ACR Anchor Method?

jktinst wrote: In a lead fall + hanging belayer + blown-leg test, sure, it makes sense that with the belayer's weight clipped directly to the anchor's central point, you'd get a higher load multiplication than if it was tethered through a dynamic tether. Basically, this makes the case for always tethering with a dynamic tether, be it the rope, a Béal dynaconnexion, a Petzl Connect, a length of dynamic rope with a Kong Slyde, etc. (which is what I do and which has been discussed elsewhere on this thread). I also still feel that it confuses the issue to use as your reference point for load multiplication in a blown-leg scenario the simple static weight of the hanging belayer. From what I understand, Ioad multiplication as a result of anchor extension usually means the multiplication factor between the fall getting arrested on the full/intact anchor and the same fall arrested with one anchor leg blowing. Talking about worst case scenarios: how about comparing a quad, a cordelette and a statically-equalized dynamic rope anchor (à la rgold) on a 2-leg anchor with legs of significantly different lengths and a 5kN fuse on each leg (with the anchor pre-equalized for a straight-down load, not pre-equalized in a direction other than down)? We're not talking about manufacturing gizmos that may fail or be misused here. Just having useful information to help understand in what circumstances some types of anchors might be safer than others. Finally, the main reason I started posting on this thread (despite determinedly abstaining from it when it first started) is because I am still trying to understand why your load distribution results (no blown leg) for the simple two-leg sliding X are different from the DAV's. Several years back, you reported a 1:3.7 distribution ratio for the sliding X pre-equalized in a direction other than down and tested on a drop tower with the weight on a track, which I found mind-bogglingly uneven. Could it have been a typo? On this thread, your numbers are more reasonable: 1:1.7 or 1:1.9, depending on the post, and you're also reporting (for the first time, as far as I know) a comparison between different equalization testing methods. With the DAV tests, from what I can see without understanding German, it seems that they got a 1:1.3 distribution when the weight mimicking the leader was dropped from 3m up and 1m to the left of the sliding X's central point (edited to add: again, with the weight mimicking the belayer hanging statically from the anchor) . Any discussion of the possible reasons for the differences would be most welcome. So would details of the 4 different methods you compared for the two-leg sliding X. PS: I think I've seen your "3-leg sliding X" referred to as a "sliding W" before. Has anyone else?

The reasons for not testing with dynamic elements added into the belayer/belay system is that a) this is not worst case, b) there is no universal standard for the dynamic element c) the objective is to test belay system effectiveness, not dynamic elements which may or may not be included in the system.

The numbers given in this thread are the actual load splits on sliding systems derived from a force which makes the belay equalise. That is if the force comes from an unknown direction when you build the belay. If you "pre-equalise" the belay by anticipating the direction of force you will get lower values BUT you have not proved the belays ability to equalise, only your ability to "pre-equalise". If a belay is claimed to be "self-equalising" then it must do exactly that.

You´ll have to give me a reference for the previous post, my memory isn´t that good!

The DAV test is with the belay to some extent pre-equalised and so they get lower results, they are not the only people to have done this and claimed the belay equalises better than it really does. With practice you can set the karabiner in a sliding X wherever you want to achieve whatever result you desire up to the real limit. After about ten tries you can get perfect equalisation.

The different test methods all have different advantages and disadvantages and with the exception of drop-testing give the same results within reason, to test the ability of a belay to dynamically equalise the tilt test is the best by far, the offset drop is very unreliable if not virtually worthless since the end result is dependent on the initial positioning of the central karabiner.

The link formatting here seems to chop off the rockclimbing.com URL so I don't think that the active link to the precise page would work.

The post I was referring to is on p.2 of The Lab forum. Thread title: Stupid Simple Elette (last post on thread: Nov. 10, 2011).

Regarding the DAV climbing wall drop tests, with the weight mimicking the belayer hanging from the anchor's central point and the one mimicking the leader falling from 1m off to the side, I don't see how it can be claimed that their pre-equalizing the anchor skewed their results towards a more even distribution. I understand that having dynamic elements in the system (dynamic rope and tether, swinging and bouncing of the leader weight, etc.) introduces more parameters that are difficult to manage and standardize but I would think that the main point of this exercise is to have tests that are as representative as possible of real-life situations and, in that respect, it seems to me that those DAV tests are very representative.

