[Pyotr]

Hi everyone,


I've run a simulation of gas flowing through a multi hole solid body (CFX). Owing to the massive size of the body and to the high number of channels, and due to the fact that there are thermal energy sources in the solid body, it requires extensive computational capabilities and a long time to complete the simulation, as it was necessary to implement a mesh with a huge number of elements.


These demanding requirements are mostly due to the 3D CFD simulation of the gas flow through the channels.


So, in order to save time and computer resources, I'd like to couple a 1D gas flow in these channels, together with a convective heat transfer correlation (that of Dittus-Boelter, Seider-Tate, or Taylor), to a 3D heat conduction within the solid.


I don´t need detailed data from the boundary layer along the channels, such as velocity and temperature distribution within it. I need data like 3D spatial temperature distribution in the solid body and the axial variation of the gas bulk temperature in the channels.


Is it possible to accomplish that (thermally couple 1D gas flow to 3D heat conduction in the solid) using CFX or Fluent? If not, is there any software that is able to do what I need?


Many thanks!

[ghorrocks]

Have a look at pipe flow modelling software like Pipe-Flow (https://eng-software.com/products/pipe-flo/). I know some of these can be coupled to 3D solvers like CFX and Fluent.

Alternately you could do your model in user fortran and impose it as a boundary condition on the solid. Obviously this would require considerable development work.

[Pyotr]

Thank you, Glenn!!!

[Pyotr]

Hi everyone,

I simplified a lot the geometry. Now it consists of a hexagonal prism solid body with a hole (channel) along its axis (z direction). The hexagon face is at x-y plane. At the vertices of this body there is heat generation.


Firstly, I ran a simulation of fluid flowing through the channel and collected all data from the flow in order to obtain a Nusselt number heat transfer correlation.


Now I am running this simulation considering the channel as a simple hole (instead of using a fluid, I’m only modeling the heat transfer through the body face).


At the body inner face (i.e, at the interface body / inner hole) I set the boundary type as “Wall” and in the “Boundary Details” tab, Heat Transfer box, I set “Option” to “Heat Transfer Coefficient”. I’m expected to insert an expression for “Heat Transfer Coefficient” and “Outside Temperature” (which took the role of the bulk temperature of the first run).


I intend to get the Heat Transfer Coefficient from my Nusselt number correlation, and the Outside Temperature (I call Tbulk_z) from an expression like “Tbulk_z = Tbulk_0 + q*z/(MassFlux*cp*L)”, where Tbulk_0 is the temperature at z = 0 (which I can choose); q has the meaning of areaInt(Wall Heat Flux)@Interface_GC; MassFlux, cp and L are constants that I already have. The origin of this expression is q(0 to z) = MassFlux*cp*(Tbulk_z-Tbulk_0), and I assumed constant q over the whole geometry (which is a coarse aproximation, but it was the easiest way I found out).


I have basically 2 problems:

1) The Heat Transfer Coefficient I got from the Nusselt Number correlation is a function of the temperature at the wall, which is a variable whose values will be found during the simulation.

2) The Outside Temperature is a function of the Heat Flux through the wall, which is a variable whose values will be found during the simulation. The expression “areaInt(Wall Heat Flux)@Interface_GC” doesn’t work, as these values are not know from the beginning.


Does anybody know how I can handle this situation? How can I get expressions that give me the values of the wall temperature and wall heat flux as a function of z? Is it really necessary to do a model in user fortran?


As I explained in a previous post, I want to solve this problem as 1D fluid flow issue, in order to save time.

Thanks a lot

[ghorrocks]

Q1: FAQ - https://www.cfd-online.com/Wiki/Ansy...ficient_in_CFX

Q2: You need to have a think about how you are modelling this and how it couples together.

I do not know what you are modelling and why you think it complex. If you describe what the complexities are then I might be able to give a more considered answer.

[Pyotr]

I’m sorry I wasn’t clear enough. I’ll try to make myself clearer.

To go straight to the point, my aim is to decrease the time it takes to complete this kind of simulation.

There are 2 cases.

In case nº 1, I used a geometry which consisted of a hexagonal prism solid body with a fluid channel at the center, along its axis (image on the previous post). From the results obtained, I collected the Nusselt number and other data, using the expressions I wrote here: Nusselt Number calculation in Ansys CFX

In case nº 2, I want to use a slightly different geometry: there is no fluid at all in this case, no mesh in the hole located at the center of the solid body. To get the proper results of the temperature distribution within the solid body, I modeled, in CFX-Pre, the heat flow out of the inner face of the solid body by setting this face as a boundary of the type “Wall”. In the “Boundary Details” tab, Heat Transfer box, I set “Option” to “Heat Transfer Coefficient”. Now it appears 2 fields in which I need to insert expressions for “Heat Transfer Coefficient” and “Outside Temperature”.
All the body outer surfaces were set as symmetry boundaries.

I have a expression for Heat Transfer Coefficient, which I got from the Nusselt number correlation I obtained from the case nº 1. It’s of the type h = (k/Dh)*Re^a*Pr^b*(Twall@z/Tbulk@z)^c. Twall@z means Twall at a position z between 0 and L (the length of the body).

I have expressions for thermal conductivity and dynamic viscosity as functions of Tbulk. I have values for the constants a, b and c. So, once I have Tbulk, I have all the parameters of the heat transfer coefficient expression, except Twall.

In order to obtain an expression for “Outside Temperature” along the axis of the body as a function of z, I’m emulating an imaginary fluid and I’m using the expression q_0_to_z [J/s]= MassFlow*Cp*(Tbulk@z – Tbulk@0) to obtain the following expression: Tbulk@z = Tbulk@0 + q_0_to_z /(MassFlow*Cp). I have the values of the constants Tbulk@0, MassFlow, and Cp.

q_0_to_z is the rate of heat transfer out of the inner face, from z=0 to an arbitrary point z.

For the sake of simplicity, I considered the heat flux to be uniform along the inner face, so q_0_to_z = (z/L)*q_0_to_L = (z/L)*q_total, and Tbulk@z = Tbulk@0 + q_total*z /(MassFlow*Cp*L).

The problem is that I don’t know how to obtain q_total or Twall to insert in these expressions. q_total needs to be something like areaInt(Wall Heat Flux)@Inner_face, but this kind of expression is not accepted in CFX-Pre, because Wall Heat Flux is a variable whose values are not known in advance. The same problem happens to Twall, and that’s why I cannot use the expression from the link you posted.

Is there any way CFX could solve these expressions together, in a coupled manner?

I intend to test this kind of simulation with this simple geometry, and after that I want to expand to the case where there are many channels and heat being generated in multiple parts of the body.

I want the temperature distribution within the body in case nº 2 to be as close as possible from that in the case nº 1, so I can have similar results in much less time.

Sorry the long text, but I tried my best in explaining the situation.

Thanks a lot