[greybeard]

Very tempted to feed the trolls, but I don't think I will.
John

[Fireman11]

Has anybody made a machine component from this stuff yet? If so how did it work?

[veteq]

Has anybody made a machine component from this stuff yet? If so how did it work?

Yes multiple people have made one, worked out ok, you should read back a few pages, dont remember what page it was..?

[ckelloug]

Hi John,

I don't think Kevin is trolling, he just sounded like it. He made a good point.

Kevin,

The simple models like Isahi-Cohen(I've misstated that this was based on Hashin Shtrikman in the past as I got confused), which is actually what the graphs I've posted are based on, assume no interaction between particles and that has been the theory I've based my arguments on. I have in the past interpreted the models to mean no interaction was actually present but on further review based on your question, the correct interpretation is that no interaction was considered in the models, not that the interaction wasn't present. I have to concede that Romanlini and others are right that interactions between particles play at least some role at higher packing densities-- Having pulled Isahi and Cohen's original paper today, the modulus model I've used in my previously posted graphs is only rigorously valid for packing densities below 52% which is about where jamming starts to occur.

You are right that my argument for no particle contact has inconsistencies. They come from a change in my own understanding as time has gone on. My comment about spheres covered in a uniform layer of epoxy is really only an explanation of the Hashin-Shtrikman model, not a correct description of what happens in a real composite, as I had thought.

I would contend at this point that whether there is direct particle contact or not, as the epoxy layer thickness goes to a small value (not necessarily even zero), the behavior of the composite changes and approaches that which we visualize as direct particle contact whether it is literally correct or not. What actually happens in the composite is proving quite messy!

The following paper, which I haven't read in detail yet because the math is very difficult presents the case for what happens when interactions between the particles are accounted for.: SpringerLink - Acta Mechanica, Volume 215, Numbers 1-4 Note that this paper is from 2010 which suggests that exactly what happens in this class of composite is still not well understood.

I'll try this argument out on you guys:
Assume the Hashin-Shtrikman equations present a valid lower bound for modulus in cases without interaction between particles. This is generally accepted in the literature. It appears safe to assume that interaction between particles produces an increase in modulus based on the graphs in the attached paper. This would suggest that the lower Hashin-Shtrikman bound still provides a good barometer for the lower bound on the bulk modulus of the composite whether or not particle-particle interaction occurs.

I'll try to post some new graphs of predicted modulus based on the actual Hashin-Shtrikman bounds when I can find some type of empirical estimate for the shear modulus of epoxy given the flexural modulus I have. At worst, I may have to develop a way of measuring shear modulus based on the instrumentation I have.

I've exceeded my time quota for E/G today so it's back to the silicon mines for me.

Regards all,

Cameron

[brunog]

It has been my experience that E/G with very high aggregate to epoxy ratios is quite difficult to mix adequately. Something like the industrial equivalent of a kitchenaid mixer is probably going to be required if we don't come up with a better way.

Cameron,
go back to post #3971, the Scott Baden video shows mixing equipment than can be easily replicated using 5 gallon plastic pails. The beauty of the machine is that the mixing blades do not move,the container is set on a turntable, also some vacuum degassing can be done while mixing. Simple yet efficient equipment that, in my opinion doesn't add much air to the mixture.

I wonder what Larry thinks about adding and mixing the hardener after mixing the resin with the particules??

Best regards

Bruno

[brunog]

From the perspective of the two models I work with*, it's much more difficult to get a high packing density with crushed particles. A given size of crushed particles tends to get Beta Packing Density Coefficients in the high 50's to low 60's. Round particles tend to be between the mid 60's and the theoretical max around 71%.

Cameron,
There is something I don't quite understand, I'll try to put it in perspective:

Consider round particules and crushed particules of the same density.

The information provided by Hexion shows that the finished density of the crushed particule mix is higher than the round particule mix. (245Kg/l vs 235Kg/l) Young's modulus and compressive strength are also higher on the crushed particule mix.

Doesn't that imply that the crushed particule mix has a higher packing density coefficient than the round particule mix? Or have I missed something somewhere

Best regards

Bruno

[ckelloug]

Bruno,

Agree that the mixer in the Scott Bader video would be a good way to go I forgot about it. I merely meant in my last post that it would have to be something powerful, not just a bit of tumbling. A full vacuum mixer like that might be a bit expensive.

