Causal Mechanisms of Webbing Anchor Interface Failure Thomas Evans a, Sherrie McConaughey b, and Aaron Stavens c

Transcription

1 Causal Mechanisms of Webbing Anchor Interface Failure Thomas Evans a, Sherrie McConaughey b, and Aaron Stavens c a Montana State University, Department of Earth Sciences, PO Box , Bozeman, MT , cavertevans@gmail.com b sherrie.mcconaughey@gmail.com c aaron.stavens@caves.org Introduction and Background: Two common webbing anchors used for rope rescue systems are wrap three pull twos (W3P2s) and basket hitches. Like many components of technical rescue systems, there is little readily available published data on the properties of these anchors, including the mechanisms of failure. Evans and Stavens (2011) were the first to measure breaking strengths of both these anchor types using statistically significant sample sizes and publish the results to facilitate access to the data by others. Their results show that basket hitches are stronger than W3P2s (in slow pull tests), though both are stronger than 11 mm nylon rope, so both are acceptable rescue anchors. It was noted that both anchor types broke in different ways and behaved differently during the slow pulls, causing their ultimate failure. These observations raised questions concerning the mechanisms of failure of each anchor type; understanding the variables that increase or decrease the strength of each anchor type can inform a rigger how to rig safer systems. Consequently, a suite of experiments were designed and implemented here to test the effects of different variables on the strength of webbing anchors. Here we investigate the causal mechanisms of failure for both W3P2 and basket hitch anchors. Through experimentation, we determine if: 1. Knots in the limbs of anchors reduces anchor strength, 2. Compression of one strand of webbing below another strand reduces anchor strength, 3. The curvature of connectors (carabiners, screw links, etc.) used as webbing connectors alters anchor strength, and 4. Knots in the limbs of anchors reduce anchor strength more than the effect produced when one webbing strand is compressed below another. In addition, we replicated the study from Evans and Stavens (2011) to determine if the results were reproducible and as a control for the experimental treatments. Materials: Ten spools of new unused one-inch PMI tubular webbing were purchased from lot number 4550 and loom 514. The connectors used to collect the strands of each anchor were half inch diameter four inch tall steel screw links purchased from a hardware store, a half inch diameter steel NFPA carabiner, or a rigging ring (God Ring) made from half inch diameter bar stock, depending on the treatment. Measurements of breaking strengths were conducted on a Baldwin universal testing machine with DP41 digital load deflection upgrade electronics with an internal load cell range of 0 to 200,000 lbs, at the College of Engineering, Montana State University. The universal testing machine was last calibrated on 3/10/2011 and measurements took place during the summer of Methods: Eight feet of webbing was used to build each basket hitch and nine feet was used for each W3P2 anchor. To ensure no effect was observed due to the spool of webbing used, lengths of webbing were cut from each spool alternating between basket hitches and W3P2 anchors. Each pair of anchors (a basket hitch and a W3P2) was assigned a treatment, and the treatments were rotated such that they were roughly evenly apportioned to each spool of webbing. Samples were given a unique sample number consisting of five parts; the spool number the webbing came from, type of anchor tied, the number of the piece of webbing along the length of a spool, a treatment code, and finally the test number. For example 3-B-14-K-28 corresponds to webbing from spool number three, it is a basket hitch, it was the fourteenth length of webbing cut from spool three, the knots were located in the limbs during

2 measurement, and it was the twenty-eighth measurement performed. All knots were tied by one person (T.E.) to retain consistency, and anchors were built around a 4 inch diameter smooth steel pipe filled with concrete. In addition to a control series, four treatments were applied to both W3P2s and basket hitches. The controls were anchors built identically to Evans and Stavens (2011); basket hitch knots were placed behind the load and strands of webbing were gathered on a screw link while W3P2s were tied with the knot facing the load with strands gathered on a screw link. The treatments were as follows: Treatment 1. Anchors were tied normally with screw links as the upper attachment point however the knots were positioned along one limb. The treatment code for this treatment is K. Treatment 2. Anchors were tied normally with the knots protected (behind the load for basket hitches, and