By comparison, drop tower tests where you manually pre-equalize the anchor in a direction other than that of the weight's vertical track to force the anchor to equalize while aligning with the track under load do not seem as representative to me. It's as if the leader fell on the anchor in a direction other than down. Or, to be a bit more realistic, it's as if the leader fell from precisely above the anchor's mid-point (ie the lowest possible position that the central point can reach) while the belayer, anticipating this fall, jumped or swung himself off to the side to get out of the way, thereby pre-equalizing the anchor in a direction other than straight down. Not impossible but not a particularly typical scenario either.

jktinst wrote:The link formatting here seems to chop off the rockclimbing.com URL so I don't think that the active link to the precise page would work. The post I was referring to is on p.2 of The Lab forum. Thread title: Stupid Simple Elette (last post on thread: Nov. 10, 2011). Regarding the DAV climbing wall drop tests, with the weight mimicking the belayer hanging from the anchor's central point and the one mimicking the leader falling from 1m off to the side, I don't see how it can be claimed that their pre-equalizing the anchor skewed their results towards a more even distribution. I understand that having dynamic elements in the system (dynamic rope and tether, swinging and bouncing of the leader weight, etc.) introduces more parameters that are difficult to manage and standardize but I would think that the main point of this exercise is to have tests that are as representative as possible of real-life situations and, in that respect, it seems to me that those DAV tests are very representative. By comparison, drop tower tests where you manually pre-equalize the anchor in a direction other than that of the weight's vertical track to force the anchor to equalize while aligning with the track under load do not seem as representative to me. It's as if the leader fell on the anchor in a direction other than down. Or, to be a bit more realistic, it's as if the leader fell from precisely above the anchor's mid-point (ie the lowest possible position that the central point can reach) while the belayer, anticipating this fall, jumped or swung himself off to the side to get out of the way, thereby pre-equalizing the anchor in a direction other than straight down. Not impossible but not a particularly typical scenario either.

That part of that thread was discussing Neoshades bizarre claim that the setup on the right-hand side of the photo below, ".... a handy little truly-self-equalizing leg on a cordalette (or equalette in this case). :-
Neoshade
I´m not so suprised I got such poor load sharing, that idea suffers from the same problems as the Alpine Equaliser in that to move the cord has to slide through ALL the karabiners (or in his case karabiner and coats hooks).

I shall repeat (or try to make clearer). I tested the ability of dynamically equalising anchors to distribute the load across the points of the anchor. You can test any "real life" situation you wish.

Copied from RC.com:

"...As usual (for us anyway) we set up a 60° two-point anchor and start pulling with the centre-point offset to the side (and line of pull) and see what the loads are on each anchor point. This gives the ratio of the unevenness of the load you will get in practice with dynamic equalisation (if that´s the word to use).
I pulled at 6kN with 6mm cord.

Simple equalisation (just on a single strand, no X or whatever) - 1.53:1
3 karabiner sliding X as in your example - 3.71:1
Funny 8 - 7.25:1

So now you can put this into your diagrams and work out the loadings you get in reality when you start your equalisation from one side or the other. And then you will appreciate why 95% of the systems are worthless and the rest not much better. And why "nearly true equalisation" has to be accompanied by "using a collection of large and very expensive pulleys throughout the system"."

I thought that it was pretty clear that these numbers were for your standard two-leg anchor testing of three different basic set-ups: single strand (as in a quad clipped in a single strand), standard sliding X and "funny 8".

This 1:3.7 figure for the sliding X seems to have been obtained with 6mm cord. Your more recent 1:1.7 / 1:1.9 figures were with 8mm? And the DAV tests were with some sort of sewn webbing slings but I don't think I ever worked out exactly which kind.

jktinst wrote: Regarding the DAV climbing wall drop tests, with the weight mimicking the belayer hanging from the anchor's central point and the one mimicking the leader falling from 1m off to the side, I don't see how it can be claimed that their pre-equalizing the anchor skewed their results towards a more even distribution.

