Load Testing Results

[estarzinger]

Watching the samples fail, they always start to fail from webbing end (where the webbing tail is) rather than from the loop end. The tail curls/pulls up and you can hear some pinging . . . then when a couple of those stitches fail the whole thing goes bang. This suggests putting more stitches near the webbing end.

[thinwater]

^^ Ya think?

Look at stitching in the failed tethers in this study and some correlation to what you have learned seems obvious.

http://offshore.ussailing.org/Assets...ea+Studies.pdf

I wonder if the failures had even been pull tested? Certainly not drop tested, since there is no stated requirement.
Also a lot of unsuitable clips. Amateur hour from commercial suppliers. I'd feel safer with knots.

It bothers me that EVERY climbing equipment bar tacks, and these folks use untried stitching. I don't get it. They could sub the construction for a few dollars to someone who knows how. But this study is old and many have stepped up.

[Seaworthy Lass]

Originally Posted by estarzinger

Sample 3: 95% efficiency - 3 rows down plus 5 across

Sample 2: 109% efficiency - 3 rows down plus 2 across

Comments/suggestions?

The efficiency difference between the above two samples makes no sense to me, so the inconsistency of stitching may possibly be affecting things.

For the perpendicular lines, could I suggest sewing a complete line across, leaving a few cm of loose thread at the ends and trying a simple knot just to lock the ends in place rather than doing back and forth randomly? You can do the same for the 'down' stitching by pulling the top thread down to the bottom (or vice versa) and again tying a simple knot. Also sewing machines generally do not sew the same way in reverse, so swirl the webbing around rather than going backwards.

To be able to make better conclusions, once the stitching is consistent I would suggest trying something like this:
- Start tests with one perpendicular row, then two with one exactly on top of the other, then three on top of each other etc.
- Then add another perpendicular line 'x' cm away, and vary the number of rows as above.
- Then add a third row of stitching.
- Then try one single line 'down' between two perpendicular rows, then between three perpendicular rows.
- Then two lines down as above.
- Then three.

Test for variability.
If you have not run out of enthusiasm, vary 'x' keeping the other variables constant.

I know it is a huge amount of testing , but I think the results will yield more reliable info.

PS Although you may not be varying the stitch length on the machine, your stitches in places are very different in length, indicating different speed of movement of the webbing through the machine either because you are pushing or pulling variably, or because the machine is gripping the webbing variably. This could be dramatically affecting results.

Also with material stretch, any stitching across will be stronger when straight, but any stitching down will be stronger when zig zagged. The density of the zig zag is critical.
The stitch length will also affect strength.
All this starts to add a LOT of complexity though.

[JPA Cate]

Seaworthy Lass:

Everything you posted above coincides with my experience! Yes!

Most of the webbing I sew, I use #92 UV treated dacron thread on. Never had stitches fail, though for lifelines, Jim uses hand sewn, multistrand threads.

This is just so very interesting! I am so glad Evans and thinwater are doing it. It feels like being on the verge of a serious breakthrough (sorry, guys) in science, so exciting.

Ann

[Seaworthy Lass]

Originally Posted by Ann T. Cate

Seaworthy Lass:

Everything you posted above coincides with my experience! Yes!

Most of the webbing I sew, I use #92 UV treated dacron thread on. Never had stitches fail, though for lifelines, Jim uses hand sewn, multistrand threads.

This is just so very interesting! I am so glad Evans and thinwater are doing it. It feels like being on the verge of a serious breakthrough (sorry, guys) in science, so exciting.

Ann

Good to get feedback. Thanks Anne.

Having seen how UV treated Dacron fails with UV exposure (it just turns to powder after several years), I use original Gore Tenara that is clear to avoid the need for colour matching for all my stitching that will have outdoor use (TR style thread is now available from Gore with greater stretch than the original - 17% compared to 5%).

Although the thread is expensive, it is usually an absolute pittance compared to the cost of other materials and 'cost of labour', and then there is the nuisance or safety issues to consider with stitching failing. I think it is worth every penny. The simple tips from Sailrite make it very easy to sew with once the correct needle and needle insertion and tension and thread distance are used.

From memory the strength of the standard 92 a bit less than V92 when new, but within months of UV exposure it exceeds it. In clear it is available in 69, 92, 138 and 207. In coloured thread it comes in 92 and 138 (Sailrite call the 138 "heavy").

I am a huge fan of Tenara .

[JPA Cate]

Originally Posted by Seaworthy Lass

Good to get feedback. Thanks Anne.

Having seen how UV treated Dacron fails with UV exposure (it just turns to powder after several years), I use original Gore Tenara that is clear to avoid the need for colour matching for all my stitching that will have outdoor use (TR style thread is now available from Gore with greater stretch than the original - 17% compared to 5%).

