Hello,

I try to analyse the age of air inside a room with mechanical ventilation using CFX5.4. This is my setting:

1.model a room with inflow duct and outflow duct. 2.use addition variable, say VAR 3.set the inital VAR as 0.01 kg/m^3. 4.set inlet BC with 0.0 kg/m^3. 5.run transient simulation. 6.Check the concentration of VAR in my result. --> it didn't converge.

Do anyone have idea or experience of analysing aga of air in a room? Can you give me some suggestion?

Regards

(1). Your question is important, but it is hard for us to answer your question directly. (2). The application of a commercial code(s) can be divided into two areas, one is using it to solve a similar problem, and the other is using it to solve a new problem. (3). My suggestion is: try to find an existing tutorial sample case first, and make sure that you can "repeat" the sample case and obtain the same result. If you can pass this stage, then try to relax (adjust) the geometry or boundary conditions one-at-a-time, to see whether the solution is stable. (4). By following this approach, you can "zoom in" your solution. (5). For example, you can make the room as the size of the inlet , and compute the flow through a simple duct. Once you have the flow through a duct, you can make the room larger than the inlet. It becomes flow through a sudden enlargement. In this way, you can find out whether it is the limitation of the code, or the hidden error of the modeling. (6). The fun part in CFD is the failure of the brain exercise.

Why would you need a transient calculation to converge? If you want to simulate the true transient development of the flow, with an accurate timestep, then it's only necessary to run enough coefficient loops within a timestep to achieve a suitable level of residual reduction.

If you are using a timestep which is on the order of the physical behaviour you want to resolve, then only 2 or 3 coefficient loops are necessary within a timestep.

Dan.

Why does your variable for age of fluid have dimensions of density?

The equation concerning age of fluid is that the substantial time derivative of the age of fluid is 1 (as you follow the fluid motion and move with it, its age increases at the same rate as actual time).

You may find it easier to work with the equation after multiplying both sides by density. There may be several ways to define the new variable in CFX-5.4 depending on what you do with density and what class of additional variable you choose, but it will always include a dimension of time. Be sure that your specify the source term (the 1 or the fluid density depending on your choice) correctly.

For the diffusivity of the age of fluid in turbulent flows, specify only the turbulent eddy viscosity (which is of convective origin) and omit the molecular viscosity.

The inlet boundary condition of age = 0 s is correct. The default initial age of 0 s for the fluid in the domain will be fine. All walls should be "adiabatic"!

And as Dan Williams pointed out, this should converge very quickly.

Hi there,

interesting problem. I have a similar problem with CFX5. I have an additional scalar source in a room and an outlet. But everything I tried with the k-epsilon model, it didn't converge. There are no problems with the differential stress model. Now I'm interested in the turbulence model you have used in your simulation.

Bye,

Martin