Compression testing pultrusion

Homebuilt Aircraft & Kit Plane Forum

Help Support Homebuilt Aircraft & Kit Plane Forum:

Joined
Nov 14, 2009
Messages
8,497
Location
Rocky Mountains
I have some fiberglass pultruded rod I'd like to test for compressive strength. It's going to be between 1/4" (6mm) and 3/8" (9mm) and I'm expecting total loads of between 4000 and 8000 pounds. The manufacturer supplies data for tensile strength and modulus but nothing for compressive. I'm not so much interested in good stress/strain as I am simple compressive strength. My plan it to load the device below into my press with a pressure gauge and dial indicator.
Am I missing anything important with the planned test rig - like a top plate with a ball bearing between the plate and my ram head?
Red = test coupon
yellow = resin to restrain the coupon cast into a couple of slip fit sleeves for the guide cylinder.
compression jig.JPG
The view port is just for entertainment value. The left side is lopped off for a section view.
 
Last edited:

vhhjr

Well-Known Member
Supporting Member
Joined
Jun 29, 2003
Messages
245
Compression testing thin columns is a bit tricky as is applying loads to the ends of pultrusions. I use carbon fiber arrow shafts often and have encounter splitting on the ends when loaded against something like a ball bearing. Because all the fibers in pultrusions are parallel the epoxy holding them togther is the weak link. It's much like splitting fire wood. I would suggest putting end caps on the samples to avoid this problem. I see from your diagram that you have this covered.

Does your test setup look like your application? Will the rod ends be that well restrained?

The free length of the sample makes a big difference. Have a look at a strength of materials handbook in the short and long column buckling sections.

Vince Homer
 

wsimpso1

Super Moderator
Staff member
Joined
Oct 18, 2003
Messages
9,136
Location
Saline Michigan
+1 on Vince. Compression strength testing is ... difficult. Fixturing and sample constraints are everything.

In actual use in compression, these pultrusions usually have either constrained ends or constraints along their length or both. Spar caps come to mind, with shear web material both between rods and wrapped

If your first samples come in way low, with splitting etc, one scheme is to lay up substantial BID reinforcements on each end tapering down to one ply through the test section.

The way to do it with way more reality is to build a short section of something that looks like your intended structure. A section of beam can then have sections cut from it with caps with web and all wraps, then laminate beefy ends and compression test those samples.

This data will have to be corrected by computing stiffness (EA) of rods and other laminates in the test section. You can then deduce that the load in the rods at first fiber failures is the total load times EArods/EAtotal.

Billski
 

Rob de Bie

Active Member
Joined
Feb 7, 2021
Messages
34
Like the others, I think you will find this to be rather difficult to execute. Heck, even standardized compression testing of composite materials (like ASTM D 3410) is difficult to perform. I would study this standard and ASTM D 6641 (and maybe others) to learn as much as you can. It's not clear to me whether your collars just provided lateral support for end loading of the specimen, or that they are clamped to introduce the force through shear.

One more comment: long time ago I read something that Hart-Smith (Boeing composites guru) stated that *any* test method that gives you the highest numbers is the right one. If you feel a standard test method gives you lower than expected numbers, due to limitations of that test method, feel free to modify the test method to get better numbers. As an example: I remember putting a wire mesh between the specimen and the ASTM D3410 fixture in an attempt to improve the shear load distribution. I think I also experimented with deleting the bonded aluminum tabs, which saved some work. But it's 25+ years ago, I don't remember the outcome of those experiments.

Yet another comment: if you get a lot of spread in the numbers, you would get a very low B-value, that you would use for your design work. And I expect a lot of spread when doing compression testing. That's another reason why it's so difficult to do properly.

Rob
 

ragflyer

Well-Known Member
Joined
Apr 17, 2007
Messages
447
All good responses above. After all there is a reason manufacturers often just give tension numbers. Do you need a generalized compression number? Often in practice is not that useful due to buckling failures that occur at much lower loads than for the full compression strengths. An alternative option that may help is to test in a manner that you intend to use the material in the structure rather than getting generalized results. This is particularly useful in composites due to unexpected failure modes.

