I am modeling flow and heat transfer in a microchannel and need to compute the local Nusselt # as a function of axial position. Is there a good way of getting this from Fluent directly? If not, how I can extract the necessary data (local wall and local bulk fluid temperature) and compute it outside Fluent?

Thanks.

<font face = "courier new">Best way to do is to use a user-defined scalar which will act as a 'storage' for the Nusselt number. Follow this method:

(1) solve your temperature profile with momentum etc.. excluding the user-defined scalar.

(2) Once you have obatined the solution in the first step, you have to write a UDF lopping all the cells u need for calculating nusselt number distribution and the you will be required to many of te macros referred in the UDF manual. You will have to assign this value of local nusselt number to the user-defined scalar. Solve the problem for 'all eqns' including the scalar for 1 iteration.

(3) Now if you plot the UDS u will have Nu distribution.

Note: I have done exactly the same thing in STAR-CD. I am just as familiar with FLUENT 6 tho I have not done it. I am sure this is the way out.

Regards

Murali</font>

Will try that. Thanks!

As I am not too familiar with UDS, can someone be kind enough to guide me along? If you have a similar example to show me, that would be awesome

Thanks in advance...

Fluent 6.0 >Plot>XY Plot>Turblence>Turbulent Kinetic Energy 'effective Prandlt number' can be written along axis function and knowing the 'Reynolds number' can one calculate 'nusselt #'.

Not sure if this will work as I am mainly modeling laminar flow. Can you explain what you mean by "knowing the 'Reynolds number' can one calculate 'nusselt #'"? Do you mean using a Nusselt # correlation? The flow in my case is mainly thermally developing...

For calculating Nu you need to know the thermal boundary layer thickness. I wonder how Murali did the job? Did you just take the wall nearest Temperarture to calculate the gradient?

Nu is nothing else than charcteristic length L divided by the boundary layer thickness. (Nu=heat-transf-coeff*L/thermal-conduction and heat-transf-coeff= thermal-conduction/thermal-boundary-layer-thickness)

Remember that Nu is just a value which helps and helped to carry out analytic or empirical heat-transfer calculations. For CFD it is a post-processing value with all its perfidities.

Fluent 6.0 >Plot>XY Plot>Y axis Functions: Wall Fluxes>Surface Nusselt number Is this will work

this won't work. the surface Nu# obtained this way depends on the reference values you assigned. as i am after the local Nu#, it will be extremly tedious to do it this way...

<font face = "courier new">Jake:

Refer to page 4.5 in the UDF manual. Example 2 uses DEFINE_ADJUST macro to store the value in a UDS. This is the way I think I would use the UDS to store Nusselt number values:

(1) Just specify to FLUENT that you want one UDS. Run your problem without solving for UDS (dont define any flux just use default) and no define_adjust.

(2) Run your problem with the UDS for 1 iteration. No define_adjust

(3) Hook up define_adjust, do the Nusselt number calculation and store in UDS and do "NOT" solve for UDS in this iteration. DO just 1 iteration only and ensure that you loop on all cells.

My guess is that the second step allots the memory for UDS. The third step assigns the value to UDS. And since u will not solve for the UDS, the value of UDS which was specified in the define_adjust would remain the same even after calculation (iteration).

BTW, this value of Nusselt number you store better a "cell" value and not a face value. This way you can pretty much plot the Nusselt number distribution.

Murali</font>

Hi Murali,

Thanks, you are a great help !

Jake