Discussion Topic |
|
This thread has been locked |
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Topic Author's Reply - Feb 12, 2007 - 11:35am PT
|
Raymond, it is certainly possible to dismiss all such considerations as "irrelevant." So let me recast my thoughts.
I was trying to estimate the rate of anchor failure in a "practical" situation, that is, as a climber does it on a route. I have two inspirations for this, one is a now forgotten Italian climber who had expert climbers place pitons, and then tested how good those placements were (much worse than expected) and jstan's test of passive gear, which ultimately convinced him, a self described conservative climber, that such gear was trustworthy.
The ANAM has at least a brief description of the accidents, even in cases where the victims died as post accident analysis, while incomplete, can help in understanding the causes. Also, the ANAM has a very long history with many accidents described. I think it is a must for every climber interested in avoiding similar accidents, it should be read as "lessons learned." It is invaluable.
Here is my reasoning: most anchors are placed and never tested in a "worst case" fall scenario. ANAM has a statitic on how many accidents occur because of a "slip or fall on rock". These slips and falls result in injuries severe enough to require rescue of the climbing team and notification of the authorities.
Assumption 1, if you are climbing on rock in a situation where you could fall, you are using roped protection.
So in these accident cases I assume that all of the protection held, otherwise, the accident would be reported as caused by a "pulled anchor".
Assumption 2, the forces involved in a "pulled anchor" caused accident were the same (or very similar to) the forces involved in a "slip or fall on rock" accident.
So, the ratio of the "pulled anchor" to the sum of the "pulled anchor" and "slip or fall on rock" gives you the fraction of anchors set on lead that failed under forces large enough to cause injury.
Not all the accidents read like this of course, but I am trying to make an estimate.
Now on the psychology of psychological gear, we all place gear like that at one time or another in our career, but you know that if you stop to put gear in that you should really attempt to get something in that will hold, otherwise you've wasted energy and set yourself up for a potentially dangerous situation of doing something that will cause you to fall on gear that wont hold. One of my all time favorite Largo quote is "if you're going to go for it, make sure there is an 'it' to go for," which I would expand to include an anchor onto which you can make a "gravity assisted" retreat. Placing pro and saying "that's psycological pro" is all too common, sort of like not wanting to sit back on your anchor at a belay.
Many people put in bomber belays every time, they are lucky to be climbing in areas where it is possible to do. Many people think they are putting in bomber belays. How many people actually do put in a bomber belay? how many people actually put in bomber pro?
I would suggest that we do it often, but not better than 99 in 100. That is my estimate based on the rather broad set of assumptions made and the statistics as they are available. 1% failure rate on anchors is probably much larger than it needs to be, or is "safe."
I don't think this is the end of the story. I think that there is a well defined path, pointed out by our unknown Italian and jstan, to try to determine just how well we set anchors. After all, most of us are self taught. We could use a little critique, we could stand to learn the limits of our own teachings.
|
|
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Topic Author's Reply - Feb 12, 2007 - 12:01pm PT
|
I think I may have overstated the precision of the analysis in the above post...better to say that:
the statistical analysis suggest that the anchor failure rate for anchors subject to high force falls is in the range from 1-in-10 (10%) to 1-in-100 (1%).
Many would decry the 10% number, but I would challenge them to explain their objection. I think the 1% number might be a low end estimate.
Any other estimates out there?
|
|
jstan
climber
|
|
|
Feb 12, 2007 - 03:00pm PT
|
A calculation I did for our good friend G Gnome and posted earlier may be worth repeating here. On the other hand it may not.
If:
1. you climb every day for thirty years
2. you take ghastly falls on two different anchors each day
3. you want to know with a confidence of between 95% and 98% that you will not get the chop,
then each of your anchors has to have a chance of failing less than or equal to one in a million.
Under these rather onerous conditions, if the chance of failure gets much above one in a million, you start dying. On the other hand if you rarely, if ever, take ghastly falls it is a very good thing. Another point. When consequences are as serious as they are here, people generally insist upon having confidence greater than 95% if at all possible. My personal opinion: any time you look at an anchor and think it has a chance of failing greater than 1%, however you determine that, start thinking about going for a beer(or a latte).