Although the thread is expensive, it is usually an absolute pittance compared to the cost of other materials and 'cost of labour', and then there is the nuisance or safety issues to consider with stitching failing. I think it is worth every penny. The simple tips from Sailrite make it very easy to sew with once the correct needle and needle insertion and tension and thread distance are used.

From memory the strength of the standard 92 a bit less than V92 when new, but within months of UV exposure it exceeds it. In clear it is available in 69, 92, 138 and 207. In coloured thread it comes in 92 and 138 (Sailrite call the 138 "heavy").

I am a huge fan of Tenara .

Both our dodger and our boom bag were sewn with Tenara. I, too, am a fan.
But unlike you, I have never sewn with it! Now I'm considering trying to source it here. Thank you.

And now, back to Evans and the testing........

Ann

[estarzinger]

Originally Posted by Seaworthy Lass

The efficiency difference between the above two samples makes no sense to me, so the inconsistency of stitching may possibly be affecting things.

My tentative conclusion is that there is diminishing marginal effeciency as you add stitches. And I can create an explication why that us, looking at the failure mode . . . There is a stress riser where the stitching loop ends (at the web tail end) and the stitches at that stress riser fail first (and are thus most important and other stitches less efficient).

However, there is also obviously quite some variability. I noted that in the other guy's data set also. He had the same machine. My tentative conclusion is that is just the way it is with this machine.


Also sewing machines generally do not sew the same way in reverse, so swirl the webbing around rather than going backwards.

I have been doing that.

Also with material stretch, any stitching across will be stronger when straight, but any stitching down will be stronger when zig zagged. The density of the zig zag is critical.
The stitch length will also affect strength.

Interestingly neither of those two points is supported by the data. My stitch effeciency was not greatly effected by either stitch length or zigzag. The data is noisy, but if either of those effects exist they are clearly smaller second order effects.

All this starts to add a LOT of complexity though.

I am going to get some V138 thread, and wait for the polyester webbing.

.........

[Seaworthy Lass]

Originally Posted by estarzinger

My tentative conclusion is that there is diminishing marginal effeciency as you add stitches. And I can create an explication why that us, looking at the failure mode . . . There is a stress riser where the stitching loop ends (at the web tail end) and the stitches at that stress riser fail first (and are thus most important and other stitches less efficient).

I would still suggest aiming for more consistency in the stitching, particularly at the web end where the failure is occurring first.

I have enlarged a row at the web end in one of your samples. The stitching does not go from end to end, in some spots along the row there is just one line of stitching, in others three. Also although you think the stitch length does not matter (it must, imagine the strength of just a couple of long stitches instead of twenty smaller ones), it is still important to keep it even. Some stitches are approx 50% bigger than others.

This is posted not to be critical, but to help reach good conclusions and reduce variability in results .

[Seaworthy Lass]

I think this would be a better way to test the strength of a "single row" (I have left long tails and just lightly knotted the ends so that the stitches are "locked" without needing extra rows back and forth) :

[estarzinger]

You do not comment on the increased tension . . . do you think it is right or better? It is pulled/sunk into the webbing on both sides.

Originally Posted by Seaworthy Lass

imagine the strength of just a couple of long stitches instead of twenty smaller ones

Please do realize when I am doing the efficiency calculations I am counting stitches. 20 smaller ones will be, and will count as expected to be, stronger than fewer longer ones. That has all been factored into the efficiency calculation.

To be clear, I have NOT found or said that smaller stitches are not stronger per inch. They are stronger per inch. I have found and said that (a) they are not stronger per stitch. And (b) that there is some advantage to longer stitches as they allow the stretch to more evenly load all the stitches. and (c) there is diminishing marginal returns to increasing stitch count. But (d) there is some advantage to getting more stitches near the tail (where smaller stitches would in fact help).

Originally Posted by Seaworthy Lass

Some stitches are approx 50% bigger than other

Yes. there are two 'walking feet' on the machine. When the stitching gets near the edge, it goes off one of the walking feet, so gets pulled only half as well. This is a testing consistency problem, but it is also a real world effect. I am not sure it makes sense to do special or unusual things to try to make the test stitches more even, when they will not be, in my 'real' sewing. As I am trying to learn about the behavior of my 'real' stitching.

There is a 'experimental delemma' - do you do things in the test which increase consistency but reduce 'reality'? For instance, I believe it would greatly increase result consistency if I tested "end to end" (rather than end loop) samples (because there would then not be the one sided stress riser on the end stitches), but I mostly use sewn loops and rarely sewn end to end webbing (when sailing, the climbers do more end for end webbing). So, I have concluded I would rather learn more about stitching in the loops I do use, than be more 'pure' about the stitches but miss something important .

