CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)

 Anurag February 2, 2005 16:08

Hi All, I am doing a transient simulation of a porous media in which a very small quantity (compared to the size of the domain) of fluid (A) is injected into the domain. For this I am using multi-component option, where dmain fluid (B) is treated as constraint and the fluid (A) is injected as transport equation.

In another attempt I am injecting the same fluid (A) into the domain (B) as an additional variable. I am using "opening boundary condition" and at the boundary I specify value of additional variable fluid (A) = 0 to make sure that all the fluid (A) remains inside the domain (Similar to tutorial # 4).

I am using the same diffusion coefficient in both the cases. I was expecting the same result in both the cases but it is coming out to be different.

My questions:

2) What is the reason behind the difference between the two results?

3) When we define fluid (A) as an additional variable, we don't have any option of defining the other properties of the fluid (A), such as physical state, density, molecular weight, thermal conductivity .. etc. how to define these properties ?

4) In tutorial # 4 smoke is used as an additional variable, which is injected into the "Air" domain. Do we treat "smoke" as a material? Do we take the properties of smoke same as that of Air?

5) Is it possible to inject a material into the domain as an additional variable? This way i can assign other properties also.

Anurag

 Rui February 3, 2005 13:57

Hi,

In Tutorial 4, when the Domain is created, the chosen fluid is Air at 25 C. Therefore it is the only fluid existing in the Domain, and at any boundary. So, the fluid entering through the Vent boundary is Air at 25 C, whether there is or there isn't an additional variable.

The additional variable smoke is modelled by a transport equation (see CFX-5.6 Manual, Solver Theory, Basic Solver Capability Theory, page 32), and it does not influence the flow field. That is, the flow field will be the same with or without the additional variable. This is not true if you set the fluid (Air at 25 C) properties as function of the additional variable.

In CFX-Post, at the Vent boundary, when you chose the variable smoke (hybrid) you can see that it has the value 1 kg/m3, which is what you set as the boundary condition for smoke. When you chose smoke.Specific (hybrid) you see it has the value 0.84, smoke.Specific=smoke/ro = (1kg/m3)/(1.185kg/m3); 1.185 kg/m3 is the density of Air at 25 C. You may regard that value, 0.84, as the smoke mass fraction, only if you assume that the smoke density is the same as the air density.

Imagine that the smoke density is 4 kg/m3. If in 1 m3 you have 1 kg of smoke, it occupies 0.25 m3, and the rest of the volume, 0.75 m3, is occupied by air. As the air density is 1.185 kg/m3, on that volume there will be 0.889 kg of air. The mass fraction of smoke will be therefore 1kg/(1kg+0.889kg)=0.53. If the smoke density was that of the air, 1.185 kg/m3, its mass fraction would be 0.84.

If you model the fluid as a mixture of components, the density used in the momentum equations will be obtained by Eqn.12 (in CFX-5.6 Manual, Solver Theory, Basic Solver Capability Theory, page 30). The viscosity also used in the momentum equations, as well as other fluid properties, will be obtained by Eqn.13. So, you will only obtain the same results with an additional variable and with a mixture of 2 components, if those 2 components have the same properties. Note that this is not allways true as the limit for the mass fraction of any component is 1, while there is not limit for smoke.Specific.

About your question 5, I think it shall be possible, if you create an expression for the mass fraction of a component as a function of the additional variable, and the fluid properties as function of that mass fraction. But if you model the problem as a mixture of components, in CFX-Pre in the Material Editor window, when you create your mixture, check the Advanced Property Override box and then you can define any mixture property as a function of mass fractions, and/or of other variables. The way CFX calculates by default the mixture properties is described in CFX-5.6 Manual, Solver Modeling, Basic Capabilities Modeling, pages 39-44.

By the way, when you have an Opening B.C. and specify the value of additional variable fluidA=0, it doesn't prevent fluidA from leaving the domain through that boundary, it just prevents fluidA from entering through it. That is, if the fluid is flowing into the domain through that boundary, the quantity of fluidA flowing in is zero.

Regards,

Rui

 Anurag February 4, 2005 17:45

Dear Rui,

Thank you very much for detaied reply. it answered all of my doubts.

i put the same question to CFX tech support also. i am sending their reply. it might be useful for others.

# An additional variable is treated like a neutral tracer, i.e. a fluid that has the same properties as the primary fluid but can diffuse within it. Its presence or absence does not affect the properties of the fluid and there are no buoyancy effects associated with it.

Consider a domain filled with air. If we tag a certain volume of air by making it radioactive, say, then we have a tracer that has the same properties as the main fluid but has a distinct identity that we could track. We could do this accurately with an additional variable- that's the approach that's adopted in Tutorial #4.

# You're introducing slightly more overhead when you treat the flow as multicomponent, but you are also making fewer approximations. If you want the concentration or mass fraction of A to affect the fluid properties or if the densities are different such that buoyancy effects are important, then you should treat the problem as a multicomponent flow.

# A specific additional variable has implied units of kg/kg while a volumetric additional variable has units of kg/m^3. The two values are related by: