[lerman]

Someone a short ways back asked how to determine the amount of epoxy needed to fill the voids in a compressed mix. One way to get a rough answer is to take the volume of the compressed mix and subtract the volume of the components. That will represent the amount of void space.

So, after compressing the mix, decompress it. Then measure the volume by adding the components to a bucket of water and seeing how high the water rises. Of course, the water is less viscous than epoxy, but still has some viscosity. So, you can try using something less viscous than water in your bucket. [Anyone have a bucket of liquid helium handy?]

Alternatively, put the mix in a sealed container. Weigh the container. Then evacuate the container to as high a vacuum as you can provide. Weigh the container again. Convert the weight of the air removed to volume. Given the total volume of the container, what remains is the volume of the composite.

That would probably accurate if you had a high precision scale.

Ken

[ckelloug]

Someone a short ways back asked how to determine the amount of epoxy needed to fill the voids in a compressed mix. One way to get a rough answer is to take the volume of the compressed mix and subtract the volume of the components. That will represent the amount of void space.

This is a good idea though the implementation is vague. Building on what you said however, Here's how I'd do it (no epoxy of course): If you measure the weight of each component in the mixture before it is mixed, you know how much material is in there from the specific gravity of the material. Now compact the material via your compacting method and measure the volume of the compacted mixture. Subtract the volume calculated via the specific gravity and the weight measurements and voila: you have the packing density.

So, after compressing the mix, decompress it. Then measure the volume by adding the components to a bucket of water and seeing how high the water rises. Of course, the water is less viscous than epoxy, but still has some viscosity. So, you can try using something less viscous than water in your bucket. [Anyone have a bucket of liquid helium handy?]

The viscosity of the water won't matter as long as there is much more water than material and the whole thing is stirred enough to ensure that all of the aggregate is wet.

Alternatively, put the mix in a sealed container. Weigh the container. Then evacuate the container to as high a vacuum as you can provide. Weigh the container again. Convert the weight of the air removed to volume. Given the total volume of the container, what remains is the volume of the composite.

That would probably accurate if you had a high precision scale.

Ken

I like this last method as it is a direct measurement but an analytical balance and a vacuum chamber are two instruments that most of us are lacking. If I ever come by an analytical balance, it's a good thing.

These volume measurements you suggested do mean however that it is well within the means of the lab equipment challenged to check the epoxy requirements to avoid dry mixes.

Awesome.

--Cameron

[lgalla]

John or Greybeard,I like your spincast.This is similar to a centrifuge,which is used to enrich uranium,or used by denturists to eliminate air or voids in fillings.Iam reassured my many fillings are void free.BTW silane is used in fillings.I will now keep my mouth shut on silanes.My dentist is not keen on me using his vibration table for E/G.
Larryard De Perogie,fried/sourcreme

[ckelloug]

After a day of working on it, I have developed an aggregate formula that uses mostly agsco stock aggregates without sieving and run it though the model. Under vibration with pressure compacting, according to the model, the formula should be able to achieve 91.6%! Assuming I didn't make a mistake in proportioning the mixture up into all of the 24 grading fractions etc, the formula ought to be this. I'll do more work on this but I'm excited to have actually gotten an output. I think that this formula can be improved upon by doing more work with the quartz aggregates in the smaller sizes and also by automating the analysis process further.

I'm not sure about the silica fume choice here and the nanopox provides just under all the epoxy necessary although no hardener so there's calculations that would have to be done to actually use this mixture. This also doesn't take into account any of the nifty additives we've been discussing.

What I've learned from this is that getting a formula that can compact to 92% density is a non trivial exercise using real materials and is both dependent on the sieving and what was sieved. It's also crucial exactly the right materials be used and that no skimping is done on the smaller particles. This last sentence is especially important because it's the tiny particles like G200 zeeospheres and silica fume and nanopox that make the last few percent. It's also probably quite difficult to make a 92% formula by trial and error although I am sure that's how the countertop guys did it.


The not-quite usable hypothetical trial formula is by volume:

10.71 part by volume #6 Agsco Brown Aluminum Oxide
10.71 part by volume #4 Unsieved Agsco Quartz
10.71 part by volume #2 Unsieved Agsco Quartz
10.71 part by volume #2/0 Unsieved Agsco Quartz
10.71 part by volume 3M G800 Zeeospheres
25.00 part by volume 3M G200 Zeeospheres
5.355 parts by volume Cabosil TS-530 silyated silica fume
13.39 parts Nanoresins Nanopox F-640 for 5.335 parts 20nm SiO2.


At any rate, it's late and I've spent way too much time on this today.

--Cameron

[walter]

Speaking of countertop guys. I took a closer look at one of their Quartz products. Pretty amazing job I must say. Now here's the thing: what appears to the untrained eye to be a matrix resin is in fact a very fine aggregate. My eye is pretty trained by now- I know a fine powder when I see one. This aggregate is so fine it looks like milk, simply amazing. (My silica is 2.9 micron average particle size and still looks grainy..) There's so little resin in these countertops (5%), it's a marvel that it's technologically possible. The machinery is said to be able to heat, vacuum, squeeze and vibrate at the same time. I'm sure it's a 100ton press. And the cost is like $20mil. Amazing stuff.


Cameron, good work on the formula. I really like the #6 Aluminum Oxide idea. But I can't seem to find that silica fume.. All they have is Aerosil product from Degussa. What size should I be looking for?
_

[greybeard]

I managed to get the test cast out of the tube and found an interesting variation in the cross section composition.
The cast end photo shows a increase in the sand toward the middle, so I broke off the other end of the test piece as it was very starved of resin. The result showed an unexpected sudden change in resin/particle size about half way out toward the center.
I'm not sure if it's just a sudden change in the resin percentage or a variation in the particle size of the aggregate or both.

The only observation that might be relevant was that there would be an initial excess of resin at that end of the mold before spinning commenced.
Once it started, the resin would have migrated outward and along the outer face of the mold, and drained that part of the center. This would leave resin bound to the surface of the sand sufficient to bind that part into a mass, but probably with a fair bit of air present, showing up as a lighter color in the central part of the sample.

Having had a bit of a problem extracting the sample, I'm going to have to come up with a design for a split mold. That should be fun.
John

[granitedoug]

L S Starrett Tru-Stone Divison can custom machine a granite surface plate to your print. With threaded inserts, tee slots, groves, ect... They can do just about any thing with granite, at a resonable price.
www.tru-stone.com

[greybeard]

L S Starrett Tru-Stone Divison can custom machine a granite surface plate to your print. With threaded inserts, tee slots, groves, ect... They can do just about any thing with granite, at a resonable price.
www.tru-stone.com

And I'm sure they would do a good job. But I'm here, and they're there, an one man's reasonable price is another man's "HOW MUCH?".

If it suited everyone, IMHO, I don't think this thread would have reached 1708 posts.

John

[ckelloug]

granitedoug,

Thanks for posting here. Starrett is well known for making some of the best solid granite surface plates etc. Unfortunately, solid granite has a relatively low tensile strength and the vibration damping characteristics are not as good as the composite material as well as the fact that it can't be molded.

While what we're working on uses various sizes of granite, the product behaves differently from actual granite and has a broader applicability than the solid material. Granitan is a commercial product along the same lines.

Regards, Cameron

[greybeard]

De Larrard gives a log size vs. sieve passage chart and he has divided the log axis that represents particle diameter into 9 intervals covered by 10 points with a range of 1 to 10000 on the log axis. He called the grading range 10e-4. His sample curve would covers between 3um and 3cm not
3um and 3mm as the bogus original version of my post had it.

His motivation may be that French Standard Sieves which are tenth root of 10 apart although these curves are all computed from models.

... If any other errors are spotted, please post so they can be corrected as I'd hate to be the one who posted bad info from a relatively authoritative and well written book.

--Cameron

According to my Ceramics Handbook, the French Sieve range goes up in steps of the cube root of two, ~ 1.26, so that each sieve lets through particles double the volume of the previous one.
Some of our readers may think this is like angels dancing on a pin head, but it seems to make more sense to me in my struggle to turn numbers into something I can visualize.

I've had good news from Lawrence Industries, here in UK, who are sending me samples of G-200 and G-800, so much thanks to them.

I'm hoping to duplicate Walter's work, but in the round. (chair)

John

[ckelloug]

John,

The tenth root of 10 and the cube root of 2 are both very very close to 1.26! De Larrard cites the former as the ratio for sieves in the Renard series while you cite the latter.

Smashing on the zeeospheres (hard to do at 60ksi 413MPa) and I still owe you a PM. Thankfully the local building officials have approved my plan to electrify my shop albeit in a slightly different manner than I had planned.

Regards,

Cameron

[greybeard]

John,

The tenth root of 10 and the cube root of 2 are both very very close to 1.26! De Larrard cites the former as the ratio for sieves in the Renard series ....

So they are !
Still, I wonder what the logic behind his(or Old Foxy's) choice was.

Re my pm. If you took one particle of each size, starting with first being one unit volume, in an infinite sequence that halves the volume of each particle each time (see my choice of sieve), and pushed them all together, they would exactly fill a cube of volume 2 units. :wee:

Mind you, I haven't yet worked out how you fit them together, that's a mere detail.

