Learning Forum

[Emma CFD]

To study particle deposition in a 90° bend pipe. I used the RNG turbulence model along with the EWT,  particles were tracked by the DPM, including the turbulent fluctuation effect using the DRW model.

The problem is within the obtained results, the RNG and DRW are known to overestimate the particle deposition rate.  However, the turbulent dispersion effect is neglected in my case. For a particle with dp=2µm,  instead of getting a value of 10% or a value beyond this value, I get 5% as deposition efficiency.

In my simulation, I used a hexahedral mesh with an inflation layer, y+=0.6.

for the RNG model, I used the swirl-dominated flow and the differential viscosity modification options but the results were the same.

 

Can anyone provide help or guidelines?

Thank you!

 

Best regards.

[Rob]

How are you adding particles to the system? How are you measuring deposition?

[Emma CFD]

I inject particles in a surface located at a distance from the bend inlet. Deposition is measured as a deposition efficiency which is equal to the ratio between trapped barticles in the bend section and the total number of particles injected.

[Rob]

 

Number of parcels trapped v injected? Did you use scale by area on the injection? The reason for asking is scale adjusts the parcel mass per facet so you may find the parcels have a different weight so you need to monitor mass. If you don’t scale, you may also be adding more streams around the perimeter (inflation mesh) so skew the results. It’s not a simple application to model: the set up is simple, the physics somewhat less so. 

There's also the issue of small particles penetrating the viscous sub layer, and that wall contact is only checked once the particle is in the near wall cell. So, well resolved (y+ = 1ish) meshes may not be what you want. 

 

 

[Emma CFD]

Thank you for the reply!

No, I inject particles from a surface located in the core region of the flow and I didn't use the scale by surface area option. 

I tried to inject particles from the whole cross-section of the tube and the number of trapped particles raised but I still didn't reach the wanted values.

[Rob]

I suspect you're getting the correct result in Fluent for the release, it's just then trying to compare with the experimental data that's the problem. Are you using a constant size injection or a distribution?