[rusel]

greybeard
John
The heading "Phenolic/basalt" is a little misleading as we change resin type just before manufacturing commenced. We are now using a vinyl ester resin generally used for infusion moulding as I felt this had the right properties and pricing for this job over the phenolic. We are still using basalt rock for the fill as this is ease to get and it packs down well.
Hope this clear up any miss understandings

johnohara
John
Yes I have seen this site and it was one of a few that help to encourage use to move to making usable parts. His part are very well plan out , made and the finish is perfect looking as nothing beats smooth and shiny.
On are next run we are looking at making a whole mill using this product. We will be paying much more attention on the finish this time as we are now comfortable with the manufacturing side.

Russell

[RomanLini]

Cameron at 3g/cc for the granite and 1.1g/cc for the epoxy doesn't it look like this;
12.5% 125g epoxy /1.1g = 113.6cc
87.5% 875g granite /3g = 291.7cc
total 405.3cc therefore 113.6/405.3 = 28.0% epoxy by volume?

It's probably my memory playing tricks but I thought your optimal ratio had MUCH less epoxy volume than 28 percent?

Cameron and Brunog- sorry I didn't know that type of centrifugal machine had already been discussed. I was bringing it up more for examination of the principle, ie using 30 or 40 G force to mix the epoxy granite rather than a mixing "stick". It might not be that hard to build a larger one that can do a couple of kg at a time.

I saw some demo photos a few years back where they put 2 colours of solid clay in one (as two big lumps) and after 30 seconds it was fully mixed... It's pretty impressive what 30 G does, the clay flows like water.

[Kevin_Johnson]

...So zero epoxy there equals maximum stiffness, and you can still have plenty of epoxy in the voids. You could even have large particles only (re one of my first posts) physically compressed to form a matrix of non compressable material which would have a very good stiffness modulus with the epoxy just filling the voids and gluing it together. From that point on the more smaller particles you put in the voids it basically just gets stronger, but probably not much stiffer.

As an analogy consider a ball bearing. Lubricant separates the balls from the bearing race with a film of lubricant always between the hard components. But with application of enough pressure the lubricant is squeezed out and solid contact of the components occurs.

Yes, exactly. But think about the loaded ball bearing with no lubricant present. With no movement the lubricant will not wet the contact point.

[Kevin_Johnson]

... This is the crux of the problem in making E/G I think. As explained below, I think the answer is that you can't get the epoxy between the particles to reach zero regardless. The Hashin Sthrikman model for modulus in a particle composite suggests that if you could, the modulus of the resulting composite would be near that of the base material.



In de Larrard's book he shows that there in an asymptote that no reasonable combination of vibration and compression can get the voids below for a given mixture. Thus he says the packing density never approaches 100% for real mixtures. I think this means in practice that you never reach the point where there is zero epoxy between particles. I think we all agree however that the more aggregate and the less epoxy we have, the stiffer the resulting part. From deLarrard's book, it appears that 92% is the best obtainable packing density with vibration and compression with the dry particles. The asymptote is several percent higher.

I've been using de Larrard's packing model with the parameter K set to 9 which equates to vibration at around 4g with a pressure applied by a weight at 15kPa. This provides probably 5% less than the unreachable asymptotic value and is pretty much the extent to which you can reasonably compact a mixture without going to high pressure presses etc.

In practice, you CAN reach the point where there is zero epoxy between SOME of the components. This is an apparent limitation of deLarrard's packing model. It does not appear to apply to a preloaded matrix of large aggregate which does not yet have epoxy applied to it. You will need to make your preload sufficient to resist movement during vibration.

Much, much smaller aggregate than the preloaded matrix will behave similar to a liquid in filling the voids under vibration.

Likewise, a thin epoxy can then be used to wet out the resulting loaded aggregate mix.

Order of operations is important when evaluating theoretical limitations.

Edit: I see the preloaded matrix is the theoretical basis for infusion molding.

Edit 2: I am musing/guessing that a sort of inverse of the deLarrard packing model would be that there exists a probability of a given percentage of unloaded aggregate such that the piece will "easily" (qualitatively) move in at least one plane of motion. This is trivial in the case of much, much smaller aggregate being added to a preloaded matrix. It is more significant when considering the initial matrix.

[ckelloug]

Hi Roman,

You are correct that there is a hell of a lot of epoxy in my last few samples. I wasn't shooting for optimal, I was trying to recreate Nivea's results with my own aggregate and the additives.