It is an interesting point that you make about the Hexion datasheet reporting lower properties for the round mix. I totally missed that. Because the round mix has a much smaller maximum size than the crushed mix, it's not an apples to apples comparison so it's hard to judge.

I almost wonder if they confuse which mixture was formula 1 vs. formula 2 between the various graphs and tables. I say this because they show the glass transition temperature being higher for Formula 1 yet the modulus as being lower which seems totally inconsistent with the way that this normally works. A piece whose glass transition is higher usually has a higher modulus than one whose glass transition is lower.

Now I'm really confused.

Regards all,
Cameron

[lgalla]

Bruno,
I wonder what Larry thinks about adding and mixing the hardener after mixing the resin with the particules??

Larry says bad idea.That video is calcium carbonate or CaCo3 with polyester resin.The ratio is 50/50.Notice how pourable the mix is.You would never be able to mix E/G in that system.Mixing the epoxy resin first with the aggregate and then the hardner will result in uncured or under cured epoxy.I have posted many times the correct way to mix epoxy and aggregates,so I will not repeat unless asked to.
I am confused also.The crushed aggregate may have an aspect ratio which would explain the higher modulus or strength.Possibly the rounded aggregates are not perfect spheres resulting in lower modulus and lower packing density.Why not have a mix of the two?
Larry

[ckelloug]

Hi lgalla,

Good to see you around.

I've seen a few papers that suggest that slightly oblong aggregate actually pack better than spheres. I'm working with crushed right now because crushed is by far the most common form.

More books=More Questions!

Regards all,

Cameron

[RomanLini]

I think the solution to some of these problems is going to be a lot more practical and a lot less theoretical.

It's all well and good to theorise on epoxy layer thickness and particle size but the solution I think will be to provide sufficient compressive force (as has been said before) while vibrating. This will expell the excess epoxy (reducing epoxy %) and at the same time provides a greater amount of direct contact between particles as the forces at the contact points will be way beyond the van der Waals forces etc. There's a huge difference between particle+fluid behaviour at 1g and at 50g or 500g.

Personally I'd be doing a lot more compressing, curing and testing and a lot less reading. And I say that with respect, it's just that I think the reading part is about the first 10% of the process...

[Nivea]

$$$ouch$$$
Regards, and good luck
John

You weren't kidding I see!



RomanLini,

Couldn't agree more regarding the practical nature of this process. Then again, I don't have a choice. If I were as talented and dedicated as Cameron with the books, well at least I'd have the option to theorize. I have no choice but to get my hands dirty to try crack this nut. It reminds me of room acoustics for studios and concert halls. What "should" sound right according to simulations and theory very rarely does. Too many variables.

Regarding compression, I just sliced a cured sample I made with compression. I was quietly confident it was going to be void free...... :tired:
All that talk about 100 ton presses etc, well I set up a simple cylinder with a piston jig in a large G-clamp and as I just started to compress the EG under the piston it could be heard loud and clear, tink, tink, tink. Aggregate breaking! I continued to swing anyway hoping at least to squeeze air out. Nope, still voids. Lesson learned- aggregate should not be "compressed" in the hope that it will gently pack. Very lightly continually compressed under vibration makes more sense now.
So I spent the remainder of the afternoon making a small vibration table to take my vacuum canister. In 30 mins I will have an epoxy rich sample vibrating under vacuum in the hope that it compacts well and de gasses. But somehow, I just know it wont......

[greybeard]

Cameron, you're quite right re trolling.
Being pedantic myself triggers an adverse reaction when I see someone else making a statement that I consider pedantic, so I hope Kevin will accept my apology for any inferred criticism of his post.

John

[johnohara]

Interesting story on SlashDot yesterday.

DoE Develops Flexible Glass Stronger Than Steel
"The Department of Energy Office of Science recently collaborated with the Lawrence Berkeley National Laboratory and the California Institute of Technology to develop a resilient yet malleable new type of glass that is stronger than steel. The material can also be molded, and it bends when subjected to stress instead of shattering. The glass is actually a microalloy and features metallic elements such as palladium. This metal has a high 'bulk-to-shear' stiffness ratio that counteracts the intrinsic brittleness of glassy materials. The team that developed the material believes that by changing various ratios, they could make it even stronger."