In order to determine my specification for the force an anchor had to hold, I did tests in addition to those mentioned above. I built a little load cell using 8mm perlon and calibrated it on a tensile machine. Then I threw a 165# dufflebag full of shale off the cliff and caught it with 70 feet of rope after it had fallen free about 40 feet. (At that particular cliff any fall of 40 feet will probably result in Substantial Personal Customization[SPC].) When using a waist belay the force on the top anchor was about 500#. When I tied the rope to a tree the force doubled to 1000#. At that point I was sold on two things:
1. A waist belay is inherently dynamic. I would never go directly into a pin with a belay device.
2. My design target of 3000# was adequate - at that cliff.
This was done using Mammut ca. 1972 or thereabouts.
There is no reasonable way actually to measure failure probabilities so to a degree this is all numerology. There is a conclusion however. If you want to climb for a long time, think very hard before you get deeply into a situation you perceive as "Iffy". Your perception is your alarm bell.
|
|
Jingy
Social climber
Flatland, Ca
|
|
|
Feb 12, 2007 - 08:32pm PT
|
Dude, Russ... That decent was totally calculated. It didn't look like it but it was, indeed, completely controlled, except that part where I came off the pad at the very end.
|
|
TradIsGood
Happy and Healthy climber
the Gunks end of the country
|
|
|
Feb 12, 2007 - 08:59pm PT
|
Ed, and jstan, I think you have a great message here. We aren't dying in great numbers because we aren't falling in great numbers - at least on trad anchors. And those who are falling a lot are probably doing it on solid anchors.
Ed, any stats on survivors of blown anchors? I'm guessing it's pretty much game over.
BTW Ed, my brother (M.D.) did some research on climber injuries in WV. IIRC, it was based on a survey (or hospital reports?). But his focus was on types of injuries, including death. I think he was looking at the probability distribution of injury given an event and not the probability of an event, given a climb.
If you want that type of info let me know...
Another thought here. The falls involving blown pieces may not be over-estimate the probability of failure of an anchor component, since most would choose a bomber location for building an anchor, but might toss in that psychological piece from an iffy stance. That opinion is based on lower grade climbing at the Gunks where most anchors are built on a nice safe ledge.
|
|
raymond phule
climber
|
|
|
Feb 13, 2007 - 05:36am PT
|
"Many would decry the 10% number, but I would challenge them to explain their objection."
I would challenge you to actually show some meaningsfull numbers.
What do we actually try to show? I place all kinds of anchors. I consider some bomber, i.e. blue camelot in perfect crack and I considers some as maybe body weigth, i.e. tied of knifeblade. Everybody place protection on lead that they dont belive is going to hold a long fall.
So what do we want to know? Do we want to have some kind of statistics that estimate how likely it is that percived sound anchors actually hold a fall? I want to know this. I already know that my tied of knife blade or small nut is not going to break a 10 m fall. I have pulled some gear but I have not trusted a single one of those to actually hold a fall.
The statistics given include all kind of anchors some percived good and some percived bad. You also have no idea how many pieces that actually hold falls. Many climbing scenarios is that a fall is going to result in no injury if the anchor holds but result in injury if the anchor pulls.
We simple dont have enough information to make a meaningsfull statistic.
I agree with you though that we simple dont know how many belay or lead anchors that actually would hold. The falls are way to few in trad climbing to actually get any real info.
|
|
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Topic Author's Reply - Feb 13, 2007 - 08:47pm PT
|
Raymond... you are determined not to accept the estimate assumptions, so be it...
I'm not sure what your objections are, but I'll try:
What do we actually try to show? I place all kinds of anchors. I consider some bomber, i.e. blue camelot in perfect crack and I considers some as maybe body weigth, i.e. tied of knifeblade. Everybody place protection on lead that they dont belive is going to hold a long fall.