I ran several more experiments on Sunday involving zeeospheres. Due to the aggregate I have on hand, the model suggests that 18% zeeospheres by volume are required to achieve 88% packing density. A Non-optimality I measured in various sizes of this angular aggregate pushes the packing lower than desired.

My first experiment with the mixture where the steel plate capsized was 300g epoxy to 1.6kg aggregate for a weight/weight percentage of 15.7%, or 37% by volume.

The second experiment I did which you mentioned in your post was 200g epoxy with 1.6kg aggregate for 11% by weight or 28% by volume.

I completed one more successful experiment yesterday which was 150g epoxy in 1.6kg aggregate for 8.5% by weight. Using the exact numbers I get 22.5% by volume. This sample was behaving like bread dough and my small mixer which is really only for liquid epoxy couldn't come close to dealing with it. I finally removed the mass from the beaker and kneaded it by hand on a scrap chunk of UHMW. The sample flowed nicely once heated and it looks like the aggregate dispersed pretty nicely with vibration, vacuuming and the cure cycle in the oven.

I also did 2 other experiments trying to disperse 295g of zeeospheres in an epoxy mixture with titante and A525. With the mixing power available, I was only able to disperse between 160g and 200g in 100ml of epoxy. I had tried earlier to disperse all 295g into 80ml of epoxy mixture but I dumped the zeeospheres in too fast all I got was coarse powder of agglomerated zeeospheres and epoxy that wouldn't mix at all.

It looks like it might be possible to make my target density with the high zeeosphere content, but it will require a much higher shear mixer than I have in house right now. On the other hand, it might be worth examining the model results and picking some different aggregate sizes so as to avoid the need for so many zeeospheres.

On a humorous note, I have the market cornered on synthetic dog turds because these are exactly what the super stiff zeeosphere epoxy logs came out of the oven looking like.

Bruno,

I plan to test the 2 normal samples and parts of the third with the plate stuck in it as soon as I can. I need to get that antique surface grinder running with a diamond saw blade in it so I can do a good job of the cutting. I'm hoping to get the tests run by the end of February.

The mixer you have looks nice but the E/G I'm mixing is way stiffer than any concrete I've ever seen. I think a much higher shear mixer than a single rotating object is required for the stuff I'm testing because the E/G simply moved out of the way of the rotating impeller of my little lab mixer. In order to get the mixture I'm working with in the current experiment series to work, I'd need a compulsory mixer perhaps like the Husky Johnohara mentioned, or a SERIOUS mixer like a Banbury or a twin screw.

Kevin,

The compressible packing model values are for dry aggregate vibrated at 4g and 10kPa of pressure. The model predicts the packing density achievable. It utilizes data from 20 years of packing experiments conducted by the French Bureau of Roads and Bridges and is semi-empirical rather than purely theoretical.

I was wrong where you quoted me in your last post as the CPM has no way of telling us that there are or aren't particles with no epoxy between them. The only thing the model itself tells us in addition to packing density of a mixture is that reaching 0% voids is not possible for any realistic compaction scheme.

I agree that order of operations is important. I can see getting aggregate to aggregate contact might be possible with vacuum infusion into a compacted aggregate mass. I still think that in the case of mechanically mixing the epoxy and aggregate, there is probably at least a molecular monolayer of epoxy on all the aggregate if you have used enough epoxy.

Regards all,

Cameron

[brunog]

Cameron at 3g/cc for the granite and 1.1g/cc for the epoxy doesn't it look like this;
12.5% 125g epoxy /1.1g = 113.6cc
87.5% 875g granite /3g = 291.7cc
total 405.3cc therefore 113.6/405.3 = 28.0% epoxy by volume?

It's probably my memory playing tricks but I thought your optimal ratio had MUCH less epoxy volume than 28 percent?

Cameron and Brunog- sorry I didn't know that type of centrifugal machine had already been discussed. I was bringing it up more for examination of the principle, ie using 30 or 40 G force to mix the epoxy granite rather than a mixing "stick". It might not be that hard to build a larger one that can do a couple of kg at a time.

I saw some demo photos a few years back where they put 2 colours of solid clay in one (as two big lumps) and after 30 seconds it was fully mixed... It's pretty impressive what 30 G does, the clay flows like water.

