rrod
having dreams about rocks
Member since December 2020
Posts: 72
|
Post by rrod on Mar 17, 2021 22:07:17 GMT -5
I was bored and decided to math-out the relationship between open space and % fill for rotary barrels. Interesting to see how much grind length opens up even with just a bit of free space in the circular cross-section case: I have results for a hexagonal cross-section as well, but they aren't too different than this.
|
|
Benathema
has rocks in the head
God chased me down and made sure I knew He was real June 20, 2022. I've been on a Divine Mission.
Member since November 2019
Posts: 703
|
Post by Benathema on Mar 18, 2021 0:15:05 GMT -5
Makes sense. It took a minute to orientate myself with the plot. That orange line is the killer. 90%+ definitely shows how easily you can overload the barrel and hinder tumbling motion.
On the other hand, there's not a huge difference between a 60 and 80% fill. There's a fair bit of wiggle room there.
|
|
Brian
fully equipped rock polisher
Member since July 2020
Posts: 1,512
|
Post by Brian on Mar 18, 2021 7:39:16 GMT -5
I just finished judging a county-wide science fair last night and would have loved to see something like this as a project!
It would be interesting to see how rock size equates to an effective grind length. Or in other words, how the packing of rocks affects the effective grind length, assuming that the rocks may not have a chance to be in motion over the entire length due to packing. Because rocks would pack, they would tend to be held in place at least momentarily which prevents them moving over the entire window in which they could move (at least in my mental model). Plus, not all rocks are located at the outer periphery and would have an effective grind length less than the maximum.
With a clear lid like you have, you could set up a barrel filled with marbles (ideally similar colors with a small percentage of contrasting colors to help track movement) and see how well the ideal model maps to an actual model.
|
|
|
Post by greig on Mar 18, 2021 9:07:37 GMT -5
So I can load a barrel to 90% and still get a somewhat effective tumble?
|
|
Benathema
has rocks in the head
God chased me down and made sure I knew He was real June 20, 2022. I've been on a Divine Mission.
Member since November 2019
Posts: 703
|
Post by Benathema on Mar 18, 2021 13:11:29 GMT -5
So I can load a barrel to 90% and still get a somewhat effective tumble? Maybe, maybe not. What Brian is saying about packing makes a lot of sense too. The rocks need to dislodge to use the grind length, and they'll need the head space in the barrel to do it. I could imagine if the entire barrel was full of 1/4" spherical rocks that you could fill to 90% and it would work. Getting into the 1"+ range, maybe not.
|
|
Fossilman
Cave Dweller
Member since January 2009
Posts: 20,709
|
Post by Fossilman on Mar 18, 2021 13:50:41 GMT -5
Following!
|
|
rrod
having dreams about rocks
Member since December 2020
Posts: 72
|
Post by rrod on Mar 18, 2021 18:04:04 GMT -5
Makes sense. It took a minute to orientate myself with the plot. That orange line is the killer. 90%+ definitely shows how easily you can overload the barrel and hinder tumbling motion. On the other hand, there's not a huge difference between a 60 and 80% fill. There's a fair bit of wiggle room there. The difference is headroom basically, which thus comes down to how aggressive you want the tumble. I saw someone mention that they do their last coarse grind nearer to 85% full to do a final bit of smoothing without chipping things up. I wonder if the steps where you are not trying to remove much material might function at the 90+% range. With a clear lid like you have, you could set up a barrel filled with marbles (ideally similar colors with a small percentage of contrasting colors to help track movement) and see how well the ideal model maps to an actual model. Marbles! I was trying to think of something perfectly circular that would work. Time to buy a couple of sizes I guess. So I can load a barrel to 90% and still get a somewhat effective tumble? Maybe, maybe not. What Brian is saying about packing makes a lot of sense too. The rocks need to dislodge to use the grind length, and they'll need the head space in the barrel to do it. I could imagine if the entire barrel was full of 1/4" spherical rocks that you could fill to 90% and it would work. Getting into the 1"+ range, maybe not. I seem to recall reading that it was the layer just below the top where the main grinding action happens, and it would make sense that if the top isn't moving then nothing else is either. Also worth nothing that the headspace given here is the max at the middle of the empty circular segment.
|
|
Benathema
has rocks in the head
God chased me down and made sure I knew He was real June 20, 2022. I've been on a Divine Mission.
Member since November 2019
Posts: 703
|
Post by Benathema on Mar 19, 2021 1:31:03 GMT -5
Interestingly, I decided to head over to google scholar to see if someone had scienced the heck out of rock tumbling already. Yes and no. So for our hobby where we're trying to polish, not so much. However, in the area of SAG mills and ore processing and whatnot, yes, very much so. There may be some extendibility to stage 1 coarse grinding. I stumbled onto this article from 2003 describing the use of a discrete element method (DEM) to simulate grinding in SAG mills. Literally simulating material removal due to impacts. Curiously they likened it to running molecular dynamics simulations, something I have 5+ years experience doing. The timestep they use is 10^-4 seconds, so to get the equivalent of hours/days/weeks worth of simulation would take a fair amount of computational time, which translates into significant real world wallclock time (for reference, the timestep for molecular dynamics is on the order of femtoseconds (10^-15 s) and most simulations run for nanoseconds (10^-9 s) and that could take literal weeks of real world time). Just to not leave anyone in the dust on what a timestep means: forces are calculated between the particles in the system, then propagated for 1 timestep amount of time, then the forces are recalculated and then propagated another timestep ad nauseum. We don't have a nice way to simulate a system continuously, so we do it in small discrete steps. The paper has all the physics laid our for what goes into the simulations, which is something that's been brought up many times on this forum - all the factors and variables that go into grinding, like RPMS, fill levels, rock sizes, etc. At that time the particles were limited to spherical and ellipsoidal shapes. At best they have steel balls in there, but no liquid or abrasives. It's a neat read all its own. They made mention of an implementation, so I went on the hunt trying to find a recent paper that cited this one and came across this Feb 2020 article. Apparently there's been a lot of improvements since then, such as different particle shapes, and they mentioned what implementation they used. I've not played with it just yet, but it's called LIGGGHTS and utilizes the LAMMPS molecular dynamics package as a backend to carry out the simulations. These are open source, so anyone who wants to watch a simulation explode can do it. I've used CHARMM, Amber, and I touched Gromacs once, but I haven't used LAMMPS - it's probably not too much of a stretch. I'd be curious to see if they allow for any fluid dynamics or inclusion of abrasives. DOIs to the two articles incase the links break one day. DOI: 10.1016/S0301-7516(03)00032-2 DOI: 10.22044/jme.2020.9045.1793edit: 500th post --> I now have rocks in the head.
|
|