Why “Bounce Testing” should be avoided

Mark Pilate wrote:

And yes I’m saying that statistically, extrapolating from aggregated meta data of similar related testing statistics, your odds/margin is better for potential “victim guy” when encountering unknown qty X , potentially somewhere in the 2-3 range, if you follow the protocol of 1. Do heavier guy first as test, then do potential victim guy.  

Adding in a 2-3x load force pre-test, ahead of it all adds an unnecessary wild card that adds little value.  

I'm having a hard time being convinced of the first point without seeing actual test results.

But there's also a gap in the logic here: if the risk you describe is that the test load happens to approach the MBS and damages/weakens the sling, does this not hold true for doing the "heavier climber first" technique? Whether it's a bounce or a heavy climber, the risk is that you hit the bullseye of damage-but-not-rip before the last rapper goes.

I'd take the risk of a bounce hitting that bullseye (if it exists) over the risk of the last person to rap accidentally creating the highest peak load due to a slip of the foot.

I'll take my free sling now, just make sure it's bleached and crispy as hell.

I don't have years of experience like others here but I've initially formed a habit of bounce testing webbing, slings and v-threads in ice climbing context based on an intuitive feeling. Glad to be convinced otherwise by this discussion. And note that when I bounce test, I'm using other factors to gauge whether this is an anchor I would trust my life to, including a back-up for the first on rappel, and I am not going off the bounce test alone.

I also don't have any scientific knowledge here either, but using a simple example, does this not at least reject the idea that bounce testing is wrong 100% of the time?

Let's say I have two climbers that are weighting the anchor at 1.5KN. The forces at which I am most concerned about when rappelling are between 1.5 KN and 2.5KN (bouncing rappel let's say), so if I exceed those forces in a bounce test, haven't I actually checked whether the anchor is going to break in the most likely worst case scenario?

Otherwise, if I just use a back-up for the first person and then the 2nd person exerts a force between 1.5 KN and 2.5KN (a range which I perhaps did not test when the first person went down), the anchor is potentially at a risk of breaking for the 2nd with no back-up.

I feel like when I am bounce testing an anchor, I am testing whether this thing has something catastropically wrong with it - like a v-thread with a crack in it that I can't see that would theoretically break above body/rappel weight.

Mark Pilate wrote:

Kyle, post justification for what you are doing.  Don’t get in the weeds of metals like titanium and curves that are outside the discussion.

Mark, I directly quoted your post and replied to the specific points you made.  Don't accuse me of getting "into the weeds" if I am replying to direct quotes from you.

Furthermore, I specifically replied to several different claims you made about nylon specifically (ropes and slings) and provided sources for my counter-arguments.  Why didn't you address those?

Oh, also, I literally never mentioned titanium.  You're the only one who has used that word in this thread.  Can you focus your criticisms on things I actually said?

I have always been discussing old nylon and ice as examples of materials that have a large delta of results from one test to the next.

You have not provided one single actual datapoint for nylon or ice.  You're so confident, why not just cite the data or material behavior that backs up your claims?

Let’s cut to the chase and keep it simple—You evidently like to bounce on ice.  There must be a reason you are needlessly subjecting your ice anchor to 2-3 times the intended load before committing to it.

I never said I did this.  You're creating a straw man instead of discussing the topic at hand.

you must have some engineering models, test data, etc leading to your reasoning that you are therefore statistically safer with a higher load test that a “normal use” verification test.  Please show it

Why does everyone you disagree with need to show extensive data and models, but you can make wild assertions that disagree with basic materials engineering without proof?

  Edit that got cut off:  I did reasonably extensive (couple dozen) firsthand tests on marginal v-threads over Covid.  (Not scientific or statistically significant) but still more validating than your intuitive gut feel with so far no engineering backing to justify your bounce procedure on fragile ice.

If you have test data, why have you been arguing with people for 2 pages without posting it?  Post up your test data that shows that a v-thread will fail at a load below the load it successfully is bounce tested to.

Example from real world.  You come to an unknown piece of tat of dubious strength and for whatever reason, you need to rely on it (let’s say it’s at the edge of useful life and if pull tested would fail at 3.5 kN like the accident example)  What is the best algorithm to follow to provide you the widest margin for the rap?  (For arguments sake you just can’t leave your back up here)

Choice A-  Bounce it with close to failure load, then rap, or Choice B- back it up while first heavier person raps carefully, then second does the same  

answer is choice B

Any of you pro bounce people, Show me any analysis or data from engineering, physics, mathematics etc that says you are statistically safer with the 2-3 x load bounce test preceding your rap, vs the careful confirmation rap of your buddy.  

If it’s not near failing, it doesn’t matter, the bounce was superfluous.  If it’s near failing, the bounce leaves you more exposed statistically than before your “test”

You're asking other people for "analysis or data from engineering, physics, mathematics etc" and yet you provide none of that.