Regards
John

EDIT
This has a minor detail missing - it's wrong.
While the total volume is 2 units, you can't even fit the first two particles into a 2 unit cube. The sum of their diameters = the diagonal of the cube, so they will bulge out through the faces of the cube.
Ho hum.

[lgalla]

Walter you cannot substitute aerosil as cabosil 530 claims not to be as thixotropic as other fumed silicas.I still say a thickener will cause more wetout problems.



B.E.T. Surface Area 205–245 m2/g
Carbon 4.25 ± 0.5%
pH (4% slurry; 1:1 v/v 4.5–6.5
isopropanol/water)
Bulk Density* 4.0 lb/ft3 max.
*At time of packaging.
Test methods available upon request.
The typical properties of TS-530 include:
Appearance Fluffy white powder
Specific Gravity 2.2 g/cm3
Wt. per gallon 18.3 lb
Refractive Index 1.451
X-ray Form Amorphous
Average Particle 0.2–0.3 microns

A10lb bag is 2'X2'X4'.This appears to be a lightweight filler but is not.The weight/gal states 18lbs assuming a solid,but the fluffy nature of the material contains maybe 90%air.From the specific gravity and the bulk density,a gallon
of fumed silica is about 10 Oz.
Cameron on the5.355parts by volume,would that be based on the SP at 18.3lbs/gallon or the bulk density at 10 Oz?

[ckelloug]

The model I got from de Larrard bases everything on bulk volume so the 5.355 parts is by volume uncompressed as it would be as poured from the bag. Probably good to weight out the quantity poured so that you can use a mass basis to repeat the formula. Silica fume is a bit of a problem like carbon black in that it for the best effect, it needs to be dispersed in a ball mill for a while before using it (which will effect the bulk density) but I digress.

Why oh why do you guys only read my formulas and consider duplicating them when they're prefaced with "this won't work but. . ."?

The main reason that this formula has silica fume and so much nanopox is that the hand calculation for the smaller grading fractions was a minor botch job but I found tweaking with the simulated fractions that it could be corrected by adding some silica fume and nanopox. I'm working on some math to optimize the quartz mix so we should be able to do better than this last formula with more easily available materials although it's tough sledding for me as I retrain my brain in math I haven't done lately.

Regards all,

Cameron

[greybeard]

I thought it would be a problem acquiring different materials to put together an experimental mix, but I now see a far greater problem is dealing with the supplier's data.

Agsco quotes % retained through standard US sieves so I can work out what size ranges are present in what approximate quantities.

However, 3M Zeeospheres data sheet quotes in "percentile". I recognize what that is, but I'm having difficulty in converting this into figures that I can use.
For G-800, they give -
95th percentile 200 microns
90th percentile 100 microns
50th percentile 40 microns
10th percentile 12 microns

Then the Fillite data sheet is laid out differently again. By quoting what passes through the`sieve, their figures seem to indicate what isn't there, rather than what is.
passing 500 microns 100 %
" 300 microns 85 - 100%
" 150 microns 30 - 80%
" 106 microns 25 - 55%
" 50 microns 2 - 10%

At least 3M has given an average figure, but with the Fillite figures, you could assume anything you like, and leaves me baffled.

Any help, much appreciated.

John

[ckelloug]

I thought it would be a problem acquiring different materials to put together an experimental mix, but I now see a far greater problem is dealing with the supplier's data.

Agsco quotes % retained through standard US sieves so I can work out what size ranges are present in what approximate quantities.

However, 3M Zeeospheres data sheet quotes in "percentile". I recognize what that is, but I'm having difficulty in converting this into figures that I can use.
For G-800, they give -
...
John

John,

This is very similar to what agsco does, it's just written up in fancier language.
I've modeled this in a couple of different ways. I'm currently interpreting the zeeosphere data sheets to mean 5% retained on a 200 micron "sieve", 5% retained on a hundred micron "sieve", 40% retained on a 40 micron sieve, 40 % retained on a 12 micron sieve and 10% smaller than 12 microns. I don't think this is a hideous model but it's probably not the best way. My code is even worse because it's actually implementing the equivalent of 5%=200microns 5%=100 micron 40%=40 micron 50%=12micron because I don't have an easy way to model the less than 12 portion.

A better way would be to say that if 10% is between 0 and 12 microns then just assume that 10% of the mix is 6 micron particles. To truly get medieval on the problem, one should probably fit a normal curve to the data and perform interpolation but that's farther than I'm planning on going.

