[peteeng]

Hi Thomas - Is this the correct interpretation of your data? Could you draw the rough values and graph please? Ta Peter

[badhabit]

1. Ideally a repeatable way of applying the force. A weight on a string, released from a repeatable location. Needs to hit the target once only.
2. Need to check metals yourself. I would not expect your numbers to be absolutely comparable to other testing numbers given the uncertainty in your testing method, however relative numbers should still be useful.

Are you sure about #1? In my tests it seems like the force applied only changes the maximum amplitude and does not affect the frequency nor the decrement. However, the "noise" i see in my data might be because the impact varies so much, so you might be right...

I see some experiments where they attach a load to the sample and then "cut the string to the load" which causes a single excitation and it is very repeatable... i could do the same with a spring and a string in my vertical setup...i think i'll try that and see what happens...

Regarding #2: I agree, i need a baseline of sorts to be able to compare things relativly. I'll have a look at what i can find that is "well known"(like steel or alu) and have a "practical size" AND something that i'm still able to excite/measure...

(...) It seems to be more of a comparative effort more than a defined science. Although the maths are clear the results are always variable. Each researcher ends up with different numbers even in metals. (...)

edit - I've added more info here. Please note Professor Slocum from MIT states - "material and joint damping factors are difficult to predict and are too low anyway" see page 11

I agree about the comparative effort. I'll try to examing a "known material", i'll be at the shop tonight and i'll have a look at what options i have at hand...

Regarding the paper from Slocum: Interesting, my numbers(factor of 0,014ish) matches with his interval for EG: 0.01-0.05...in the low end, but still somewhat matching... it might just be pure luck though :-)

Is this the correct interpretation of your data? Could you draw the rough values and graph please?

Not quite. Each testline is a different excitation, so 5 different "hits" on the sample and 5 different oscilations and 5 different captures on the scope. My "high" and "low" values for each test was selected on the osciloscope capture where the oscilations had "a visual nice/pure form"(no noise) and as far apart as possible in both time and voltage. The "n" is the number of "oscilation tops" between the top of the "highest" peak without noise and the top of the "lowest" peak. Does it make sense? I can try to make it capture the raw data and put it on a USB-stick to show if you'd like...

/Thomas

[peteeng]

Hi Thomas - The decrement should be the same, what is the voltage peaks in this image from your test? The decrement calc can be across two peaks as per decrement 1 image. Or it can be across several peaks. It should be the same figure no matter how you do it? Peter

[badhabit]

Hi Thomas - The decrement should be the same, what is the voltage peaks in this image from your test? The decrement calc can be across two peaks as per decrement 1 image. Or it can be across several peaks. It should be the same figure no matter how you do it? Peter

My scope was able to save this image:
Attachment 505794
I've marked the "high" and "low" peaks i was talking about. Their values is 1110mV and 496mV in this case. I ignored the first part of the curve because i think i'm hitting the limitation of either the sensor or the scope or both, so i started "a bit down the curve" where it becomes more "sinusoidal"...I did this 5 times in the spreadsheet in my previous post....makes sense?

/Thomas

[peteeng]

Hi Thomas - Now I know what you mean by high/low Ta. I like pictures... Yes your maths checks out... So your bar has a zeta (loss factor or damping factor ) of around 0.01 a bit to go to get to ideal of 0.7!! Thats why the conclusion that material damping is not enough.... Peter

would have been good if Slocum quoted what aluminium and steel factors where...

[badhabit]

(...) So your bar has a zeta (loss factor or damping factor ) of around 0.01 a bit to go to get to ideal of 0.7!! Thats why the conclusion that material damping is not enough.... (...)

Yeah, it matches very good with what i saw in the first tests:
Attachment 505718
...but it will, as you said, only be valid as a relative measurement, so i have to test other materials to get something to compare it to....

...i tried a solid 20x40mm hotrolled steelbar yesterday, but it gave very weird results(loads of harmonics and my scope did all sort of weird stuff) even though it oscilated physical "nicely", i suspect the problems i have in the beginning of the EG oscilation was muuuuch larger in the steel sample, maybe due to it being much more stiff. I also have a solid 15x15mm coldrolled bar i'll try next, but i have a business trip the next couple of days, so that will have to wait a bit...but i'll post it where whatever the measurements are...

