Is Shock loading a myth?

hillbilly hijinks wrote
So, is the take away that in a climbing application when dynamic connections are used there is no peak impulse being missed in the posted set up and the extra forces when one piece fails are low?

Well, you need to be a lot more careful and specific to get a conclusive answer.  If we're referring specifically to the setup in this video, Jim's concern (which I share) is that the frequency (and signal filtration) of the measurement devices might cause us to miss the actual peak force, especially in the test(s) likely to have a sharper force curve (the ones with less or no dynamic rope in the system).  There's no way to conclusively know if this is happening without looking at a plot of the data, and measuring it at a higher sampling frequency if possible (some educated interpretation of a data plot may be sufficient).  The problem is that simply recording that isn't trivial, because high frequency data acquisition isn't a cheap or simple tool.

The good news is that we know there is a practical limit on the frequency response of the squishy human body, so the force generated in a "shock loading" scenario cannot be unbounded.  Whether that unbounded force is above safe limits driven by gear capacity and health risks is the important question; I've done some back-of-the-envelope numerical simulation (posted elsewhere on MP) that indicates that it might be below maximum load capacity for climbing slings, etc., but that's not the most robust evidence.

but I think that other than that the take away from the video is that this idea of extension being a deal breaker is over hyped?

It's situational.  If you have a follower being belayed directly from the anchor with 100ft of rope out and the anchor extends 1 foot, it doesn't matter at all (at least, in terms of loads on the remaining anchor points).  But that's kind of a given.  There are other scenarios, however, where extension could be a lot more insidious, the classic one being a climber belaying a second off their harness, while being tied to the anchor with static materials, and the anchor extends when the second falls off and has the rope stretched out.  This will yank the mass of the belayer against the anchor with a possible higher-than-1g acceleration and could cause a very high load on the remaining anchor pieces.  It's a complex mechanical system to simulate or test though.  There are lots of ways to guard against this, including (but not limited to) tying in with a sufficient length of dynamic material and knots, belaying off the anchor, building anchors that don't extend, separating the belayer's mass from the anchor, etc.  Not all situations are conducive to all solutions.

Note that getting blasted off your stance by an extending anchor has other risks too, even if the anchor doesn't ultimately fail.  For example, having your partner drag you over and/or pin you against the edge of the belay ledge would not be that cool.

Ma Ja wrote:

Could you propose to Ryan how they could have done a more specific test to gain the knowledge that wasnt gained from his test?

please don't misunderstand me. i am not knocking what the video producers are doing. it's a cool, real world test. but, we should take pause and consider what the video really is. it's sharing the results of some hap-hazard testing of a single rope system...nothing more and nothing less. 

If you really wanted to answer the question "do shock loads matter," i would expect testing of multiple variables (different rope systems, terrain, friction, etc.) and some kind of statistical process to be used so we can isolate the results (like design of experiments or similar). obviously this would take a lot more work, but that is why organizations like ITRS exist and perform more scientifically oriented experiments.

http://www.itrsonline.org/archives/

hillbilly hijinks wrote:And we are then back to Jim's iconic statement of just "get strong placements and tie yourself to them" as being the effective holy grail in anchor building.

at the end of the day, this is basically what all anchor building boils down to. well said...

I’m still missing the takeaway. Ryan is tied in with a dynamic rope and falling a short distance onto a static sling attached to a bolt. He didn’t get hurt or break his back. Is that it?

A few thoughts regarding force vs. impulse...

When you're catching a falling climber, the impulse is the same, regardless of elasticity in the system. (Assuming that you've successfully arrested the fall.) Impulse is just the total change in momentum; in this case, bringing the momentum of the climber to zero. In our case, we're concerned with peak load, since that is what will cause the gear to break. This is exactly why falls on static systems are worse than those on dynamic systems. The more elasticity is present in the system, the more that impulse is spread out in the time domain, and the lower the peak load will be.

In fact, the total impulse on gear that failed is likely to be less on a static system that fails than on a dynamic system that catches the fall. (For the same theoretical fall.) In the failed static system, the force exceeds the strength of the system, which fails before the fall is arrested. So the total impulse (change in momentum of the climber) is lower than it would be on the dynamic system that brought the climber to a stop.

The same is true for an injury to the climber. Total change in momentum (impulse) isn't what kills you. It's the rate at which it changes (acceleration). We can impart huge impulses onto humans; we routinely accelerate astronauts to thousands of mph. The problems start to happen when you try to do it too quickly.