You also skipped over the most relevant scenario.  When the 2 members in the party went on rappel they exceed the material's 3.5 kN strength (and sadly perished).  If they had bounce tested the anchor to 2X the load they were going to apply, which would be 7 kN, the material would have failed, they would not have relied on it, and they would have left something stronger.  In the exact scenario you have concocted, a bounce test might have saved their life.

The claim you're trying to make is that a material will hold a bounce test at some force, and then second later it will fail at a much lower force.  That's not logical.  If a sling (or v-thread) held 4 kN 30 seconds ago, the odds are quite good that it's going to hold 4 kN again now (or, in the scenario you've put forward, half that).  What the heck is the mechanism where a piece of material holds X force, and then 30 seconds later doesn't hold 1/2 X?  Why don't you post data that shows the strength of a piece of material going down by 50% in the minute or two between loads?

This is like arguing with a creationist.  You're just throwing assorted wild theories at the wall, and every time someone provides a counter-argument, you say that wasn't the important part, and you have 10 more random ideas that they now need to disprove.  There's no possible way for this to be a beneficial discussion, because you're going to continue to sidestep any of the legitimate issues people bring up, and just keep moving the goalposts so only you can hit them.

Kyle, I'm coming around to your line of thinking here. I'll agree with you that the recent tragedies just might have been prevented with bounce testing, and no you will not find data where material strength is cut in half after a couple load cycles.  I can also see the value of bounce testing when you find yourself in the dire circumstance of needing to conserve gear, likely on an unusual/completely off-route rap line on a long climb. 

However, the field application is extremely complex. In any other scenario than the one above, I would leave my own gear and construct a bomber anchor myself. I will leave whatever is needed to do so, cams/nuts/slings/webbing, the whole shebang. There are just too many heartbreaking instances where anchor failure has resulted in death and I am not seeing how a quick field test can confirm the 100% reliability of the in situ anchor.

To take a step back here ... being familiar with descent options for your route, and bringing a tag line can save you from disaster on longer, more committing climbs. I'll link an accident report that really stuck in my mind:

Cathedral Peak accident

The team ended up making ~15 off-route rappels down Cathedral Peak, chopping their ropes when they got stuck, and finally rapping off a single marginal cam when they ran out of gear. It held part of the way down their simul-rap. UNTIL it didn't. The anchor failed and they slid into a snowbank, miraculously surviving. This team could have constructed a gear anchor on the summit block and made a single rappel toward the notch on the mountaineer's route and walked off.

Mark Pilate wrote:

I’ll send the first person here, $100 worth of slings of your choice if you can prove the validity of bounce testing unknown sketch. 

The data is right in the accident report for the incident you're referencing.  SAR pull tested the broken tat 3 times.  Their results:

Pull #1: 2.5 kN

Pull #2: 3.6 kN

Pull #3: 3.2 kN

The first pull test stressed all of the webbing to a high load (2.5 kN), and found one weak point.  When re-tied, the next pull test was slightly higher, indicating that the first pull didn't cause a significant reduction in strength of the rest of the webbing.  After the second test, it was retied and pulled again, and it failed at 3.2 kN.  So, again, the strength of the webbing wasn't significantly reduced after the 2nd pull test either.

These results show that an old crusty sling has pretty similar strength, even after pulling on it very hard (near/to failure) a couple times.  In other words, the strength of the material is not significantly affected by 1 or 2 initial hard pulls, and the strength is fairly consistent across tests.  This completely disproves your theory that a hard bounce test will cause the material to then fail at a much lower force.

You can donate my $100 to the ASCA, please provide proof of donation.

https://rmru.org/2022/09/28/2022-28/

James W wrote:

Similar ~20% strength changes on nylon-6 in this paper (73 MPa at stain rate 5, 57 MPa at 0.05): https://deepblue.lib.umich.edu/bitstream/handle/2027.42/34829/10484_ftp.pdf?sequence=1

If you have more data around the effect of strain rate on the types of materials used in climbing soft goods that is better, different, or disproves what I said above, please post it up, I'd love to be more knowledgeable.

Kyle Tarry wrote:

The funniest part of this whole thing is that the data is right in the accident report for the incident you're referencing.  SAR pull tested the broken tat 3 times.  Their results:

Pull #1: 2.5 kN

Pull #2: 3.6 kN

Pull #3: 3.2 kN

Static pull tests don't reveal the impact of strain rate on applied stress, which can be quite significant...

Kyle Tarry wrote:

SAR pull tested the broken tat 3 times.  

Your point is good, but it is muddied by the likelihood of the different break locations being where a rope was pulled over the webbing.  Pull #0 was likely the weight of the victims.

To clarify: In my tests PLENTY of slings broke below body weight, at body weight, barely above body weight etc.. maybe half or so of a hundred or so samples. Those that survived bouncing just ALWAYS survived more bouncing-- except those that showed obvious tearing from a bounce. 

