DEM application possibilities

Hi Peter,

Firstly, you don't need a 64 bit computer to use ESyS-Particle but you will be limited to run simulations of less than 1M particles on 32 bit computers. Of course, regardless of which CPU(s) you have, you will need linux installed to use ESyS-Particle.

Regarding your bucket elevator application, ESyS-Particle has not (to my knowledge) been used for materials handling applications similar to this to date. ESyS-Particle is generally best at applications in which the particle flow is constrained to a relatively small volume in space. To gain benefit from the parallel engine of ESyS-Particle (and hence to be able to use a large number of particles) it is best if particles do not move large distances en mass.

Having said that, I think you probably can use ESyS-Particle to simulate the bucket elevator. There would be (at least) two significant challenges to face though. The first is to import the bucket geometry as a triangle mesh and to move that mesh in a manner consistent with the operation of the bucket elevator. This challenge shouldn't require any changes to the ESyS-Particle source code but is a little challenging to implement a suitable Runnable in python to move the bucket mesh geometry in a suitable way.

The second challenge is that ESyS-Particle currently doesn't have any frictional particle-wall interactions. I'm not sure this is terribly important for your application but if friction between the bucket and the particles is considered important, you will need to modify the ESyS-Particle source code to implement that.

In summary, I think you can use ESyS-Particle to simulate granular flow in a bucket elevator but you need to be prepared to experiment to get a desirable result. This is basically the same for any of the DEM codes. Patience and persistence are always required to adapt a DEM code to a specific application.

I'm interested to hear/see how you go and if you have any other questions, please feel free to ask.

Cheers,

Dion.

Many thanks for your response.

I've got to this point because I've had commercial DEM software quoted at around £90,000 for purchase (and maintenance) or £4,000 for a single run.(common DEM applications are the flow of bulk materials through chutes). In my area of bulk materials handling those are not viable numbers!

I came across ESyS-Particle and thought it might provide maybe a partial solution, thinking I could start with just a few particles and single bucket.

But I'm not a Linux person (other software I use AutoCAD, Inventor etc running under Windows) so I was intending to use the live dvd version which I've downloaded to avoid having to install Linux and this requires a 64 bit processor.

I was thinking that I would start with a single particle on a flat plate, the plate would be horizontal at the 12 'o clock position then rotate the plate around a centre corresponding to the wheel centre at different speeds and see what happens. However your statement that ESyS-Particle currently doesn't have any frictional particle-wall interactions brings me to a full stop as the motion of the particle starts when there has been sufficient rotation of the plate that the friction between the particle and plate has been overcome.

So if there is no way round this friction problem then I'll have to do a simple analysis with paper and pencil :)

Thanks for your input.

Regards

Peter

Hi, Dion:

I am a little curious how difficult (or how much labor would require) to implement the frictional wall? What is the major difficulty? Because I am thinking of implement this during my spare time (after work).

Feng

Hi Feng,

Implementing frictional interactions is not terribly difficult but it is more challenging than one might first expect. The main challenge is that you need to keep track of transitions from one state to another e.g. from static to sliding friction, particles going into or out of contact with a wall etc. I imagine things will be somewhat more complicated for frictional interactions with triangle mesh walls but not too bad for simple planar walls.

I discussed wall friction briefly with Steffen a few weeks ago. The Model/SoftBondedWallInteraction.cpp probably provides a good starting point because it implements different elastic stiffnesses parallel and perpendicular to the wall. Steffen suggested looking at the friction implementation in Model/FrictionInteraction.cpp as a guide for adding the frictional part.

If you are keen to try implementing frictional wall interactions, please have a go. I'd be happy to help where/if I can. A number of people have expressed a desire for frictional walls but I cannot find the time (unless my own research absolutely requires it...).

Cheers,

Dion.

Hi Peter,

As a follow-up to your last message, I have an ESyS-Particle liveDVD for 32bit processors but haven't uploaded it to the FTP site yet. I'll do that next week at work (when I have faster internet access!). Will let you know when it is available. At least that way you can try out ESyS-Particle, some of the tutorials etc. before deciding whether to invest the time to adapt it to your specific problem.

Until frictional wall interactions are implemented, you might be able to do your test of particles on a rotating plate by bonding a number of particles to the surface of the plate. As explained in the ESyS-Particle Tutorial, this is a simple way to simulate frictional walls that is usually quite effective.

Cheers,

Dion.

I really would appreciate a liveDVD for 32 bit processors. At some point I will have to upgrade my PC, but money is a bit tight at the moment. :(

Peter

Hi Peter,

I've uploaded the 32bit ESyS-Particle LiveDVD image. You can download it from:
ftp://ftp.esscc.uq.edu.au/pub/esys/ESyS-Particle-2.1-rc1-x86-32bit.iso

Let me know if you have any problems with it.

Cheers,

Dion.