[m.reza]

Hi everybody.
I wanna define the conduction heat transfer in colliding between two particles or between particles and walls. I've obtained some correlations in this regard, but I was wondering to know how can I define the following simple formula in UDF:
Temperature Difference=T(i)-T(j)
i= particle one
j= particle two or wall

regards,

[vinerm]

Assumption for a DPM simulation is that the particle never collide with each other because the volume fraction is not enough. So, first you have to define collision. You can work on the lines of O'Rourke where collision is assumed based on a probability density function for all those particles that are within same cell. Then, you need to define two time scales, one for heat transfer and one for collision. And heat transfer should only be possible if time for heat transfer is smaller than collision time.

[m.reza]

Originally Posted by vinerm

Assumption for a DPM simulation is that the particle never collide with each other because the volume fraction is not enough. So, first you have to define collision. You can work on the lines of O'Rourke where collision is assumed based on a probability density function for all those particles that are within same cell. Then, you need to define two time scales, one for heat transfer and one for collision. And heat transfer should only be possible if time for heat transfer is smaller than collision time.

Thanks for your reply, Vinerm.
In Fluent 2019 when the DEM option is activated, we can determine the type of collision (damping, Spring dashpot , ...). As I asked Ansys support team, Fluent ignores the heat dissipation in the collision. On the other hand, when we use nanoparticle, I think the volume fraction of them is enough to have collisions.
Moreover, there a function in UDF "DEFINE_DPM_SPRAY_COLLISION.

regards

[vinerm]

That is correct. DEM is required to be included only when the particle volume fraction is high enough that particle-particle collision become important. For that, volume fraction should be more than 10-12%. Do note that it is based on volume fraction and not mass fraction. It is also correct that DEM does not model the heat transfer across particles. It is important to know that DEM is a soft sphere approach and is valid for droplets only under certain conditions and not always; it is meant to be used for solid particles.

As far as DEFINE_DPM_SPRAY_COLLIDE is concerned, it has got nothing to do with DEM. You can use it with or without DEM. This is meant to replace inbuilt O'Rourke model. O'Rourke model is valid only for droplets but user can define a model for any type of particle.

Volume fraction is not really a function of diameter. For nano-particles, you don't even need two-way coupling. If the volume fraction is very high and size is of the order of nano, then better would to use Granular model under Eulerian instead of DPM.

[m.reza]

Originally Posted by vinerm

That is correct. DEM is required to be included only when the particle volume fraction is high enough that particle-particle collision become important. For that, volume fraction should be more than 10-12%. Do note that it is based on volume fraction and not mass fraction. It is also correct that DEM does not model the heat transfer across particles. It is important to know that DEM is a soft sphere approach and is valid for droplets only under certain conditions and not always; it is meant to be used for solid particles.

As far as DEFINE_DPM_SPRAY_COLLIDE is concerned, it has got nothing to do with DEM. You can use it with or without DEM. This is meant to replace inbuilt O'Rourke model. O'Rourke model is valid only for droplets but user can define a model for any type of particle.

Volume fraction is not really a function of diameter. For nano-particles, you don't even need two-way coupling. If the volume fraction is very high and size is of the order of nano, then better would to use Granular model under Eulerian instead of DPM.

Thanks for your reply in detail. And, sorry for bothering
My maximum volume fraction of nanoparticle (Al2O3 with diameter 100nm) is 5% in a microchannel (Length is 4mm). I simulate it with DPM with consideration of all forces such as Thermophoresis, Saffman, Virtual mass and etc. The results are different from those of the single-phase model. Hence, we decide to develop our study on DEM model to consider the interaction of particles. However, you believe that VF should be greater than 10%.
On the other hand, defining O'Rourke model, two time-scales, the collision in UDF is very difficult and eventually converging residuals maybe is more difficult.

I will highly appreciate if I know your feedback on my project.
regards

[vinerm]

For 5% volume fraction, you may not expect any particle-particle interaction. My suggestion would be to first use inbuilt model, i.e., O'Rourke to see how much collisions it shows. If it shows that the collision frequency is high, then you should spend time in writing a UDF for heat transfer. If the collision frequency is not high, then writing a UDF will not be justified.

Defining multiple time scales will not affect the stability or convergence. These time scales are not to be used for simulation, just to determine if the heat transfer should take place or not.

[m.reza]

Originally Posted by vinerm

For 5% volume fraction, you may not expect any particle-particle interaction. My suggestion would be to first use inbuilt model, i.e., O'Rourke to see how much collisions it shows. If it shows that the collision frequency is high, then you should spend time in writing a UDF for heat transfer. If the collision frequency is not high, then writing a UDF will not be justified.

Defining multiple time scales will not affect the stability or convergence. These time scales are not to be used for simulation, just to determine if the heat transfer should take place or not.

Thanks for your feedback. It sounds good, using inbuilt model in the first step.

[vinerm]

Instead of doing a full simulation, you may also do some back of the envelope calculation to determine if collision is really important or not. For doing this, refer the following

https://www.afs.enea.it/project/nept...th/node279.htm

[m.reza]

Originally Posted by vinerm

Instead of doing a full simulation, you may also do some back of the envelope calculation to determine if collision is really important or not. For doing this, refer the following

https://www.afs.enea.it/project/nept...th/node279.htm

Hi again,
I found your point. There are no collisions between particles in my problem. If I want to develop my knowledge on DEM investigation of heat transfer in a drum mixer, except running my own code, which software can simulate the interaction of granular materials? Just some of the articles in this field mentioned that they used in-house DEM code.

[vinerm]

There are a plenty of codes that have DEM model, such as, LIGGGHTS, EDEM, Rocky, etc. But I doubt if any of these model heat transfer across particles because collision time is always much smaller than heat transfer time except when particles are tightly packed. LIGGGHTS probably has it, but I am not sure

[m.reza]

Originally Posted by vinerm

There are a plenty of codes that have DEM model, such as, LIGGGHTS, EDEM, Rocky, etc. But I doubt if any of these model heat transfer across particles because collision time is always much smaller than heat transfer time except when particles are tightly packed. LIGGGHTS probably has it, but I am not sure

Thank for your reply and help.