EC225 crash near Bergen, Norway April 2016
Join Date: Sep 2007
Location: USA
Posts: 601
Likes: 0
Received 0 Likes
on
0 Posts
Very detailed report. Will read all of it when time permits. After a quick scan, one thing that caught my eye was table 5 listing the difference in calculated inner/outer race surface contact stress levels between the FAG and SNR designs. Would expect to see lower design contact stress levels on the inner race than on the outer race, since the loaded sector of the fixed (with respect to load) inner race surface sees more load cycles.
It was also interesting to read the problem began with surface spalling on the outer race. Surface spalling of hertzian contacts usually originates from a subsurface shear failure. This type of subsurface shear failure tends to propagate at a 45deg angle out to the nearest surface from continued rolling contact cycles. Normally, the location of max subsurface shear stress in a hertzian contact, like this planet gear outer bearing race surface, should be well within the carburized case depth.
It was also interesting to read the problem began with surface spalling on the outer race. Surface spalling of hertzian contacts usually originates from a subsurface shear failure. This type of subsurface shear failure tends to propagate at a 45deg angle out to the nearest surface from continued rolling contact cycles. Normally, the location of max subsurface shear stress in a hertzian contact, like this planet gear outer bearing race surface, should be well within the carburized case depth.
Join Date: Sep 2007
Location: Birmingham
Posts: 244
Likes: 0
Received 0 Likes
on
0 Posts
I understand the sentiment, but what if the next rotor to detach is the other brand of gearbox with a TTIS of 35hrs?
They have taken steps they think might mitigate the likelihood of a reoccurrence, but if they don't know why it happened, they can't know if their recommendations will have any mitigating effect whatsoever.
They have taken steps they think might mitigate the likelihood of a reoccurrence, but if they don't know why it happened, they can't know if their recommendations will have any mitigating effect whatsoever.
Also agree with helicrazi when he says (a) the industry has moved on and (b) super mediums are probably the future (oil and gas fields are more automated and require fewer people). Maybe the larger cabs of the future will be mostly military.
As to the H225 Gouli also has a point - more metallurgy needs to be done but it does seem that it may simply be that the design of the gearbox has moved too close to the limits of the materials.. I'm not competent other than to ask the question but it is clear from this excellent report that those who are don't believe they have the answer either
Last edited by birmingham; 29th Apr 2017 at 13:39.
Join Date: May 2017
Location: US
Posts: 1
Likes: 0
Received 0 Likes
on
0 Posts
Failure is rare. This implies an unusual tolerance stack occurring. Normally the stacks work. On a very rare occasion, an extreme stack is occurring. This is statistical.
This leads to the conclusion that the failure is either due to a) a load sharing problem among the planet gears or b) a vibration problem. Solutions are to improve load sharing if necessary; and reduce vibration by adding damping coatings/rings to the gears. An all new helical planetary (assuming it is spur) would be beneficial. Use damping as an interim solution, and the helical planetary as a long term solution.
This leads to the conclusion that the failure is either due to a) a load sharing problem among the planet gears or b) a vibration problem. Solutions are to improve load sharing if necessary; and reduce vibration by adding damping coatings/rings to the gears. An all new helical planetary (assuming it is spur) would be beneficial. Use damping as an interim solution, and the helical planetary as a long term solution.
Join Date: Dec 2013
Location: Norfolk
Age: 67
Posts: 1
Likes: 0
Received 0 Likes
on
0 Posts
Current best engineering practice is to design gear mechanisms that mesh in such a way that all parts of the system continually present different wear faces to other components. This ensures that all components wear evenly. Statistically with such a system there will come a time when all the weak spots in the components of such a gearbox will momentarily come into alignment. Similar to all the planets in the solar system lining up on one side of the sun. It happens very rarely, but when it does, that is the time that cracks might be initiated in highly stressed components.
Join Date: May 2016
Location: Western Europe
Posts: 15
Likes: 0
Received 0 Likes
on
0 Posts
Vibration seems not to be the cause, according to the HUMS data, no abnormal vibration has been detected until 2 seconds before separation.
Fatigue prediction and lifetime calculation are statistics. Taking into account scattering in materials resistance, industry take margins that keep the functionning points far from the real limit.
This kind of tragedy shouldn't happen but it's a nice dream to believe that it's all about design matters. It's a shame that the fractured gear of REDL wasn't recovered (where the crack started)...
Concerning the future of heavies, it still exists and what we know today of O&G will change.
Fatigue prediction and lifetime calculation are statistics. Taking into account scattering in materials resistance, industry take margins that keep the functionning points far from the real limit.