Roman,
you are right there is close to 30% epoxy by volume in the mix at 8% by weight it's close to 20%

The centrifugal mixers are pretty impressive indeed. I believe i've seen a video a while back and there are photos of white and red clay at different mixing stages. I wonder how many G force is produced by a 5 gallon paint shaker.

Best regards

Bruno

[Kevin_Johnson]

Kevin,

The compressible packing model values are for dry aggregate vibrated at 4g and 10kPa of pressure. The model predicts the packing density achievable. It utilizes data from 20 years of packing experiments conducted by the French Bureau of Roads and Bridges and is semi-empirical rather than purely theoretical.

It is a limited model, this is certain. My initial simple experiment exceeded the pressure applied 10 fold (a mere one bar). I understand that applying one bar of pressure to a random curing concrete mass is unusual even over 20 years of collected data. Not so unusual for pavement using a different binder (think superpave) that has been around for a while too.

I was wrong where you quoted me in your last post as the CPM has no way of telling us that there are or aren't particles with no epoxy between them. The only thing the model itself tells us in addition to packing density of a mixture is that reaching 0% voids is not possible for any realistic compaction scheme.

He later moves to virtual packing density where 0% voids is just a special case of particle size/shape and the container dimensions.

I agree that order of operations is important. I can see getting aggregate to aggregate contact might be possible with vacuum infusion into a compacted aggregate mass. I still think that in the case of mechanically mixing the epoxy and aggregate, there is probably at least a molecular monolayer of epoxy on all the aggregate if you have used enough epoxy.

I don't agree with the latter except as a possible class of solutions. It is a demonstrably weak assumption for a given system every time you see a bubble pop at the surface of the epoxy or easily cause this with a prick from an instrument. Think about it.

[antsals]

Hi,
Been reading the many years of posts some which go back to early 2000. Now I'm really intrested in making myself a "epoxy granite" bed for a CNC Milling Machine.

I've had a look at the German example thats exactly what I want to do.

I want to do some test peices first but based on all the information in the forum, I'm guessing there are a few of you that are well down the line with knowledge in the product.

I'm based in the UK and am having problems on the products I need to use.

I've read totally different uses of aggregates and sand etc. From the experience attained where should I start and what products should I get hold of to begin testing? There is loads of different epoxys and I'm not sure what grade and where I should get the stuff from and what and where should I get aggregate from?

Local building merchant have some Grante Chips - C & W Berry Ltd - 6mm to Dust Grano - 25kg bag - approx

Any help would be good,
Ant

[RomanLini]

...
I completed one more successful experiment yesterday which was 150g epoxy in 1.6kg aggregate for 8.5% by weight. Using the exact numbers I get 22.5% by volume. This sample was behaving like bread dough and my small mixer which is really only for liquid epoxy couldn't come close to dealing with it. I finally removed the mass from the beaker and kneaded it by hand on a scrap chunk of UHMW.
...

Cameron, I'm surprised at your results with 28% and 22.5% epoxy by volume. I have made mixtures of various things and kept an eye on the epoxy percentage by weight and by volume, and at 25% epoxy by volume and especially 28% it should be *quite* wet and mix and pour fairly well.

I'm going to go out on a limb here and speculate that your fine particles are causing a big problem, making the mixture "fluff up" as all the small particles get an epoxy film on them and it gets solid like pastry. I've never used those very tiny particles.

Have you considered that your packing density figures might go totally out the window once you mix the epoxy in?

Here's an experiement, eliminate all aggregate particles under size X (1.5mm diameter?) then add epoxy and vibrate until you have finally added too much and you get evidence of epoxy excess on top. At that point you will have found the exact % volume of epoxy needed to fill the voids between your larger particles. I think you'll find 20% to 25% by volume will be plenty.

At this point you can surmise for packing density to be good, that you should be able to reduce epoxy volume and replace it with fine particle volume, ie the voids between large particles *will be the same size* but will be filled more with particles and less with epoxy.

Now try again with some or all small particles added, and again add epoxy (and knead it if you have to) but eventually get to the point where you get excess epoxy. At this point you can analyse the volumes of all the components, and see if the small particles filled the voids and your packing density worked, or if your packing density went fubar and you got a mass of tiny particles each with a sizable epoxy film around it.

ie; you will be able to actually measure the amount it has "fluffed up" if any.

My guess here is that there is no way you can use the really tiny particles in any significant quantity without destroying your packing model.

[Kevin_Johnson]

Hi,
Been reading the many years of posts some which go back to early 2000. Now I'm really intrested in making myself a "epoxy granite" bed for a CNC Milling Machine.

