Tensile Strength Testing

[subscriber]

Bob,

It is good to do basic volume/pressure exchange calculations as a reality check for LDC designs.  I have a spread sheet set up to do those for me, but confess I only used it when the airgun FPE is over 60 FPE for rifle length barrels.  I used the calculator recently for the Huben pistol mufflers I designed.  That has a very high muzzle pressure because of its short barrel and high FPE.  It also has a propensity for the valve to stick open, if you shoot until the reservoir pressure drops too low. 

I certainly prefer using an aluminum or CF tube for anything over 50 FPE.  But have fully printed designs for the Huben pistol where most walls are just 2 mm thick.   I had help testing conditions that emulate a stuck valve with a short barrel.  Shall we just say that my printed Huben pistol mufflers are not rated for stuck valves.   To do that would require walls perhaps 3 or 4 times as thick.  It would be so heavy and expensive that no one would want one.   And I make that clear to anyone that wants to use one.  I am happy to report that should such a long valve duration event occur, no bits are projected towards the person holding the airgun.  This is based on more than one data point in controlled trials, rather than customer hands.  Also, the mounting threads do not appear to be a weak link in the chain.  Assuming they are printed properly, rather than with a overly shallow thread form.

Commercial mufflers are not made as string as the air tanks that feed them.  So, while a lot stronger than thin poorly printed plastic, some commercial might not be able to contain a stuck open valve event.  One might even argue that coming apart at 500 PSI is better than coming apart at 3000 PSI, due to the severity of the event.  It was suggested to design in what amounts to a hinged flap in the first expansion chamber.  Something that would open a relief air path, but not create a free fragment that might go somewhere unexpected.  A fail safe, if you will.

My muffler designs tend to get larger in proportion with FPE, and the first expansion chamber with it.  Larger for backyard friendly than hearing safe.What I really would like to know, is how much the peak air pressure in the first chamber drops, due to the flow out of the baffle bore.  What I can tell you is that I design my baffle bores with generous projectile clearance, so that extra flow relief should help.  The loss in muffling performance is not as much as one might estimate when oversized bores are used.   Take the Marauder for example. The stock baffles are the same for .177, .22 and .25; using 5/16" bores for all of them.  None of them are loud at typical power levels for caliber.

Anyway, back to thread testing...

[rsterne]

I got a chance to go back and test the Vertical (Z) coupons with Overture PLA this evening, and plotted them along with the previous Horizontal (X) curve.... Here they are....



The vertical Yield strength peaks at 220C, instead of the 230C for the horizontal Yield.... at about 57% of that for the horizontal at 220C.... I am extremely pleased with the consistency of results I am getting with my coupons.... Based on these results, I will be printing the thread sleeves at 220 for the PLA and 240 for the PETG (both Overture).... I think I will do 5 or 6 of each, so that I can test various thread depths....

Bob

[subscriber]

That 2:1 strength differential is very telling.  I wonder how that compares to wood, along and across the grain?  Of course, there are many types of wood, and it varies a lot within one type. 

If plywood is a good structural material, is it possible to 3D print an interweave of horizontal and vertical printing.  My gut says no. 

Perhaps printing the pull test coupons at 45 degrees off horizontal is the best way to produce more even strength?  Assuming that a given part needs even strength.

[rsterne]

I have seen threads printed on a 45 deg. angle, and they end up in between vertical and horizontal.... Using a line or zig-zag infill is about as close to plywood as you can get, right?....

I am printing five PLA thread test cylinders right now.... I changed the design, separating the long and short threads with a 14mm cylinder....



The long, upper portion has a 45 deg. taper at the top, the other three ends of the threads are square, no countersink.... The upper section is 25mm long, the cylinder 5mm, and the lower section 10mm, for a total length of 40mm, and a diameter of 26mm (increased a bit to give a full 6mm wall around the 14mm diameter center cylinder.... I have put 5 models into Cura, and set 4 of them below the Platen by 2, 4, 6 and 8 mm, giving me coupons with 2, 4, 6, 8 and 10mm of test thread.... This will give me an engagement depth to work with, and then I will try different layer heights.... Just over 6 hrs. to print all five.... 

