Mountain Project Logo

What's the actual strength of your carabiners?

Original Post
MacM · · Tucson/Preskitt, AZ · Joined Feb 2010 · Points: 675

I just recently completed a Solid Mechanics course, basically fundamentals of Structures. During the summer we are assigned a project, where we get to choose the topic of our choice that is related to structures. Naturally I am a climber so I figured, why not carabiners?

Original proposal for the project was to conduct testing for all three published ratings on the carabiner and find the Factor of Safety that is used in those ratings. Then, conduct torsion tests on the carabiner, along the major axis, and find what the published rating would be including the F.S. found.

Unfortunately, it's a compressed summer course so we were only allotted 1.5 weeks for testing and analysis, versus the standard 3 weeks as if it was the regular semester. Ideas needed to be cut. The torsion test unfortunately was cut first because we would have needed to design the test and then conduct the test, along with analysis, which would have taken too much time. That was the one test I was actually interested in and would have been most beneficial.

What was finalized for our experimentation was standard tensile testing along the major axis with the gate open, then determine the Factor of Safety in the published rating.

The carabiner tested was the BD Neutrino, 2012 model. If you check their website now the major axis strength rating has been increased by 1kN for the 2013 model.

All the information is below for your enjoyment, and to satisfy your curiosity.

Analysis Paper:
docs.google.com/file/d/0B9n…

Project Presentation(Summary):
docs.google.com/file/d/0B9n…

Neat Video (Visual Summary):

http://www.youtube.com/watch?v=lTcvl4n6qnM

If you have any questions PM me or post them on here and I'll reply as soon as I can.

Also, if you would actually really like to have a look at the MatLab coding I can send that your way. (Input 3 test results, get F.S. and 3 Stress Diagrams)

-Mac M

Larry S · · Easton, PA · Joined May 2010 · Points: 872

Mac - Climbing gear like this is rated to it's failure strength with a 3 sigma (99.87%) certainty - meaning that 99.87% of a particular item will fail above it's rated strength. In other words, at most 1 out of 1000 pcs will fail below the rated strength. If you were to test and plot enough of these on a histogram, you should see a nice normal curve centered well above the spec limit.

Crossing · · Breinigsville, PA · Joined Apr 2010 · Points: 1,621

Mac, to echo Larry's point you also only have a sample size of 3, this adds uncertainty of estimating the mean from your sample data and ultimately reduces the the minimum value at which you can claim your true mean to fall into. Your 90% confidence interval is 1945.6 to 2081.5 - so you can be 90% confident that the true mean is somewhere between these values based upon your sample size.

Larry S · · Easton, PA · Joined May 2010 · Points: 872

Just wanted to interject so you don't take Ryan and I the wrong way, because we probably sound pretty negative - This is good work and I'm betting you learned alot from it. I loved watching the stuff break. We're just trying to convey how their quality system really works, that there's alot more to product quality than FOS, and in this case, you should expect some tiny percentage of samples to fail at or below the spec limit.

Jim Titt · · Germany · Joined Nov 2009 · Points: 490

A karabiners most vulnerable state is not as you suppose gate open but when it is nose-hooked,they snap with little more than body weight. This is something better designed karabiners than the Neutrino avoid.

Safety Factors are not particularly a description climbing gear manufacturers prefer to use, we may prefer to overshoot the certified strength to take into account any manufacturing variations (which a sample of 3 won´t identify since you didn´t test the next billet delivered for example) OR there are strength overshoots due to the requirements to pass other parts of the tests.
Some of my products are 3 or 4 times stronger than the certification requirements due to the materials used and other customer requirements such as durability and have nothing to do with "Safety Factors".

MacM · · Tucson/Preskitt, AZ · Joined Feb 2010 · Points: 675

You guys (Larry and Ryan) are definitely correct. Unfortunately the scope of this experiment did not allow for enough carabiners to be tested in order to reduce the uncertainties. We felt that the second test was not precise due to the carabiner slippage in our set up (noted numerous times in the report). Not to mention the set up used wasn't exactly specific to our test, as seen in the video when we tested a spare carabiner with the gate closed and it was stronger than the bolts used. So to base this data on accuracy would not be the best idea.

I wish I could have destroyed enough carabiners to produce results that established even remotely some consistency and establish a histogram to obtain a 3 Sigma value.

These tests were done merely as an extreme generalization of the possible results. More of an "idealization" than anything.

Thanks for your input!

EDIT: Larry, that was the main concern within our group; the inability to find results that are really worth anything due to the lack of tests.

-Mac

MacM · · Tucson/Preskitt, AZ · Joined Feb 2010 · Points: 675

Jim,

I believe when I wrote that statement I meant that a carabiner's most vulnerable position when tensile forces are applied strictly at the corners along the major axis (shown in CATIA model). If a force is applied at the nose it wouldn't surprise me if the carabiner broke at near body weight, that moment would be tremendous.

