Why “Bounce Testing” should be avoided

A recent accident discussion and analysis has revealed a widespread practice and myth of the “Bounce Test” that should be widely called out and discontinued to avoid discussing the same situation in future accidents.  

This is a classic case of confirmation bias due to the low statistical nature of the outcomes, but these accidents are real and they keep happening.   The goal here is OPTIMAL decision making and practices, to give people the widest margin for error in a sport that has drastic consequences for sub-optimal decision making and practices.

Situational definition and context

• This does not apply to testing an anchor piece (pro)  that can be re-adjusted or re-placed to a better orientation or location
• The purported reason for a “Bounce Test” here is that you have a sketchy anchor that you don’t trust as in the context of in situ “tat” similar to the recent rappel anchor failures in the Accidents thread.
• You have no way of knowing its real ultimate strength
• You think conducting a “bounce shock test” will give you useful information on the strength of that anchor for when you intend to use it.
• Many erroneously assume, without evidence, the victims obviously didn’t bounce test their anchor, because they erroneously assume that a “ “bounce” type shock test definitive.
• A bounce test here is defined as a guesstimated  2-4kN shock load (especially if done from static tethers) to attempt to verify a suspect anchor (tat, ice v-thread, etc) 

Logical Argument against

• To “test” it, you have a safe back up available and you employ it to safeguard your test
• If your test “fails”,  meaning your bounce test results in original anchor failure, then you need to construct and leave your back up anyway.
• So you HAD a back-up alternative to a solid anchor but chose to go with the sketchy one because you now think it’s “safe” to do so. 
• A bounce test means you have a choice.  So this test becomes essentially an economic test – to save leaving a bit of gear.   Bad choice.  
• Any arguments against these assertions?

Economic Argument Against 

• Your life is worth more than the cost of some climbing gear.   Leave the backup. 
• Any arguments against this assertion?

Technical Arguments Against 

• Fatigue/deterioration/aging is a process that has a degree of randomness often showing considerable scatter even in seemingly identical samples in well controlled environments.
• The greater the applied stresses (especially shock loads),  the shorter the life and greater the variation the results of materials tests.  
• Expected material life variability and material properties changes increase, the longer fatigue/deterioration/ageing processes continue 
• Damage is irreversible. Materials do not recover with age.  Load effects are cumulative in the decay of material properties over time.  
• Fatigue /useful life is influenced by a variety of “aging” and deterioration factors,

•               temperature cycling

•               surface finish/ damage (rap ropes being pulled, rock edges rubbing, etc)

•               oxidizing and UV damage, ozone attack 

•               scuffing contact (fretting), etc.  and normal fiber wear due to internal grit abrasion (not visible) 

•               Knot and equipment stresses, and load angles

•               Moisture content (wet nylon is weaker than dry and has a degraded shock response and recovery)

• Macroscopic and microscopic discontinuities are common locations at which the fatigue/deterioration process begins and the longer a material ages in it’s “lifetime” the greater the delta and unpredictability between the results of “test N” and “test N+1”.
• See more principles in thread page 6
• Any arguments against these assertions?

Takeaways

• Obviously never stake your life on old deteriorating tat materials in the field.  But if you absolutely need to for some emergency reason…don’t use shock load techniques to test it. 
• Uncalibrated “Bounce” shock load testing may be misleading,  and may provide less reliable future use “safety gauge” than a smoother/softer static test. 
• Regardless of the results, you likely just reduced its energy absorption capability in an emergency during use.   There’s a chance you may have just “bounce-burned” your one emergency shock load those old materials have in it.  
• DO NOT bounce test ice anchors!  For example, a v-thread in good ice is bomber as is.  Ice is a brittle medium that can change after a shock load with subsequent loads…way worse than old aging nylon.  Even though it’s still likely to be fine, you may have just introduced the defect you tried to detect. 
• If you absolutely need to use a sketch anchor for whatever reason….and want to test it first;
• Edit : Let’s cut to the chase and what can climbers easily take away from this:   Learn from other industries who live and work this stuff daily — Riggers.  Due in large part to exactly what I have been saying all along (in thread below)federal and industry regulations specify that slings that support a shock load should be immediately taken out of service.   A proper proof test is a backed up, smooth pull to 2x working load.   A “bounce” shock test would immediately render the sling unusable due to its being too unpredictable.   OSHA, ASTM, ASME, and rigging companies base their best practices on the science. 