I assume that, in general, you will place "bomber pro" if you have the opportunity. I am not sure about your last statement, however, about everyone places pro they "don't believe is going to hold." What would you estimate the number of times you have done this? compared to how many placements? And how many times have you fallen on that gear?
So what do we want to know?
What I was trying to estimate is the actual failure rate of anchors. Simply put, look at all accidents involving falls on rock, then see how many of them had an anchor failure. I have made no assumption regarding the quality of the anchors, however, where the anchor failure was a primary contribution to the accident, it was recorded as such in the ANAM.
The statistics given include all kind of anchors some percived good and some percived bad. You also have no idea how many pieces that actually hold falls. Many climbing scenarios is that a fall is going to result in no injury if the anchor holds but result in injury if the anchor pulls.
The assumption is that those falls which were severe enough to result in injuries were falls on gear. In those falls the anchors did not fail. If you say that many of these accidents are not appropriate to include in this estimate, then the estimated rate of anchor failures goes up.
We simple dont have enough information to make a meaningsfull statistic.
Is that an assertion, or are you interested in exploring the issue? The whole point about making an estimate like this is to not only see if the number makes sense, but also to define what assumptions lead to the conclusion, then question those assumptions.
The assumption about whether a leader knowingly puts in good or bad pro is an interesting one, and may be a part of the statistic in the numerator. But it doesn't matter what the answer to that question is if the leader falls and the pro pulls.
I am interested in the question regarding my placement, if I put in "bomber pro," just how bomber is it?
The accidents listed in the numerator most likely have a mix of such placements, known bad, and assumed good.
One way to test this would be a la the Italian climber's scenario, ask a bunch of experienced climbers to place pro with a variety of types in a variety of placements and then subject them to test.
When you think about that proceedure, you realize that the historic accident data might have done just that to an unwitting participant. Some of those participants paid with their lives, I feel it is important to try to learn as much from those accidents as I can, it is their final legacy to climbing, and it might help to save future lives.
|
|
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Topic Author's Reply - Feb 14, 2007 - 01:13am PT
|
Something odd struck me about Raymond's statement above:
"...Everybody place protection on lead that they dont belive is going to hold a long fall..."
"...I already know that my tied of knife blade or small nut is not going to break a 10 m fall..."
and Peter Haan's comments about Bridwell:
"Productive yet very careful and wise, he really set a great example for us younger climbers, who might have been willing to be far riskier in our endeavors. He never seemed to teeter between going on living and risking the current moment for a final stab at deepest meaning. In other words, he never solo-unroped anything at all to my knowledge and figured out how to protect stuff somehow, always."
When you get to a point where the pro is jingus you can do 1 of 3 things:
1) continue and hope you don't blow it,
2) retreat,
3) pound a bolt in (only if you're not on an established route).
It's not written anywhere that you must continue.
|
|
raymond phule
climber
|
|
|
Feb 14, 2007 - 08:26am PT
|
"Raymond... you are determined not to accept the estimate assumptions, so be it..."
I cant accept your assumption 2.
An example, take Yosemite 94.
10 accidents and 1 anchor failure. Lets assume that all accidents where reported and all accidents where due to roped falls. This might make a large error but we ignore it.
Thus we know 9 accident where the anchor hold and 1 where the anchor failed.
We have no idea how many anchors that actually hold a fall. The forces in those falls could have been higher than the falls that resulted in accidents. One common accident in Yosemite is that someone fall on the nutcracker mantle. Not a lot of forces on the anchor in that case. Accidents is not just dependent on the forces in the fall, ledges are probably a more common problem.
So we know about 1 anchor failure but have no idea about the number of actuall falls. Why should I belive any statistics taken from this data when we know almost nothing?
The next problem I have is that we dont know why the anchor failed. This could be solved though by reading reports.
The failed anchor could have been a failed beak pic on Reticent. It could have been a old pin on space bable. It could have been a small nut placed an a run out route on glacier point apron. It could ofcourse also be a red camelot placed on Moby dick.
