possible? - study extrusion process of 200nm spheres

Chris,

I am relatively new to ESyS myself, but I can offer a few insights for you.

First, ESyS has very nice capabilities for creating complex geometries and moving walls. Thus, is would be relatively straight forward to implent an extrusion barrel with one or more die holes at the end of the barrel and a moving piston to force the particles through the die. It might also be possible to implement rotating screws which would be more realistic for most extruders, however someone else would have to talk about that. I don't know how to do that yet.

Fundamentally, ESyS is a soft sphere model as far as I can tell. Thus, collisions are not instantaneous and particles are allowed to overlap to simulate deformation during a collision. This should be fine for your spheres, but you are quite right that some "stickiness" would be useful/necessary for your particle size. Currently, a Hookian force model is used to simulate elastic collisions with no stickiness, and a viscocity interaction can be specified for energy dissipation. Stickiness can be implemented somewhat artificially by bonding particles together by specifying a breakDistance, that is, a separation distance beyond which the Hookian spring breaks. For my simulations, I chose to modify the normal force model to be a JKR type interaction which accepts an interfacial surface energy term to account for stickiness. I have not done any comparisons, but I suspect my force model is more computationally expensive than simply specifying a breakDistance.

Also, 200-300 nm is pretty small so a very small time step will be required which will increase computation time. Additionally, you will have a tremendous number of particles present even in a miniature extruder so you may have to define some periodic boundary conditions. I don't know if that will be possible with your geometry though because I don't have experience implementing periodic boundaries.

Finally, all walls are frictionless unless you glue particles to the walls, which can be done. Depending on your process, friction in the barrel and especially the die are likely important.

I hope this is helpful. If you haven't done so, check out Dion Weatherley's tutorial (? 52366 on the anwsers page). I found it to be very helpful, and it is a great start to determine if ESyS will be able to meet your needs.

Best regards,
David

Hi Chris,

Thanks for your interest in ESyS-Particle. Thanks also to David for a very comprehensive and accurate answer to your question. I just wanted to add a couple comments of my own.

Firstly, there are a number of ways to implement "stickiness" in DEM models. David's suggestion to use non-rotational bonds with a suitable breakDistance is a simple way that requires no changes to the ESyS-Particle force calculations. The down-side is that you must specify which particles will be stuck together a priori. A nicer approach might be an interaction law in which touching particles experience a weak force that acts to hold the particles together that gets balanced by the normal elastic repulsive force as the overlap of the particles increases. I can imagine this sort of force might be a bit like static electricity, with touching particles tending to hover together in unbonded clusters. It would require quite a lot of testing and tuning to get the right balance. This is often the case with DEM models as no one interaction law can capture all the possible types of interactions one might desire for different applications.

My other comment refers to your question about the practicalities of modelling 200nm spheres and David's response regarding timestep sizes. The radii of the spheres is really not that important as you can rescale your model units so that a sphere of radius=1 is equivalent to a real sphere of radius=200nm. By also rescaling other model units such as bond stiffnesses and density, the timestep increment is not necessarily prohibitively small. I suspect your main constraint will be the relative sizes of your spheres and your extruder/container. If the container is extremely large compared with the spheres then you will need to consider some tricks like periodic boundaries as David suggested, to reduce the total number of spheres in your model.

Although I am not familiar with the topic of extruding polystyrene spheres, it would seem that DEM is a suitable choice for modelling the process although some work would be required to implement suitable interaction laws and to rescale units correctly. In that case ESyS-Particle might be a good candidate for your research, as you have access to the source code if changes to the interaction laws etc. are required.

Cheers,

Dion.