I would make strands of ~10 webbing samples, and test till breaking, then combine the survivors, ad infinitum. Probably a minimum 10 drop cycles for the weakest sections, to over a hundred for the strongest (unbreakable from small weighted drops)

Ice may be different, given it's plastic deformation (melting) under pressure. 

Edit: "don't just trust tat: test tat" 

Double edit: "don't trust tat"

My take away. For the sake of those that follow, please don’t leave webbing as a rap anchor. Carry cord. 

abandon moderation wrote:

• try to verify that it's not certainly unsafe, make some best guess, do some primitive test, and accept that there's some chance you could be wrong.

isn't that how all of us should approach climbing? we all think we know it all, but if there's one thing I've learned from years of recreating in the backcountry, confirmation bias is real...and it can be really deadly...

approach everything with skepticism, question your partners and maintain situational awareness...

You send the fat Hobbit down first.  If he doesn't make it, now is as good a time as any to reflect upon the poor life choices that brought you to this moment.

That Cathedral Peak story ^^^^^ is amazing! So glad they survived. 

philip bone wrote:

That Cathedral Peak story ^^^^^ is amazing! So glad they survived. 

Yes, but also one of them died of hypothermia on the hike out.

Oh damn! I starred it to finish later. 

Greg R wrote:

My take away. For the sake of those that follow, please don’t leave webbing as a rap anchor. Carry cord. 

Good idea, but it's not full proof.  There was a fatal accident in the southeast about 10 years ago. Anchor consisted of full thickness climbing  rope around large boulder. The unseen  part of cord had been gnawed on. I've run into that situation a couple of   times. Once in a bad storm,  we rappelled without checking.  Every now and then I'll think about how stupid that was.

Mark, you have zero experience in testing or otherwise working with textiles. You've repeatedly failed to site sources that actually apply to textile testing, and just pulled other stress-strain curves and whatnot from other application. Why are you trying to pass your opinion off as fact? 

You're full of shit

Kyle, you're a godsend. Thanks for taking the time to write that all out and refute a bunch of that nonsense.

Big Red wrote:

I'm having a hard time being convinced of the first point without seeing actual test results.

But there's also a gap in the logic here: if the risk you describe is that the test load happens to approach the MBS and damages/weakens the sling, does this not hold true for doing the "heavier climber first" technique? Whether it's a bounce or a heavy climber, the risk is that you hit the bullseye of damage-but-not-rip before the last rapper goes.

Good point.  I’ll hunt up the data, they’re out there, but I don’t need to.   You win on logic alone, no technical basis needed.    Thesis (that I stand behind) was that the higher force “shock” load of the bounce at “N” on non-uniform, highly variable materials resulted in greater subsequent variation and unpredictablilty at N+1 vs lower a more static load with the rope involved in the other case resulting ina statistically better out at N+1    

The fatal flaw in my logic was assuming there was a test N, then the N+1 was always the “money rap”, in both cases.  But with a bounce or pretest, there is N, N+1 (first guy can be backed up) and it’s N+ 2 - the last guy  that counts.  Therefore, there is actually a belt and suspenders approach and I’d totally agree, that if a bounce passed AND then a regular test load test passed with the first person, then the risk to the last guy is ultimately lower than if no bounce.  

I'll take my free sling now, just make sure it's bleached and crispy as hell.

Pm me for shipping address.

edit to Ian and Kyle: post and time limited, but you are mistaken. Follow up later. It’s easy to be a google engineer and look things up and post. Harder to immediately collate decades of experience and data. Find and post stuff you did 5yrs ago vs look it up now. You both missed the logical solution (as I did) in your fools errand to disprove the irrelevant semantics of the technical 

Mark Pilate wrote:

Still challenge anyone to make a coherent argument why a higher load bounce on already sketchy material is the best practice. 

Its been said, but to reitterate - the goal is to break any sketchy tat that would fail at bodyweight, whilte it is backed up.

Do you drill a new bolt when you encounter a Leeper on lead? Or do you give it a yank and look to see if its cracked and won't hold up your draw? (thereby giving you the option of choosing to downclimb or trying even harder to not fall on the next section) Imperfect data can still be useful data. 

Matt N wrote:

Its been said, but to reitterate - the goal is to break any sketchy tat that would fail at bodyweight, whilte it is backed up.

Do you drill a new bolt when you encounter a Leeper on lead? Or do you give it a yank and look to see if its cracked and won't hold up your draw? (thereby giving you the option of choosing to downclimb or trying even harder to not fall on the next section) Imperfect data can still be useful data. 

The point of a bounce test is to confirm that tat is strong enough to REMOVE THE BACKUP. 
 
Simply the POSSIBILITY that a bounce test could weaken tat would prove the bounce test unsafe.

It is on the proponents of bounce testing to prove that it DOES NOT weaken tat without a doubt because it only takes once to die!

Let they who has not rapped off tat, cast the first stone.