in front of the load for W3P2s) however the upper attachment point had two screw links with one leg of each anchor on each screw link. The treatment code for this treatment is S. Treatment 3. Anchors were tied with the knots positioned along one limb however the upper attachment point had two screw links with one leg of each anchor on each screw link. The treatment code for this treatment is D1. Treatment 4. Anchors were tied normally with the knots protected (behind the load for basket hitches, and in front of the load for W3P2s) however the anchors were tied through a rigging ring (God Ring). One strand of webbing was compressed under the other just like when a screw link was used as the collector. The treatment code for this treatment is D2. In addition, sixteen simple webbing loops were broken with knots half way along a limb, to compare the location of breaks for loops versus basket hitches and W3P2s. Six loops were broken with screw links as both the upper and lower attachment points, while the remaining ten were broken with a screw link as the upper attachment point and the steel pipe as the lower attachment point. Each anchor was built and the upper strand marked with a red sharpie pen at the location where the webbing crossed the upper connector(s). Anchors were quickly loaded at a rate of ~82 lbs per second up to ~7000 lbs then the rate of loading was decreased to ~14 lbs per second, until breakage occurred. All trials were photographed prior to initiation and force and deflection data recorded to create a permanent record of both qualitative and quantitative observations. The anchor internal angle at the initiation of loading was measured from anchor photographs taken looking down the short axis of the screw links or carabiners used. The location of the break (at the upper connector, at a knot in the limb, or along one of the limbs), number of breaks each anchor experienced, and the maximum load at breakage were recorded in addition to any notes or abnormalities observed during measurement. The measured raw breaking strengths were multiplied by the force multiplier determined by the internal angle of the anchor to calculate the load experienced by the anchor. These scaled data were used for all statistics. Descriptive statistics (average, maximum, minimum, range, and standard deviation) were calculated for all trials as well as a subset of those trials in which no abnormalities were observed. To test the null hypotheses that two treatments had the same breaking strength a two-tailed Z-test was performed for all the data as well as the subset of tests in which no abnormalities were observed. Two-tailed Z-tests were performed comparing all treatments to controls, and the results from W3P2s were compared to basket hitches for each treatment. All anchors broken were saved and archived for later study and can be accessed by contacting the authors. In addition, copies of the electronic data (photographs, and Excel files) can be provided upon request or downloaded from the ITRS 2012 web page ( Results: Table 1 displays the raw breaking strengths, scaled breaking strengths, number of breaks, location of breaks, which strand broke (if applicable), and any notes and observations made during measurements. Table 2 displays a synopsis of each treatment, including the sample size (N), average breaking strength, standard deviation, maximum breaking strength, minimum breaking strength, and

3 range. Table 3 lists the P-values for each of the comparisons made. When the populations were compared with all the data versus with the conservative data set without trials in which abnormalities were observed, different correlations were deemed statistically significant. Consequently, P-values are only reported from comparisons of conservative data sets, the data sets excluding abnormal trials. Generally basket hitches were stronger than W3P2s (Table 2, averages column), though which anchor type was stronger depended on the treatment. Usually basket hitches had a smaller range of breaking strengths (Table 2, range column), as well as a smaller standard deviations (Table 2, standard deviation column). When knots were placed in the limbs (treatment K), basket hitch strength was reduced to that of W3P2s, and failures occurred in the lower webbing strand at the screw link not the knot (N=31), though sometimes two failures occurred simultaneously at the screw link (N=3). Knots in the limbs of W3P2s (treatment K) did not change the anchor breaking strength (P=.0830), and failures occurred in the lower strand at the screw link not the knot (N=28, with two exceptions), or two failures occurred simultaneously at the screw link (N=4). When no webbing strand was compressed under another strand of webbing (treatment S), both basket hitches and W3P2s were significantly stronger (P= for basket hitches, P= for W3P2s), and