It's because the master point is already set in the "ideal" position (or close to it). Where as in many situations the biner will have to move to this location. This doesn't happen instantly, nor does it occur before load is applied.

The time the load is applied is usually faster than the reaction time of the anchor, so the peak load is seen before the master point biner can achieve the ideal location, and thus the load distribution is more offset than expected.

I believe the explanation for this is due to friction in the system. Lower friction systems, like say a quad - with no twist around the master point biner - will be better than higher friction systems like the ACR, sliding X - which have a twist - but they wont be perfect. The number of sliding elements also impacts this, and I expect the anchor material does as well.

Yes the belayer weight held the central point at its lowest position, which would have been ideal to obtain an even distribution if the leader weight had dropped in the same alignment. With the leader weight dropped 1m to the side of this alignment, the initial impact of the fall pulled the central point towards that side. Then the leader's weight would have swung towards the other side (but by then the load will already be quite a bit lower), etc. If anything, it seems to me that the belayer's weight, by resisting the pull of the central point to one side then the other, would have made the sliding X anchor behave more like a static one, with less equalization, or am I completely misreading what happens in this kind of situation? If I got that more or less right, then isn't it also possible that the inertia of the belayer weight could have evened out the load distribution for the static cordelette-type anchor, yielding a more even distribution in this situation than it might have if the belayer weight wasn't hanging from the anchor?

Each time the anchor is subjected to a lateral pull, the pro that gets the higher load is the one on the opposite side. I believe that the swinging of the leader weight (while realistic and representative of a real-life situation) also contributes to making the cordelette-type anchor look not-too-badly equalized with an overall distribution of about 1:1.9, compared to the sliding X's 1:1.3. However, of course, with the leader weight swinging from this static anchor, the whole load applied at any given time is either almost completely on one pro or on the other in rapid succession, whereas the sliding X does some sharing of the load between the two at all times and provides a more gradual transition from loading predominantly one pro to loading predominantly the other.

The DAV tests had the two "pros" of the anchor at the exact same level, making a completely symmetrical anchor. I believe that this is a fairly standard set-up in Europe but I rarely see it where I climb. A set-up where the longer arm would have been on the side of the leader weight would have predominantly loaded the shorter arm, probably yielding more uneven distributions for all anchor types. The opposite set-up would have attenuated these differences and skewed the results towards more even distributions.

I also believe that the "long arm towards the fall" set-up would have resulted in more pronounced differences between the sliding X and the cordelette-type anchor. With the sliding X, the less loaded long arm would be able to feed material through the central biner to the shorter, more loaded one, as demonstrated by the Long/Gaines/Ewing tests.

jktinst wrote:I thought that it was pretty clear that these numbers were for your standard two-leg anchor testing of three different basic set-ups: single strand (as in a quad clipped in a single strand), standard sliding X and "funny 8". This 1:3.7 figure for the sliding X seems to have been obtained with 6mm cord. Your more recent 1:1.7 / 1:1.9 figures were with 8mm? And the DAV tests were with some sort of sewn webbing slings but I don't think I ever worked out exactly which kind.

Well the guy understood!
I said the three karabiner system as in his diagram (it was a photo in fact and used coat hooks) and the "funny 8" was his first attempt (his initial post) where he expected the rope to slde through a fig 8 knot to equalise.
The difference between the different materials tested as a single strand as a 90° Vee is;-

9mm Rope 1.74:1
8mm Cord 1.65:1
7mm Cord 1.53:1
6mm Cord 1.48:1
4mm Cord 1.40:1
12mm Dyneema Hybrid 1.48:1
10mm Dyneema Hybrid 1.40:1
8mm Dyneema Hybrid 1.39:1
16mm Nylon Tape 1.67:1

Brian L. wrote: It's because the master point is already set in the "ideal" position (or close to it). Where as in many situations the biner will have to move to this location. This doesn't happen instantly, nor does it occur before load is applied. The time the load is applied is usually faster than the reaction time of the anchor, so the peak load is seen before the master point biner can achieve the ideal location, and thus the load distribution is more offset than expected. I believe the explanation for this is due to friction in the system. Lower friction systems, like say a quad - with no twist around the master point biner - will be better than higher friction systems like the ACR, sliding X - which have a twist - but they wont be perfect. The number of sliding elements also impacts this, and I expect the anchor material does as well.