DoE Develops Flexible Glass Stronger Than Steel - Slashdot

Maybe the Dept. of Energy has been reading this thread.

~John

[Kevin_Johnson]

Cameron, you're quite right re trolling.
Being pedantic myself triggers an adverse reaction when I see someone else making a statement that I consider pedantic, so I hope Kevin will accept my apology for any inferred criticism of his post.

John

Accepted.

[Fireman11]

I think the solution to some of these problems is going to be a lot more practical and a lot less theoretical.

I wasn't being snide. I think some more experimentation would be a good way to move forward. Some problems would inevitably turn out to be small. and perhaps some new one would crop up.

If one is trying to move a DIY process forwards, thats the way to go. If you are trying to duplicate the engineering process of large companies, its sort of tough.

I ask because it is potentially a great thing for DIY folks, but if you need industrial tables and autoclaves etc its isn't really a DIY process. From were you guys are now could I reasonable make a "low stress" part like a router table. Could I make a "high stress" part like a mill column or gantry.

Keep up the good work.

(Also) Kevin

[Kevin_Johnson]

Cameron,
go back to post #3971, the Scott Baden video shows mixing equipment than can be easily replicated using 5 gallon plastic pails. The beauty of the machine is that the mixing blades do not move,the container is set on a turntable, also some vacuum degassing can be done while mixing. Simple yet efficient equipment that, in my opinion doesn't add much air to the mixture.

I wonder what Larry thinks about adding and mixing the hardener after mixing the resin with the particules??

Best regards

Bruno

The video shows larger versions of equipment that have been used for many, many decades in the dental business to mix dental stone. One trade name is Vac-U-spat.

If you do a patent search on ways to deaerate mixtures one worthwhile and common technique is to apply the vacuum near the central axis of rotation. This is because gases normally migrate to the center being less dense (again, look at hydrocyclones). It is practical too if you want to avoid sucking epoxy into your expensive vacuum motor.

In the 928 Porsche the engineers incorporated such a vent into the oil pump albeit sans vacuum. The input shaft clearance was used as a metered orifice with a diversion back into the sump. Porsche never claims the use of this technology in this engine. It is left to the interested student to analyze the design and deduce this.

Drawing again from automotive design it would be wise to have an intermediate container that would allow fluid epoxy inadvertently drawn into the vacuum line a place to collect and later be disposed of. This is known as an oil catch can for a crankcase ventilation system.

[ckelloug]

Roman,

I agree totally that there is much experimentation required beyond the theory. I have quite a lab here at this point and have done a fair number of experiments to confirm basic properties of some materials I'd like to use. I can't post all the experiments I do because many involve data which was provided to me under NDA from vendors. I make it a point to share all the theory knowledge I collect in the process and any practical knowledge that isn't proprietary to my other projects.

I have to admit that part of my bent for theory comes from an ass-chewing I got many years ago from my materials science professor, Dr. Joseph King at Harvey Mudd College. We were trying to measure shot peening damage done to aircraft skin coupons that had been CO2 pellet blasted. When we went into the lab to try to quantify differences in surface hardness, we hadn't done the calculation needed to understand whether polishing for a Knoop microhardness test would obliterate the layer whose properties we were trying to measure. I argued that I had no idea what the layer would look like so calculating it was pointless. Dr. King said something to the effect of, "Real engineers calculate the approximate value of what they're trying to measure before they try to measure it." I flailed my way through the microhardness test since we barely polished the specimens in hopes of not destroying what we wanted to measure. In the end, we performed what was probably not the world's best microhardness test and I learned the lesson that in serious endeavors that it pays to know as much as you can about the problem before you hit the lab hard.

The theory I've studied tells us the following:

Packing Density is the only strongly dependent parameter in composite modulus thus maximizing packing density is the key to achieving a high modulus.
Achieving a modulus in keeping with commercial products based on quartz requires a packing density that is almost certain to be greater than 85%. At high packings, a 1% difference in packing density can make for a 10% or more change in modulus. Maximizing packing requires the use of a wide range of aggregate sizes. The optimal size distribution for the particles is not predicted by any of the common concrete design models. Spheres pack better than crushed aggregate as shown in dozens of measurements by de Larrard on actual aggregate. Once particle density gets higher than 52% for spheres, jamming starts to occur: while the exact onset of jamming is not predictable for a given mixture the fact that it will occur has implications on mixing strategies. Finally, coupling and deairing agents can improve strength, modulus, and mixibility to a surprising extent.