I still have to get and machine the end bolts....

Bob

[subscriber]

Crisscross 100% infill is like plywood in the X and Y axis, but it does not improve the Z-axis layer to layer adhesion.  It was the Z-axis plywood I was aiming for.  Unobtanium, unless you glued more that one part together.  At least I can't think of a way to print it, but I am not a printer expert.

[rsterne]

Well, today was an interesting, wasteful, but eventually productive day....

I started up the five threaded cylinders in PLA that I mentioned above in Reply #83, and then went out to the shop to hopefully make the 1/2"-20 threaded adapters to fit them to my testing setup.... I had some Grade 8 bolts the right length, so I tried milling flats on the side, and after making a shallow initial cut, my HSS mill just polished the surface and wouldn't cut any deeper.... I didn't have anything of less quality, but I did have some coarse threaded rod, and it milled fine (with the same mill, fortunately I didn't cook it).... This week I will pick up some fine thread rod (hopefully they will have it in town), barring that I will try and get some "hardware store quality" bolts.... I have decided to use 5/16" pins, and to thin the bolt by milling flats to insure it will pull closer to the center, and not tilt the test pieces....

Anyways, back to take the finished prints off the printer, and the outside was fine, but the threads were a real mess.... I had forgotten to print outer walls last, and to use the "Exclusive" setting for the slicing of the sloped surfaces of the threads.... plus I only printed at 0.2mm layer height.... There were strands of filament running from point to point inside the threads, from say 2 to 4 o'clock, then 3 to 5 o'clock, etc. as each layer printed.... Not all layers were like that, but the inside was sure a mess.... I ran a tap through them, and it cleaned them up OK, but there were chunks missing from the threads, like it had been machined with a dull tap....

Anyway, I decided that the 2mm (which is just over 1.5 theads) was a joke, and the 10mm was likely too strong, so I repeated the printing, still in PLA, with a 0.12mm layer height, exclusive on the profile setting, and outer walls last (which makes overhangs a lot better, so really improves the thread profile).... Printing time stayed at just over 6 hrs, but for only 3 instead of 5.... They turned out beautiful, and a bolt threads in by hand with no cleaning with a tap required.... So, once I get the threaded rod and make my ends, I will have 4, 6 and 8mm sections of thread to test.... If they all work, I will have a 3-point graph of pull-out force vs thread count.... Wouldn't that be nice?....

Bob

[subscriber]

Bob,

Your hassle with thread printing and your solution make me not even want to try it

I rely heavily on the other Bob to print threaded parts I design.  He has it all figured out.  At this point, the prospect of learning the software settings to make a printer run right, makes me feel more like a mechanic than an electrician.   You seem to have mastered all this trickery rather fast.

Good to hear about success after you got the kinks out of the way.  It sounds like that grade 8 bolt was springing your mill.  Assume the HSS cutter did not get dulled.

[WobblyHand]

Interesting to hear of printing thread issues.  Can't say that I have experienced any issues printing 1/2-20 threads in PETG at 0.2mm height with nominal PrusaSlicer settings.  Not much stringing, relatively well formed threads.  A tap only removes the tiniest amount of fuzz.  Maybe I was lucky?  Anyways, I chuck the LDC in the lathe and use the tail stock to center the tap,  line up the threads and turn the tap by hand.  It takes little effort to go through, since the threads are only the slightest bit undersized.  Since tapping takes so little time compared to constantly reprinting coupons, I haven't bothered to refine the threads anymore. 

I am encouraged by Bob's tests, they should reveal things of interest for the airgun community. 

[rsterne]

Bruce, I think the biggest issue was printing outside to inside, as the protruding edge of the thread profile has nothing to sit on, and ends up as a straight line from where it has no support to where it touches again.... When you print inside to outside, the last part of the thread to be printed is the overhanging tip of the profile, and it has something to at least stick to, if not sit on.... By using a thinner layer, but the same wall thickness, the "next" layer printed is a more flattened oval, and has a significant overlap to sit on, and I don't get the stringing.... It was just a matter of not changing the settings back after printing my tensile test coupons, where I had it printing outer wall first to get more accurate and consistent dimension on the thin test section....