/Thomas

[peteeng]

Hi Thomas - I think wacking the beam is the issue. Why not just bend it by hand a bit and let it go? Peter

[jackjr-123]

So...the results are in...short answer: 22-23Gpa...for a 12% by-weight Silimix282 with R&G ResinL + GL2 Hardner. Cured for 2 weeks at 30c, no postcure treatment.

Thanks for doing the experiment.
IMO 12% is too much epoxy. It should be < 10%. Rampf Epument uses 9% epoxy.

I'm trying to do the same experiment with a sample of UHPC but I'm struggling getting good results. I think the test rig is potentially a huge source of error and the measured stiffness ends up always lower than expected.
Did you test the same setup with a material of known stiffness? (e.g. a beam of aluminium or steel)

[badhabit]

Thanks for doing the experiment.
IMO 12% is too much epoxy. It should be < 10%. Rampf Epument uses 9% epoxy.

I'm trying to do the same experiment with a sample of UHPC but I'm struggling getting good results. I think the test rig is potentially a huge source of error and the measured stiffness ends up always lower than expected.
Did you test the same setup with a material of known stiffness? (e.g. a beam of aluminium or steel)

I don't think 12% vs. 10% matters much since i think it'll just create more cavities, but "for the sake of objectivity" i'm open to try to make a new batch with say 9-10% and retest in the same way... and i think you are right: i need to test a known material in the same testsetup, so i'll do that aswell while i'm at it. Coldrolled steel should be fine, right? i would expect 210Gpa, but I'm unsure if the processing of it will affect it...

Is there such a thing as "good results" when the criteria is that it must match ones expections? :-) I was hoping for much(!) higher modulus than my tests shows, but they are what they are, nomatter what i want the to be :-) What stiffness did you get from your UHPC tests? Do you have a picture of your setup?

/Thomas

- - - Updated - - -

I think wacking the beam is the issue. Why not just bend it by hand a bit and let it go?

I think you are correct. I'll have to revise the setup, i'll be back

/Thomas

[jackjr-123]

For reference, here's a piece I casted using Epument some years ago. It is straight out of the mold without any post-processing. There were very few air bubbles (I vibrated the mold with a motor) and overall the piece looked great.
I have some Silimix & RG epoxy on hand, maybe I should try casting a sample and compare the results.

Cold rolled steel is fine. It doesn't matter. All steel alloys have basically the same stiffness (within 10%), no matter the state.

I don't have any results for my UHPC stiffness test yet. There was too much deflection coming from the test rig. Instead of trying to measure an absolute stiffness value, I will simply try to compare the stiffness with an aluminium beam of same dimensions. This should eliminate any error coming from the test rig.

[peteeng]

Hi Jack Jr - In my deflection tests I also use a piece of aluminium and could correctly predict the stiffness of the aluminium. Here's some shots of the set up. I used a reference beam above the cantilever. This deflects if the bench deflects and since the foundation of the test piece and ref beam are the same then the foundation deflection should be accounted for. Make sense? Peter

[jackjr-123]

Peter, yes that makes sense but I can't use anything like this for my case: my sample is only 350mm long. In hindsight I should have made it at least twice longer, but then I couldn't fit it in the oven for heat treatment.

[peteeng]

Hi Jack - Have to be careful with "short or deep" beams vs slender beams. Short beams can be shear deflection dominant and will not follow the usual slender beam formulas. If you have FE available then use second order elements and check that the short beam follows the slender beam equation... If your using equations to back calculate the modulus.

If shear deflections are applicable then the relationship between its tensile modulus and the shear modulus has to be tested. Its assumed to be G (shear modulus) = E/(2x(1+v) where v=Poisson ratio for isotropic material. I have not seen any academic test that this is correct for EG. In my work with long fibre composites (survey analysis for composite vessel structures) we have done many shear modulus tests and often this elastic relationship does not hold. Even so for purposes of design we ignore that as beams deflect close to the correct equation... many unknowns to understand.

Plus with EG I personally believe the tensile modulus and the compressive modulus are different. This means the neutral axis is not the geometric NA in flexure of a beam. So maybe that's one reason under test the flexural modulus calculates low. But under small strains I can't see this being too far off a good assumption ie its Ok to assume the material is isotropic for machine design purposes... long story short... Peter