With an offset drop test AND some belayer weight there is never enough sideways force applied to move the central karabiner, the force required to move the karabiner is less than that required to pull the falling weight sideways. You can work out the force required from the karabiner factor of the central point for any given material and karabiner wrap from the measured load split, for example the horizontal force required to move a weighted central karabiner in a sliding X and 8mm cord is 38% of the belayer weight. Once the central weight is approx 1/3rd the falling weight nothing moves.
I actually measured the vertical and horizontal forces involved and the horizontal was only 28% of the vertical so not suprisingly nothing happened.

Jim Titt wrote: With an offset drop test AND some belayer weight there is never enough sideways force applied to move the central karabiner, the force required to move the karabiner is less than that required to pull the falling weight sideways. You can work out the force required from the karabiner factor of the central point for any given material and karabiner wrap from the measured load split, for example the horizontal force required to move a weighted central karabiner in a sliding X and 8mm cord is 38% of the belayer weight. Once the central weight is approx 1/3rd the falling weight nothing moves. I actually measured the vertical and horizontal forces involved and the horizontal was only 28% of the vertical so not suprisingly nothing happened.

Right, that makes perfect sense. My explaining was in the case on an unweighted belay - which i think was the condition of som of the testing reported in this thread (not the DAV article).

Right, I´ll put the graphs up for an offset drop test on an unweighted belay in the morning, it´s midnight here!

Jim Titt wrote: Well the guy understood!...

You mean Neoshade understood that you had tested his various 3-arm systems? Maybe he did but I didn't see any clear indication of that in his subsequent posts.

Jim Titt wrote: I said the three karabiner system as in his diagram...

Wow! I really did not understand this to mean a test of the entire 3-arm system. A standard two-arm sliding X requires 3 biners; the Central biner on the photo was on a standard sliding X; this photo (3-pro "stacked binary" system yielding a theoretical distribution of 1:1:2) showed only 2 biners but if you replaced the coathooks by biners, it would have 5; and your results in "1:x" format were clearly indicative of two-arm systems.

You even added...

jktinst wrote:...now you can put this into your diagrams and work out the loadings you get in reality...

...As if the numbers were for various standard two-arm components and he would need to use these numbers to work out how his 3-arm systems distribute the load.

Finally, if the tests were on the 3-arm system but you reported only the two-way load split at the main central point, it would have been critical to know if your pre-equalization offset was towards the "2" side or the "1:1" side.

Regardless, I'm happy to accept (and am quite reassured) that when testing an offset 2-arm standard sliding X for equalization, your test methodology yields load distributions in the 1:1.5-1.9 range rather than the 1:3.7, as I previously thought.

Now this:

Jim Titt wrote: With an offset drop test AND some belayer weight there is never enough sideways force applied to move the central karabiner, the force required to move the karabiner is less than that required to pull the falling weight sideways...

Is quite interesting too.

I still don't know for sure what you meant by Drop Test, Slide Test, Tilt Test and Offset Load Static test earlier and would really like to. Nevertheless, I assume that this "offset drop test with belayer weight on the anchor" means tests like the DAV's. The fact that the central biner does not shift during this type of test confirms that the belayer weight does make the sliding X behave a bit more like a static anchor (though not completely since in the DAV tests, the sliding X did distribute the load better than the static systems). In any case, that's something to keep in mind when setting anchors for ledge vs. hanging belays.

The discussion was ONLY regarding the central and right hand points and the system used to link them, the rest of the anchor was anyway not worth wasting time on.
The system linking the right hand two coat hangers is neither topographically nor functionally a sliding X in that the capturing the central karabiner in the fig. 8 knot forces ALL the cord to slide through ALL the karabiners and vastly increases the friction.

