Learning Forum

[louai]

Dear all

when defining injection for DPM calculation and activating the Discrete random walk model a time scale constant will be required

i noticed that increasing this number will increase the erosion rate and this may be explained by that the particles will have more time in the domain so more compacts and hence more erosion(at least my understanding)

in Fluent theory guide it was mentioned that for K-e models a value of 0.15 is recommended while a value of 0.3 is recommended for RMS model

i want to know what is the basis of this values and during real simulation how can we choose best value for our model or geometry

is this constant geometry dependent if so how can we estimate it

if any has a good reference to read besides the fluent manual please share with me

Regards

[Rob]

Have a look at how the particles are behaving. Is the erosion a high energy impact or lots of particles repeatedly hitting the same spot due to being trapped by the flow?
The values are mentioned here https://ansyshelp.ansys.com/account/Secured?returnurl=/Views/Secured/corp/v212/en/flu_th/flu_th_sect_pt_turbdispersion.html and there's a reference which may provide additional information.

[louai]

Dear Rob
thanks for your reply , but still wondering whether we still need to stick with Ansys numbers or can change it to calibrate erosion profile for example
( in Fluent theory guide it was mentioned that for K-e models a value of 0.15 is recommended while a value of 0.3 is recommended for RMS model)
i saw some publications run sensitivity on it up to value of 1 so makes me wondering how can we chose its value properly
Regards

[Rob]

You also need to consider the mesh and parcel sensitivity for erosion. If your cells are small & particles big you can get very high local erosion rates. Larger cells or more parcels tends to smooth this out. Hence, the turbulence effect may be less critical in commercial simulations. Add in the uncertainty regarding material properties, sand loading etc and it's common to use erosion rate calculations to estimate rates rather than predict with the level of accuracy we associate with the flow results.
If you're trying to mimic an experiment start with (surface) mesh and stochastic tries before worrying about the model coefficient.

[ai0013]

My feeling is that the time-scale constant CL ≈ 0.15 is an approximation of CL = Cmu^(3/4), where Cmu = 0.09 for the standard k-epsilon model [1]. In OpenFOAM the DRW model uses CL = 0.16431 instead.

For the RSM model I'm not sure why they use CL ≈ 0.30, but I believe that the rationale behind it may be similar to the one of the standard k-epsilon model. They might have used some constant of the RSM model instead.

 

 

[1] Gosman, A. D., and E. Loannides. "Aspects of computer simulation of liquid-fueled combustors." Journal of energy 7.6 (1983): 482-490.