I've had a look at the German example thats exactly what I want to do.

I want to do some test peices first but based on all the information in the forum, I'm guessing there are a few of you that are well down the line with knowledge in the product.

I'm based in the UK and am having problems on the products I need to use.

I've read totally different uses of aggregates and sand etc. From the experience attained where should I start and what products should I get hold of to begin testing? There is loads of different epoxys and I'm not sure what grade and where I should get the stuff from and what and where should I get aggregate from?

Local building merchant have some Grante Chips - C & W Berry Ltd - 6mm to Dust Grano - 25kg bag - approx

Any help would be good,
Ant

You might consider buying some glycerin and practice the manipulation of the material you are considering. This allows you to develop handling technique with a very safe material that will not set up and could be rinsed away from your aggregate if you wanted to. Then move to the expensive epoxy resins that people will recommend. If you're already experienced in mixing plaster, mortar or concrete you can easily skip this step. Glycerin is used as a binder in many products.

[ckelloug]

It is a limited model, this is certain. My initial simple experiment exceeded the pressure applied 10 fold (a mere one bar). I understand that applying one bar of pressure to a random curing concrete mass is unusual even over 20 years of collected data. Not so unusual for pavement using a different binder (think superpave) that has been around for a while too.

The framework of the model handles other compaction conditions besides 10kPa and vibration. This is encoded into the variable K. K=9 corresponds to the above conditions. A higher K value would correspond to more pressure. One would have to do a series of tests to determine the K value for another condition. Even as K goes to infinity, the packing density doesn't go to 1.

In theory, you can find Beta from easily run tests at K=9 then use the results from calibrating your compaction technique's K value to predict the packing under those conditions. I used deLarrard's assumption from concrete that designing the aggregate for K=9 is sufficient. This is possibly a bad assumption and I will have to go back measure packing densities in something like C12-C14 glycidyl ether and see how the numbers work out as Roman suggested a while back. This is messy.

He later moves to virtual packing density where 0% voids is just a special case of particle size/shape and the container dimensions.

Virtual packing density is never attainable in the CPM framework. It is a theoretical construct helping to describe Beta for deriving actual packing density from the remainder of the model framework. CPM does come up with ways of accounting for container shape but these are important mainly when the container is less than 5 times the largest aggregate size.

I don't agree with the latter except as a possible class of solutions. It is a demonstrably weak assumption for a given system every time you see a bubble pop at the surface of the epoxy or easily cause this with a prick from an instrument. Think about it.

This is a fair statement but I think it conflates theory and experiment. If you have adequate mixing and sufficient epoxy, I don't see how you can fail to get a monolayer of epoxy on all the particles in theory. In practice, mixing may turn out poorly and things may go wrong but my own perspective is that models are helpful in telling what happens when things go right.

[ckelloug]

Cameron, I'm surprised at your results with 28% and 22.5% epoxy by volume. I have made mixtures of various things and kept an eye on the epoxy percentage by weight and by volume, and at 25% epoxy by volume and especially 28% it should be *quite* wet and mix and pour fairly well.

I'm going to go out on a limb here and speculate that your fine particles are causing a big problem, making the mixture "fluff up" as all the small particles get an epoxy film on them and it gets solid like pastry. I've never used those very tiny particles.

I Absolutely agree that the fine particles are causing a problem as far as raising the viscosity too much for the mixing to be easy. They might be fine when I find a higher shear mixer. The problem goes back to that Gupta paper cited eons back where he states that the ultimate measure of epoxy is how much is required to form a minimal layer on all the particles. This is proportional to the surface area of the particles. Zeeospheres have areas in meters squared per gram which makes it easy for them to mess things up.

Have you considered that your packing density figures might go totally out the window once you mix the epoxy in?

I wouldn't quite surmise that the model breaks but parameters do change. As I said to Kevin, I'm going to have to do the experiment you suggested and run wet packing tests. In CPM terms, adding the epoxy increases K. This will change the particle size distribution required somewhat but my instinct is that the epoxy should enhance the compaction rather than hinder it.

The small particles are a problem because they do not dry pack well due to entrained air as a result, the measuring piston approach used for other particles doesn't work for zeeospheres. I'm using Beta packing coefficients in the model for them of .66 based on their spherical shape. This assumption may in fact be awful so without wet pack tests, it's not really possible to know what's going on.