all failures for both anchor types occurred at the screw link (N= 34, for basket hitches, N=35, for W3P2s). When anchors were built without compression of webbing strands and knots were placed in the limbs (treatment D1), there was no change in breaking strength relative to anchors built with no compression (treatment S), (P=.562 for basket hitches, P=.297 for W3P2s). However, in a comparison between treatment K (knots in limbs) and treatment D1 (knots in limbs and no compression), there was a significant increase in strength (P=.0234 for basket hitches, P=.0144 for W3P2s). When anchors were tied around a rigging ring with a much larger internal radius than a screw link (treatment D2, less webbing pinching), there was an increase in anchor strength for both basket hitches (P=.0282) and W3P2s (P=0). W3P2s showed a marked increase in strength (~2400 lbs) while basket hitches showed a modest strength increase (~300 lbs). Of the six webbing loops broken with screw links as both upper and lower attachment points, all failed at one of the screw links. Of the ten webbing loops broken with the steel pipe, eight broke at the knot, one broke at the screw link, and one broke mid strand of one limb. Scaled breaking strengths for some of the webbing loops are not reported since it was impossible to photograph the internal angle for many of the webbing loops, making the conversion impossible to make. Additional Observations It was observed that in the D1 treatments (knots in the middle of a limb with no compression of webbing strands), all basket hitches broke at the screw link in the webbing strand with the knot while all W3P2 anchors broke at the screw link in the webbing strand without the knot. This observation is baffling, however, it may hint at underlying mechanisms behind anchor failure. We observed four steel NFPA 46KN rated carabiners break after being loaded three (X2), four, and five times each. A loading cycle consisted of pulling a webbing anchor to failure, then repeating the loading with a new anchor. All carabiners failed when the nose broke off after the spine deflected significantly. In trials where carabiners failed, the webbing strands remained intact. Sources of Error: All measurements have an associated error. In this case the error inherent in the Baldwin universal testing machine was low since it was recently calibrated. More importantly, the error is on the order of plus or minus a few pounds, much smaller than the effects observed. The error in cutting the lengths of webbing was on the order of a millimeter or two. The variability in tying hitches and their internal angles are the largest source of error in this suite of measurements. This variability was small enough that, when measured, the internal angles for each anchor type (basket hitch or W3P2) were consistently the same. Internal angle measurement error was on the order of half a degree. Thus the

4 sources of error are small enough that the conclusions reached are not affected by the inherent uncertainty in measurement (error bar). It should be noted that the experiments described here were slow pull trials, and it is presently unclear if the results from slow pulls can be extrapolated to dynamic events. Thus, further research is required to constrain how applicable these results are to dynamic events. In addition, we utilized screw links as connectors, which have a much smaller radius of curvature than carabiners, which suggests the webbing anchors would break at higher strengths if carabiners were used, since carabiners have a larger radius of curvature than screw links. As such, the values reported should be used in comparison with other trials utilizing screw links, and not as absolute strengths of each anchor type. Conclusions: 1. Comparing the controls from this research to the data from Evans and Stavens (2011), we replicated all their results, thus demonstrating that anchors behave relatively consistently given the same conditions. 2. The weak point in basket hitches and W3P2 anchors is the connector (carabiner, screw link, etc.) and not the knot even when the knots are in the limbs. 3. With a narrow angle and a single strand or loop of webbing, the connector is the weakest point, and not the knot (based on N=6, so a small sample size!). 4. With a wider angle, and a single strand or loop of webbing, the knot may be the weakest point (N=8 out of 10, so a small sample size!). 5. Knots in the limbs of basket hitches reduce anchor strength, yet the knots are still not the weakest point in the anchor. 6. Knots in the limbs of W3P2s do not change the strength of the anchor. 7. When anchors are built without compressing one strand of webbing under another, the anchor increases in strength. 