**Do Note that that the numbers you have given are For G850 zeeospheres not G800**

Looking over the fillite data sheet, the fillite material corresponds to 3M's glass bubbles product which is a bit unlike zeeospheres. The zeeospheres being looked at have a crush strength of around 60,000psi while the fillite bubbles are only around 3000psi. It's conceivably possible to have the fillite spheres fail in a load bearing application although it's probably unlikely with the intended use. The zeeospheres are much stronger.

As for their grading curves, the reason they give ranges in ranges is likely because their material is made from coal power plant fly ash. Since it is a natural material, it has a lot of variation and isn't as well characterized as zeeospheres.

Can't we just look at this as retained=1-passing for the fillite data? I'd pick the average of the two limits they give for each fraction's percent passed and assume that a group of particles in the middle of the size range at a percentage of 1-passing was retained. It may not be perfect model but if it looks like it would work on that basis then you have to redo the calculation at the upper and lower limits and see if it still works.

Don't worry about having difficulty visualizing, the grading curve work is plenty hard to visualize. I've got a 9x27 matrix that represents the properties of #6 AlO2, all the agsco quartz fractions down to 2/0, and G800 and G200 zeeospheres. I'm now feeding this to a numerical linear optimizer program (octave with glpk, they're both free) and asking it to produce the same percentage of each of the 9 grading fractions from my logarithmic grading curve with the constraint that if multiple material sizes contribute to that fraction then to assign each one 1/n of the volume in that fraction.

When this calculation gets through, I am then feeding that result into the compressible packing model simulator I wrote. My current result from the compressible packing simulator is 88.5% for vibration under pressure compaction for the following mixture of agsco quartz plus G800 and G200 zeeospheres.

#6 Brown ALO2 0.070167
#4 Quartz 0.134937
#3 Quartz 0.075202
#2 Quartz 0.117888
#1 Quartz 0.101797
#1/2 Quartz 0.060450
#2/0 Quartz 0.088723
#G800 0.175418 (Zeeospheres)
#G200 0.175418 (Zeeospheres)

Oddly enough, this simulation result tends to mirror Walter's data that lots of zeeospheres seem to be necessary. This is because the bulk of the zeeopsheres in both the G800 and G200 varieties influence only one of the fractions on the bottom of the logarithmic grading curve and all fractions are supposed to be equal. Note, that it's the actual size ranges that are supposed to be equal, not the amounts of bulk sand. The optimizing software I spoke about picked the percentages of the 9 raw materials that best optimizes all of the 27 fractions with respect to the grading curve.

It still may be easier to get sieved sand as this mixture is likely to have some tendency to segregate according to de Larrard because there is not enough sand in the mid range size fractions.

I'll do many more sim runs before settling on a mixture or getting one sieved as I think there are a couple things that can be tweaked in the optimizer to do a touch better.

--Cameron

Note: I haven't tested the above formula in the real world. It is output from mathematical modeling software I wrote which attempts to predict the packing density of aggregate using a model developed by the French Government and documented in "Concrete Mixture Proportioning, a Scientific Approach," by Francois de Larrard.

[greybeard]

John,

This is very similar to what agsco does, it's just written up in fancier language.
I've modeled this in a couple of different ways. I'm currently interpreting the zeeosphere data sheets to mean 5% retained on a 200 micron "sieve", 5% retained on a hundred micron "sieve", 40% retained on a 40 micron sieve, 40 % retained on a 12 micron sieve and 10% smaller than 12 microns. .

Thank goodness. I came to this result as well after I'd posted.

**Do Note that that the numbers you have given are For G850 zeeospheres not G800**

Your right - tired eyes.

Looking over the fillite data sheet, the fillite material corresponds to 3M's glass bubbles product which is a bit unlike zeeospheres. The zeeospheres being looked at have a crush strength of around 60,000psi while the fillite bubbles are only around 3000psi. It's conceivably possible to have the fillite spheres fail in a load bearing application although it's probably unlikely with the intended use. The zeeospheres are much stronger.

As for their grading curves, the reason they give ranges in ranges is likely because their material is made from coal power plant fly ash. Since it is a natural material, it has a lot of variation and isn't as well characterized as zeeospheres.

I found it a bit off-putting when my supplier couldn't tell me what grade he'd supplied. He hadn't even noticed on his own site that his data sheets give 6 different ones. So I went to the Fillite web site where they give an impression of very tight control over the particle size specs, then give data like that !

Can't we just look at this as retained=1-passing for the fillite data? I'd pick the average of the two limits they give for each fraction's percent passed and assume that a group of particles in the middle of the size range at a percentage of 1-passing was retained. It may not be perfect model but if it looks like it would work on that basis then you have to redo the calculation at the upper and lower limits and see if it still works.