Similarly, I understand your suggestion about the knots, and if you go back to the second set of samples (I think I posted about them but can't remember) I machine stitched I did much shorter stitches (in order to more carefully test 1" across vs 2" across vs 1" down vs 2" down) and did exactly the sort of hand knots you are now suggesting and got the same general results I am now getting with much longer stitches just machine locked. My conclusion from that early work was that it was better to do stitching that is more similar to the lengths likely to actually be used and locked off in the fashion actually likely to be used, rather than doing something special (like hand knotting) that might or might not somehow produce results that will not correlate with 'real' stitching.

Now, I will certainly admit I could do a neater job on the across stitching. I have done neater in prior samples. But now I am mostly just using it as a turning point to move the long/down rows across the webbing. I will be more careful when I get the dacron webbing (and even more so if I try in spectra webbing), but (the failure mode and break load data suggests that) in nylon the material stretch seems to compensate almost completely for sloppy stitching placement.

If you look at the samples the other guy sewed up, he was neater, but arrived at the same results (in nylon) that the stitching shape did not effect the breaking efficiency. Again, I believe, because the nylon stretch compensates.

Anyway . . . I think I have done what I can with V69 thread and nylon webbing. My overall conclusions from that are in my pp3 above. I am going to wait for some heavier thread and dacron webbing to continue.

[Seaworthy Lass]

Hi Evans
Just tossing around my thoughts. Something I write may be of use in your experiments. I have little theoretical knowledge of strength of stitches, just practical experience (a few decades of sewing of domestic fabrics in some of my free time, and in the last few years a fair bit of marine stuff).

So here are a few things I have found in practice:

I can report that number of stitches is not the only factor to strength. Whether or not they are on top of each other (or very close) or well separated makes a huge difference. So 3 rows separated will be much weaker than 3 rows on top of each other.

Fabric (as opposed to webbing) is weakened by the holes the needle makes, but I am guessing this is not much of a consideration for webbing. It may still be a factor.

Zig zag stitches are definitely stronger than straight stitching when sewing parallel to the stretch. This is not primarily due to more stitches per cm with the same stitch length, it is due to the fact that the stitching will stretch out to a straight line as the material is stretched, so more force can be applied before the load is put on the stitches rather than the fabric.

One row of stronger thread is better than multiple rows of weaker thread, even if theoretically the multiple amount adds up to the same strength. I would always try and use the thickest thread I have on hand when security is at stake.

Originally Posted by estarzinger

You do not comment on the increased tension . . . do you think it is right or better? It is pulled/sunk into the webbing on both sides.

If the stitches have sunk equally into the webbing on both sides, the tension you are using looks about right as far as I can see (the stitches do not look too loose with loops sticking up, nor are they digging into the webbing which would mean they were too tight).

Originally Posted by estarzinger

Please do realize when I am doing the efficiency calculations I am counting stitches. 20 smaller ones will be, and will count as expected to be, stronger than fewer longer ones. That has all been factored into the efficiency calculation.

To be clear, I have NOT found or said that smaller stitches are not stronger per inch. They are stronger per inch. I have found and said that

(a) they are not stronger per stitch.

Have you found this from your tests? I think smaller stitches may be stronger per stitch as given they need to sink into the webbing, more thread is being used. There is therefore more stretch possible in the fabric before load is put on the stitches (just guessing here).

Originally Posted by estarzinger

And (b) that there is some advantage to longer stitches as they allow the stretch to more evenly load all the stitches.

It will probably not make much difference if a shock load is applied (it wont have time to stretch out the stitches evenly as a slow loading would do). And the opposite may well apply, because there is proportionally less thread per stitch used (due to thread not sinking into the fabric as many times).

Originally Posted by estarzinger

(c) there is diminishing marginal returns to increasing stitch count.
But (d) there is some advantage to getting more stitches near the tail (where smaller stitches would in fact help).

Agreed.

Originally Posted by estarzinger

Yes. there are two 'walking feet' on the machine. When the stitching gets near the edge, it goes off one of the walking feet, so gets pulled only half as well. This is a testing consistency problem, but it is also a real world effect. I am not sure it makes sense to do special or unusual things to try to make the test stitches more even, when they will not be, in my 'real' sewing. As I am trying to learn about the behavior of my 'real' stitching.

I see the logic with this, but it does make conclusions really hard to reach. If you look at the line near the end of the webbing (that you have seen is the first to fail) it is variable in length (some go all the way across the webbing, some are really short), variable in number of rows (with the backstitching you have 3 rows in some spots, one in others and nothing consistent between samples). If this is where the stitches are failing first, followed be rapid failure of the rest, then this first row is really important. A shock load is probably applied to the rest once the first row goes.

