[walter]

I was wondering, before you go to the next step, you should do another sample of E/G with larger aggregates of about 5-6mm (3/16"- 1/4") in size.

Bruno,

I will definitely do that. I was hoping to find pure quartz but might use pea gravel instead.

Your sizing theory paper really gains importance here- my test shows that vibrocompaction is absolutely crucial and for that we need tight packing!

Greybeard also posted something on this topic, so I thank you both. I need to get my math straight or else...


Larry,

Good to hear from you.
Thanks for the mixing tips!



Jim,

Welcome to the thread.
Polymer concrete does have some shrinkage and is probably better suited for filling weldments- we're trying to reverse engineer the man made granite. Or the newer product known as Quartz/Epoxy composite which can rival cast iron.


Cameron,

I appreciate the effort.
2.9 ksi would be awesome given the poor mixing technique and massive air entrapment. Or maybe it's 0.29
That would really burst my bubble lol..

I'll hit The Cary Co, they sell IMSIL® microcrystalline silica fillers with average particle size 2.1 micron and surface area 2.0 m^2/g
_

[djshop]

Crap, I've been washing that drill paint mixer in lacquer thinner. Could this weaken the samples?

I always clean up epoxy with denatured alcohol. It works much better than any other solvent I have tried.

[martinw]

Washing-up liquid and really hot water works well to clean-up tools. Don't even think of using cold water unless you are particularly fond of vile yukky gloop.

Some people swear by white vinegar, but I've never tried.

Best wishes

Martin

[brunog]

Bruno,

I will definitely do that. I was hoping to find pure quartz but might use pea gravel instead.

Walter,
Take a look a your local pet shop, the stuff they have is mostly quartz ans granite.

The larger aggregate i got locally is a supplies Hagen.

Best regards

Bruno

[brunog]

Cameron,
That PDF file from Reichold is very instructive!

Best regards

Bruno

[ckelloug]

Bruno,

How does the Reichhold PDF compare with the Fuller's parabola formula? I think this paper follows closer to the niss paper graybeard posted than the parabola. Also, did you see the post where I commented on your question about gantry geometry and not quite holding .0001 but closer to .00015?

[speed33317]

http://www.concreteexchange.com/cata...3F32BEF0C802DA

Concrete vibrators, good website if you also want to cast your own
concrete countertops.

[ckelloug]

I have attached a graph showing my best estimate of young's modulus for Walter's 22.5 ft*lb Sample. I don't know the exact number of whole and fractional clamp turns that Walter used to break it but I estimate Flexural modulus at about 0.8 ksi whereas the NIST paper's least effective mix was around 2 ksi. I estimate the maximum stress at failure to be about 1.3 ksi or a bit under half the minimum value given in the nist report.

If only we knew what was in the sample.

<h5>Attached also please find a PDF paper describing how I computed the modulus and please tell me if you spot mistakes. </h5>

[brunog]

Bruno,

How does the Reichhold PDF compare with the Fuller's parabola formula? I think this paper follows closer to the niss paper graybeard posted than the parabola. Also, did you see the post where I commented on your question about gantry geometry and not quite holding .0001 but closer to .00015?

Cameron,

Here are the differences

Aggregate size .....Reichhold .....Fuller
<0.06 mm....... .....12% ...........10%
0.06 to .6 mm .......12% ...........22%
0.6 to 1.0 mm .......15% ...........9%
1.0 to 1.5 mm .......20% ...........9%
1.5 to 3.0 mm .......35% ...........21%
3.0 to 6.0 mm .......6% ............29%

As for the NIS report I need a bit of time to digest it.

As for the E/G beam this will give you an idea how the 1x2 fb are incorporated in the E/G beam itself the flat bars will be anchored in the E/G. .0015" deflection is good, could I get .0001 deflection with a 15" beam instead of 12"??

Best regards

Bruno

[brunog]

Sorry Cameron,

Here is a bmp representation of the E/G beam and the position of the 1x2 FB

Bruno

[walter]

I'm not sure how many turns, probably 2, but this is just a wild guess.

I believe it was my standard "high compaction" mix pictured in post #853, page 72 (first picture).

These mixes used only 3 components and no pigment. I don't think I used more than 35% for the largest size.

Did you calculate the 22.5 ft*lb sample or the 32.5 ft*lb sample pictured on page 99, post #1181?

Thanks for plugging in the numbers, much appreciated!

[lgalla]

Upon reviewing past posts,there seems to still be confusion on epoxy ratios.The magic 10% or so epoxy ratio by the machine builders refers to weight,not volume.
Volume,or parts/per ratio's are very difficult and will be too thick.A gallon of sand weights more than a gallon of 1/4"pebbles.Volume percentages cannot be accurate especially with different aggregate sizes.Use a scale.
larry

[ckelloug]

Bruno,

I'm attaching a png of what I thought you were trying to do which isn't what you were trying to do. The diagram I'm sending is the maximal strength that can be added via the steel pieces of that size. Moving the steel from the bottom to the placement in your diagram will probably up static deflection to from .00025 to about .0003 inches.