Kyle Tarry wrote:

Well, you need to be a lot more careful and specific to get a conclusive answer.  If we're referring specifically to the setup in this video, Jim's concern (which I share) is that the frequency (and signal filtration) of the measurement devices might cause us to miss the actual peak force, especially in the test(s) likely to have a sharper force curve (the ones with less or no dynamic rope in the system).  There's no way to conclusively know if this is happening without looking at a plot of the data, and measuring it at a higher sampling frequency if possible (some educated interpretation of a data plot may be sufficient).  The problem is that simply recording that isn't trivial, because high frequency data acquisition isn't a cheap or simple tool.

The good news is that we know there is a practical limit on the frequency response of the squishy human body, so the force generated in a "shock loading" scenario cannot be unbounded.  Whether that unbounded force is above safe limits driven by gear capacity and health risks is the important question; I've done some back-of-the-envelope numerical simulation (posted elsewhere on MP) that indicates that it might be below maximum load capacity for climbing slings, etc., but that's not the most robust evidence.

It's situational.  If you have a follower being belayed directly from the anchor with 100ft of rope out and the anchor extends 1 foot, it doesn't matter at all (at least, in terms of loads on the remaining anchor points).  But that's kind of a given.  There are other scenarios, however, where extension could be a lot more insidious, the classic one being a climber belaying a second off their harness, while being tied to the anchor with static materials, and the anchor extends when the second falls off and has the rope stretched out.  This will yank the mass of the belayer against the anchor with a possible higher-than-1g acceleration and could cause a very high load on the remaining anchor pieces.  It's a complex mechanical system to simulate or test though.  There are lots of ways to guard against this, including (but not limited to) tying in with a sufficient length of dynamic material and knots, belaying off the anchor, building anchors that don't extend, separating the belayer's mass from the anchor, etc.  Not all situations are conducive to all solutions.

Note that getting blasted off your stance by an extending anchor has other risks too, even if the anchor doesn't ultimately fail.  For example, having your partner drag you over and/or pin you against the edge of the belay ledge would not be that cool.

So, "get strong placements and tie yourself to them" and "build the anchor and your connection to it with the rope and not static materials" and you have the holy grail found for anchor security as far as I will ever be concerned.

ps. I think your "kind of a given" about extension when belaying the second not mattering is something that newbies worry about immensly due to the unfortunate use of acronyms (ie SERENE) for anchor building. It is this introduction of complexity that is worthless, imo. It all comes down to the strength of the placements and the use of the dynamic features of the rope and that's it.

No one can build a good anchor out of bad placements. You either know from experience that they are totally strong or you better climb like its "psychological" at best and don't make believe you can build a solid anchor out of crap.

I think ryan quickly says something about his load cells.  From his other videos he has load cells that are suitable for this type of measurements.  In particular i think one of the load cells is ​Rock exotica load cell​​​ 

Al Pine wrote: I’m still missing the takeaway. Ryan is tied in with a dynamic rope and falling a short distance onto a static sling attached to a bolt. He didn’t get hurt or break his back. Is that it?

Well not really since the proposition was "is shock loading a myth?". We see clearly that a sliding X in their scenario had absolutely no functional benefit and when it failed tripled the force on the remaining piece. So "shock loading" as they understand it isn't a myth is the takeaway.

hillbilly hijinks wrote: So, "get strong placements and tie yourself to them" and "build the anchor with the rope and not static materials" and you have the holy grail found for anchor security as far as I will ever be concerned.

ps. I think your "kind of a given" about extension when belaying the second not mattering is something that newbies worry about immensly due to the unfortunate use of acronyms (ie SERENE) for anchor building. It is this introduction of complexity that is worthless, imo. It all comes down to the strength of the placements and the use of the dynamic features of the rope and that's it.

No one can build an anchor out of bad placements. You either know from experience that they are totally strong or you better climb like its "psychological" at best and don't make believe you can build a solid anchor out of crap.

^^^^^

In the video, I think they said the sampling rate was 500 Hz. We don't know what the filtering is but it would be usual to low pass filter to 500 Hz or something less. Roughly, If I was designing the instrument, I would try to low pass filter at something like 250 Hz. To detect a signal you need to digitally sample at twice the frequency of the signal so sampling at 500 Hz will detect if there is a 250 Hz signal but can't give you the magnitude. To identify that signal you would need to sample over some time window and look for it in the frequency domain by doing a FFT. However, to have some idea of the magnitude, you need to sample at least 10x the frequency so the instrument should be able to measure the magnitude of a 50 Hz signal or the peak for everything up to 50 Hz combined. Again, if you wanted to look for just the 50 Hz signal, you would need a time history. The natural frequency of the human body is in the 5 to 10 Hz range depending on posture, size, ect. So with a 500 Hz sample rate and a 10 Hz natural frequency, they have a good chance of getting a good approximation of the peak. Digital sampling never captures the absolute peak but the error is probably negligible.