I would say that my current problem doesn't make me thing the model is broken yet, but it does make me think that this particular mix design is pretty bad. I chose a number of graded aggregates but the way I chose them made the choices farther and farther from ideal as particles got smaller. The optimizer thus relied on the smaller particles to fill in all those tiny gaps. As a result, it called for a lot of tiny particles. Since I don't know really know how the tiny particles behave in epoxy and just assume they pack like spheres, the mix design may be way off in this case.

Here's an experiement, eliminate all aggregate particles under size X (1.5mm diameter?) then add epoxy and vibrate until you have finally added too much and you get evidence of epoxy excess on top. At that point you will have found the exact % volume of epoxy needed to fill the voids between your larger particles. I think you'll find 20% to 25% by volume will be plenty.

At this point you can surmise for packing density to be good, that you should be able to reduce epoxy volume and replace it with fine particle volume, ie the voids between large particles *will be the same size* but will be filled more with particles and less with epoxy.

Now try again with some or all small particles added, and again add epoxy (and knead it if you have to) but eventually get to the point where you get excess epoxy. At this point you can analyse the volumes of all the components, and see if the small particles filled the voids and your packing density worked, or if your packing density went fubar and you got a mass of tiny particles each with a sizable epoxy film around it.

ie; you will be able to actually measure the amount it has "fluffed up" if any.

My guess here is that there is no way you can use the really tiny particles in any significant quantity without destroying your packing model.

This test series wasn't so much about meeting my density predictions but rather finding out what was mixable. I think with high enough shear mixing, the problems I am seeing will go away. Better choices in the larger aggregate would lower the small particle requirement and ease the problems too.

One technique I've seen in the literature is to place a carefully weighed sample in a crucible and burn the epoxy off in a furnace. Carefully weighing the remains will give a good picture of what was actually obtained.

Keep in mind that almost all of my aggregate is below 1.5mm in size. Fewer fines will definitely make the process easier but I want to find out if the predicted mix works given good enough mixing.

I will take wet packing measurements and see what changes. I appreciate the time you took to comment on my results and look forward to posting some more results when I get some more lab time.

Regards all,

Cameron

[Kevin_Johnson]

It is a limited model, this is certain. My initial simple experiment exceeded the pressure applied 10 fold (a mere one bar). I understand that applying one bar of pressure to a random curing concrete mass is unusual even over 20 years of collected data. Not so unusual for pavement using a different binder (think superpave) that has been around for a while too.

The framework of the model handles other compaction conditions besides 10kPa and vibration. This is encoded into the variable K. K=9 corresponds to the above conditions. A higher K value would correspond to more pressure. One would have to do a series of tests to determine the K value for another condition. Even as K goes to infinity, the packing density doesn't go to 1.

I understand that the compaction index varies by condition.

In theory, you can find Beta from easily run tests at K=9 then use the results from calibrating your compaction technique's K value to predict the packing under those conditions. I used deLarrard's assumption from concrete that designing the aggregate for K=9 is sufficient. This is possibly a bad assumption and I will have to go back measure packing densities in something like C12-C14 glycidyl ether and see how the numbers work out as Roman suggested a while back. This is messy.

He later moves to virtual packing density where 0% voids is just a special case of particle size/shape and the container dimensions.

Virtual packing density is never attainable in the CPM framework. It is a theoretical construct helping to describe Beta for deriving actual packing density from the remainder of the model framework. CPM does come up with ways of accounting for container shape but these are important mainly when the container is less than 5 times the largest aggregate size.

I well understand that there is common sense resistance to reductio ad absurdum arguments in engineering. De Larrard uses perfectly spherical particles in his explication of the definition. Playing fast and loose cuts both ways.

I don't agree with the latter except as a possible class of solutions. It is a demonstrably weak assumption for a given system every time you see a bubble pop at the surface of the epoxy or easily cause this with a prick from an instrument. Think about it.

This is a fair statement but I think it conflates theory and experiment. If you have adequate mixing and sufficient epoxy, I don't see how you can fail to get a monolayer of epoxy on all the particles in theory. In practice, mixing may turn out poorly and things may go wrong but my own perspective is that models are helpful in telling what happens when things go right.

No, the context of the comment assumes your mono-layer being present. The point is that this layer can be easily pierced. The context of the earlier remarks about the very low compaction force in use (to derive K=9) is that relatively unremarkable and common force levels in forming the aggregate skeleton will pierce it (say at 1 bar for a dominant coarse grain).

