Comet Metal Fatigue
 

A few years ago, I wrote my autobiography and posted on here about the Comet in which I flew in the early sixties. I mentioned that the early Comet had "square" style windows and it was at the corner of these that the problems of metal fatigue became evident. Here is the link to Wikepedia which will verify the existence of the square style windows which led to the castatrophic failure of the airframe on more than one Comet.https://en.wikipedia.org/wiki/De_Havilland_Comet

It is a very interesting account and shows just how much has changed in the past 50 - 60 years.

No disrespect to Wikipedia, but the fact that early Comets had square windows and that those were implicated in the disasters isn't really one that needs to be "verified" 60 years later. :O

Thank you Dave.
Of course there's a bit more to the story than that.
The OP might like to peruse the previous discussions on here re the causes.
I recommend the 'Search' function.

Worth mentioning that DC 10s also had square windows and the Sud Aviation Caravelle (which shared the Comet's nose) had triangular ones without problems. The shape was not the only issue: when Redux was used (invented by unsung hero Norman de Bruyn at Duxford) things were OK, when additional corner plates were riveted into place in Comet windows stress fatigue was initiated.

To describe the Caravelle windows as "triangular" is to miss the point, joy ride (pun unintended). They were considerably-rounded triangles.

On the early Comet series, it was not so much the square windows but a combination of a brittle alloy and slightly inadequate manufacturing practices. Many openings were not drilled or cut but rather punched, leading to microscopic cracks around the opening. Also, despite the plans calling for Redux glue in this area, rivets were installed for some reason that also had their holes punched instead of drilled.

With G-ALYP, the investigators seem to have formed the opinion that the destruction of the fuselage has consequently started not on the windows but on the mentioned ADF hatch. On the subsequent ALYY accident, it seems that insufficient wreckage has been found for a similar analysis but the board had concluded from the similarity of the surrounding circumstances that the destruction of the aircraft occurred in a similar way.

The ADF hatch was formed in the same way as the windows. All the windows and hatch openings had 90 degree corners and it was established that the cracking stemmed from the corner, not just on the wreckage of 'YP, but on the airframe involved in the pressure tests at Farnborough.

Whilst square windows and hatches on pressurised aircraft were not a problem on piston engined airliners, the pressure differential at the altitude the Comet operated at combined with the thin fuselage skin and the various methods of fixings used were a lethal combination which had neither been appreciated nor tested for in depth in the development of the type.

The Caravelle, as stated above, had triangular windows with rounded corners an th DC10's windows certainly were not square but were rectangular with rounded corners.

Not that it makes much difference to what we're talking about here, but the Caravelle's windows were described at the time as "lozenge-shaped".

Agreed, Caravelle and DC 10s windows were well rounded! Still, it seems to me that the shape was more like a contributory factor rather than the initiator of the failures.

Early welded Liberty ships often cracked and split around their cargo hatches (sometimes completely bisecting the hull!) until these had their corners radiused, and I think this example has led many people to make a simplistic comparison between Liberty Ships problems and the Comet's (pointy-cornered!) square windows! However, the cause and effect was not the same in both instances : it was the whole design and build of the Comet's windows, skin, added rivet plates etc.which introduced the problems, rather than just the shape.

There is a Finnair Caravelle standing at ARN. I took some pictures of it, but they are all very bad. Here is a good one from sudaviation. Great shot of the windows.

http://sudaviation.com/wp-content/ga...e/img_5057.jpg

Here is a picture that shows the propagation of the cracks in G-ALYPs breakup as determined by the investigation board.

https://aviation-safety.net/photos/a...0110-1-C-3.jpg

On that drawing the window corners seem to be slightly curved!

Tu. 144 wrote earlier: "Also, despite the plans calling for Redux glue in this area, rivets were installed for some reason that also had their holes punched instead of drilled."

It is my understanding that Chief Designer Bishop specified JUST Redux, but conservative management at de H could not bring themselves to fully trust this new process and material, and went for a "belt and braces' approach.

I've read all the reports that originated from the RAE about the Comet disaster and the observation that shocked me was that there were some cracks in the window (passenger and ADF) that had been stopped drilled but continued to propagate. The problem was one of material and production methods (not helped by the shape of the opening). How many metal fuselage designs had de havilland built before the dh106? They were learning the art whilst pushing the envelop.

