Hello all,

I am trying to simulate a bag filter assuming bank of bags as a porous media. The measured pressure drop across the bags is 100 mmWC [around 980 Pa]. I had tried permeability as well as quadratic resistance options. But I got DP=20 mmWC [around 200 Pa]. Velocity profiles are as expected. So I tried general momentum source term= -180 (kg.m^-2.s^-2)in Y dirn along with permeability value. Using this, I got DP=80 mmWC.

However the vector direction in the bank of bags is reverse. It means, flow is traveling from bags towards the inlet. In actual, it is not possible. Flow should travel towards the outlet. Also flow pattern is extremely turbulent, with unacceptable high velocity values.

Flow is incompressible & isothermal flow. Boundary conditions are velocity inlet and pressure outlet, Std K-E model was used. Mesh used was hexahedral mesh [0.6 million cells].

Could anybody tell where did I go wrong?

Any suggestion would be highly appreciated.

Regards,

Sanyo

HELP ! HELP !! HELP !!! PLZ.......

Hi,

The resistance option is likely to be the best option. How have you tuned the resistance constant? Looks like you need to adjust the constant to give you the right resistance.

Glenn Horrocks

Hi,

At the outset, thank you very much Mr. Glenn. I had calculated the permeability using Darcy's law. The value used was of the order (E^-8). I also tried increrasing this resistance by using the values of the order (E^-10 or E^-11). The increase in DP is not significant(increase of 3 to 4 mmWC). However beyond the value of (e^-11), porous medium blocks all the flow & solve fails due to linear overflow. Same thing happens with quad. resistance coeff. Beyond certain value solve fails. I think, when I use these constants, somehow solver readjust these values to keep the certain velocity patterns & I am not able to obtain increased DP.

So I thought this might be the limitation of porous model & so I tried the momentum source term which is giving me unrealistic results.

Is there any parameter or any trick I do not know?

Please help.

Regards,

Hi,

Don't you just need to adjust the constant somewhere between the values you are using? You have done too much and too little, so isn't the correct amount somewhere in the middle?

Glenn Horrocks

Hi,

Thanks for your reply. As I stated earlier, The measured pressure drop across the bags is 100 mmWC [around 980 Pa]. Using permeability value of the order e^-8; I got DP=20 mmWC [around 200 Pa]. Using permeability value of the order e^-11; I got DP around 23-24 mmWC. If I use this value of order e^-12, porous medium blocks all the flow & solve fails due to linear overflow. So it is pointless to use the values between e^-8 & e^-12, as it wont produce expected pressure drop. I am unable to understand the treatment of porous media. Any suggestion?

Thanks for your kind assistance.

With warm Regards,

Sanyo.

Hi,

How are you implementing the region? Is it a thin 3d region or a 2D region?

Glenn Horrocks

Hi,

Bag house contains around 1024 bags. Certain no.of bags are grouped in a bank. This bank is modeled as a porous block by defining a subdomain . So there are around 10 blocks, all grouped in a single porous domain. The permeability value is assigned to the single porous domain consisting all these banks. Space between two blocks is fluid space.

Plz let meknow if u need any additional info.

Thanks for your support.

With regards,

Sanyo

Hi,

So it boils down to a convergence problem when you start using the amount of resistance you need. Have you read the section in the documentation about obtaining convergence? This also gives some tips: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

Also what form of momentum source are you using? Isotropic loss or directional loss?

Glenn Horrocks

Hi, Thanks for the link.

I had not tried the double precision solver. I will try it. What covergence criteria should I look at? I had set convergence criteria max residual <1e-4. But I never acheived this criteria. I will try the suggestions now.

And I have used Isotropic loss model.

Thanks again.

Regards,

Sanyo.