[ckelloug]

greybeard,

If you would like to maximize pain, you could go look up the structural formulas for epoxy and amine and then figure out the change in gibbs free energy when it polymerizes. You can then compute using the specific heats of epoxy and quartz how much this should heat the mixture either of neet epoxy or epoxy and quartz.

Ok, so this isn't the most practical of solutions.

I know that the published data from Reichhold only gives the gel time of 37-127 with hardener 37-614 in a 4 ounce paper cup at 25 C. It's 14-17 minutes.

I suppose we have to take measurements and establish the gel time ourselves perhaps using some of the above logic to try to figure out the temperature it could achieve.

--Cameron

[greybeard]

.. as I feared.

Though I still hope there may be lurker with the practical background experience......:wave:


On second thoughts, I wonder how far out you might be if you simply took account of the % of the filler as being propotional to the heat capacity of the mix, and Le Chatelier principle of rate of reaction proportional to absolute temperature ?

mmmm?
On third thoughts, with such a small amount of epoxy with the large heat sink that the quartz will represent, I strongly suspect that the rate of reaction of the resin will be controlled much more by the ambient temperature, and hardly affected by the exotherm.

[I've just done a couple of castings, one yesterday with the mold at room temperature(~18degrees C), then onto the storage heater overnight(~40), and one today, straight onto the now cooling storage heater (~30). The speed of the latter cure was very obvious.]

[jhudler]

On third thoughts, with such a small amount of epoxy with the large heat sink that the quartz will represent, I strongly suspect that the rate of reaction of the resin will be controlled much more by the ambient temperature, and hardly affected by the exotherm.

Spot on as you lot say. The best results seem to be; allow the EG to cure 24 hours. Then elevate to 50C for a few hours to take care of those little free radical polymer buggers that didn't cross-link!

[lgalla]

Jack here are some spec's on the Moglice.
COF .05
Compressive 24000
Hardness 89 Shore D
If your part incorporated some piping for coolant,drain,wiring etc,you could post cure or cook it from the inside out by using hot water.Your home hot water supply is capable of 120F or 50C.
Larry

[ckelloug]

Spot on as you lot say. The best results seem to be; allow the EG to cure 24 hours. Then elevate to 50C for a few hours to take care of those little free radical polymer buggers that didn't cross-link!

Reichhold actually recommends 24 hours at ambient, 2 hours at 121C on the data sheet! It will make a difference in glass transition temperature and some in stiffness.

[acondit]

Another Lurker,

I have just spent the last few days reading all 2800+ posts. I have read various posts over time but this was an effort to absorb as much as possible.

I am truly impressed by the effort that has gone into figuring how to make this work. I have done casting of parts in cast iron and aluminum before and I believe that E-G casting has real promise. I am glad that the naysayers didn't kill off the "I can do it!" attitude. I am in the process of designing a machine to use the E-G castings. I won't even attempt to name names because there are so many who have contributed to this threads success, but thank you all.

Alan

[ckelloug]

Welcome Alan and the other lurkers! I used to be a lurker too.

Greybeard et. al,

In explanation of the concept of a cost function, imagine this. . .

Imagine Billy-Joe-Bob has just asked your daughter Sarah-May to the big shindig by the river. You'd like Billy-Joe-Bob to mind his manners but you know that Sarah-May and he will still have some fun. Being a sadistic mathematician, you give Billy-Joe-Bob a cost function and let him optimize it under the constraint he will be grounded forever if he gets home late. (pretend travel time is zero).

1. Bring Sarah-May Home Late, 2 minutes of lecturing per minute late.
2. Drive mud covered Pickup Truck to shindig, 7 minutes of lecturing.
3. Get Naked at Party, 25 minutes of lecturing and call to parents.
4. Get Sarah-May Pregnant, Interview with Mr. Shotgun, 6 hours in ICU.

If Sarah-May must be home by 10:30 pm and Billy-Joe-Bob is constrained because he has to be home by eleven or his parents will ground him "forever", how do you go about optimizing this situation if you are Billy-Joe-Bob.

From the cost function, we can see that Billy can't get Sarah pregnant because 6 hours in the ICU will have him grounded for life.

We can also see that Billy gets grounded for life if he brings that despised pickup truck and gets naked at the party.