Since this is for an "Aerospace Engineering" course, at an Aerospace Engineering school, sending students directly into the field of Aerospace Engineering. The course covers "Factor of Safety" applications standard to industry. Which is actually most commonly at least 1.5. So unfortunately that is what I needed to relate the tests to, if this was back 2 years ago in a lab I had it would have been all about the Sigma System.

What you are describing falls in line with Larry and Ryan's points, which I totally agree with. I didn't exactly think that the climbing industry utilized a F.S., I would have guessed that it would follow the Sigma System.

Thanks for all the constructive criticism so far, I wish I had submitted this to you guys before I turned in the report and gave my presentation.

-Mac

Larry S · · Easton, PA · Joined May 2010 · Points: 872

So... below is probably just a bunch of mental masturbation... I plugged your numbers into Statgraphics, and we can estimate that distributions (making a big assumption about normality). Doing so, i get:

95.0% confidence interval for mean: 2013.0 +/- 100.814 [1912.19, 2113.81]
95.0% confidence interval for standard deviation: [21.13, 255.055]

So we can be 95% sure the mean is between 1912 and 2013, and that the standard deviation is between 21 and 255.

I've plotted those distrubitions below with the worst and best case means and Sdev's. (red, pink, black, brown)

We can also work backwards from the lower spec limit (LSL) of 1573 lbs (7kn) open gate, and assuming your mean to be correct at 2013 lbs, and their 3 sigma testing to be perfectly accurate, we see a difference between the mean and LSL of 440 lbs. With the assumption of a normal distribution (and probably assuming many other things) we can divide that by 3 to get an estimate of the maximum standard deviation they could have while acheiving 3 sigma performance - 440/(3sigma) = 147 lbs. I plotted that below (in blue) too.

Possible distributions

So there's your possible best and worst distributions based on the 3 data points, and the blue one is an guess of what they actually make them too.

MacM · · Tucson/Preskitt, AZ · Joined Feb 2010 · Points: 675

Larry, you've got some great mental masturbation!

Thanks for throwing together those diagrams and calculating some more data. It still isn't super significant in meaning because there are only 3 data points, but it definitely helps visualize the experimental data in relevant terms. Also, that really gives more conclusion to our results by providing a better, more definitive, result of ~147lbf.

Thanks!

-Mac

Woodchuck ATC · · Unknown Hometown · Joined Nov 2007 · Points: 3,280

Somebody here at MP did a bunch of destructo-testing on biners we sent in to him a year or so ago. You might be able to find the old post, or his name and email he left on the old post. He posted up pics later of our busted out biners and exact numbers they failed at. Was very cool to see.

Richard Delaney · · Unknown Hometown · Joined Feb 2014 · Points: 5

We broke 100 carabiners 2 weeks ago at RopeTestLab facebook.com/groups/RopeTes…

My hypothesis was that aluminium carabiners passing a visual check would be ok regardless of history.

Method:
- Number each one and photograph both sides
- Drop all biners 5.5m onto concrete 5 times
- give 20 of these some extra special treatment with a lump hammer (enough to fail a visual)
- Rephotograph all, both sides
- slow pull test and record failure strength.

Results:
- the 20 'special' biners all broke at or above rated
- of the 80 others, all broke at or above rated except 1 which broke 14% below

Interpretation:
The sample size is still too small and the biners were a mix of brands so there is limited value in detailed statistical analysis.
The one that broke at -14% could have been the 1/1000 outside 3-sigma as discussed previously and may well have broken low when brand new.

Additional test: 3 x alu biners were dropped 275m onto plate steel over concrete and these too broke at or above rated.

My take away (and you are all encouraged to differ!):
Aluminium biners that pass a visual are most likely ok regardless of any unknown history. The manufacturing unknowns and variations would seem to be just as likely, if not more, to be responsible for a biner breaking below it's rated strength.
I prefer to use biners that have been proof loaded to 1/2 their rated strength as part of production to rule out the lowballs from the bell curve.

Locker · · Yucca Valley, CA · Joined Oct 2002 · Points: 2,349

IMHO these types of threads RULE!

Good stuff!

20 kN · · Unknown Hometown · Joined Feb 2009 · Points: 1,346
Larry S wrote:Mac - Climbing gear like this is rated to it's failure strength with a 3 sigma (99.87%) certainty
Not all gear is 3-sigma certified. Although some manufacturers advertise 3-sigma testing, most dont.
Guideline #1: Don't be a jerk.

General Climbing
Post a Reply to "What's the actual strength of your carabiners?"

Log In to Reply

Join the Community

Create your FREE account today!
Already have an account? Login to close this notice.

Get Started.