Blackbird From , joined Dec 1969, posts, RR: Posted (5 years 4 months 2 weeks 3 days 20 hours ago) and read 3295 times:
I know the primary source of metal-fatigue with the comet was the fact that the window-mounts were riveted, punched, not drilled instead of glued with adhesive as planned. But I've heard other things about the skin still being too thin and even then the plane would have failed long before it's predicted service life...
That part I have trouble understanding -- they subjected the airframe to comprehensive tests regarding repeated pressurization and de-pressurization and it passed. Did they inaccurately gauge the amount of loads the plane would encounter? Did they miss some kind of variable that incorporated into the service life?
I've heard lots of comments about the shape of the windows themselves being a problem having very squarish edges. I've looked at them up close... they are curved around the edges. So that ain't it... Maybe they're a bit too big for the skin thickness they used...
TristarSteve From Sweden, joined Nov 2005, 3694 posts, RR: 34 Reply 1, posted (5 years 4 months 2 weeks 3 days 20 hours ago) and read 3288 times:
Remember that pressurised fuselages was a new art in those days. The famous Farnborough water tank which proved the Comet 1 pressurisation failures was not built until after the aircraft was grounded.
2H4 From United States of America, joined Oct 2004, 8950 posts, RR: 62 Reply 2, posted (5 years 4 months 2 weeks 3 days 19 hours ago) and read 3284 times:
AIRLINERS.NET CREW HEAD DATABASE EDITOR
Quoting Blackbird (Thread starter): I've heard lots of comments about the shape of the windows themselves being a problem having very squarish edges. I've looked at them up close... they are curved around the edges. So that ain't it.
How could a visual inspection possibly conclude that the corners are 'too squarish' or 'not squarish enough' with regard to structural loads and fatigue?
LongHauler From Canada, joined Mar 2004, 4281 posts, RR: 36 Reply 3, posted (5 years 4 months 2 weeks 3 days 19 hours ago) and read 3275 times:
When the Comet 1 was designed, it exceeded design pressure requirements almost twofold. A very new science in the day, no one had ever envisioned metal fatigue.
Never gonna grow up, never gonna slow down .... Barefoot Blue Jean Night
Blackbird From , joined Dec 1969, posts, RR: Reply 4, posted (5 years 4 months 2 weeks 3 days 18 hours ago) and read 3245 times:
But if they did an elaborate repeated load testing... how did the fuselage NOT break up during the testing (which was also done before the water-tank test when the comet was being built) and yet it broke up Jan 10, and Apr 8, 1954?
Jetlagged From United Kingdom, joined Jan 2005, 2452 posts, RR: 17 Reply 5, posted (5 years 4 months 2 weeks 3 days 18 hours ago) and read 3236 times:
Quoting LongHauler (Reply 3): When the Comet 1 was designed, it exceeded design pressure requirements almost twofold. A very new science in the day, no one had ever envisioned metal fatigue.
Not true, de Havilland were aware of metal fatigue. The design was tested using the techniques available at the time and fatigue was accounted for by testing a section of cabin structure. What they didn't allow for was the additional cyclical stress at the wing attachments and the effect that had on fuselage fatigue life. No one fully understood this at the time. Unfortunately for de Havilland, they were the pioneers of high altitude, pressurised jet transports.
The Farnborough water tank allowed them to test an entire airframe, including loading and unloading the wings. Using water meant testing to destruction would be easier and safer, avoiding the possibility of explosive decompression. In this way the source of the fatigue cracks was confirmed.
Quoting Blackbird (Thread starter): I've heard lots of comments about the shape of the windows themselves being a problem having very squarish edges. I've looked at them up close... they are curved around the edges. So that ain't it... Maybe they're a bit too big for the skin thickness they used...
Stress raising factors such as small radius cutouts, rivet holes which are not finished properly, etc. were not so well understood then either. So local fatigue life was drastically reduced. Non-rounded corners are worst, but small radii increase stress locally too. The larger the radius the less the increase in local stresses.
The glass isn't half empty, or half full, it's twice as big as it needs to be.
Blackbird From , joined Dec 1969, posts, RR: Reply 6, posted (5 years 4 months 2 weeks 3 days 18 hours ago) and read 3235 times:
So they only tested a small section of the cabin? And they didn't factor in both the repeated pressurization AND aerodynamic flexing at the exact same time?
Tdscanuck From Canada, joined Jan 2006, 12709 posts, RR: 80 Reply 7, posted (5 years 4 months 2 weeks 3 days 15 hours ago) and read 3204 times:
Quoting Blackbird (Thread starter): I know the primary source of metal-fatigue with the comet was the fact that the window-mounts were riveted, punched, not drilled instead of glued with adhesive as planned.
Stress concentration due to the tight corner was a problem too...that would have happened regardless of the fastening method.
