de Havilland Comet Design changes 1

[jpaero]

I am trying to learn the specific design related issues with the de Havilland Comet. I have read the investigation report which spoke of:
1. Underestimation of stresses at the corners
2. Overloading the fatigue specimen before the fatigue test itself
3. Testing only the fwd fuse and effect of end restraints
etc

What I am interested however is the specific changes that were made to the design to make it airworthy again. Possible amends I could think of (based on my limited knowledge) were:

1. Reduction of general stress levels by making the skins thicker
2. Better materials in terms of crack resistance/growth (fracture toughness?)
3. Changing some of the cutout profiles to reduce the Kts.
4. Adding stiffners around windows to allowing stress away from the cutouts

But I was hoping for some definitive answers rather than guessing. An ideal answer would be one which compares the old and new design almost side by side. While a lot seems to written about what led to the accident, I couldn't find much by the way of specific design changes that were incorporated to make it airworthy again. Any pointers in that direction would be great! I will also welcome less than ideal answers .

 

[berkshire]

This is from memory with no document references. one increase the size of the radius at the window corners. Two change the rivet edge distance from 1.5D to 2D ,remember this happened almost 70 years ago.
B.E.

You are judged not by what you know, but by what you can do.

 

[rb1957]

maybe added dblrs around the cut-outs. I'd get hold of the inquiry that resulted from the accidents.

I don't think "Underestimation of stresses at the corners" is correct (though it may be the commonly held opinion). I think, remembering reports I've read a long time ago, they ran their fatigue test on the static test airframe ... and this was a (the) major problem with their approach (in the 50s).

I'd've thought there was something out there discussing the differences between the Comet 1 and the Comet 4. But the details you want are probably buried deep in the engineering data.

another day in paradise, or is paradise one day closer ?

 

[verymadmac]

This covers the details nicely but leaves me with one question, anyone know exactly what punch riveting is?

https://www.raes-hamburg.de/media/files/text_2019_01_24_Comet-2.pdf

 

[crackman]

Hi jpaero

If you want to see the changes, you can look at the comet 4 or nimrod structural ipc and srm and compare to the comet 1. There are also RAE papers showing the stress investigations around cutouts.
As for other significant sources, you could always contact The Hub at farnborough and get a copy of the full original accident investigation as I did many years ago. Beware, it is quite large (over 1000 pages) and expensive to obtain a copy.

By the way, little advertised fact is that the structure recovered from the accident aircraft had stopped drilled cracks in numerous areas around the adf cutouts. The stopped drill cracks where a result of the punch drilling opertion dehavilland employed on the early produced aircraft. It is discussed in the public hearings but was written off as standard practice. Note the subsequent full scale fatigue test by RAE did not replicate the failures at the ADF cutouts in the crown of the fuselage but instead showed failures at the cabin windows and the test article did not have any stopped drilled cracks. The punch drilling operation induced damage in the holed thru "ripping" of the material.

If you search online, I presented a paper at AAS 2017 which shows the stop drilled cracks and also presented some of it at ASIP 2021, you might be able to find the papers online. Also, Dr Paul Withey has presented several papers on this topic and confirmed the presence of damage in the original panel at the ADF cutouts. Not to say high stresses were not significant, which they obviously were, but if you include at least 4 or more stopped drilled cracks in the same area, the stresses would be even higher.

Anyway, the comet is a great case study and still has some valuable information for learning.

good luck in your efforts.

 

[crackman]

found a quick snapshot from the accident report.

 

[rb1957]

"punch drilling" is probably like cutting a lightening hole, particularly if you're dimpling the skin (like the flange of the lightening hole).


another day in paradise, or is paradise one day closer ?

 

[rb1957]

that nicely reinforces my idea that the general public only got the Coles Notes version with a very simple cause. the real cause is more difficult and simpler at the same time !

another day in paradise, or is paradise one day closer ?

 

[rb1957]

is page 8 really saying the skin panels were "brazed" together (using Redux) ??

another day in paradise, or is paradise one day closer ?

 

[rb1957]

page 15 is funny ! happens all the time ... objects in the pic are not anchored and move (London has moved to North America, about Edmonton.

another day in paradise, or is paradise one day closer ?

 

[rb1957]

it's sad to realise that since they did the fatigue test on a limit (or overloaded) pressure loaded specimen ... then if they had done this test on each production aircraft, everything would have probably been ok.

another day in paradise, or is paradise one day closer ?

 

[LiftDivergence]

Some interesting info in here:

https://web.archive.org/web/20091026195856/http://geocities.com/CapeCanaveral/Lab/8803/fcogalyp.htm

Especially starting at the heading "PART XI INVESTIGATION OF THE ACCIDENT TO G-ALYP AND G-ALYY"

Keep em' Flying
//Fight Corrosion!

 

[rb1957]

too bad they don't have the appendicies

another day in paradise, or is paradise one day closer ?

 

[Sparweb]

Thank you for the link, Mac.

The life of ‘YP’ can be calculated as 1272 cycles.
The actual number of pressurised flights was 1290.

And there you see why the Paris equation is in the prescribed methods today.
Not a lot of presentations take the historical evidence and carry through with it in a fully developed analysis. That presentation does it, bridges the gap between general discussions and a tech report. Could be useful...

 

[jpaero]

Just wanted to note here that there is a youtube video by Prof. Paul Withey on the accident pretty much on the lines of the presentation in verymadmac's link above.

https://youtu.be/K5HqEwbp4GA

While I find it immensely useful, it falls short of the design measures used to prevent the kind of crack propagation observed.

 

[3DDave]

Nevermind. This thread is useless. Being argumentative about answers is unacceptable.

 

[jpaero]

@3DDave:"The majority of efforts aren't to stop propagation, it's to never start one to begin with and to detect any that have started."

I thought you can almost reliably never stop a crack from initiating. All you can do is to ensure that when one starts, you have designs in place to slow them down until the inspections pick them.

So, on the Comet, I think it's the fast and unarrested propagation that was the problem rather than the apperance of the crack
itself cause, I think one way or the other, a crack would have appeared with increased usage either due to manufacturing defects or otherwise.

In summary, given the structures sensitivity to crack initiation and given that cracks would have appeared with usage, I doubt if preventing cracks from appearing during manufacturing would have gone enough in the way to make it airworthy again.

Another pointer in that direction is a 707 marketing video released during that time which demonstrated slow crack growth when fuselage was penetrated by steel harpoons and I don't know if the Comet could have demonstrated that (slow propagation) given how sensitive it's to crack initiation itself.

I am willing to be wrong on the above prognosis but I think there is substantially more to Comet redesign than stop making cracks during manufacturing.

 

[Compositepro]

rb1957, Redux is a brand name for a line of structural adhesives for aircraft.

https://en.wikipedia.org/wiki/Redux_(adhesive)

 

[rb1957]

@WindWright,
"And there you see why the Paris equation is in the prescribed methods today." ... Paris is good as a first cut but not a good final match to crack growth, if you're in the threshold region.

Personally, I look at the correlation and think ... I wonder if they tuned their analysis to suit the result they wanted ?
But then I'm a cynic ...

another day in paradise, or is paradise one day closer ?

 

[rb1957]

@compositepro,

ok, and de Havilland have good experience with bonding ... but it is a very tricky process to get right.

another day in paradise, or is paradise one day closer ?