This kind of tragedy shouldn't happen but it's a nice dream to believe that it's all about design matters. It's a shame that the fractured gear of REDL wasn't recovered (where the crack started)...
Concerning the future of heavies, it still exists and what we know today of O&G will change.
1.16.8.3 Oil cooler, has this statement:
Surely that's 2.34 mm2 ?
....particles of 16NCD13 were found, notably one particle with a surface area of 1.8 mm2 (length 1.8 mm, width 1.3 mm)
On second thoughts, if it's other than a rectangle it could be correct. Describing a width and length kind of infers a rectangle but a trapezoid could be smaller. It caught my eye because the 330J accident was largely due to misinterpreting the defined maximum surface area of debris. The AMM said 7 square mm and it was read as 7 mm squared - 49 square mm!
Join Date: Sep 2007
Location: USA
Posts: 601
Likes: 0
Received 0 Likes
on
0 Posts
That statement about finding such a relatively large piece of gear steel debris inside the heat exchanger core seemed odd to me at first, since it would have had to pass thru the oil pump inlet screen to end up there. However, after looking at this image of the oil pump inlet screens it does seem possible that such a relatively large piece of debris could pass thru due to the coarse mesh size.
One thing that AH might consider is using a finer mesh size for the pump inlet screens. To me, the existing inlet screen mesh seems too coarse. Obviously using a finer mesh size would require increasing the screen area to maintain acceptable flow characteristics. Logically, the size of the screen mesh should be capable of stopping any debris particles that could not pass through the pump without causing unacceptable damage to the steel pump elements. I don't believe it could be demonstrated by analysis that this would be the case with the relatively large piece of gear steel debris found in the heat exchanger.
Would be interesting to see the condition of the oil pump elements, assuming they were recovered.
One thing that AH might consider is using a finer mesh size for the pump inlet screens. To me, the existing inlet screen mesh seems too coarse. Obviously using a finer mesh size would require increasing the screen area to maintain acceptable flow characteristics. Logically, the size of the screen mesh should be capable of stopping any debris particles that could not pass through the pump without causing unacceptable damage to the steel pump elements. I don't believe it could be demonstrated by analysis that this would be the case with the relatively large piece of gear steel debris found in the heat exchanger.
Would be interesting to see the condition of the oil pump elements, assuming they were recovered.
Join Date: Mar 2009
Location: offshore
Age: 64
Posts: 24
Likes: 0
Received 0 Likes
on
0 Posts
I think you are missing the point completely Riffraff. There shouldn't be debris of this magnitude anyway to warrant any change of filter design. The root cause needs addressing, not the consequences
Join Date: Sep 2007
Location: Birmingham
Posts: 244
Likes: 0
Received 0 Likes
on
0 Posts
Comment in this week's Flight Global ...
"But it seems clear that opportunities were missed. The frankly startling revelation that parts sourced from one of two suppliers have a failure rate three times as high as the alternative is troubling; that no one thought to analyse this data before last year’s Norway crash is almost beyond belief".
Yes, the H225 complies with all the certification requirements, but that begs the difficult question as to whether those standards are still fit for purpose.
"But it seems clear that opportunities were missed. The frankly startling revelation that parts sourced from one of two suppliers have a failure rate three times as high as the alternative is troubling; that no one thought to analyse this data before last year’s Norway crash is almost beyond belief".
Yes, the H225 complies with all the certification requirements, but that begs the difficult question as to whether those standards are still fit for purpose.
Yes, the H225 complies with all the certification requirements, but that begs the difficult question as to whether those standards are still fit for purpose
@Pitts: if all you are interested in is providing a wind-up it might be best to avoid hitting the "post" button. (I have more frequently of late overcome the urge to post, which I find is a healthy thing).
It is troubling to me, the idea that of two suppliers for a precision component in a drive train/transmission, one had 3x failure rate (or significantly less reliability) and yet was still retained as a supplier/vendor by the manufacturer. While there may be more to that story, on the face of it someone didn't take reliability as seriously as they ought to have done, or, a batch/lot of material of considerable size was produced and distributed and it took a long time for people to discover and try to remedy that.
It is troubling to me, the idea that of two suppliers for a precision component in a drive train/transmission, one had 3x failure rate (or significantly less reliability) and yet was still retained as a supplier/vendor by the manufacturer. While there may be more to that story, on the face of it someone didn't take reliability as seriously as they ought to have done, or, a batch/lot of material of considerable size was produced and distributed and it took a long time for people to discover and try to remedy that.