The 22 kN generally required for climbing anchors has nothing to do with the forces a falling climber can generate but with the survivable force a climber can endure. As mentioned earlier, this is in the 12+ g range. 80 kg person at 13 g is (80*9.8*13) about 10 kN. For an anchor, you need to double this rope force.

I don't think it is possible for a climber to generate a 10 kN force on the rope in any reasonable configuration. I think the tests in the video generally agree with this opinion. If I followed the video right, they got a maximum of about (2.4/0.8) 3 g's. Much less than the 13 g's used for climbing gear design.

In the video, they mentioned survived rocket sled tests of 46 g's. This is front to back with a fully supported body. In the vertical, 46 g's would collapse your spine and probably scramble your soft tissue innards. Ejection seats produce 13 g's but usually cause spinal compression injuries.

The hassle is a lot of the the strain guages made for weighing/lifting applications may have quite high sampling rates but the signal is reduced to a speed that the display can cope with and that´s the digitally recorded peak, for one I have it´s every 10secs refreshed so easy to miss the peak in short duration events . My other two sample at 500hz but log at 100hz so you can get much nearer the true peak just on the data, starts to work my laptop though!

Dan Merrick wrote: In the video, I think they said the sampling rate was 500 Hz. We don't know what the filtering is but it would be usual to low pass filter to 500 Hz or something less. Roughly, If I was designing the instrument, I would try to low pass filter at something like 250 Hz. To detect a signal you need to digitally sample at twice the frequency of the signal so sampling at 500 Hz will detect if there is a 250 Hz signal but can't give you the magnitude. To identify that signal you would need to sample over some time window and look for it in the frequency domain by doing a FFT. However, to have some idea of the magnitude, you need to sample at least 10x the frequency so the instrument should be able to measure the magnitude of a 50 Hz signal or the peak for everything up to 50 Hz combined. Again, if you wanted to look for just the 50 Hz signal, you would need a time history. The natural frequency of the human body is in the 5 to 10 Hz range depending on posture, size, ect. So with a 500 Hz sample rate and a 10 Hz natural frequency, they have a good chance of getting a good approximation of the peak. Digital sampling never captures the absolute peak but the error is probably negligible.

The 22 kN generally required for climbing anchors has nothing to do with the forces a falling climber can generate but with the survivable force a climber can endure. As mentioned earlier, this is in the 12+ g range. 80 kg person at 13 g is (80*9.8*13) about 10 kN. For an anchor, you need to double this rope force.

I don't think it is possible for a climber to generate a 10 kN force on the rope in any reasonable configuration. I think the tests in the video generally agree with this opinion. If I followed the video right, they got a maximum of about (2.4/0.8) 3 g's. Much less than the 13 g's used for climbing gear design.

In the video, they mentioned survived rocket sled tests of 46 g's. This is front to back with a fully supported body. In the vertical, 46 g's would collapse your spine and probably scramble your soft tissue innards. Ejection seats produce 13 g's but usually cause spinal compression injuries.

Actually Nyquist says you can also get amplitude, in addition to frequency and phase,  by sampling by more than twice the signal frequency.  Your 10X sample rate makes seeing the signal on an oscilloscope easy.

climber pat wrote:

Actually Nyquist says you can also get amplitude, in addition to frequency and phase,  by sampling by more than twice the signal frequency.  Your 10X sample rate makes seeing the signal on an oscilloscope easy.

Nyquist is right. Sampling at twice the frequency of a continuous signal can identify the frequency and amplitude of the signal so long as you deal with the possibility of aliasing. The signal has to last long enough (be continuous) and you have to take enough samples to run an accurate FFT. In the drop test, the signal is a damped pulse rather than a continuous signal and without oversampling, you are very likely to miss the peak with 2x sampling. Whenever you need accurate waveform reproduction your only recourse is to oversample at a rate beyond Nyquist's 2 times minimum, typically 10 times or more. 

The dynamometer isn't running a FFT looking for a signal frequency and amplitude, It is just discretely sampling, displaying the current value and storing the maximum value.

I can't find in the specs what the Rock Exotica Enforcer sample rate is. The Linegrip unit states that the high sample rate is 40 Hz which might be a bit low.

Please correct me if I'm wrong.

Dan Merrick wrote:

Nyquist is right. Sampling at twice the frequency of a continuous signal can identify the frequency and amplitude of the signal so long as you deal with the possibility of aliasing. The signal has to last long enough (be continuous) and you have to take enough samples to run an accurate FFT. In the drop test, the signal is a damped pulse rather than a continuous signal and without oversampling, you are very likely to miss the peak with 2x sampling. Whenever you need accurate waveform reproduction your only recourse is to oversample at a rate beyond Nyquist's 2 times minimum, typically 10 times or more. 