[RomanLini]

...
This test series wasn't so much about meeting my density predictions but rather finding out what was mixable. I think with high enough shear mixing, the problems I am seeing will go away. Better choices in the larger aggregate would lower the small particle requirement and ease the problems too.

I'm not sure that any type of mixing will stop those tiny particles collecting epoxy and fluffing up enormously.

...
One technique I've seen in the literature is to place a carefully weighed sample in a crucible and burn the epoxy off in a furnace. Carefully weighing the remains will give a good picture of what was actually obtained.

Interesting! I have access to a kiln too. But I still think weighing and volume measuring the stuff you put in works just as well.

...
I will take wet packing measurements and see what changes. I appreciate the time you took to comment on my results and look forward to posting some more results when I get some more lab time.

No worries. It's just when you mentioned 22.5% epoxy and it turned to bread dough that really rang alarm bells!

As an example when I was first experimenting with garnet particles about 1mm to 1.5mm size I could get a measured volume of them, and after adding the right amount of epoxy to fill the voids and vibrating, de-gassing etc it would be almost identical volume at the end, that was one way I initially tried to ensure there was no excess of epoxy to ruin the stiffness.

Have you tried that type of test with your mixture? ie get a measured volume of the packed-down dry aggregate, then add X percent epoxy volume, and mix, vibrate degass etc then measure the resultant total volume? I'll bet it's fluffing up a lot... I'd really like to see that figure!

Also I'll second Kevin's idea of using a washable epoxy substitute, but I prefer honey. It's cheap and you can mix it with warm water to get any viscosity you like (measure wth a drip gauge) and it's 100% washable (ok it was for my larger particles... I'm not sure about your pixie dust!).

[ckelloug]

I'm not sure that any type of mixing will stop those tiny particles collecting epoxy and fluffing up enormously.

Interesting! I have access to a kiln too. But I still think weighing and volume measuring the stuff you put in works just as well.

I weigh each fraction individually. I get the volume by dividing weight by density. The numbers I quote are measured epoxy volume vs. aggregate solids volume. The non-uniformity of the tops of the samples in the 9x9 cake pan I've been making samples in makes it difficult to get an accurate measurement of the volume of the finished product for true density measurements.

The reason to do a burn-out test rather than assuming what is measured beforehand is that the material isn't necessarily homogeneous once it is mixed.

No worries. It's just when you mentioned 22.5% epoxy and it turned to bread dough that really rang alarm bells!

The mix design I am using for this test series based on model results is the following:

295g G200 Beta assumed 65%
14g #600 SiC Beta measured 56%
237g #320 SiC Beta measured 67%
254g #70 SiC Beta measured 66%
303g #36 SiC Beta Measured 65%
507g #6 SiC Beta Measured 56%

I don't think this is a particularly good mix design as too many fines are unmixable. Even if the model results are right in this case, and I don't believe they are, a different mix design would almost certainly have to be used in practice.

My Beta assumption for G200 is probably optimistic. The measured Beta value is around 45% for g200 however entrained air and van der Waals forces make the measured value unreliable for mixtures where the air is displaced. Likewise, the 10 micron size of the #600 makes its 56% likely to be pessimistic. I know from literature and experiments that 10 micron and below particles suffer from Van der Waals effects and don't behave nicely in the standard CPM packing piston test.

The problems above are why I think that I need to repeat the dry packing measurements wetted with an epoxy-like substance and look for deviations and then rerun the model with those numbers.

As an example when I was first experimenting with garnet particles about 1mm to 1.5mm size I could get a measured volume of them, and after adding the right amount of epoxy to fill the voids and vibrating, de-gassing etc it would be almost identical volume at the end, that was one way I initially tried to ensure there was no excess of epoxy to ruin the stiffness.

Have you tried that type of test with your mixture? ie get a measured volume of the packed-down dry aggregate, then add X percent epoxy volume, and mix, vibrate degass etc then measure the resultant total volume? I'll bet it's fluffing up a lot... I'd really like to see that figure!

I haven't run a test with a measured packing of the dry aggregate. I could whip up a small batch of the current aggregate and see what happens in the larger of my two packing cylinders.

