Learning Forum

[yew511]

I am trying to simulate microchannel heat transfer using Ansys Fluent. It is a simple simulations where I want to investigate the Nusselt number of the heat transfer. But before this, i need to verify the simulation by comparing the simulation result with the formula q= mass flow rate X specific heat capacity X (temperature difference). The temperature difference I take is temperature fluid outlet minus temperature fluid inlet. Q, mass flow rate, and specific heat capacity is the input value, and I found that the temperature difference is not the actual value from what I calculate.

Geometry dimension

10 mm length X 0.25 mm width X 0.35 mm height for solid domain and 10 mm length x 0.1 mm width x 0.2 mm height for fluid domain, heat flux is 1e6 W/m2

Based on this the heat flux supplied is 2.5 W, the velocity is 1.41 (based on 187 Reynolds number) and mass flow rate becomes 2.82x10-5, specific heat of water is 4183.56

The temperature outlet from fluent simulation is 319 K

The temperature inlet is 293 K

The temperature difference is 26 K

If we use the formula Q=mcT , the temperature difference should be around 21-22K

Anyone can assist what is the issue for this problem?

It is just a simple heat transfer problem

[Swathi V. V.]

Hello How are you finding the temperature value in Fluent?
Verify that the residuals have converged and that the flux imbalance is less than 1% of the least flux contribution from the boundary surfaces.
Ensure that the Y+ range corresponds to the turbulence model you're using.
Preferably use K-W SST with Y+ around 1 and 10-15 layer of cells to resolve the boundary layer and to capture the thermal effects properly. You can also do mesh independence study.

[Rob]

Also check if the model is actually turbulent before turning turbulence on! It's a micro channel so aim for a more uniform mesh but with a lot more resolution than you'd think as you may be resolving a laminar flow profile.
Re the the temperature did you report mass or area weighted average?

[yew511]

Thanks for comment from everyone. For your information, when I choose turbulent model to run the simulation, I found out that the answer is near to the formula. But, may I ask if the situation is laminar flow, why do I need to turn on turbulent model for accurate solution. If I just use laminar model, the answer would be wrong. Can you all enlighten me because I am keen to learn.

Besides that, may I ask how does Fluent actually calculate the surface heat transfer coefficient. I have read from some source that they said the surface heat transfer coefficient can be calculated by taking wall temperature minus by reference temperature, in here I put 293, but the wall temperature is 310, the surface heat transfer coefficient reported from fluent is 36000, can anyone tell me actually how does Fluent calculate this? Or is there anyway that we can display what is the wall temperature that Fluent used to calculate the surface heat transfer coefficient.

[yew511]

The temperature I report for wall is mass average. For heat transfer coefficient, I use the area weighted average.

[Rob]

OK, for the flow use mass average to get the outlet temperature. For the wall, use area average of whatever you're wanting. Read up on the definitions for the reasons.
HTC is calculated in several ways, you need to check the field functions for the definitions, and which reference values are used. https://ansyshelp.ansys.com/account/Secured?returnurl=/Views/Secured/corp/v221/en/flu_ug/flu_ug_fvdefs.html

[yew511]

Hi sir, I couldn't download the document, can sir email to removed

[Rob]

https://forum.ansys.com/discussion/22563/using-help-with-link

[yew511]

Hi sir, may I know how do we make an inflation layer around the rectangular channel? Because I cant make it across the length of the channel, it only presents in the inlet and outlet region, while across the length there isn't any, and when I run the simulation, there will be error. But inflation can be done for circle channel.

[Rob]

Read up on Sweep in Ansys Meshing. I'll also refer you to my earlier comment about refining ALL of the mesh.

[yew511]

Hi, sir. I have done the inflation layer as what you have mentioned. But I think the laminar problem only be accurate when solved with turbulent. Do sir know is there any other issue for this? I have even try for different geometry such as from rectangular geometry to circular geometry and the results also same.

[Rob]

How well resolved is the mesh? I've worked with a few micro scale problems (and supported users in this field) with good agreement with the experiment.