Is it really suprising that beaks fail once in a while, that a small nut fail on a run out route? Does this really say anyting about the failure rate of a belay anchor if the anchors is considered good?
The obvious answer to all this according to me is that we cant draw any meaningsfull statistic from data given.
"I assume that, in general, you will place "bomber pro" if you have the opportunity."
Of course but do we always have that opportunity?
"I am not sure about your last statement, however, about everyone places pro they "don't believe is going to hold." What would you estimate the number of times you have done this? compared to how many placements? And how many times have you fallen on that gear? "
I sometimes place pro that I belive could hold a short fall but not a long fall. I try to place pieces often in this case. Doesn't everyone that have climb a not perfectly protected climb placed gear that the are not sure is going to hold a long fall? I dont know how often. No I haven't fallen on that gear free climbing but have riped pieces aiding.
I am suprised that you seem to disagree with me on this as I am sure that you are climbing climbs with less than perfect protection.
So what do we want to know?
"The assumption is that those falls which were severe enough to result in injuries were falls on gear. In those falls the anchors did not fail. If you say that many of these accidents are not appropriate to include in this estimate, then the estimated rate of anchor failures goes up. "
You convienently ignore all falls that dont result in accidents.
"Is that an assertion, or are you interested in exploring the issue?"
I am interested about the issue.
"The whole point about making an estimate like this is to not only see if the number makes sense, but also to define what assumptions lead to the conclusion, then question those assumptions. "
Yes, I have questioned the assumptions and results. I see no useful info in the numbers, sorry.
"The assumption about whether a leader knowingly puts in good or bad pro is an interesting one, and may be a part of the statistic in the numerator. But it doesn't matter what the answer to that question is if the leader falls and the pro pulls. "
This depends on what question you want to answer. I want to know how often good looking pieces actually fail. Not how often people climb above there head on badly protected climbs.
"I am interested in the question regarding my placement, if I put in "bomber pro," just how bomber is it?"
I agree, see above.
"One way to test this would be a la the Italian climber's scenario, ask a bunch of experienced climbers to place pro with a variety of types in a variety of placements and then subject them to test. "
Could be a really good test and the result would be interesting.
|
|
jstan
climber
|
|
|
Feb 14, 2007 - 11:08am PT
|
RP:
I think this is very useful thread because of what I have seen climbers say on another site. They ask each other whether or not they have yet, “taken a real whipper”, and they seem to feel “ you are not climbing to your maximum if you aren’t falling.” Since the two questions appear together I also suspect they mean regular very serious falls. Perhaps people feel since ropes and protection are supposed to hold falls, you can assume they will. Sort of cut and dried like in the closely controlled environment of a climbing establishment. I think none of us feels this is good and many attribute it to people going into the out of doors untrained for that environment.
Now the people on ST, by and large, have been around a long time and have seen and done just about everything(several times even). When I started climbing people, like those on ST, taught me how to think about risk. Frankly knowing how to think about the risk is more critical to climbing than is knowing how to use your feet. And way more critical than what routes you have been climbing. I think people coming out of a climbing establishment sometimes don’t even know risk is an issue.
So in this thread we have very experienced people discussing the ways they themselves handled that issue. There are many unknowns. But that is what life is. Learning how to handle unknowns. It may even be that learning how to handle risk while climbing, teaches us how to handle risk in other parts of life? So what is the output of all this? The youngsters showing up are probably going to learn what they need to know from people like yourself and the others on ST. The more we have thought about this the better prepared we will be to help them.
I have, again, to thank Chloe(?) for making me do a better calculation. If you want to climb a long time and you want to have serious regular falls on your anchors, you have to have reliability beyond anything you can measure. Ergo: Not falling is your best safety. That is what I was taught. I think we need not to forget it.
PS
I have even seen discussions where "not falling" was considered. "an old paradigm that is no longer applicable".
|
|
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Topic Author's Reply - Feb 14, 2007 - 11:12am PT
|
good comments, thanks....