I've been making samples that are designed for D790 flexural tests. The 9x9 inch flat plates I produce make it difficult to get good data of that you actually got but here's a pathetically crude measurement of the achieved densities obtained by measuring the cured samples in a random place with a set of calipers:

For the lower epoxy amount of my last two samples
8.8*8.8*.5 =38.7 in^3 = 634cm^3 total sample volume
504/634=.79 => 79% aggregate by volume
(150ml of epoxy)

634-150=484ml
484/504 => Aggregate appears to occupy 4% less volume than dividing mass by density of the dry components.

For the higher amount
8.8*8.8*.55=42.6 in^3= 698cm^3 total sample volume
504/698=.72 => 72% aggregate by volume
(200 ml of epoxy)

698-200=498
504/498 => Aggregate appears to occupy 1% more volume than dividing mass by density of the dry components.

I think that this says that with large amounts of epoxy, the material fluffs up but as the amount of epoxy goes down, the materials compacts. This goes back to a worry I had a long time ago that the particles might repel each other and separate as far as possible in a given epoxy volume. The above numbers are a suspect as my pan has a few dents and isn't a precision mold but I think it provides food for thought.

Also I'll second Kevin's idea of using a washable epoxy substitute, but I prefer honey. It's cheap and you can mix it with warm water to get any viscosity you like (measure wth a drip gauge) and it's 100% washable (ok it was for my larger particles... I'm not sure about your pixie dust!).

Kevin's idea of an epoxy substitute is good but I'd want to match surface energy as well as viscosity. That's why I'll probably take the messy step of using one of the reactive dilutants used to thin our epoxies.

Regards all,

Cameron

[RomanLini]

Thank you very much for taking the time to post those figures Cameron I appreciate it.

Just to make sure I understand your words;
"504/634=.79 => 79% aggregate by volume"
I am assuming the 634 is the measured final volume, and the 504 is a calculated volume of the initial aggregate (no epoxy) portion?

The trouble with that calc is that it's no good for measuring the "fluff up" caused by adding the epoxy. Sorry for making up terms like "fluff up" but I don't know the correct term for this phenomenon although I understand it and have worked around it in my own experiments.

To measure the fluff up you need an actual volume measurement from the vibrated and packed dry aggregate, this includes the real-world voidage. Then ideally after adding enough epoxy to fill the voids (and vibrating/compacting) the volume will be the same. Any volume increase over that initial volume is fluff up and means the epoxy is not just filling the voids but is now increasing total volume, making an epoxy layer around each particle and the final result will start to behave more like epoxy and less like granite.

I'm sure all of this is stuff you already know, I'm just trying to quantify what I mean by the term "fluff up" and why I think it is so important as a measurement. It's the difference between filling the known (empirically measured) voidage and increasing the total volume. Filling the voidage is good, increasing the total volume is very bad, although some tiny increase will be unavoidable.

(edit) I whipped up a diagram of the "fluff up", with 3 crude examples. I'm not saying this is totally accurate or that any of these 3 models are better than the others, but it is a start.

[ckelloug]

Hi Roman,

Thank you very much for taking the time to post those figures Cameron I appreciate it.

Just to make sure I understand your words;
"504/634=.79 => 79% aggregate by volume"
I am assuming the 634 is the measured final volume, and the 504 is a calculated volume of the initial aggregate (no epoxy) portion?

The 504cm^3 is the actual solid volume of the aggregate indirectly measured by dividing mass of the solid components by the density of the solid components. It is invariant with respect to the packing.

I also have the actual packing density of each aggregate measured by vibrating a known mass of the aggregate and measuring the volume that it occupied under the packing piston. What I don't have is a measured packing of all the components mixed together. I can go measure that this weekend.

The trouble with that calc is that it's no good for measuring the "fluff up" caused by adding the epoxy. Sorry for making up terms like "fluff up" but I don't know the correct term for this phenomenon although I understand it and have worked around it in my own experiments.

You're right. I thought about those numbers last night and all I probably measured was the amount of material that I couldn't get out of the mixing container.

To measure the fluff up you need an actual volume measurement from the vibrated and packed dry aggregate, this includes the real-world voidage. Then ideally after adding enough epoxy to fill the voids (and vibrating/compacting) the volume will be the same. Any volume increase over that initial volume is fluff up and means the epoxy is not just filling the voids but is now increasing total volume, making an epoxy layer around each particle and the final result will start to behave more like epoxy and less like granite.

I agree with you that it would be good to get the actual packing measurement for the dry aggregate. It's a start but I'm not sure that it actually is as meaningful as we would like because the lubricating action of the epoxy and van der Waals forces change the packing characteristics of the aggregate. With the small particles, the changes appear quite large.