Unfortunately, I dropped a factor of two in my last calculation and it looks like at the design load of 100lbs, your deflection is pegged at .00025. Nothing seems to make it better except for making the beam hollow by leaving about a 1 inch wide 6 inch high slot in the middle or reducing the design load. Making the beam larger as a solid seems to increase weight faster than it increases strength.

Since this is all static deflection, post tensioning could be used to null this pesky .00025 and then under design load, the thing would be exposed only to cutting forces and probably be good to a gnat's arse.

I've attached a diagram of what I interpreted your last request to mean and also a little pdf with the governing equation for the situation although it isn't quite batteries included since it lacks descriptions of all the terms. I've also got the gnuplot code I've been using to prognosticate but it runs on linux and it's confusing.

Walter,

The sample calculation that I did was for your 22.5 ft lb sample although the results might be closer to rubbish than results. It's a bit guess work as I forgot to mention the need for a turns count hopefully good to an eighth of a turn and I haven't heard from either fyffe555 or Geof on whether they can help check the calculation.

Right now, I think that the flexural modulus figures figures I have given you are probably erroneously high because I had thought I was plotting in turns on the bottom of the graph I posted but I was plotting in radians. Redoing the calculation by hand where pretty pictures don't serve to distract, my instinct is to say that I believe the correct number lies around 0.17 kips for your two good samples but I think I'm delirious today after having stared at too many papers and numbers. So with all of the certainty I can muster at 1:21 AM I think you're between a factor of 2 and 20 away from the nist paper's lower range for E/G modulus though I need to do some more research to make sure that the equation I presented isn't neglecting anything crucial.

What I think the equation says is you really really want is a sample that breaks at 350 ft lbs in 2 turns.

If you have a piece of aluminum or mild steel Walter, could you record the torque for me at 4-8 places per clamp turn for about 1 turn so that I can see if I get a modulus for metal that approaches an accepted value?

Larry,

Most of the physics equations work on a volume basis and have to be converted to mass basis for mixing.

[walter]

Thanks for doing the calculations on Bruno's beam. That validated my hollow shaft idea.

Jig calibration results should be ready later this week.

Thanks!

[fyffe555]

haven't been following the thread recently, got pm from cameron and another member who suggested I had a look, I'll have a go at answering the questions. This is of course entirely my opinion, you may not agree, and you don't have to.

Bruno's 2 material beam; Camerons formula is correct. Or more specifically it is the solution for a simply end supported beam with centre load for two (or extendable to more) disparate materials in a composite beam, AND the formula ignores entirely any shear tension or other loads between the materials when under load.

Being pedantic here because the thread seems to have taken a turn to selectively pluck and use formula when it suits, which doesn't really help the big picture. As suggested earlier in the thread when Larry posted on the subject, it's pragmatic engineering to simply ignore intermaterial shear and stress and you can simply ADD the individual strengths for each material xsection. The above formula does the same and simply calculates the moments and modulus products for all materials in one go. That might be more than is necessary.

For Bruno's idea ( I assume the steel are embedded mounting points?) it can get very complicated very quickly determining the values for moment of irregular sections to put into the formula for the e/g beam with insets for the steel. Are the steel bars fully flush or proud? The 'add' method can simply assume the moment is an area and you can add (or subtract) from it. For all practical purposes, as the strength is a product of the moment you can do the same with strength to so it opens up the quick calc methods.

Assume the e/g beam is a rectangular or other regular section, ignore the steel and calc the moment. Then calc the moment for the steel beam(s), either seperately or maybe easier, as one large rectangle and deduct the unwanted centre to get the combined moment. Then determine what percentage by area the e/g and steel beams intersect and deduct by that proportion the steel moment from the e/g and there you have the moment for e/g. Stick that into the formula and away you go.

Or, often quicker, if have your calc strengths for e/g and steel seperately and then adjust for intersection by proportion and simply Add you'll get the same answer.

Walters Stress test; Not sure what the Flexural modulus calc derivation is or what you're supposed to do with it in this context, other than use it for subjective testing or you're planning to do the rest of the extensive math that goes with it. Once you have it, sure, you can compare it to published lab results or production materials but I disagree if 1/ you think you'll be able to make a material to those strengths at home, or 2/ even what those strengths actually are in practice. More on that later. The formula together with the info in the thread from Walter ( nice job BTW) together doesn't give all the bits you need to determine the 'flexural modulus'. Specifically you need to know *accurately* the displacement of the piece under load, or accurately determine the point of failure AND know the torque actually moving the part to the point of measure. I don't see either in the thread.

More importantly as we know from the discussions of screw efficiency elsewhere there's a substantial amount of torque 'wasted' in over coming friction in the rig. As the load increases so does the amount of torque required to simply move the screw. Add the means of measuring the torque allows at least a +/- 5lbs error at least then there's not much point in applying them to such a complicated calculation and expecting an answer closer than +/- 50% or so.

HOWEVER - for Walters situation it does give a very specific, ingenious and not to say cheap and useful means of comparing samples. Good job. It's a decent qualitive indication of the material strength, unfortunately by my guesstimates it's still a lot under the 4.5x10^5 we talked about some weeks ago.

