[wadoud31]

HI all,
I try this simulation to know the mass flow rate of air at the bottom in the furnace, I used the knonw boundary conditions wich are:
1 the speed, mass flow rate and the pressure at the inlet of the ejector.
2 at the bottom of the furmace (atmospheric openning), we have only the atmospheric pressur, normaly is the inlet of atmospheric air.
3 at th top (duct) of furnace is the same (atmospheric openning), normaly is the exit (out).
at the and of this simulation we must confirm the mass flow rate at the bottom;
I am trying to simulate an existing device(ejector) used to purge an fired furnace, but the issue is how we must define the boundary conditions.
I tried the boundary conditions below:
Reference pressure: 1 atm
Inlet of ejector: static pressure 7 bar and velocity 48 m/s
In the bottom of the furnace I tried 2 options:
1 inlet of furnace with a static pressure 0 atm
2 as opening with a static pressure 0atm
Out of the furnace I tried 2 options:
1 outlet of furnace with a static pressure 0 atm
2 as opening with a static pressure 0atm
The simulation did not converge.
NB: the geometry and the conditions is like the site (real case).
I need your help and your advice; thank you in advance.
after trying a lot of different setups, it seems that I need some help with the boundary conditions of my problem. I'm looking for a stable (steady state?) solution
I am using ANSYS CFX 14.5.
thanks, I am very grateful

[wadoud31]

here the geometry, thanks .

[ghorrocks]

The FAQ has lots of advice: https://www.cfd-online.com/Wiki/Ansy...gence_criteria

In your case, I can see that your boundaries are too close to the area of interest. You will need to move your boundary conditions further away. Also, as your geometry has lots of bluff edges I would expect they will shed vorticies which will make getting a fully converged steady state result challenging. The FAQ describes some tricks to deal with this.

[wadoud31]

Dear ghorrocks;
thank you for reply, as usual you are helpfull.
coudl you advice me how I can move the boundary conditions further away.
wich the suitable configuration for the boundary conditions.

thanks you

best regards

[ghorrocks]

You need to put boundary conditions at locations where the flow is simple and known. So you want to stay away from sharp edges or separations. So keep adding pipework (from the real object) to a point where the pipes are straight, the flow reasonably developed and you know the flow rate or the pressure.

For choice of boundary condition read the documentation "Choosing boundary conditions".

[wadoud31]

Dear ghorrocks;
thank you for reply,
I changed my mesh (refine it), the simulation crash after several iterartions (arround 600)
I used the known boundary conditions wich are:
1 the speed, mass flow rate and the pressure at the inlet of the ejector.
2 at the bottom of the furmace (atmospheric openning), we have only the atmospheric pressur, normaly is the inlet of atmospheric air.
3 at th top (duct) of furnace is the same (atmospheric openning), normaly is the exit (out).
at the and of this simulation we must confirm the mass flow rate at the bottom;

best regards

[ghorrocks]

General advice is in the FAQ: https://www.cfd-online.com/Wiki/Ansy...gence_criteria

If you want help in your specific case you will need to some images of what you are modelling and your output file.

[wadoud31]

Dear ghorrocks;

I can get a converged solution by using upwind scheme. However, when I change the upwind to high resolution, no converged results come out. Could you please tell me how to solve this problem?

Thanks.

[ghorrocks]

Read the FAQ, it covers this. Move the boundaries further away and consider transient simulation.

Also double precision numerics, better mesh quality, smaller time step, better initial condition and better mesh quality. Mesh quality gets two mentions because it is super-important.

[wadoud31]

Dear ghorrocks;
thanks for your help, I will try and I will share the results.
Best raegards

[wadoud31]

Dear ghorrocks;
I tried timescal change physical timescale 0.1 S the similation run and converged.
Best raegards