Not a wind up. The issue here is fuelled by people, some of whom would be a useful voice within the industry who in the past have shown what one might call ambivalence. As your post goes on to say it is troubling and yet that process can not be a sudden revelation to those embedded in the various organisations. Your attention should be less upon my voice and more upon those whose concern in the recent past extended to selecting a large size of under crackers.
Join Date: Jun 2016
Location: Aberdeen
Posts: 90
Likes: 0
Received 0 Likes
on
0 Posts
It is troubling to me, the idea that of two suppliers for a precision component in a drive train/transmission, one had 3x failure rate (or significantly less reliability) and yet was still retained as a supplier/vendor by the manufacturer. While there may be more to that story, on the face of it someone didn't take reliability as seriously as they ought to have done, or, a batch/lot of material of considerable size was produced and distributed and it took a long time for people to discover and try to remedy that.
The two catastrophic failures were by fatigue crack propagation beyond the hardened depth. Not all spalling results in deep fatigue cracking, as evidenced by the 20 FAG bearings that have spalled ‘harmlessly’. Remember that 7 SNR bearings have spalled ‘harmlessly’ too. Given the total production numbers of bearings, most of both types would actually appear to have either been still in satisfactory service or have reached design SLL when these statistics were gathered. In that sense the behaviour of all except the REDL and LN-OJF bearings has been ‘reliable’ in that they followed the expected end of life degradation process.
You can look at those same statistics presented in Table 6 a very different way. Look at the total number of units of each type and how many of them DID NOT produce spalling. That would give “reliability” scores of 99.35% and 99.76% for FAG and SNR respectively. Not so clear cut now is it? Remember also the definition of L10 life for bearings (see 1.6.8.2).
You cannot equate that minor difference in reliability simplistically with Hertzian contact stresses of 1800MPa and 1550MPa (86% of 1800MPa) respectively. There is far more to it than that. A lower Hertzian contact pressure (Edit: by widening the path) in a spherical raceway can involve more slippage and, depending on lubrication, generate more surface friction that is more likely to turn a crack inwards.
Further investigation should not only focus on metallurgical differences but the whole tribology of these bearings and their operational conditions.
Last edited by Concentric; 16th May 2017 at 11:13.
Join Date: Jun 2016
Location: Aberdeen
Posts: 90
Likes: 0
Received 0 Likes
on
0 Posts
Possibly of greater interest than comparing FAG bearing spalling with SNR bearing spalling statistics over the same period, is to compare the information in 1.6.11.4 with that in 1.6.11.5. Excluding the crash items, between 2001-2016, 27 bearings have been removed in service due to spalling or micro-pitting but (according to Airbus) only 2 of those were between 2009 and 2016.
FAG stated (in 1.6.8.3) that there have been no differences to the design or manufacture during the production life of their bearings, yet for the first 8 years 25 bearings (incl. both types) were removed (average 3.125p.a.) and for the next 7 years only 2 were removed (0.286 p.a.), despite any heightened concerns or vigilance there may have been following the REDL crash in 2009. I can’t quite figure that reduction of 91% out.
What, if anything, has changed materially in that time and in particular perhaps in, or shortly before, 2009? Was there any change to lubricants, additives or to internal spray nozzle arrangements?
FAG stated (in 1.6.8.3) that there have been no differences to the design or manufacture during the production life of their bearings, yet for the first 8 years 25 bearings (incl. both types) were removed (average 3.125p.a.) and for the next 7 years only 2 were removed (0.286 p.a.), despite any heightened concerns or vigilance there may have been following the REDL crash in 2009. I can’t quite figure that reduction of 91% out.
What, if anything, has changed materially in that time and in particular perhaps in, or shortly before, 2009? Was there any change to lubricants, additives or to internal spray nozzle arrangements?
Join Date: Sep 2007
Location: USA
Posts: 601
Likes: 0
Received 0 Likes
on
0 Posts
One photo in the report shows an obvious difference in roller face width between the FAG and SNR bearings. But the difference in face width alone may not be responsible for the different hertzian contact stress levels noted. The roller/race osculation ratios can also have a significant effect.
The M50 material used for aircraft bearing rollers/races typically allows a higher material reliability adjustment factor to be used in a conventional rolling element bearing fatigue life analysis than a race using carburized VIM-VAR 9310 gear steel would.
The M50 material used for aircraft bearing rollers/races typically allows a higher material reliability adjustment factor to be used in a conventional rolling element bearing fatigue life analysis than a race using carburized VIM-VAR 9310 gear steel would.
