|
adatesman
May 3, 2008, 2:55 AM
Post #26 of 38
(2306 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
curt
May 3, 2008, 4:30 AM
Post #27 of 38
(2298 views)
Shortcut
Registered: Aug 27, 2002
Posts: 18275
|
adatesman wrote: I have no idea
OK--we're in agreement on something. You are ignoring the entire basis of my earlier post. I'm a material scientist by training--and, I'm telling you that the force put on the threads by tightening the quicklinks 1/4 turn beyond finger tight can not be simply subtracted from its ultimate failure strength. That's simply not how it works--and to weaken the link, you do indeed need to overtighten it to the point where you are weakening the metal itself.
Curt
|
|
|
|
|
adatesman
May 3, 2008, 1:57 PM
Post #28 of 38
(2285 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
jt512
May 3, 2008, 5:44 PM
Post #29 of 38
(2282 views)
Shortcut
Registered: Apr 12, 2001
Posts: 21904
|
adatesman wrote: On a related note, I just heard back from Tradklime and we'll be doing the 10 remaining quicklinks with 2 completely open, 4 at +1/4 turn and 4 at +1/2 turn.
Perhaps I should have posted this earlier, but it is a good idea to base sample size decisions on a minimum difference you want to be able to detect and say is statistically significant. With samples sizes of n=4 per group, then there would have to be about 1000 lb difference in the breaking strength between two groups to be able to get a statistically significant result. If, say, there is only an 800-lb difference between groups, then, it is unlikely that you will be able to conclude that the difference is due to how the links were tightened, versus random error. Remember that the standard deviation in the original sample was 434 lb, and an 800-lb difference is less than 2 standard deviations, and so could easily be due to random variation. So, if you are interested in being able to draw conclusions from differences between groups of less than 1000 lb, then you need a larger sample size. If you forgo the test of the two open-gate links, and add these links to the other groups, to give you n=5 per group, then, the minimum difference drops from 1000 to about 800.
If you are not locked into a total of 10 quick links, and you can specify a minimum difference you want to be able to detect, I could calculate the sample size you need. Unfortunately, I'm off for the weekend to go climbing. Shoo might be able to do such a calculation as well.
Technical note: the above calculations use the standard deviation from the original tests, with 80% power.
Jay
|
|
|
|
|
curt
May 3, 2008, 5:46 PM
Post #30 of 38
(2281 views)
Shortcut
Registered: Aug 27, 2002
Posts: 18275
|
adatesman wrote: As for whether +1/4 turn is enough to make a difference, I can't say without doing some testing. I suspect that Petzl spec's their torque so low because it does make a difference at some point, so I'm curious where that point is. I also suspect that this point is sufficiently high that it is not encountered except under extreme circumstances, so perhaps you are correct that +1/4 turn is safe. We just don't know that for a fact and that practice is contrary to the manufacturer's instructions. On a related note, I just heard back from Tradklime and we'll be doing the 10 remaining quicklinks with 2 completely open, 4 at +1/4 turn and 4 at +1/2 turn. I'll probably get a torque reading on one and then set the others to that torque to get better consistency, but chances are that they'll all end up with about the same rotation. I'll measure just to be sure though, and will probably break them tomorrow. -a.
Good--I think testing is the only way to settle this. By the way, why test any of the links "completely open" when we really want a comparison between finger-tight and finger-tight plus 1/4 turn?
Curt
|
|
|
|
|
adatesman
May 3, 2008, 9:49 PM
Post #31 of 38
(2259 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
curt
May 5, 2008, 5:55 AM
Post #32 of 38
(2227 views)
Shortcut
Registered: Aug 27, 2002
Posts: 18275
|
So where's the data on the quicklink pull testing? Since you have probably done the testing by now, I would hate to believe that you are withholding any data that did not align properly with the incorrect assumptions you posted earlier.
Curt
|
|
|
|
|
adatesman
May 5, 2008, 8:35 PM
Post #33 of 38
(2175 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
adatesman
May 6, 2008, 1:53 AM
Post #35 of 38
(2129 views)
Shortcut
Registered: Jul 13, 2005
Posts: 3479
|
|
|
|
|
|
gunkiemike
May 11, 2008, 11:50 PM
Post #36 of 38
(2050 views)
Shortcut
Registered: Oct 1, 2002
Posts: 2266
|
adatesman wrote: Snipped results of 8 cheap-o links all holding 20+ kN.
The more I see here, the happier I am to rap on a SINGLE CHINESE screwlink. Finger-tight or otherwise.
(This post was edited by gunkiemike on May 12, 2008, 12:18 AM)
|
|
|
|
|
curt
May 14, 2008, 4:03 AM
Post #37 of 38
(2030 views)
Shortcut
Registered: Aug 27, 2002
Posts: 18275
|
adatesman wrote: Hey Curt, In case you don't see it, over in the other thread Bigo pointed me to an explanation why the strength goes up when the nut is tightened. Machinery's Handbook page 1400 (25th Edition). Makes sense now that I read it a couple times, but rather counter-intuitive at first. There's a copy of it here if you're interested, but watch out as its a 5mB PDF. The section is on page 26 of the pdf. -aric.
Aric,
Thanks, I'll take a look and let you know if I have any meaningful comments. I do appreciate all the hard work you have put into this.
Curt
(This post was edited by curt on May 14, 2008, 4:07 AM)
|
|
|
|
|
stymingersfink
May 14, 2008, 4:08 PM
Post #38 of 38
(2003 views)
Shortcut
Registered: Aug 12, 2003
Posts: 7250
|
Ya know, I'd be willing to bet that Petzl already has a ton of data on the mean breaking strengths of their mallion rapide, going back years.
On the flip side, somehow I doubt that Hank Moon would be able to email you any of that info, what with the proprietary nature of such data and all.
Perhaps someone from the Petzl QA lab might be willing to post up some more generalized information gleaned from their own observations on the break-o-tron. Couldn't hurt to ask, could it?
|
|
|
|
|
|
) is that 7 of the 8 quicklinks in this test all failed by the lower threads shearing off. In the earlier test (where they were all only finger tight) 5 of the 6 samples failed by the upper threads under the shoulder shearing off.