I am thinking I have way too much wall thickness, and I will probably print some today with smaller OD's to speed up the printing process.... I'll try a 20mm, and if that fails, go to 22mm, etc.... Currently I am at 26mm (full 6mm walls at the 14mm cylindrical thread break)....

Bob

[rsterne]

Well, I machined up the adapters today.... I annealed the Grade 8 bolts by heating the shank and head to cherry and letting it cool slowly.... I was STILL having problems with the mill not cutting.... I found out that 0.020" was not deep enough cuts, and when I went to 0.030" cuts, it worked fine.... So, once I had the adapters done (and 5/16" quick pins), I put the 2mm of 1/2"-20 thread sample in and it took 51.0 x 4 = 204 lbf. to break it!....   

I then tried a 4mm thread sample, and I stopped pulling at 110 x 4 = 440 lbs. (200 kg.), which is the limit of my force gauge.... One thread is 0.050" = 1.27mm, so the number of threads was 4 / 1.27 = 3.15 threads!.... Now we know that three 1/2"-20 threads in PLA will stand over 400 lbs!.... I am SHOCKED, that is over 100 lbf. per mm of engagement!.... Here are the samples and an assembled test piece (with only a 20mm OD, plenty strong enough)....



Top left is the 2mm sample after failing.... Center is the 4mm sample, and it looks a bit beat up, but that was what it looked like after I chased the threads, which were done at 0.20mm layer height.... Right is the 0.12mm layer height sample, which has also been subjected to 200 kg. of pull, and it looks and screws onto a bolt like new.... NO permanent damage....

I'll print up a few thin engagement samples in PETG to test, but all the rest of the samples I made are too much thread engagement for my setup!....

Bob

[subscriber]

Amazingly strong!

[WhatUPSbox?]

Great results Bob.

If you want to try higher loads, Using some longer bars (1" square tubing?), still connected to your lathe and load cell, could get you higher gain without overloading your set up.

[rsterne]

Stan, while that would certainly work, do you really think we need more information?.... I'll test some PETG, which I expect to be weaker, and I'll probably print samples at 2, 3 and 4 mm for that, with the 20mm OD.... They won't take long to print, I can do them all at the same time by setting them below the Platen....

Sub, anything more you would suggest?....

Bob

[WhatUPSbox?]

No, I think you captured the data relevant to the airgun application.
I just thought, since you printed them, if you Really wanted to brake them, then there might be a simple configuration.

[rsterne]

I have redesigned my thread test cylinder to 20mm OD, 25mm tall, with a 4 thread high (5.08mm) test section.... I will be printing 1, 2, 3 and 4 threads in PLA and PETG, using a layer height of 0.158mm, which gives exactly 8 layers per thread....



The printing time is just over 2 hrs. for 4 cylinders, 1 of each length.... The object is to produce a strength graph for each filament, with force to fail plotted vs number of 1/2"-20 threads.... It should be a straight line, of course.... If I need more data points, I will do test cylinders at 1.5, and 2.5mm of threads....

Bob

[subscriber]

Bob,

Do you print one test piece at a time; or four of them together?  I assume that printing one at a time will be stronger because the welding of subsequent layers will occur at higher temperature - less cooling time between layers, if the part is effectively smaller with no jumps between parts.

Perhaps this effect would be small for four 20 mm tubes that are almost touching on the platen?

[rsterne]

I print several at a time, different lengths by setting them different depths below the platen.... I did 5 for the PLA (1, 1.5, 2, 2.5 and 3 threads), and for the PETG I will do 7, adding 3.5 and 4 threads.... The PLA printed great, I am having some stringing with the PETG, so I'm currently running some test towers to hopefully refine the tune.... It may be time for a new nozzle....

I just read that with PETG, for cosmetics you should run 60-100% fan, but for more strength 30-50%.... Anyone care to comment?....

Bob

[WobblyHand]

Default settings in PrusaSlicer for PETG is 30-50% fan speed.  Fan disabled for first 3 layers.  Seems to work for me.

[TorqueMaster]

I just read that with PETG, for cosmetics you should run 60-100% fan, but for more strength 30-50%.... Anyone care to comment?....