8. There is no change in the breaking strengths between anchors without compression between webbing strands that have knots in the limbs and anchors without compression but no knots in the limbs. This suggests compression is more important in determining breaking strengths than knots in the limbs. 9. An increase in the pinch (radius of curvature) of a connector reduces anchor strength. This effect is much more prominent in W3P2s than basket hitches. 10. W3P2s breaking strengths are more dependent on how they are built than the breaking strengths of basket hitches which are less variable and more consistent even when configured differently. 11. Generally, basket hitches are stronger than W3P2s, and they have more consistent breaking strengths (smaller standard deviations and ranges) even when built differently. 12. Large sample sizes were absolutely necessary to observe the magnitude of the effects discovered, and the variability of the systems studied. 13. Both anchors are strong enough for rescue loads, and should be used when appropriate, acknowledging that W3P2s show greater variability in breaking strengths given different conditions. 14. While statistically significant, many of the trends observed here may not be practically significant, since all anchors were stronger than an 11 mm nylon rope. Discussion: When interpreting the findings presented here it is important to keep in mind that these results apply to anchors tied in the configuration tested and under slow pulls. It is presently unknown if the

5 results would be similar if the tests were performed dynamically. Dynamic tests are needed to verify the applicability of these results to dynamic events in system use. While many of the comparisons yielded statistically significant results this does not suggest the results have a practical significance. Keep in mind that all of the configurations tested were stronger than an 11 mm nylon rope, so the results may have no practical rigging significance. However, if tests with used webbing demonstrate the same trends, then some general rigging guidelines, or rules of thumb, can be drawn from the research. Rule of Thumb #1: When building basket hitches protect the knots. This is not because the knot is the weakest part of the anchor, since they are not, but because having a knot in a limb reduces the strength of a basket hitch at the connector by ~800 lbs in new webbing. Rule of Thumb #2: It probably does not matter where the knot is in a W3P2, at least in new webbing. Rule of Thumb #3: If you can build an anchor without compressing webbing strands under each other, this will increase the strength of the anchor. Rigging plates or rigging rings could facilitate such configurations. Rule of Thumb #4: Using connectors with larger radii of curvature to collect webbing strands will increase anchor strength. It is often taught that the weakest points in webbing anchors are the knots, which is solidly contradicted by the present data. This disconnect between common knowledge and experimental observation is striking, raising the question of how this misconception developed. The results of the single webbing loop tests where the knots were indeed the weakest points, suggests a possible scenario of how this misconception formed. It is possible that previous tests with loops of webbing showed that the knot was indeed the weakest link, and these results were extrapolated to basket hitches and W3P2s. Alternatively, it is possible that knots are the weakest link in used webbing anchors, which could be determined by repeating this study using used webbing. If the results presented here are valid for dynamic events, then the compression of webbing reduces anchor strength, as does the pinching of webbing in connecting devices. Both the effects of pinching and compression have a larger effect on breaking strength than does the presence of knots in limbs, suggesting that the material properties of webbing are more important in determining webbing anchor breaking strength than location of knots in anchors. Acknowledgements: Dr. Mike Berry, College of Engineering, Montana State University provided access to the testing equipment and lab space necessary to complete this work. In addition, Brian Zirbel opened the building on the weekends, Josh Norquist helped set up electronic data acquisition, and Timothy White provided essential technical support when the Baldwin ceased functioning. Cathy Lash provided invaluable help in cutting and labeling webbing strands. Sarah Truebe edited the document for both content and grammar, greatly improving the final draft. The Pacific Northwest Region of the National Cave Rescue Commission generously provided support for this research. Bibliography: Evans, Thomas, Stavens, Aaron, 2011, Empirically Derived Breaking Strengths for Basket Hitches and Wrap Three Pull Two Webbing Anchors, Proceedings of the International Technical Rescue Symposium, Fort Collins, Colorado, November 3-6, 2011, 8 pages