Again I agree that this is the easiest way to cope with the given info, but then I didn't expect life to be easy.

It still may be easier to get sieved sand as this mixture is likely to have some tendency to segregate according to de Larrard because there is not enough sand in the mid range size fractions.

Yes, I think this is where I've reached.

I've done a series of sievings of the batch of "builder's ballast" that is my starting point, and done some experiments along the lines mentioned earlier of getting quarts into pint pots with different sizes of sand fractions.
I now tend towards the idea that I will have to make a "size separator", not only to find out what is in the mix, but to make good its deficiencies.
The design is now quite simple, so if I get a chance to build it this weekend, I'll post up pics.

One thing though has come up in the course of my deliberations.
I think both Larry and Walter have mentioned aggregates that "look good" from the way they are tightly packed.
However, I think that when you look at a mixture, the eye will concentrate on the distribution of the biggest particles, and the rest will tend to merge into a simple background mass. I know this isn't strictly true, but go with it for the moment.
If we are looking for a mix containing say 10% of the biggest range, and the next range is typically 1/root2 times smaller in diameter(the sieve sequence), and so on, even though all sizes are actually present, I feel that the eye will only notice the biggest. This leaves 90% of the volume taken up by smaller particles. I think the ideal mix will actually look as though the biggest particles are isolated from one another.

I calculated the distance between the surface of two identical spheres in a space where the particles took up 10% of the available volume, and the result is 0.75 of the diameter of each - quite a gap.
This was based on cubic packing rather than tetrahedral, as it was simpler, and would exaggerate the result, but is the point valid ?

Regards
John

[walter]

I think both Larry and Walter have mentioned aggregates that "look good" from the way they are tightly packed.
However, I think that when you look at a mixture, the eye will concentrate on the distribution of the biggest particles, and the rest will tend to merge into a simple background mass.

Here's the picture of a "good looking" mix (from compaction standpoint).

Mix on the left looks random and not very exciting, but when you press it forms incredibly solid void less block. The mixture was created purely by stirring and feeling how tightly it packs. Stirring the mix on the left is effortless, the one on the right- almost impossible. Once compacted it takes a lot of energy to bring it back to uncompacted state. That's my definition of a good mix.


Here's something about Zeeospheres..

You're all assuming that they are there for filling purposes. They're not. They create flow- most important property of your mix. They make things possible that otherwise wouldn't be. Their size/hardness is just an added bonus.
_

[Zumba]

Anyone have any idea what the strength/rigidity differences are between a mix with say 7-10% epoxy vs. 15-18%? Some people have mentioned cost savings as one of the reasons to go for the lower %. However, beyond a certain point, there is no doubt an increase in cost of equipment and labor.

Also, I'm very curious as to the process by which the countertop guys get the epoxy qty down to 5%-7%. I'm thinking, mix it all up at a HIGHER epoxy percentage, maybe 15%, to ensure that there aren't any dry spots, and then compact it down in the expensive equipment. Excess epoxy would "squeeze out" and float at the top where it can get scooped up. It'd work well for producing a good product, but obviously saving money on epoxy would not be one of the goals, since much of it would be thrown away.

Someone here posted way back in this thread a PDF link to Anocast products... I believe they were E/G machine frames made by GE or someone. Anyway, the young's modulus of their product is 5250 ksi, about half that of Aluminum. I wonder how Walter's cool samples compare?

[ckelloug]

Walter,

Thanks for the good pictures. Near as I can tell, my theory results tend to be agreeing with you despite the fact that our packing theories came by very different means. Zeeospheres are indeed round which means that they improve flow properties, at least when there are enough of them to surround the larger particles. The other interesting thing about zeeospheres is that they are very small and they are also well characterized as to their size. As a result, they are an excellent and very high strength filler in the lower size ranges. I don't trust the agsco 3/0 and 4/0 quartz because most of the particles are in the Pan class which is not characterized and as a result, I am not using them.

I've been running my simulations using zeeopsheres to fill in the fractions at low size range because they are the right size. Because they will be one of the largest fractions the way the optimizer is solving the problem, their ball bearing effect will be maximized. I must say that I doubt that the mix in the last formula I published will flow much if at all because the solids density is so high. I've been trying to think of ways to construct pieces of this material but this is one case where until seeing the stuff live, there's no way to figure out any more. I'm actually contemplating crazy stuff like alternating ribbons of epoxy and dustings of material in the mold and then compressing the hell out of it and vibrating.

If I knew what you had in stock, I'd try running the optimizer on it and see what comes back.

Best Regards,

Cameron