Bob

I used to run the low-to-mid fan since it was on in the defaults, and recommended for dimensional accuracy, but lately I run no fan, and have not noticed any loss of cosmetic appeal or sizing.   I think you just showed that less fan doesn't always mean more strength.  It sort of makes sense the faster it cools, the less time it has to ooze, though.  The more I read online articles from websites that should know, and anecdotal results from individual users...the less I trust the websites that should know.  Too many variables involved, too many formulations of PETG filament, for blanket recommendations like that fan speed one to necessarily hold true.

[rsterne]

I ran a set of thread tests today, 1/2"-20 threads printed at a layer height of 0.158mm (8 layers per thread) in both PLA (Overture Grey) and PETG (Overture Black).... I tested 1, 1.5, 2, 2.5 and 3 threads in PLA and those plus 3.5 and 4 threads in PETG.... I could not test more in the PLA because it would have exceeded the limits of my Force Gauge.... Here are the results....



The PLA took about 150 lbf. per thread to break, and the PETG took about 80 lbf. per thread.... A single thread is not very reliable for data, as one might expect, but from 1.5 threads up, the data is pretty linear, as you would expect.... The dotted lines give the Shear Strength in psi, based on the calculated theoretical area of threads in shear (number of threads x pitch diameter x PI x (pitch/2)).... The last number is the theoretical thickness in shear of one thread at the pitch diameter.... For 1/2"-20TPI that works out to 0.0368 sq.in. per thread in shear.... In theory, those should be straight, horizontal lines (the shear strength should be constant, independant of area).... They are pretty close, so that validates the testing method....

I took photos of the broken test cylinders, and the results were interesting.... First the PLA....



and then the PETG.... Both photos had a single thread on the left, increasing in thread count to the right.... Click on the pics to enlarge....



In all cases, other than the very thin thread samples, the failures occurred right where the last thread stopped, where a 14mm diameter internal cylinder provides a precise location where that occurs.... The test threads were between that 14mm ID and the outer end of the test samples.... Now one might think that showed that the wall of the cylinder at the 14mm diameter was too thin, causing the part to fracture there.... In fact, with the PETG I made the OD larger as the thread count increased to help avoid that, to no avail....

I went back and checked my calculations on that portion of the cylinder, and recalculated the theoretical breaking force, using my measured tensile strength in the "Z" direction for these two materials.... The area of that cylindrical portion is (20^2 - 14^2) x PI/4 = 160 mm^2 = 0.248 sq.in.... The Overture PLA had a tensile strength of 4750 psi when printed at 220C (which is what I used for the thread test), which should take (4750 x 0.248) = 1178 lbf. to break, requiring nearly 4 times as much as the 300 lbf. it took to break the middle sample, which had 2 threads.... The Overture PETG had a measured tensile strength of 3650 psi at 240C (the thread printing temperature), which would take (3650 x 0.248) =905 lbf. to break, over 9 times the 100 lbf. it took to break the 1.5 thread long sample.... Additionally, if it was the wall thickness of the 14mm ID section that was the weak point, it would have failed at the same force in every test, which it did not.... Also, the thicker cylinders I used for the longer threads in the PETG samples, would have showed an increase in force required to break them.... The largest OD samples I used were 26mm, having a cross sectional area at the 14mm ID more than double that of the 20mm cylinders....

So, how are the samples breaking?.... If you look closely at the thicker thread samples in PETG (right end) you will notice that the samples broke along a cone shape, and did not simply rip between the layers.... None of the samples (except the very thinnest ones, which stripped out) broke anywhere but at the inner end of the threads, which of course is where the stress is the highest.... Although every thread imposes load on the cylindrical wall outside of it, those loads add up and peak where the threads end at the 14mm ID recess.... Therefore, that is the point that tears first, and once that rip starts, it just zippers across the outer tube, following the line of maximum stress, which is supported less and less as each layer of the print breaks....

The fact that the data behaved exactly as it should, and that the Shear Strength was less than the Tensile Strength of the material (in the Z direction), I think these are valid and reliable tests.... How thick an outer structure would you need before you would see even 4 threads pull straight out the end?.... I dunno, but way more than the 6mm wall in those 26mm OD samples....

Bob