An offset drop test is what it says really, instead of dropping the weight straight onto the anchor the weight is offset to one side in an attempt to force the anchor to move and "equalise."
The DAV result of a load split of 1:1.3 is merely an artefact of human involvement, i.e. they put the central karabiner in a position which gave that result, the anchor in itself did not adjust to give this result.
We can see the effect of the human factor by arranging an anchor (in this case a 90° Vee from a single strand of 16mm nylon tape) and dropping a weight from 45° upwards and away from the left side. By moving the unwieghted central karabiner to the left, centre or right and measuring the force on each piece we can see that the initial positioning of the karabiner decides on the end result, not the anchor equalises.
The blue force curve is the piece on the right side i.e furthest from the drop weight and the red curve the piece nearest.

Left

Central

Right

Clearly the anchor is not "equalising" and the person positioning the karabiner is pre-disposing the forces on each piece.

As we see that with either the central karabiner weighted or unweighted the belay itself does not equalise we can search for a scenario where the force is applied to ensure the anchor moves. The only realistic one is where the faller side rope is through a piece of gear off to one side. In a top-rope situation this is purely academic as the already minimal forces will be further reduced. In a lead fall the consequences could be serious but regrettabl it is effectively impossible to arrange a dynamically equalising anchor so that it can cope with a downward fall and a sideways one without an enormous penalty in potential extension.
Since dynamic equalisation does not occur, the loading on the pieces is increased by arranging them as a Vee and the undoubled effects of extension if one piece fails there is nothing to recommend any sliding system for building belays.
A redundant, no extension system based on single pieces each capable of withstanding the force is the gold standard.

Have any of the tests of sliding vs. non-sliding systems looked purely at equalization based on unequal arm lengths?

Some of the early tests that got people nervous about the cordelette showed that in a fall perfectly aligned with the anchor's rigging, the shortest arm took most of the force. The greater the discrepancy in arm length, and the higher the modulus of the cord, the greater the effect.

The fact of this didn't surprise anyone who'd taken high school physics, but the degree of it raised some eyebrows.

I'm curious to know if the sliding systems slide enough to compensate for this, or if they're as effectively unmoving as they seem to be in off-axis falls.

You really only have to look at work like Marc Beverly´s on multi-piece fixed anchors and compare with the results for whatever sliding system you are interested in. Obviously the moment the stretch in the longer leg(s) put´s a higher load split on the pieces than the sliding system gives then the dynamic system will move to maintain that load split.
However the problem of extension in a dynamic system cannot be removed and the problem of extreme unequal legs in a static system can easily be solved as noted by Beverly by using a more dynamic material (nylon) to build the anchor and low-stretch material to roughly adjust the leg lengths (Dyneema/Spectra slings, quickdraws etc to extend the piece(s) down).

I've seen a lot of different test results for multipiece fixed anchors, and see how unequal arm lengths reduce equalization. Do Beverley's tests come to different conclusions? I only see his work on ice screws and abalakovs.

What i don't see is tests that specifically address unequal arm lengths with on-axis loads in a sliding anchor (x, acr, alpine equalizer etc.). I'd be curious to know if the system is able to compensate for the different length "springs" of the arms, or if the system binds under load and acts like a fixed system.

I understand that extending pieces with high-modulus materials can reduce the effect of unequal arm length. In practice I find this kind of rigging slow and tedious, and don't think very many people will actually do it. Especially on long routes, alpine terrain, etc.

paulraphael wrote:I've seen a lot of different test results for multipiece fixed anchors, and see how unequal arm lengths reduce equalization. Do Beverley's tests come to different conclusions? I only see his work on ice screws and abalakovs. What i don't see is tests that specifically address unequal arm lengths with on-axis loads in a sliding anchor (x, acr, alpine equalizer etc.). I'd be curious to know if the system is able to compensate for the different length "springs" of the arms, or if the system binds under load and acts like a fixed system. I understand that extending pieces with high-modulus materials can reduce the effect of unequal arm length. In practice I find this kind of rigging slow and tedious, and don't think very many people will actually do it. Especially on long routes, alpine terrain, etc.