If for example you are using it as a spar then test it in bending. This is much easier to do than pure compression. Or if you are testing it as truss member, build model trusses and test to failure and back track by calculation the allowable stresses. You can then scale these results for other sizes as long as they are geometrically similar (if you are careful) or test in full scale. Stan Hall had a good article on how to do this in sport aviation many years ago.

The benefit of this method is that it allows you test in a manner that it will be used and avoids unexpected failure modes. It is particularly useful for amateur designers but you can apply it only in the limeted context of what you tested. Even some professional designs back in day (Mosquito bomber of WW2 fame) were designed in a similar manner.
 

Bill-Higdon

Well-Known Member
Joined
Feb 6, 2011
Messages
2,272
Location
Salem, Oregon, USA
I have some fiberglass pultruded rod I'd like to test for compressive strength. It's going to be between 1/4" (6mm) and 3/8" (9mm) and I'm expecting total loads of between 4000 and 8000 pounds. The manufacturer supplies data for tensile strength and modulus but nothing for compressive. I'm not so much interested in good stress/strain as I am simple compressive strength. My plan it to load the device below into my press with a pressure gauge and dial indicator.
Am I missing anything important with the planned test rig - like a top plate with a ball bearing between the plate and my ram head?
Red = test coupon
yellow = resin to restrain the coupon cast into a couple of slip fit sleeves for the guide cylinder.
View attachment 117570
The view port is just for entertainment value. The left side is lopped off for a section view.
Depending on design set up you may want to protect your measuemnt instruments against a sudden failure of the sample.
 
Joined
Aug 7, 2019
Messages
282
Location
Harleysville, PA
spent 36 years in materials evaluation/testing at Boeing in advanced composites. all the prior statements are accurate with respect to the highest number is the "realest" one as composites, especially unidirectionally reinforced composites, are exceptionally sensitive to all variations.
in my prior life I as sent to labs around the orld to help them learn ho to beat this sensitivity. Compression testing required far greater squareness and parallelism than the ASTM (and Boeing specs!) specify. with pultrusions the parallel issue (specimen geometry to fiber direction) goes away. what is left is the end finish and squareness of the specimen and the test fixture. we found that wet polishing the ends of the specimens at 600 while held in the vee of a machinists vee block worked very well as a repeatable method that pretty nearly anyone could master. If you are using a support fixture, the specimen overhang is also critical as instability is the enemy of uniformity. a specimen support like used in ASTM D695 helps but the best results occur when using a subpress (Compression Subpress (ASTM D695) – Wyoming Test Fixtures) to provide perfect axis alignment between applied and reacted forces and the specimen neutral axis. https://nvlpubs.nist.gov/nistpubs/TechnicalNotes/NIST.TN.1679.pdf
quite easy to make the subpress and specimen support double cruciforms, but the fit ans slip of the sub press piston is critical as is the need to use hardened surfaces against the ends of the specimens. We found that hardness testing calibration check disks above and below solved that problem effectively.
 
Joined
Nov 14, 2009
Messages
8,497
Location
Rocky Mountains
Exactly the kind of response I was looking for. Thanks all!

I'm now rethinking the time and trouble of doing the testing. I expected a lot of scatter in the numbers and wondered how critical the set up of the jigs would be to reduce scatter and to get the 'best'est' numbers. Based on DLDs experience above I think I'll just use what published numbers I have found (compression about 70% of tension) for similar material to use for design, then build a spar and test. I already have most of the tools to load test the planned spar to destruction.

Just to answer the question I know some may ask - Why fiberglass and not carbon pultrusions?
Because I found a source of high quality US produced rod that is available locally for a very good price, even considering the post processing I will need to do.
Boku's China source for carbon pultrusions may not be all that reliable in the future. My spar depth is so large the weight gain is a quite acceptable trade.

And yes, I had planed to place the measurement instruments such that a total sudden failure wouldn't squish them. I'm so cheap I'd even protect the Harbor Freight dial indicator. ;)
 

wsimpso1

Super Moderator
Staff member
Joined
Oct 18, 2003
Messages
9,136
Location
Saline Michigan
If for example you are using it as a spar then test it in bending.