  It seems relatively obvious and intuitive that uncalibrated shock loading of degraded and unpredictable old tat would be bad, and that for a last ditch safety proof test, you’d want to test smooth and easy to a higher safety margin (2x intended use) and then use it in a very smooth and careful easy fashion

  Not sure why all the pushback below from the the high school science class analysis led by Kyle.  Apply The concepts of good proof testing principles, and the principles developed by an industry that abuses nylon slings daily.  
  1. don’t create the problem you’re trying to detect and leave yourself more at risk, and…
  2.  new material doesn’t equal old degraded material in expected performance to testing, and…
  3. shock loading doesn’t equal static loading for tests in such materials,  should be plain.  
  
  And as always If you have a chance to back it up, and leave it,  DO IT!

 
Open for discussion  if you disagree. 

edit: Rev 2 modified per discussions below for clarification.  And again “rev3” on 10/14.  -removed figure 1 to avoid confusing any new people checking in

Coming at you:

1. I think the purpose of a bounce is to generate loads in excess of what might happen during a typical rap - the result is some meaningful, though not definitive, information. If you have generated 6kn on a bounce and the anchor has held, it's reasonable to expect that another load under 2kn won't cause failure. Not sure I agree that you've weakened an anchor if the bounce stress is in the elastic deformation zone. You'd have to be pretty unlucky to bounce it into its plastic zone without having it fail. Also I believe these slings aren't failing due to fatigue and cycling, they're failing due to damage changing their stress/strain curve. So extra bounces aren't causing more damage if they're below the current yield.

2. I won't argue with your economic point, but I do think it's unreasonable to leave your own tat on every tat anchor you find vs leaving it on the sketchiest ones. Of course, how do you know which one is sketchiest? If I carry 10 ft of backup tat and I need to do 6 raps off slung trees, I might choose to bounce a particular anchor since it looks pretty ok, in anticipation of a subsequent anchor looking even worse. There's an element of judgement here when it comes to using your limited resources, and it costs nothing to bounce each anchor while backed up to make sure you haven't pulled the short straw in terms of unseen damage. I agree that in an ideal world everyone would replace every piece of tat with their own on every rap but that is clearly not a realistic scenario. The next best thing is to make a practice of bouncing and replacing the sketchiest ones.

3. There's a similar precedent in aid climbing. I can't imagine there's a bulk of evidence for placements failing after a good bounce test (notwithstanding shifting due to load direction changing when stepping up), but maybe the more experienced aid climbers can chime in.

4. Edit: As Greg said below, I'm curious where you got that plot - soft goods shouldn't look like that.

Not arguing with your overall point, but I think your points about fatigue are somewhat misleading. The graph you show is for how a ductile metal works, it probably would not apply to a corroded metal and definitely wouldn't apply to soft goods like dynema or nylon. Those wouldn't have strain hardening or necking or anything like that. Also the post bounce test strength assumes that you're getting into the 'strain hardening' part of the graph with your bounce testing, which is very unlikely. Metal goods (unless they're old ones someone made in their garage), are made so when new and uncorroded you wouldn't get anywhere near that type of load with bounce testing. Also, load effects are not cumulative unless you're getting up close to the yield strength of the metal
(at least on the order of cycles climbers subject them to). If they fail it would be due to corrosion, cracks, or other factors that won't follow the characteristics shown there.

The 'fatigue' that happens on slings and other soft goods would be a totally different process with different failure methods.

Also, load effects are not cumulative unless you're getting up close to the youngs modulus of the metal

edit: "Figure 1 below is just representative example plot of generic possible “Tat” test results" I can make any graph I want and it can be 'representitive of possible generic results'. Unless you do the test yourself or find someone who did, this is misleading.

to Puppy:  Different context for your scenario with those bolts.  Do whatever you want on the ground

To Greg:  It’s a “generic” curve.  Nylon does have similar curve and varies only trivially.  The graph is undefined in slope and values so as not to argue in these “weeds” that make no difference.

  I believe your other assertions are definitely false.  Nylon itself, and textiles made from it, such as rope and webbing, do exhibit the phenomena you mentioned.  Hardening/stiffening (very long term) necking, plastic deformation, etc.  the textile failures are just cumulative fiber or elemental failures and upon aging have a combination of ductile and brittle failures.  Example is Ropes continuously lose stretch upon subsequent loading over time (falls) -especially wet-  which is essentially same as strain hardening to an extent though not “technically”  

And load effects ARE cumulative.  Even in the elastic region.   Are you saying that a material or specimen has infinite life if loads are kept below the yield point? — in real life, over the life of the material   (Not uniform test specimens) 

Does anyone know the yield point for 2 year old weathered tat that has been rapped on an unknown number of times?