Doubt anyone has bothered to test it because the answer is fairly obvious, the longer leg will stretch more until the load split exceeds whatever the system gives and then start to slide so in that sense a dynamic system compenates for unequal lenghts but only to the limits inherent in the system. So for example the ACR will always give 28% of the load on the longer leg and 72% on the shorter in a two-piece anchor. This implies the longer leg must be over three times longer than the shorter one.
With a cordalette the central knot has a fairly major influence on unequal stretch as the higher-loaded strand cinches up more, effectively reducing the imbalance in the forces but it´s hard to define the real effect since there are too many variables in the knot itself, there´s quite a lot of informatin on this in the paper from Mike Gibbs somewhere in this collection sarrr.weebly.com/multipoint…

Sorry to bring this one back again... I'm not much of a scientist type but I'm curious to see the load distribution numbers on the eyeballed cordalette(non self equalizing) with an offset load. I read the whole thread and didn't seem to see it, maybe I missed it. It seems like, no matter what, if you offset the load on a knotted cordalette you must be dividing the load 100%, 0%, 0% with the piece opposite the fall direction getting all of the load. So while it seems like we've agreed that self equalization doesn't work as well in practice as it does in theory at least it does to some degree divide that offset load. Maybe I'm just completely overlooking something, thoughts?

Jack Mullen wrote:Sorry to bring this one back again... I'm not much of a scientist type but I'm curious to see the load distribution numbers on the eyeballed cordalette(non self equalizing) with an offset load. I read the whole thread and didn't seem to see it, maybe I missed it. It seems like, no matter what, if you offset the load on a knotted cordalette you must be dividing the load 100%, 0%, 0% with the piece opposite the fall direction getting all of the load. So while it seems like we've agreed that self equalization doesn't work as well in practice as it does in theory at least it does to some degree divide that offset load. Maybe I'm just completely overlooking something, thoughts?

Jim has posted on this previously at rc.com back in 2011. Here's what he had to say:

Jim Titt - Sep 11, 2011, 1:23 AM - rockclimbing.com/cgi-bin/fo…;post=2530879;page=3;sb=post_latest_reply;so=ASC;mh=25;

One has to differentiate between static and dynamic equalisation. When you set up a sliding X or whatever it is initially in the static mode where the climber decided where the centre karabiner was going to be. As the load direction is moved offcentre the equalisation stays static until the karabiner slips when we move into dynamic equalisation. For pretty well all the systems I´ve tested the load angle has to be about 13° before the dynamic condition is reached, until then you can just tie a knot in the middle and clip into that and one is not testing the dynamic equalising properties at all.

-----

Jim Titt - Sep 16, 2011, 12:03 AM - rockclimbing.com/cgi-bin/fo…;post=2530879;page=3;sb=post_latest_reply;so=ASC;mh=25;

What you have to remember is that as the load moves off-centre the proportion taken by each leg merely changes until either the karabiner moves or if it is knotted the load on one leg becomes zero.
For an 60° included angle system and a single cord the load proportion changes like this as you offset the load (these are test results so a bit erratic, I´ll work out the theoretical ones sometime):-

0° 50%/50% 1:1
2,5° 48%/52% 1.08:1
5° 44%/56% 1.27:1
7,5° 42%/58% 1.38:1
10° 38%/62% 1.63:1
12,5° 37%/63% 1.7:1
13° Slip
15° 36%/64% 1.78:1

After the slip point the split remains the same in a sliding system but with a fixed karabiner the split continues to increase until 30° when the load on one leg is zero. Obviously the split and the point at which it becomes 0 on one leg varies with the included angle of the pieces with narrower angles naturally being worse. With the pieces vertical it´s not only virtually impossible to equalise by hand but the slightest angular change completely unloads the upper piece immediately whereas a wider angle is much less sensitive for both initial equalising and angular movement, shame it loads the gear more though!

-----

Based on this static testing it seems that what he's saying is this: up until you get 13 degree's off the anticipated line of loading, both systems act the same. AFTER that, the "sliding" system will level out as shown, while the static system will continue to ramp until one leg is at 100%.

So, for small variations in anticipated load direction both systems act the same.

Good man, someone has been paying attention!!!!

Your final comment sums it up but it´s worth noting it´s hard to get a small (or even large) load offset unless the rope gets caught on something or there is an intermediate piece in the system somewhere as the load simply swings across under the belay without moving it.

Haha, yeah I saw those posts and thought it was really interesting data, so copied it into Notepad.

Have you done similar testing with different materials/included angle? Does it change the slip point?