I particularly like this one, and did much of my strength validation tests by laminating plates, measuring their thicknesses, and making samples of each that were then loaded in four-point bending. This loading method put the space between the inner supports in uniform bending, which is uniform tension across one surface and uniform compression across the other surface. Then you determine the first fiber failure by first noises, and look at the sample to determine if first fiber failure was compression or tension... Four-point bending also has shear and maximum bending at each of the inner supports. Since the shear is zero at to and bottom surfaces, where tensile/compressive stresses are maximum, you get a good clean look at tensile or compressive strength. Shear is low in this test scheme.

I got pretty uniform strength numbers for compression side failures with different thickness plates. Presumably the tension side was stronger than the compressive side as I did not get any visible damage on the tension side. The literature out there indicates higher tensile strengths than compressive as well. All of my numbers were higher than published values, so I used the somewhat more conservative published values for my design and failure criteria.

Billski
 

dwalker

Well-Known Member
Supporting Member
Joined
Mar 6, 2021
Messages
709
Location
Tennessee
Exactly the kind of response I was looking for. Thanks all!

I'm now rethinking the time and trouble of doing the testing. I expected a lot of scatter in the numbers and wondered how critical the set up of the jigs would be to reduce scatter and to get the 'best'est' numbers. Based on DLDs experience above I think I'll just use what published numbers I have found (compression about 70% of tension) for similar material to use for design, then build a spar and test. I already have most of the tools to load test the planned spar to destruction.

Just to answer the question I know some may ask - Why fiberglass and not carbon pultrusions?
Because I found a source of high quality US produced rod that is available locally for a very good price, even considering the post processing I will need to do.
Boku's China source for carbon pultrusions may not be all that reliable in the future. My spar depth is so large the weight gain is a quite acceptable trade.

And yes, I had planed to place the measurement instruments such that a total sudden failure wouldn't squish them. I'm so cheap I'd even protect the Harbor Freight dial indicator. ;)

Since the pultrusion for the Dragonfly tailwheel is NLA, I am very interested in finding a suitable stand-in. It is about 1x 1/2 inch and a couple of feet long. Mind sharing the information?
 
Joined
Nov 14, 2009
Messages
8,497
Location
Rocky Mountains
Fiberglass rebar.
The sand 'skin' needs to be removed for proper secondary bonding to anything but cement. That is pretty quick and easy to do by hand, but will require a jig/tooling to do the job consistently enough for spar material.
There seems to be 2 styles on the market. One is a simple linear pultrusion and the other has a spiral wrap that kind of simulates a standard steel rebar surface profile. The style with the wrap needs more ground away because the fibers near the surface are not as linear. There is obvious deformation of the pultrusion due to the wrapping. Good for rebar 'snag' in concrete but bad for compressive strength.........but in concrete, not a problem.
The simple strait extruded type looks like it has better surface fiber orientation. I haven't burned a sample of that style or put it under the scope. It's still setting next to the chicken coop.

QC seems to be good (ASTM) since this stuff is also 'man rated' in that we depend on it to keep floors and bridges off our heads.

Were the original D'fy tail springs ScotchPly? If so it is still available - though I don't know how readily. McMaster Carr?
 

dwalker

Well-Known Member
Supporting Member
Joined
Mar 6, 2021
Messages
709
Location
Tennessee
Fiberglass rebar.
The sand 'skin' needs to be removed for proper secondary bonding to anything but cement. That is pretty quick and easy to do by hand, but will require a jig/tooling to do the job consistently enough for spar material.
There seems to be 2 styles on the market. One is a simple linear pultrusion and the other has a spiral wrap that kind of simulates a standard steel rebar surface profile. The style with the wrap needs more ground away because the fibers near the surface are not as linear. There is obvious deformation of the pultrusion due to the wrapping. Good for rebar 'snag' in concrete but bad for compressive strength.........but in concrete, not a problem.
The simple strait extruded type looks like it has better surface fiber orientation. I haven't burned a sample of that style or put it under the scope. It's still setting next to the chicken coop.

QC seems to be good (ASTM) since this stuff is also 'man rated' in that we depend on it to keep floors and bridges off our heads.