So optimizing, Billy tests combinations of various factors in the cost function and settles on coming home 7 minutes late and bringing the pickup truck or 10 minutes late with some other car. He could also take the sedan, get naked at the party, and bring Sarah back not more than 1 minute late.

The cost function describes the "cost" of a given solution component and the constraints are additional restrictions that must be obeyed. In this case Billy is Maximizing fun which maximizes cost with the constraint that the cost has to be under 30 minutes and bringing Sarah home late both costs in time and has penalty lecture.

I hope this amusing anecdote clarifies rather than declarifies the situation.

Regards all,

Cameron

[BobWarfield]

I regret to inform you that in the South, where there are people named Billy-Joe-Bob, the ICU is extremely busy almost all the time, and there are many mud covered pickup trucks illegally parked out in front.

Sincerely,

Bob W.

[jhudler]

Why that's as clear as a bourbon and branch.

[ckelloug]

Sorry to post jokes about the neighbors I live in Alabama (wasn't born here) and I've been listening to too much country music.

I'm still a few equations behind in answering the questions people have asked in the last 5 pages. . .

Regards all,

Cameron

[acondit]

I have been working on a mini-mill design and I have a few questions. For a mill roughly the size of an X2 or X3 will a EG column 4"x5" likely have enough rigidity?

Do you think the THK HSR15 rails will be strong enough or do I need to plan on HSR20 rails? I have some 15's and some 25's, it seems that the HSR25's are way overkill.

Alan

[ckelloug]

Alan,

I have no experience with rails so I hope that one of the myriads of regular contributors or lurkers will give you some advice.

As for the column, I'd have to ask what your planned loads are. Flexural modulus for the material is estimated at between 2000 ksi and 4000 ksi and ultimate flexural strength is estimated between 2 and 4 ksi. I have the materials test machine but I haven't run tests yet as haphazard testing without theory seemed as though it wouldn't provide real insight. With the modulus number it ought to be fairly straightforward to take the load and plug it into a cantilevered beam equation but I haven't done it.

If you are ready to pour I'd be happy to test your material and tell you how well it does, 3/8 x 1/2 by 7 1/2 is the largest sample I can test right now. For a full ASTM flexural test, 5 samples are needed.


--Cameron

[acondit]

Cameron,

The column on a stock X2 minimill is approximately 2" x 4" by about 1/4" cast iron. I know that there is some flex in it, so I was just wondering how (for example) a 4" by 5" x 1" wall thickness EG part would compare in resistance to flex.

It is hard to come up with a design without a feel for what is needed strength wise. Based on Walter's work, is it possible to estimate size and wall thickness of E-G required to match or exceed a given size and wall thickness of cast iron?

Alan

[ckelloug]

Alan,

I saw your PM and the added question. Were I to try to answer the question I'd have to know what kind of loads to expect, the kinds of deflections that you would wish to hold and some material characteristics. The material characteristics I have right now are pretty vague and they are for mixtures that were also pretty vague.

Last set of load calculations I did assumed a 100 lb load in addition to the weight for a horizontally supported beam at the ends. see <A href="http://www.cnczone.com/forums/showpost.php?p=293843&postcount=1159"> Post 1159</A> Let me emphasize that these numbers are for beams supported at both ends with the load in the middle and sagging. I will eventually get to posting numbers of a cantilevered beam but I don't think I'll bother until I have a clue about material performance which I don't now.

I guess I'd say that I'd be a touch queasy about an inch thick E/G section without some better performance numbers. From an engineering motivation but mostly <B>Qualitative</B> measures, I'd probably want to make the beam either 6x6 or 8x8 with 2 inch thick sides. The 8x8 beam is more conservative. Unfortunately, without the strength and modulus numbers we're working on, it's pretty much a WAG what the strength and stiffness of E/G are and these calculations were likely direly pessimistic.

--Cameron

[acondit]

Alan,

I saw your PM and the added question. Were I to try to answer the question I'd have to know what kind of loads to expect, the kinds of deflections that you would wish to hold and some material characteristics. The material characteristics I have right now are pretty vague and they are for mixtures that were also pretty vague.

Last set of load calculations I did assumed a 100 lb load in addition to the weight for a horizontally supported beam at the ends. see Post 1159 Let me emphasize that these numbers are for beams supported at both ends with the load in the middle and sagging. I will eventually get to posting numbers of a cantilevered beam but I don't think I'll bother until I have a clue about material performance which I don't now.