To Big Red:  you are thinking of the delta strength between N and N+1 of say new webbing just placed…which is trivial.   No bounce test needed.  The delta between N and N+1 on old tat (or ice for that matter) is wide open for a variety of reasons and variables.  and your “Delta” could very well have been significant for old degraded materials like tat.   It’s Russian roulette you dont need to play.    Use commonly accepted proof testing principles.  Never excessively overload, especially with shock loads.  

Mark Pilate wrote:

Situational definition and context

• This does not apply to testing an anchor piece (pro)  that can be re-adjusted or re-placed to a better orientation or location
• The purported reason for a “Bounce Test” here is that you have a sketchy anchor that you don’t trust as in the contest of in situ “tat” similar to the recent rappel anchor failures in the Accidents thread.
• You have no way of knowing its real ultimate strength
• You think conducting a “test” will give you useful information on the strength of that anchor for when you intend to use it.
• Many erroneously assume, without evidence, the victims obviously didn’t test their anchor, because they erroneously assume that a test is definitive.

I'm confused.  Bounce testing is being used to test the strength of webbing/tat?  Who does that?  I've bounced tested many anchors but only to test the actual protection pieces but never to see if the webbing is still strong enough.

Mark, I agree with everything you write, except for one conceivable scenario:  Suppose a party has decided to bail off a long route that requires multiple rappels of unknown length or commitment. They leave gear where they must, but also try to conserve as much gear as possible to ensure that they can make it all the way down.  They arrive at a webbing anchor of unknown origin, but it appears to be in decent shape.  Weighing the odds, they decide that the material looks good enough that it isn't worth using more of their dwindling supply of soft goods.  

In this scenario, I think it would make sense to temporarily tie a backup, bounce test the existing webbing, and then rap off said webbing (by itself).  Your technical arguments against bounce testing webbing are still valid, but if you found yourself in this predicament, would you honestly not bounce test it?  To me, and this is based solely on intuition, I would expect that to be the statistically safer answer.  I think that's what the pro-bounce test people are arguing (even though they make the mistake of discounting your other logical/economic arguments altogether).

Not saying you are wrong, Mark, but the reason most people grab and yank on tat is to see if it rips away. Basically, they want to know if it's UNSAFE. Which isn't the same as knowing that it's SAFE. I've pulled on suspect tat and manky bolts and had them come off in my hand. Definitely learned they were unsafe. And also removed a hazard that someone else might have fallen prey to. But I agree with your point that it NOT pulling off doesn't show that it's safe. Especially in light of the Tahquitz tragedy I'll be extra careful about never rapping off a single piece of unknown web or cord. 

Mark Pilate wrote:

To Greg:  It’s a “generic” curve.  Nylon does have similar curve and varies only trivially.  The graph is undefined in slope and values so as not to argue in these “weeds” that make no difference.

  I believe your other assertions are definitely false.  Nylon itself, and textiles made from it, such as rope and webbing, do exhibit the phenomena you mentioned.  Cyclic Hardening, necking, plastic deformation, etc.  the textile failures are just cumulative fiber or elemental failures. Example is Ropes continuously lose stretch upon subsequent loading (falls) which is essentially same as strain hardening.  

Nylon as a solid material has those same properties or nylon as a cord? Since you didn't feel the need to source your graph I'm not going to spend the time to research mine fully either, but this is what a quick google gives me when searching specifically for a cord instead of just nylon as a material. Your 'generic' curve isn't generic for every material out these, and properties definitely change when you make something into a rope that is composed of fibers of the material. 

Also, I'm not saying the life is infinite, but most peices of gear don't get loaded anywhere near 20 thousand times, which is closer to the order of magnitude you need for fatigue failure in aluminum. It's my understanding that fatigue failure is usually on that order of magnitude of number of cycles to failure, please show me if that's incorrect for nylon. That of course assumes that fatigue works the same as nylon as aluminum, which is unlikely. For example, steel in fact does have a load below which it will not fatigue to failure, which aluminum has no such point, and will always fail due to fatigue if put through enough cycles. Without evidence, how can you say how nylon or dynema slings will behave? They're way farther different from steel and aluminum than those metals are from each other, and the metals behave quite differently. I suppose if some tat got loaded 5x per day every day for 10 years you would get close to that, but it would break much sooner due to weathering and uv exposure, but thats also a moot point without knowing how the material behaves.