Andrew

[ckelloug]

Fyffe555,

Thank you immensely. I haven't computed any of this in years and I don't want to put up wrong, incomplete, or misinterpretable information as I derived all of the stuff I'm putting up from first principles as a relearning exercise, perhaps myopically. There's nothing worse than not having another set of eyes on formulas. I am not a PE or a mechanical engineer but I am a decent systems engineer on sabbatical thus having time to learn and work. As for shear, like I promised in the Java applet I posted earlier which also neglects shear today although it has a not horrible UI, I'll add that to the formula for the multi-material deflection. I was being pragmatic about shear because the books I have suggested that shear is only a couple percent once L is more than 10 times h. You're right and it isn't in Bruno's beam. My bad.

Generally, with the equation I published for Bruno, I've just graphed the deflection with respect to to moments only equation and used it to try to gage whether there is any hope of .0001 deflection for the member in question given the 2000ksi modulus E/G that NIST talks about. I have a set of gnu plot equations set up to automagically handle the moment of inertia and weight as rectangle tube dimensions vary.

If you are willing to go through my derivation on the Flexural modulus estimate, Andrew, I will write that one up fully. I'm sure it won't precisely mirror lab tests but I think it's a not horrible screening test for samples to show that spending 175$ for a real lab test isn't worth it. I tried to work out this math to get something remotely quantifiable from Walter's results since some idea of the strength of this home brew E/G compared to real E/G from the nist study is essential in determining whether it is useful.

Briefly speaking, in the flexural modulus calculation, I said that the energy required to torque the plunger screw the integral of torque from zero to Walter's final torque value dtheta should be equal to the energy required to bend the beam. Since I know the tpi of the screw on Walter's jig, I can estimate total displacement if I have screw turns. (which I don't have accurately thus I've been graphing it to constrain the unknown to a curve representing screw turns vs. modulus rather than a whole plane).

In conclusion, Andrew, if you have any thoughts about calculation methods I've posted and stuff to take into account and the time to read this thread, please post them. Well meant and helpful criticism is the nicest thing one could give on a thread like this. I'm just trying to do the math on the beams and use the best theory available on the materials so that the expectations of what is being produced and what can be produced are set appropriately and met.

Finally, Andrew, I politely disagree with the notion that home shop folks can't meet the published numbers for structural grade E/G in the NIST paper without adding the qualifier yet to the statement. I do think you are right that home shop folks have no chance to make DuPont Zodiac counter tops in the basement.

I do think that it shouldn't be impossible to come up with E/G that has a Flexural Modulus of 2000ksi and a ultimate tensile strength of 2ksi. I believe it is possible and have devoted substantial time over the past several weeks to try to make it so.

I've got a rented tractor sitting outside and gotta get back to it.

Thanks again, Andrew and please criticize any aspect of my posted engineering calculations.

[walter]

Awesome post Andrew.

Pretty much in line with my earlier thoughts that commercial E/G works and hobby E/G does not.
Hope we can get closer to that commercial level.


Cameron,

I can't thank you enough for what you are doing here. Your help has been invaluable.


In any case, I have that one good sample, let me try to improve it further.

[ckelloug]

Walter,

Can you post what tests you are thinking of doing so that everybody who has suggestions can chime in? I know Bruno suggested something.

I still think based on the paper from Reichhold and the e-mail where the apps engineer sent it that we need to start working on bonding agents, deairing agents, surfactants and nanosand.

The other thing that needs to be discussed is the exact mixing procedure. Martin had comments as did Larry. I think that one of the key reasons for the low E numbers is that the epoxy is not coating all of the particles. According to Gupta, 30-32 micrometers of epoxy are required around each macro particle.

[lgalla]

Un qualified observations:
I will agree with Cameron,the particles are not wetting,resulting in large void content.This would explain why WilliamD had better hammer results with 25% epoxy by volume.Better bonding V/S dry dough like material.
I think Walter has come to the conclusion proper vibratory compaction is the answer.Yes it is but different from concrete mixes which readily pour.Water has a viscosity of 1,thin epoxy 600cps.This would require a different vibration frequency or thinner epoxy which would compromise strength.
Alternatives are heat if you have slow epoxy.I posted on this previously.Heating reduces viscosity,say from 600cps to 100 to 250cps at 100 degrees F.This would solve part of the wetting problem.
Only guessing on this one.Since the material is similar to concrete ,a concrete mixer or tumbler may be a benifit.
Since I have appeared to have named most process, the final product should be named W/C or Z/C,that is Walter/ Crete or Zone /Crete or both.:cheers: :cheers:

[walter]

Dear Cameron,

Some tips on getting the right material would be helpful.

Where's the aggregate? I don't have any..

You may recall that we moved onto the theory and white papers because we couldn't find any aggregates.

What mixing are we talking about?

I've got nothing. Some pool sand and a bag of silica? And some funny glass stuff?

That's not what E/G is made of.

I'm going back to the drawing board.
_