Philosophy aside, I've rerun the graph of minimum expected Young's modulus for quartz composite based on Hashin-Shtrikman and also did a comparison to Isahi-Cohen. Hashin-Shtrikman predicts a lower minimum modulus at high fill percentages than Isahi-Cohen ignoring the effects of particles in contact or close proximity. From what I've read, most composites tend to follow the lower Hashin-Shtrikman bound fairly closely until higher fill percentages where the modulus goes well above what Hashin-Shtrikman would predict.

Due to blind luck, the Isahi-Cohen's model's divergence from the lower Hashin-Shtrikman bound is positive and based on my very qualitative interpretations of that Acta Mechanica paper, seems likely to fairly accurately predict the modulus of the composite with particle interactions at high fill rates though the risks from extrapolation can bite us.

Nivea,

Interesting observations re compression and getting the air out.

Regards all,

Cameron

P.S. Enjoy the new graphs

[Kevin_Johnson]

I wasn't being snide. I think some more experimentation would be a good way to move forward. Some problems would inevitably turn out to be small. and perhaps some new one would crop up.

If one is trying to move a DIY process forwards, thats the way to go. If you are trying to duplicate the engineering process of large companies, its sort of tough.

I ask because it is potentially a great thing for DIY folks, but if you need industrial tables and autoclaves etc its isn't really a DIY process. From were you guys are now could I reasonable make a "low stress" part like a router table. Could I make a "high stress" part like a mill column or gantry.

Keep up the good work.

(Also) Kevin

I started my business as a student. I converted my old Makita table saw into a fixed router table by creating a poured polymer surface on a scrap piece of wood -- inspired by such tables in restaurants and lack of funds to just go out and throw money around. One of my old chemical business contacts made these tables/surfaces as well.

[Kevin_Johnson]

Aside:

In discussions of theory, one of the easiest and most devastating things you can do is to make a circular argument where you (presumably unknowingly) assume or import what you are trying to demonstrate.

To avoid philosophical situations like the cat being and not being on the mat simultaneously scientists and engineers usually retreat to probability theory.

[ckelloug]

Hi Kevin_Johnson,

Others pointing out when you are being a dumb-ass is also a great help. :cheers:

I'm currently developing a system for assigning R&D money at the company I work for based on heavily probabilistic arguments calculated rigorously.

The Compressible Packing Model for aggregate packing is semi-empirical and has a published error margin of 3%.

Hashin-Shtrikman for modulus is based on a calculation of strain energy for a representative area in the composite and has no probabilistic nature beyond the representative nature of the area for which the strain calculation is done.

The more sophisticated model in Ju and Yanase posted a few posts back uses a sophisticated probability model for near-field particle interactions to get an expected value for the composite properties but doesn't explicitly expose statistics in its output.

As far as whether spheres (or aggregate in general) directly contact or not is most likely probabilistic however arguments based on surface energy suggest that with sufficient epoxy at equilibrium, the lowest energy state is for the aggregate to be fully wetted.

I would expect at least a one epoxy molecule layer between all spheres in a well mixed mixture.

After reading 700 pages of Israelachili's book on surface forces, there can be significant interaction between aggregate particles even if the surfaces are farther apart than what macroscopic arguments would consider directly touching.

I agree with you that the actual thickness of epoxy between any two particles is likely probabilistic in this system however I think that zero is of near-infinitely lower probability than thicker layers.

Germane to the original point we were discussing with Romanlini, I would argue from my current readings that it doesn't actually matter whether there is a single molecule epoxy layer between the aggregate particles or whether they are in direct contact:

Epoxy layer or not, the particles still can interact significantly. Roman was right that there was interaction that I neglected and thus my conceptualization of the situation was wrong. I wish I could weasel out of being wrong via a statistical argument but I simply missed a force in this case.

Fireman11,

I'm currently working with those here trying to find the parameters for material that works and meets the specifications of commercial material using my modestly equipped industrial lab. I'll pay more attention to making it DIY practical later when I can reliably make material that works. My own practical attempts have contributed far less than my theoretical ones up to now. Others here have been far more successful than I have at a practical level of actually making E/G.

Regards all,

Cameron