Mikey Schaefer wrote:

I'm confused.  Bounce testing is being used to test the strength of webbing/tat?  Who does that?  I've bounced tested many anchors but only to test the actual protection pieces but never to see if the webbing is still strong enough.

Mikey, your scenario is the “correct” one 

Many have been advocating for what effectively could be called “Tat bouncing”. Here and In the accident thread 

Edit to Big Red :  

The exact shape of the curve is meaningless (all essentially have an elastic region(modulus) , a yield/inflection  point and a failure point.  It’s the movement of the failure end points between subsequent cycles that matter, and as materials age and fatigue, that delta can grow to significance.  

Kyle:  (note- he deleted most posts) Per usual, you react emotionally and respond with nonsense.   Are you saying you made it stronger by Bouncing it?  Ice is way more brittle and has way more variables than webbing.  The bounce test told you nothing you didn’t already know.  A good v-thread is bomber.   I agree that on a good v-thread, the bounce test won’t add any meaningful risk to your rappel.  But more importantly, your bounce test on solid gear/equipment was MEANINGLESS.  The more appropriate case to examine is a sketch v-thread or ice anchor for some reason.  Now suppose you bounce a sketch v-thread and it holds, but you likely just made a sketch ice anchor way worse (I’ve done it).  If it was sketch before, it’s now way sketchier. You definitely didn’t make it better.  Go do your own tests and report back before spouting off.  Study proof testing brittle non uniform materials.  
Testing of brand new equipment at the factory is a spurious comparison.  And that’s also one reason why they rate it conservatively below tested failure strength and why there are mfr suggested lifespans.  It makes no sense what you are arguing or why 

Mark Pilate wrote:

To Greg:  It’s a “generic” curve.  Nylon does have similar curve and varies only trivially.  The graph is undefined in slope and values so as not to argue in these “weeds” that make no difference.

  I believe your other assertions are definitely false.  Nylon itself, and textiles made from it, such as rope and webbing, do exhibit the phenomena you mentioned.  Cyclic Hardening, necking, plastic deformation, etc.  the textile failures are just cumulative fiber or elemental failures. Example is Ropes continuously lose stretch upon subsequent loading (falls) which is essentially same as strain hardening.  

I'm no materials expert but this doesn't seem true. Nylon's curve doesn't look like a generic metal curve, and certainly a rope or sling behaves in more complex ways than a simple metal sample. Losing stretch after repeated falls is not similar to strain hardening, since it is a largely recoverable process.

Paging Dave Custer or someone else who knows soft goods...

Edit: Note that SAR tested the material in question in the accident that prompted this thread, and their results are the opposite of what Mark is claiming.

Pull #1: 2.5 kN

Pull #2: 3.6 kN

Pull #3: 3.2 kN

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

Mark makes a variety of other claims in the first post and others, and generally speaking these claims are incorrect and display a significant lack of understanding of materials science.  See a variety of other posts I and others have made in this thread for details.

Be careful about suggestions in this thread.  Just back up rappels and don't trust your life to old sketchy stuff.

Kyle Tarry wrote:

Remember folks, many brands proof test 100% of climbing gear, which is the manufacturing equivalent to a bounce test; they load the gear to some fraction of the failure strength, which is similar (or likely greater) than the loads you will actually apply.  For example, Black Diamond supposedly pulls 100% of carabiners to 50% of rated strength.  According to this thread's author, this is a detrimental practice.  Who do you trust?

Static weighting of gear in a controlled environment != bounce testing gear in variable context. There are some things that Mark said that may or may not be speculation, but these do not equate.

I don't have the expertise to argue any of the fine details. I've personally never bounce-tested tat. This is especially true for tat that has been threaded through constrictions and cannot be viewed in its entirety.  Maybe a special case, but there's no way I'd bounce on that; too great a chance of cutting the sling (which was ok before the bounce test) on something out of sight.     

One of the naive problems is that the usual "back up for all but the last person" is also a kind of bounce test that could, according to the arguments advanced, further weaken a sling and so cause it to break.  To complicate matters, I know of two cases in which an anchor failed for the last person after being backed up for everyone else, but neither of those failures involved tat breaking; both were failures of the actual anchor and were related to the fact that the last person loaded the anchor in a different way then all the others.  (We could say that the "test" didn't replicate the actual conditions.)  I still think the backup for all but the last person is the most practically useful, understanding that although very unlikely, it isn't impossible for that procedure to fail.  