I guess I'd say that I'd be a touch queasy about an inch thick E/G section without some better performance numbers. From an engineering motivation but mostly Qualitative measures, I'd probably want to make the beam either 6x6 or 8x8 with 2 inch thick sides. The 8x8 beam is more conservative. Unfortunately, without the strength and modulus numbers we're working on, it's pretty much a WAG what the strength and stiffness of E/G are and these calculations were likely direly pessimistic.

--Cameron

So at the moment in trying to design anything, we are really just looking at WAG or it looks about right. We don't have any sense of how the strength compares to something done in cast iron. Did I get that right?

Alan

[ckelloug]

Alan,

Due to the lack of measurements, educated guessing is about as good as we can do however the range given below reasonably covers the expected numbers. Very few posters ran anything close to the type of tests which give usable numbers. I have the correct equipment for this testing but having a life and a few bills, I haven't been able to run the tests yet even though I do have the machine.

We know the flexural modulus of E/G is likely somewhere between 2000 and 4000 ksi and we know that the ultimate strength is between 2 and 4 ksi from the results that AustinT posted. These are more or less consistent with literature values. Class 20 Cast Iron by comparison has a tensile stength of 20ksi and a modulus of about 8000 ksi.

Our E/G is thus likely to be between 1/4 and 1/2 as rigid as cast iron and between 1/10 and 1/5 as strong. One point to remember in machines is that a fair amount of the iron in an oldschool machine was there for damping, not rigidity. On the damping point, the machine design article listed somewhere in the thread showed that E/G had a damping factor about 7 times that of cast iron.

In short, Alan, the WAG part of my response was the column size as I haven't done the calculation to say exactly the deflection that would be experienced by a column with the nominal 2000ksi modulus to compare with my past numbers on beams. The deflection numbers given in the post referenced earlier come from doing the calculation with the 2000ksi lower bound on modulus which was established with the little data we have that appears reliable. Based on the beam number I calculated, I'd suspect that what I said about using a 6x6 column with 2 inch walls isn't far off but unlike the numbers for beams which I ran, I haven't run the numbers for columns yet. All this is exacerbated by the fact that we do not have an actual modulus or strength value for the E/G formulations we're making: only the broad estimate as described above.

<B>I will make the offer again to anybody on this thread. If you send me 5 samples of whatever you want to use cast in the form of 1/2 x 3/8 bars 7 1/2 long with an explanation to the best of your abilities of the composition, I will perform an ASTM D-790 flexural test and post the results on this thread. Also, explain the epoxy cure method used and the ambient temperature. PM me for the address if you'd like your material tested.</B>

I will eventually test the mixtures that the simulator thinks is optimal and post results but I don't have time to do the mixing and curing etc right now due to the impositions of real life.

I also still owe folks here a bunch of answers on the model. Not tonight.

Regards all,
Cameron

[acondit]

Alan,

<snip>(snip)

We know the flexural modulus of E/G is likely somewhere between 2000 and 4000 ksi and we know that the ultimate strength is between 2 and 4 ksi from the results that AustinT posted. These are more or less consistent with literature values. Class 20 Cast Iron by comparison has a tensile stength of 20ksi and a modulus of about 8000 ksi.

Our E/G is thus likely to be between 1/4 and 1/2 as rigid as cast iron and between 1/10 and 1/5 as strong. One point to remember in machines is that a fair amount of the iron in an old school machine was there for damping, not rigidity. On the damping point, the machine design article listed somewhere in the thread showed that E/G had a damping factor about 7 times that of cast iron.

In short, Alan, the WAG part of my response was the column size as I haven't done the calculation to say exactly the deflection that would be experienced by a column with the nominal 2000ksi modulus to compare with my past numbers on beams. The deflection numbers given in the post referenced earlier come from doing the calculation with the 2000ksi lower bound on modulus which was established with the little data we have that appears reliable. Based on the beam number I calculated, I'd suspect that what I said about using a 6x6 column with 2 inch walls isn't far off but unlike the numbers for beams which I ran, I haven't run the numbers for columns yet. All this is exacerbated by the fact that we do not have an actual modulus or strength value for the E/G formulations we're making: only the broad estimate as described above.

I will make the offer again to anybody on this thread. If you send me 5 samples of whatever you want to use cast in the form of 1/2 x 3/8 bars 7 1/2 long with an explanation to the best of your abilities of the composition, I will perform an ASTM D-790 flexural test and post the results on this thread. Also, explain the epoxy cure method used and the ambient temperature. PM me for the address if you'd like your material tested.