I'll try a dry packing of the aggregate this weekend and I'll also see how much it changes after repeated application of vacuum.

I'm sure all of this is stuff you already know, I'm just trying to quantify what I mean by the term "fluff up" and why I think it is so important as a measurement. It's the difference between filling the known (empirically measured) voidage and increasing the total volume. Filling the voidage is good, increasing the total volume is very bad, although some tiny increase will be unavoidable.

Telling me things I might know is very helpful because sometimes I don't know them and sometimes I have them wrong.

I agree that it is important not to increase the volume of the mixture any more than necessary with additional epoxy.

Ideally, given the measurements I've taken for the different aggregates, the CPM should give a good approximation of the dry packing density. What I've seen so far however is that the small particles are creating anomalous low values for measured packing so I've been trying to compensate.

I'll cease the blizzard of text and try to get some more data.

Regards all,

Cameron

[Kevin_Johnson]

...
The 504cm^3 is the actual solid volume of the aggregate indirectly measured by dividing mass of the solid components by the density of the solid components. It is invariant with respect to the packing.

Be careful. Alumina (for example, see wikipedia for casual reference) has a density range of 3.95 to 4.1 g/cm^3. The error is of the same order as the effect you are trying to measure.

[Kevin_Johnson]

... I agree with you that it would be good to get the actual packing measurement for the dry aggregate. It's a start but I'm not sure that it actually is as meaningful as we would like because the lubricating action of the epoxy and van der Waals forces change the packing characteristics of the aggregate. With the small particles, the changes appear quite large.

I'll try a dry packing of the aggregate this weekend and I'll also see how much it changes after repeated application of vacuum.

Sans application of vacuum this is the general procedure used by the Superpave group.

If you are using repeated applications of vacuum -- while keeping the aggregate skeleton loaded such that shifts in structured volume (versus gases expanding a coresident liquid phase like water or epoxy, say) are necessarily negative in sign -- you could use water as a fugitive lubricant. It would be interesting to compare these progressively "dried" values with those obtained without use of water at all.

By repeating the experiment with the same actual sample over many trials the (possible) variation in density values can be better controlled for. By repeating using multiple samples from the same prime source the control becomes better. Synthetically produced aggregate probably has a much tighter variance in density but this relies on good statistical sampling practice and tight manufacturing protocols.

I understand that this is a bit much for the hobbyist.

[ckelloug]

Hi Kevin,

I haven't responded because I had to travel to give a lecture at my employer's headquarters on my research developing 3D Scanning Electron Microscope Imaging algorithms.

The idea of using water as a fugitive lubricant is interesting but water has a surface energy of 72.8 milliJoule/m^2 while epoxy has a surface energy of about 47milliJoule/m^2. Viscosity also makes a difference. Thin epoxy has a viscosity of 300cP while water has a viscosity of 1cP. Which bad things happen and when in the process they happen are likely vary quite a bit between water and epoxy. My vacuum suggestion on the other hand only deals with trapped air and not the lubrication effects so all experiments are a compromise.

I'm not worried about density variations as much as I am having only 2 significant figures in the density values published for the materials. I can repeat experiments to minimize the variance effects but currently variance isn't my problem. It's difficult to take measurements based on subtracting amounts of material if you don't know the true material amounts as you rightly pointed out.

Some materials, especially zeeospheres, are not packing the way their spherical shape should dictate and I think we really need to understand why so as to avoid having 20mm or so as the maximum aggregate size in the mix.

It will be some time before I make it back to my own lab but I did manage to make a couple of wet pack measurements before I left. Somehow, the posting was lost so I'll recreate what I can from memory.

I did two wet packing experiments on zeeospheres with C12-C14 glycidyl ether (An epoxy reactive dilutant). Thse used my packing piston for small particles that applys 10kPa of pressure from a weight and vibrated at 4g.

14.7g g200 zeeospheres plus 7.5ml C12-C14 glycidyl ether led to a one percent decrease in measured packing density from 50% to 49% (If I remember right).

14.7g g200 zeeospheres plus 5.0 ml of the glycidyl ether led to a 2 percent increase from 50% to 52%.

I don't know enough about the uncertainties on these values to be sure these are good measurements but they were done under very similar conditions and provide no indication of gross error.

When I get back, I'm going to try running the zeeospheres with the glycidyl ether through a high power ultrasonic disperser and see what the packing looks like afterwards.

Regards all,
Cameron