Were the original D'fy tail springs ScotchPly? If so it is still available - though I don't know how readily. McMaster Carr?

Had not considered glass rebar or similar products, but that is certainly an idea, and the simple straight fiber bar is probably the ticket.

The original tailspring was supplied initially from I think Featherlight and then from someone else. I never have seen one in person but the drawing has them as more of an oval than a squared off rectangle.
 

ragflyer

Well-Known Member
Joined
Apr 17, 2007
Messages
447
I got pretty uniform strength numbers for compression side failures with different thickness plates. Presumably the tension side was stronger than the compressive side as I did not get any visible damage on the tension side. The literature out there indicates higher tensile strengths than compressive as well. All of my numbers were higher than published values, so I used the somewhat more conservative published values for my design and failure criteria.
Billski

Yes, nice to hear as that his exactly what I did as wel and found them valuablel. Of course there are some potential gotchas worth noting. Just before bending failure on the compression side the material will yield transferring more load to the tension side resulting in much higher failure strength. In effect the neutral axis is moving towards the tension side and delaying failure. In materials that yield and have a significant delta between tension and compression strength you have to be careful as the compression failure values you get could be higher (not conservative ) than if you did a pure compression test in the short column range. Now this may not be a big deal in highly elastic material like carbon that does not yield. But wood is a great example where this matters.

If you did the bending test with a piece of spruce you will get a compression failure very close to the tension failure (about 9600psi). On the other hand if you did a pure compression test in the short column range, the compression strength of wood is about half of that (~5000psi). And for wooden box spars (tube like) the values will be somewhere between 5000psi and 9600psi. There are good empirical curves to estimate this.
 

wsimpso1

Super Moderator
Staff member
Joined
Oct 18, 2003
Messages
9,136
Location
Saline Michigan
When the highest value you can get in compression is probably the right one, how do you square using low value result?
 
Joined
Nov 14, 2009
Messages
8,497
Location
Rocky Mountains
Not sure I completely understand, Billski.

I think that there may be an unintentional mixing of test method/results here due to swapping from direct testing of compression to testing 'by inference' using bending - leading to this confusion?

I'd never thought about the moment of the neutral axis during the bending test but if Billski is using observation of first fiber failure is there any significant movement of the neutral axis due to effective thinning of the test coupon?
 

ragflyer

Well-Known Member
Joined
Apr 17, 2007
Messages
447
In wood you cannot detect these fibre failures, its just yielding of the fibers that transfer the load from the compression side of the beam to more of the tension side.

If you want to use the compression failure stress you got from bending in a similar load scenario ( say a spar) then there is no issue with using the higher values if you maintain geometric similarity. If you want to derive a short column compression stress from the bending test so you can apply it to say a fuselage panel in compression, then you maybe in for a surprise.

I must point out I do not know if this is an issue with fiber glass or carbon. If the deflection to load curve bends that maybe an indication that the bending compressive failure stress maybe higher than the short column stress. This is certainly an issue in wood (we talk of modulus of rupture to explain this in wood) and their are well documented semi empirical curves to estimate things.

When you design a wooden spar you use the adjusted modules of rupture (derived from bending tests) as the failure stress on the compression flange of the spar and not the ultimate compressive stress with is lower. On the other hand on a wooden fuselage truss you use the ultimate compressive stress.
 
Last edited:

wsimpso1

Super Moderator
Staff member
Joined
Oct 18, 2003
Messages
9,136
Location
Saline Michigan
Not sure I completely understand, Billski.
There was a post above that cited the composite expert at Boeing as stating the test method that produces the highest strength reading on composites in compression is the right one. The reasoning behind this is that compression testing is tough to get good results on - lots of ways to trip low readings that would then drive overbuild. In glass and carbon, some test methods produce compression strengths 75-90% of tensile strengths, others are way low. The designs done with compression strengths close to tensile have been used extensively and with good history. Use Compressive Strength that is 40-50% of Tensile and your bird will get heavy.

Billski
 

wsimpso1

Super Moderator
Staff member
Joined
Oct 18, 2003
Messages
9,136
Location
Saline Michigan
Harder than compression strengths is shear strength of composites. Ugh. You end up falling back on analyticals...
 
Top