fortuitous that the EMLUB warning system gave a false warning...
Does this perhaps explain AH's immediate interest in the bevel gear shaft in the Bergen crash?
Apologies for going off topic and if this was covered in a 2012 thread, but it is a quiet day. So far.
I think it may have been fortuitous that the EMLUB warning system gave a false warning...
Does this perhaps explain AH's immediate interest in the bevel gear shaft in the Bergen crash?
Apologies for going off topic and if this was covered in a 2012 thread, but it is a quiet day. So far.