Join Date: Jun 2016
Location: Aberdeen
Posts: 90
Likes: 0
Received 0 Likes
on
0 Posts
I agree, the difference in face width is not the only factor. But even the difference in Hertzian contact stress I believe may not be the only driving force behind this failure mechanism. Some bearings in wind turbine applications run up to 3 GPa. Normally fatigue damage is actually stress range raised to the power of 3, or higher depending on the S-N curve assumed for the damage calculation, so such a difference in Hertzian stress would produce even more contrasting results than we are seeing here. That approach would seem to apply more to dry rolling contact.
As I understand it the prevention or delay of damage in rolling/sliding contact is highly dependent on the ability to maintain a thin film of lubricant between the contacting surfaces, most significantly for ball or barrel roller bearings with their element of sliding within the Hertzian contact area. Without changing the bearing design or roller/race osculation ratios, the best way for a manufacturer to improve this would be a change of lubricant or introducing certain additives (some of which can be aggressive). That could significantly prolong surface contact fatigue life and delay the onset of spalling; producing an improvement for both FAG and SNR bearings similar to what we see in the reported statistics post 2009.
I am no expert but the profile of the crack running deeper into the material ahead of the travel of the rollers looks uncannily similar to that caused by lubricating fluid seepage into a crack with pressurization and possible entrapment effects. There are several recent research papers available here and here.
To put it in layman’s terms, this mechanism is similar to hydraulic fracturing (fracking) of highly compressed rock formations in shale gas recovery. Could it also overcome the residual compressive stress in the carburised raceway, even the higher compression on the FAG bearing?
So my big question is – did AH achieve that dramatic improvement in spalling statistics post-2009 by changing the lubricant to one better able to maintain that thin film but potentially also better able to seep into micro-defects and be pressurized behind the crack tip?
By reducing the degradation process of spalling which, at the time of certification was considered to be benign, could AH have removed a crucial indicator and (unwittingly) simultaneously introduced a lethal failure mechanism that leaves no trace?
I really hope I am wrong.
As I understand it the prevention or delay of damage in rolling/sliding contact is highly dependent on the ability to maintain a thin film of lubricant between the contacting surfaces, most significantly for ball or barrel roller bearings with their element of sliding within the Hertzian contact area. Without changing the bearing design or roller/race osculation ratios, the best way for a manufacturer to improve this would be a change of lubricant or introducing certain additives (some of which can be aggressive). That could significantly prolong surface contact fatigue life and delay the onset of spalling; producing an improvement for both FAG and SNR bearings similar to what we see in the reported statistics post 2009.
I am no expert but the profile of the crack running deeper into the material ahead of the travel of the rollers looks uncannily similar to that caused by lubricating fluid seepage into a crack with pressurization and possible entrapment effects. There are several recent research papers available here and here.
To put it in layman’s terms, this mechanism is similar to hydraulic fracturing (fracking) of highly compressed rock formations in shale gas recovery. Could it also overcome the residual compressive stress in the carburised raceway, even the higher compression on the FAG bearing?
So my big question is – did AH achieve that dramatic improvement in spalling statistics post-2009 by changing the lubricant to one better able to maintain that thin film but potentially also better able to seep into micro-defects and be pressurized behind the crack tip?
By reducing the degradation process of spalling which, at the time of certification was considered to be benign, could AH have removed a crucial indicator and (unwittingly) simultaneously introduced a lethal failure mechanism that leaves no trace?
I really hope I am wrong.
Last edited by Concentric; 19th May 2017 at 08:52.
Join Date: Dec 2013
Location: Norfolk
Age: 67
Posts: 1
Likes: 0
Received 0 Likes
on
0 Posts
Concentric
This does appear to be a plausible mechanism to explain the crack propagation in the gears. It cannot be the sole reason because a crack has to be initiated by some other event or process first, but once that opening or pit is created it would seem that eventual failure is inevitable. Every compression cycle would be the equivalent of driving an incompressible hydraulic wedge deeper into the material. As the process continues the hardness of the surface layer would assist in driving the crack propagation deeper into the softer underlying material.
This does appear to be a plausible mechanism to explain the crack propagation in the gears. It cannot be the sole reason because a crack has to be initiated by some other event or process first, but once that opening or pit is created it would seem that eventual failure is inevitable. Every compression cycle would be the equivalent of driving an incompressible hydraulic wedge deeper into the material. As the process continues the hardness of the surface layer would assist in driving the crack propagation deeper into the softer underlying material.