The dynamometer isn't running a FFT looking for a signal frequency and amplitude, It is just discretely sampling, displaying the current value and storing the maximum value.

I can't find in the specs what the Rock Exotica Enforcer sample rate is. The Linegrip unit states that the high sample rate is 40 Hz which might be a bit low.

Please correct me if I'm wrong.

It has 2 sample mode, 2 samples per second and 500 samples per second.  Rock Exotica Quick Reference Card The card says the fast most is designed for capturing dynamic events.  The company manufactures some climbing gear so I have to believe the are cognizant of the issues associated with this type of data collection.

Actually Nyquist theorem is independent of FFT, that is just the most often method of calculating these measurement because it is computationally efficient.  Conceptually you could calculate signal's characteristics another way.  In fact, DSP engineers get the phase/amplitude/frequency  for decoding symbols and bits that change as often as every cycle.  BTW, I am agreeing that you want to oversample but I suspect with enough computation you might not have too.  I also agree with your 10X oversample rate.  I lost the measurements from my own load-cell but 100 Hz is plenty to tell what is going on.  

Dan Merrick wrote:

Nyquist is right. Sampling at twice the frequency of a continuous signal can identify the frequency and amplitude of the signal so long as you deal with the possibility of aliasing. The signal has to last long enough (be continuous) and you have to take enough samples to run an accurate FFT. In the drop test, the signal is a damped pulse rather than a continuous signal and without oversampling, you are very likely to miss the peak with 2x sampling. Whenever you need accurate waveform reproduction your only recourse is to oversample at a rate beyond Nyquist's 2 times minimum, typically 10 times or more. 

The dynamometer isn't running a FFT looking for a signal frequency and amplitude, It is just discretely sampling, displaying the current value and storing the maximum value.

I can't find in the specs what the Rock Exotica Enforcer sample rate is. The Linegrip unit states that the high sample rate is 40 Hz which might be a bit low.

Please correct me if I'm wrong.

Dan, You might be interested in this.  This paper Sinusoidal Frequency Estimation Based on Time-Domain Samples shows how to estimate the frequency of a real signal with either 3 or 4 samples and a complex signal with 2 complex samples. Once you have the frequency one could fairly easily generate a sampled with a higher sample rate sine wave that passes though the original sample points and then compute the max value of the points to get the peak load.  All this assumes the load is a sine wave which I think is very likely.  

EDIT:  This might be an interesting way to estimate a load that is above the capacity of the load cell to measure.  

Dan Merrick wrote:

Nyquist is right. Sampling at twice the frequency of a continuous signal can identify the frequency and amplitude of the signal so long as you deal with the possibility of aliasing. The signal has to last long enough (be continuous) and you have to take enough samples to run an accurate FFT. In the drop test, the signal is a damped pulse rather than a continuous signal and without oversampling, you are very likely to miss the peak with 2x sampling. Whenever you need accurate waveform reproduction your only recourse is to oversample at a rate beyond Nyquist's 2 times minimum, typically 10 times or more. 

The dynamometer isn't running a FFT looking for a signal frequency and amplitude, It is just discretely sampling, displaying the current value and storing the maximum value.

I can't find in the specs what the Rock Exotica Enforcer sample rate is. The Linegrip unit states that the high sample rate is 40 Hz which might be a bit low.

Please correct me if I'm wrong.

You're absolutely right. Nyquist doesn't really havre much to do with single transient events.

Nyquist ascerts that you can capture frequency and amplitude by sampling at twice the frequency of interest, but that's really for repeating (continuous) signals. Transient spikes tend to contain very high frequency content, and to capture true peaks in a transient, you need to sample at a much higher rate. For some laboratory failure analysis (e.g. tensile testing), it might not be a big deal, since the load is applied gradually (low frequency content). But for a rapidly applied load, sample rate matters.

Andrew Krajnik wrote:

You're absolutely right. Nyquist doesn't really havre much to do with single transient events.

Nyquist ascerts that you can capture frequency and amplitude by sampling at twice the frequency of interest, but that's really for repeating (continuous) signals. Transient spikes tend to contain very high frequency content, and to capture true peaks in a transient, you need to sample at a much higher rate. For some laboratory failure analysis (e.g. tensile testing), it might not be a big deal, since the load is applied gradually (low frequency content). But for a rapidly applied load, sample rate matters.

Rapid is context sensitive.  For drop yesy 100 hz is a fast enough sample rate to see the forces.

the required sampling rate is going to depend on what you are testing.  in some cases 500hz will be overkill and in other cases it would be underkill.  for a situation like this, where they probably don't really know what they need, it helps to do multiple tests of the same situation but at different sampling rates to see what type of attenuation you are getting.