I will eventually test the mixtures that the simulator thinks is optimal and post results but I don't have time to do the mixing and curing etc right now due to the impositions of real life.

I also still owe folks here a bunch of answers on the model. Not tonight.

Regards all,
Cameron

Thank you. I think that this is the information that I was looking for.

I am aware that the strength of my mixture may vary from not only what others have done but from batch to batch due to lack of precise mixing controls and curing control. But I wanted to know that I would be in the ballpark of what I hope for with a given design. It is one thing to spend several hundred dollars in time and supplies on the reasonable possibility that you have got it right and quite something else to take that risk without the slightest idea whether it is going to work.

Alan</snip>

[ckelloug]

Alan,

For the record, batch-to-batch variation isn't the problem. The problem is that we don't yet have accurate measurements for any material resembling the ones we're discussing. I would be remiss in my work here if I didn't clarify that the numbers I've posted here are based mostly on a perceived worst case estimate, not many measurements at this point.

Mix-and-test optimization is expensive and thus I and others have done a lot of work with models. Mix and test usually doesn't tell you what you did wrong. I didn't see a need to run a huge lab test program on materials that could be eliminated via a mathematical model.

The numbers I've given in the past are specifically based on my estimate of the minimum strength and modulus for which I arbitrarily chose half the value of a published result from NIST. This guess has been an excellent lower bound for the material properties based on the minimal information I have.

Strength in a machine is a function of both the material and the way it is used. There is static stiffness which affects the deflection under load and then there is dynamic stiffness in which E/G excels describing deformation under vibration. There is also creep to contend with: creep is the tendency of a material to change dimension permanently under a load much lower than would be predicted to effect it. All of these are important in a machine design.

Just because E/G isn't as strong as iron, it doesn't mean that you can't make a stiff and well built machine that is vibrationally dead. Qualitatively, Larger cross sections parts have to be used than would otherwise be used with cast iron to get the same stiffness. The parts don't necessarily have to be solid-core, just of bigger cross section. How big is a function of the material properties.

Without knowing things like the weight of the spindle you want to hang on the column, the height of the column, and other particulars like the accuracy you would like the machine to hold, it is very difficult to give you any information which is specific. Without the material properties of the E/G you want to use, it's not possible to tell if it's either half as good as the minimum I estimated or 4 times as good.

<B>I can tell from having been here a while that people are trying things but am I speaking in geek when I talk about flexural modulus and flexural strength? It appears nobody is interested enough in getting good numbers on the material properties to send me a set of samples of what they're making so we actually can predict the required sizes of beams and columns. (Jack has made 3 sets but he hasn't sent them over yet. . .) Without testing actual material to get these numbers (for which I bought a $7000 test system and offered free testing to everyone here [Normally 250$ a pop from PLTI Inc.] ), we're all continuing to run around in the dark on WAG's rather than data.</B>

I will eventually generate modulus and strength data for the mixtures I think are good and with the additives I have gone to great trouble to acquire but I am still a few months from being able to do this due to a large backlog of commitments.

<B>
If anybody on this thread (regular contributors or lurkers alike) is wondering how he or she can contribute, please consider making and sending me a 5 sample set of your material 1/2 x 3/8 x 7 1/2 as described so that I can test it and post results. If interested folks send me samples, I'll update the tables of calculations for beams and columns I did so long ago with actual numbers and we can start taking E/G from being a curiosity to a proven material with accurate rules of thumb for designing a machine.</B>

[greybeard]

Cameron -if I spin test samples, they will have directional in their properties, stroger in one direction.

If I do this and send them over, can you produce info that might be useful in comparison to static cast blocks ?

John

Point of interest - what about the aggregate size being restricted to 1/5 minimum dimension ?
This would mean the test samples at 3/8" thick would only have 2mm max particle size ?

Not a problem for me, but what of others ?

[Gizmot]

Point of interest - what about the aggregate size being restricted to 1/5 minimum dimension ?
This would mean the test samples at 3/8" thick would only have 2mm max particle size ?

Not a problem for me, but what of others ?

pretty much what I was thinking. I also tought about doing some ASTM testing but then I realized that you probably dont test concrete the same way . There must be something somewhere about that in one of the books.