The official Air Ministry "Behaviour of Skin Fatigue Cracks at the Corners
of Windows in a Comet I Fuselage By R. J. ATKINSON, W. J. WINKWORTH and G. M. NORRIS states on
page2 para 2.2. "Local Structure at Windows and Escape Hatches. With the exceptions of the two forward escape hatches, which interrupted a circumferential frame (Fig. 6), the windows and escape hatches were positioned between the frames.
The windows and escape hatches were rectangular, their relative sizes being:
Window: 16.6 in. wide x 14 in. high, corner radii 3 in.
Escape hatch: 19.0 in. x 21.5 in. high, corner radii 4 in. (see Fig. 5).
The apertures were reinforced by peripheral members of zed section bonded to the skin with
Redux adhesive and additionally riveted by ~ in. countersunk-head rivets at the corners.

A 3 inch radii is not 90 degree as I stated in post #7 but is close enough not to matter in this context.

On page 5 it states "5.1. Origins of the Fatigue Cracks. All the fatigue cracks originated at the counter-
sunk rivet holes in the skin at the window and escape hatch corners. Those cracks which eventually
became catastrophic started at outer-row rivet holes. The few cracks that originated at holes in the
inner row grew inwards to the edge of the aperture and did not become catastrophic"

CAP 127: Report of the Court of Inquiry into the Accidents to Comet G-ALYP on 10th January 1954 and Comet G-ALYY on 8th April 1954:

http://lessonslearned.faa.gov/Comet1/G-ALYP_Report.pdf

And a three-part 1955 article from The Engineer with abstracts from the above:

www.gracesguide.co.uk/images/a/a8/Er19550218.pdf

www.gracesguide.co.uk/images/5/56/Er19550225.pdf

www.gracesguide.co.uk/images/8/8e/Er19550304.pdf

I have always understood the Comet phenomenom to be instable crack propagation due to non-connected formers and longerons, in junction with susceptible alloys. When these formers and longerons were connected to the skin and to each other, instability of crack propagation was prevented.

Check out the book 'No Highway' by Neville Shute. Apart from being a successful author he did time as a stress engineer.

I remember the film

I had something to add the last time this came up but I can't remember which book it was in.
I thought it was in The DH Story by C. Martin Sharp but I haven't been able to find it without re-reading the entire book.

Since I can't find the actual quote I can't add the exact job descriptions or positions but from what I recall, a production manager went to the design dept. saying they were having trouble achieving the specified clamping pressure when bonding around the windows/openings.
That was the only reason for adding the rivets.

Joy ride blamed conservative management last time too, I'd like to know on what he bases that accusation. Over-ruling a Chief Designer on matters of design sounds more than unlikely to me.

Obviously I'm not an aircraft designer and hindsight would prove me wrong but 3 in corner radius sounds generous to me, certainly a long way from the widely reported "square".

Interesting history of De Havilland. The Comet 1 was not only designed and manufactured by them, but it's jet engines were their own as well. Given that jet engines were a brand-new technology, and that it was the first pressurised aircraft designed in Europe, it was an amazing coming-together of aspects for what was still essentially a family business run by aviation-lifer Sir Geoffrey De Havilland, which did everything in aviation normally done by separate substantial businesses - they even did aircraft propellers (the Vickers Vanguard had those big, thundering De Havilland props). His two sons were chief test pilots for the company, both unfortunately killed in separate accidents. De Havilland had continued developing wooden-framed aircraft well after other mainstream manufacturers, and some of their other products, had switched to all-metal.

The metal fatigue accidents were only some of the early Comet 1 losses, there were a number of others. Takeoff characteristics were iffy to say the least, and among other crashes was the loss of the first Canadian Pacific aircraft at Karachi, actually on its delivery run, which claimed the lives of the senior CP execs involved in the purchase and introduction of the aircraft, who had just handed the cheque over for it. In less than two years from introduction, 5 of BOAC's pioneer fleet of 10 had been lost in accidents, and their first jet services had already been curtailed as a result. When grounded in 1954 a very large number of follow-on aircraft orders were part-completed, and had to be abandoned. What seems extraordinary nowadays is how the company managed to survive financially after all this.