It is not generally appreciated how large the forces actually are on falls. While we have an image of the "two screamer" fall ("...the climber fell so far he screamed twice...") you don't have to fall all that far to generate huge forces. In fact, most fall factor 2 falls are relatively short, occuring before the first piece is placed. There is a 41% probability of a limb fracture from a 10' fall (see more statistics here).
A fall onto a ledge is bad because the forces are large, even from the seemingly modest height of the mantel on Nut Cracker.
Reading the reports will reveal a lot about what pro failed. I haven't had the time to do it, but I intend to take it to the next level. Once again, my general impression is that the surviving victims were unaware that the piece they placed would fail, and of the consequences of that failure. In some of the accidents the pieced was placed and weighted, then blew. This is hardly a fall, but it does get at the discussion of how good the pro placements are (in that case, pretty terrible).
It's not that I want to ignore falls that don't result in accidents, it is just that I don't have any information at all on those falls. I don't fall that often. The last really big fall I took is when I blew the crux on the first pitch of Knob Job. I got the sequence wrong, I had placed a nut as high as I could in the left hand crack, and then worked right to get under the roof. I was pulling hard in a lay-back, realized I didn't have it right, and decided to back off to think about it. But that was too hard for me to pull off and off I went, probably not more than 10'. I managed to avoid hitting any of the knobs on the way down. Unreported.
My friend Steve was climbing in the same area later in the year and fell on something, hitting a knob with his ankle, spraining it. Unreported also. In both cases the gear held. I imagine that there are a lot of falls like this, but I don't know how to capture that, we don't report on that sort of thing, it is true.
I appreciate your interest in the details of the probability of a failed anchor. Setting aside the basis for the numbers, if the failure rate is as high as 1%, then actual testing and demonstration will be difficult to do, simply because testing thousands of anchor placements (which is what you would have to do to get a statistically relevant answer) probably just isn't going to happen. In the end we have to resort to some analysis to understand how often these relatively rare events happen.
How uniform is the training for placing an anchor? what factors in to a choice of a particular type of pro for protection (e.g. nuts, cams, etc.). Is experience a factor in the probability?
When such a rare thing does occur in real life, we need to extract the maximum information possible.
I have very strong opinions regarding this topic all of which may be incorrect when confronted by actual data. But by explaining the reasons for those opinions, and being criticized by patient souls like you, I can understand better what is actually happening.
Thanks for you time to expand your thoughts.
|
|
WoodySt
Trad climber
Riverside
|
|
|
Feb 14, 2007 - 12:30pm PT
|
I've not been following this thread until reading the last few comments today. From reading those comments, it's clear that unreported accidents are a significant variable and problem in coming to a satisfactory conclusion.
If I'm an example of unreported and somewhat serious falls, and if my example is common, there's an immense amount of unknown data out there. I've had four such falls in the last few years, one that put me in the hospital, that were unreported.
|
|
TradIsGood
Happy and Healthy climber
the Gunks end of the country
|
|
|
Feb 14, 2007 - 01:19pm PT
|
So far the focus has been on the probability of failure given a fall (serious fall). P(failure|fall). But, as was pointed out, we do not know the count of the set of falls. Ed's approach was to estimate the size of the fall set from the set of climbs on anchors (?).
Alternatives:
P(failure|anchored lead climb),
P(failure|climber-day), (or year)
etc. But these seem to stray from the focus on the conditional probability of a random anchor to fail given a fall.
|
|
Russ Walling
Social climber
Out on the sand, Man.....
|
|
|
Feb 14, 2007 - 01:56pm PT
|
Woddy writes: If I'm an example of unreported and somewhat serious falls, and if my example is common,
With all due respect Woody.... your ripper streak is quite amazing and I would think is not that common. If it was that common, there would be none of us left!
|
|
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Topic Author's Reply - Jan 25, 2011 - 08:32pm PT
|
bump for TrundleBum
|
|
|
SuperTopo on the Web
|
Let us know!