Directional loss model
 

I need to simulate a thin filter where the fluid can pass only in the direction normal to the inlet surface. I would to use the directional loss model in ANSYS CFX 10. However this model considers the friction losses along the inlet normal direction but it doesn't forces the fluid to pass only in that direction.How can I do to resolve my problem?

Re: Directional loss model
 

...TELL TO "REMIGIO" TRONC CONICS FANNING

Re: Directional loss model
 

Hi,

If the fluid cannot pass in the other directions, the friction losses on those directions are infinite.

Re: Directional loss model
 

Set the loss coefficient in the flow direction.

Set a "factor" in the other two directions to 100. This means that the loss coefficient in the transvers directions will be 100 x K. It's not infinite, but it does the trick.

Jeff

Re: Directional loss model
 

Thank you for the advise. I've allready tought to this solution but I did'nt know the multiplier factor. It can be that 100 is a bit small, but I will try because with very high factors (10e+10) I can't obtain a solution. I use permeability in the directional losses, thus the permeability in the transvers directions is divided by the multiplier factor. Can I use again a factor of 100?

Re: Directional loss model
 

Hi Cenci,

The Cartesian or Cylindrical compenents of the loss model are only used to specify direction of the flow, not losses. If you specify the "Streamwise Loss" as "No Loss" and a "Transverse Loss" with a "Streamwise Coeff. Multiplier" of 100, you should get the desired effect. Even a value of 10 should do.

A streamwise coefficient multiplier of 10 adds a resistance of 10x the local dynamic head, which should be plenty enough to keep the flow straight. If you set it too high, such as 1e10, you may cause other problems due to round-off.

Regards, Robin

Re: Directional loss model
 

Small correction:

Setting the 'streamwise loss' to 'no loss' and a 'transverse multiplier' of 100 means that the transverse loss will be 0*100, which is still 0.

Instead, you need to set either: - a streamwise loss of 'no loss' (if the streamwise loss is indeed zero) and some transverse loss model using appropriately large coefficients (or small permeability) - a streamwise loss model (if the streamwise loss is nonzero) and a larger transverse loss model (eg, a multiplier of 100)