Quoting Blackbird (Thread starter): That part I have trouble understanding -- they subjected the airframe to comprehensive tests regarding repeated pressurization and de-pressurization and it passed. Did they inaccurately gauge the amount of loads the plane would encounter? Did they miss some kind of variable that incorporated into the service life?
The major thing that threw the testing off was that they did some very high load tests *before* they did the fatigue tests. Although counterintuitive, a significant overload will actually significantly increase fatigue life. This is (now) called autofrettaging and is a relatively common procedure for items where fatigue can be an issue. At the time, the mechanism wasn't well understood so nobody realized that the sequence of tests would make a big difference. Had they done the fatigue test first, the may have realized they had a problem (although likely not realized the full extent).
Quoting Blackbird (Thread starter):
I've heard lots of comments about the shape of the windows themselves being a problem having very squarish edges. I've looked at them up close... they are curved around the edges.
They are curved, but they're still tight corners. Too tight, given what we know today.
Tdscanuck From Canada, joined Jan 2006, 12709 posts, RR: 80 Reply 9, posted (5 years 4 months 2 weeks 3 days 14 hours ago) and read 3186 times:
Quoting Blackbird (Reply 8): So the windows weren't round enough for the given skin thickness?
That's what I've heard. Just from looking at them, they look too tight to me, but I don't know what the actual loading was so that's just opinion. Even a perfectly round window picks up a stress concentration of 3. Anything tighter than that is worse.
Quoting Blackbird (Reply 8): How would the high load tests increase fatigue life? I thought fatigue was caused by repeatedly loading metal one way then the other?
Fatigue is caused by cyclic loading. You don't actually have to fully reverse the load, just change magnitude over and over. Once you start fatigue cracks, the crack grows with each cycle. A very large single load (an overload) will blunt the crack tip and induce a compressive stress field around the crack, significantly slowing the growth of the crack on subsequent cycles.
It's qualitatively similar to the way shot peening and autofrettaging increase fatigue resistance, although the grain-level physics are different.
Blackbird From , joined Dec 1969, posts, RR: Reply 10, posted (5 years 4 months 2 weeks 3 days 11 hours ago) and read 3162 times:
So the aerodynamic testing caused cracking in its own right? I thought the aerodynamic flexing and loading passed with flying colors? I thought it was the combination of the flexing AND the pressurization that caused the cracks to form?
Tdscanuck From Canada, joined Jan 2006, 12709 posts, RR: 80 Reply 11, posted (5 years 4 months 2 weeks 2 days 19 hours ago) and read 3098 times:
Quoting Blackbird (Reply 10): So the aerodynamic testing caused cracking in its own right? I thought the aerodynamic flexing and loading passed with flying colors?
Unless you're operating below the fatigue limit (which is rare for primary aircraft structure) cycling loading *always* causes cracking. It's just a question of how fast the crack forms and how fast it grows.
Tdscanuck From Canada, joined Jan 2006, 12709 posts, RR: 80 Reply 13, posted (5 years 4 months 2 weeks 2 days 17 hours ago) and read 3079 times:
Quoting Blackbird (Reply 12): So the cracks just form a LOT slower on planes like a DC-8, B-707, DC-9, CV-880, and B-727 because of thicker skin than say the comet?
Crack formation is probably about the same, but growth should be a lot slower.
Formation of the crack is primarily a surface phenomenon while growth is primarily a bulk material/stress phenomenon. I'm not sure that the planes you mention have thicker skins, they might just have better fatigue details, but thicker skins would help (for fatigue...they hurt weight).
Design principles around fatigue have changed quite a lot since the Comet. Very early designs were safe-life...you assumed how many cycles that the part would last before fatigue failure and then replaced the part before that happened. Then we went to fail-safe, where you design multiple redundant load paths so that you can fail a part and still carry load, then you inspect for and replace failed parts. Then we went to damage-tolerant (where we are today) where you assume you have a fatigue crack and that it grows and you inspect often enough to detect the crack before it grows large enough to be dangerous.
Tdscanuck From Canada, joined Jan 2006, 12709 posts, RR: 80 Reply 15, posted (5 years 4 months 2 weeks 1 day 11 hours ago) and read 2995 times:
Quoting Blackbird (Reply 14): So cracks form during the same time during development, and just take longer to spread to failure?
For equal manufacturing techniques and the same alloys, crack initiation should be pretty uniform. Crack initiation is mostly a surface phenomenon, which is why things like shot peening and mirror finish improve fatigue resistance. Once you actually have a crack, growth is controlled primarily by the alloy and the stress at the crack tip.
Sponsor Message: 
From , joined Dec 1969, posts, 
From Sweden, joined Nov 2005, 3694 posts, 
From Canada, joined Mar 2004, 4281 posts, 
From United Kingdom, joined Jan 2005, 2452 posts, 
Sponsor Message: 
Printer friendly format