I don't think you can simply count the number of stitches in this row and keep that constant between samples. It does make a big difference to strength how these stitches are combined.

Originally Posted by estarzinger

There is a 'experimental delemma' - do you do things in the test which increase consistency but reduce 'reality'? For instance, I believe it would greatly increase result consistency if I tested "end to end" (rather than end loop) samples (because there would then not be the one sided stress riser on the end stitches), but I mostly use sewn loops and rarely sewn end to end webbing (when sailing, the climbers do more end for end webbing). So, I have concluded I would rather learn more about stitching in the loops I do use, than be more '

Similarly, I understand your suggestion about the knots, and if you go back to the second set of samples (I think I posted about them but can't remember) I machine stitched I did much shorter stitches (in order to more carefully test 1" across vs 2" across vs 1" down vs 2" down) and did exactly the sort of hand knots you are now suggesting and got the same general results I am now getting with much longer stitches just machine locked. My conclusion from that early work was that it was better to do stitching that is more similar to the lengths likely to actually be used and locked off in the fashion actually likely to be used, rather than doing something special (like hand knotting) that might or might not somehow produce results that will not correlate with 'real' stitching.

That is a tough dilemma. Unfortunately the locking back stitch will affect results as it essentially adds another row of stitching in one spot. And in your samples sometimes this additional bit superimposes the original row (effectively increasing strength more than when the backstitching goes off at an angle). It makes it very difficult to conclude much as you don't know how much that is contributing to the outcome. Testing many samples and looking at the average would help with this, but that is very time consuming.

Originally Posted by estarzinger

Now, I will certainly admit I could do a neater job on the across stitching. I have done neater in prior samples. But now I am mostly just using it as a turning point to move the long/down rows across the webbing. I will be more careful when I get the dacron webbing (and even more so if I try in spectra webbing), but (the failure mode and break load data suggests that) in nylon the material stretch seems to compensate almost completely for sloppy stitching placement.

I believe the short perpendicular rows are the more important ones and where special care needs to be taken. I know just from hand ripping stitching that it is extremely easy for a row stitched parallel to the force and impossible for me if the row is perpendicular. This may have no correlation to your tests, but it is worth considering.

Originally Posted by estarzinger

Anyway . . . I think I have done what I can with V69 thread and nylon webbing. My overall conclusions from that are in my pp3 above. I am going to wait for some heavier thread and dacron webbing to continue.

Maybe some what I have posted may be useful in your experiments with the thicker thread.

[Jim Cate]

"Have you found this from your tests? I think smaller stitches may be stronger per stitch as given they need to sink into the webbing, more thread is being used. There is therefore more stretch possible in the fabric before load is put on the stitches (just guessing here)."

Angela, I'm puzzled here: how can the webbing (or fabric) stretch before load is put on the stitches? If the stitches were not there, or didn't take any load, what forces could stretch the fabric... they would just slide apart?

As a occasional home sailmaker and frequent hand stitcher of stuff, I'm finding this all very interesting and informative... the very first quantitative data on the subject that I've seen (not that I've looked very hard!).

Cheers,

Jim

[JPA Cate]

You guys know, thinwater's been on about bar tacking. Would one place the bar tacks perpendicular to the webbing length? or parallel? And how would that compare for strength? leaving out for the time being, the application of glue.... makes my needles dirty! then clean with turps.

[Seaworthy Lass]

Originally Posted by Jim Cate

"Have you found this from your tests? I think smaller stitches may be stronger per stitch as given they need to sink into the webbing, more thread is being used. There is therefore more stretch possible in the fabric before load is put on the stitches (just guessing here)."

Angela, I'm puzzled here: how can the webbing (or fabric) stretch before load is put on the stitches? If the stitches were not there, or didn't take any load, what forces could stretch the fabric... they would just slide apart?

Cheers,

Jim

Yes, you are right with a loop arrangement. I was thinking of two bits stretched rather than a loop.
It would still need some force to start separating the stitches, so I think a larger force would be needed before the stitches themselves receive the load, increasing strength.

[estarzinger]

Originally Posted by Ann T. Cate

You guys know, thinwater's been on about bar tacking. Would one place the bar tacks perpendicular to the webbing length? or parallel?

Basically a series of very close together but relatively long individual zigzag stitches (eg a lot of zig but not much zag ). The individual stitches are 'down' the webbing' but the bar is 'across'. This is ideal for the sort of stress riser loading I have been seeing.





By the way, to really see clearer 'stitch pattern' test results it may be that I need to get some spectra webbing, so that webbing stretch will not compensate/hide the pattern effect.