That said, there are things beyond the pale for me, and one of them would be a single 1" webbing sling. Fuhgeddabout testing, I'm adding more material to that anchor no matter what.

I disagree that the cord/slings are being damaged by repeated bounce tests, only that repeatedly testing gear in a lab versus bounce testing real gear in rock are different - not because slings are different, but the context of the terrain - meaning, the attachment to the drop tower is a lot different than a V threat / slug tree / horn / chicken head and THOSE unique contexts can change how a bounce test can weaken said cord/sling.

Otherwise, yes, I agree. Please don't be mean :( 

rgold wrote:

One of the naive problems is that the usual "back up for all but the last person" is also a kind of bounce test that could, according to the arguments advanced, further weaken a sling and so cause it to break.  To complicate matters, I know of two cases in which an anchor failed for the last person after being backed up for everyone else, but neither of those failures involved tat breaking; both were failures of the actual anchor and were related to the fact that the last person loaded the anchor in a different way then all the others.  (We could say that the "test" didn't replicate the actual conditions.)  I still think the backup for all but the last person is the most practically useful, understanding that although very unlikely, it isn't impossible for that procedure to fail.  

Got any details/links? On the surface that is scary as shit; sounds like a reasonable explanation was found though.

Big Red wrote:

 Also I believe these slings aren't failing due to fatigue and cycling, they're failing due to damage changing their stress/strain curve. 

I think this is the key point that can be argued about Mark's technical analysis. I also think Mark made things confusing by including a stress-strain curve that does not clearly contribute to his point, and people are using that to pick his argument apart. A visual that would have better fit his point would be an s-n curve and to state that bounce testing a piece of tat would be over the endurance limit, so after some number of loading cycles the tat will eventually fail (and we don't know what that number is).

I have no experience with nylon (I study metals), but I would side with Big Red in my opinion of why these slings are failing (environmental damage/degradation lowering the yield strength). However, I'm certainly not knowledgeable enough to say Mark is wrong. I don't think the failure mode really changes the main takeaway though, which is to always backup questionable anchor material. 

What's the difference between a bounce test and fully-weighting the intended anchor? From my POV they are intended to discover the same thing: a faulty anchor setup. Therefore, if "fully weighting the anchor" is a valid test method, so too is this so-called bounce test.

So the OP's thesis is rejected.

Mil spec webbing is ~4k lbs breaking - WTF is this bounce test you’re doing on the side of a cliff that is coming even remotely close to yielding that?  

For the premise of unwanted yielding of highly degraded webbing - that is flawed as well - there is no plastic region remaining - it’s brittle.

Overall - C+ for your ability to produce a diagram and notes, D- for your engineering, F for the overall recommendation.

+1 - there is no excuse to not take your time and construct an adequate anchor.  The premise of needing to do some sort of engineering test on the side of the cliff and that it would have any value what-so-ever is SO many levels of wrong-headed.

Mark Pilate wrote:

Economic Argument

• Your life is worth more than the cost of some climbing gear.

This assertion is true, but to make a rational argument of it you'd have to multiply the left side by the probability of dying from using the old tat.

(not disagreeing with the main thesis)

J C wrote:

Got any details/links? On the surface that is scary as shit; sounds like a reasonable explanation was found though.

One was a friend, the other was a friend of a friend, neither was written up as far as I know.  In one case an old piton, in another case an old bolt, pulled out.  In both cases, the last person down was positioned to exert much more of an outward load than any of the "test subjects."

Past bounce test results - it works, if done correctly.

https://www.youtube.com/watch?v=gq3_DfyHg1A
14:00  generates 2.3 kN with roughly 160 lb climber and quite small bounces on nylon cord.

14:45 generates 2.7 kN switching to spectra, still quite small bounces.

2.7 kN is 3-4 times his body weight, double what can be reached when rapelling.

more

https://www.mountainproject.com/forum/topic/107638898/bounce-testing-forces

http://www.supertopo.com/climbers-forum/1851414/Load-cell-testing-How-much-force-does-a-bounce-test-produce

So the highest force incurred using an aggressive daisy bounce test that is realistic to a wall climbing scenario is 618 lbf. or 2.75 kN.   ... a hard bounce test directly on your daisy chains will produce 500-600 lbf.