Heat Transfer validation error

TinusPilot · January 13, 2022, 03:50

Quote:

Originally Posted by Gerry Kan

Good point. The surface is reflected in its emissivity, which gets multiplied into the radiation wattage. So the Stefan-Boltzmann equation becomes:

$q = \epsilon\sigma{A}\left(T^4 - T^4_\infty\right)$ .

For most non-black surfaces $\epsilon \approx 0.9$ give a reasonable estimate, but I imagine it could be lower for polished aluminum surfaces, such as your case.

Gerry.

This is what I found on emissivity:

Aluminum Commercial Sheet: 0.09
Aluminum Heavily Oxidized: 0.2 - 0.31
Aluminum Highly Polished: 0.039 - 0.057
Aluminum Anodized: 0.77
Aluminum Rough: 0.07

I have no idea what I should use but the radiation heat transfer is much lower with these kind of values. I think the 'Aluminium Commercial Sheet' is a good approximation for the emissivity.

LuckyTran · January 13, 2022, 06:26

Yeah this is exactly why you run heat leakage experiments. It's really not cool having an uncertainty of 100% in a quantity that you've assumed until now to be 0.

TinusPilot · January 13, 2022, 07:13

Quote:

Originally Posted by LuckyTran

Yeah this is exactly why you run heat leakage experiments. It's really not cool having an uncertainty of 100% in a quantity that you've assumed until now to be 0.

I get why its not 'cool', but I don't have the experience, time or equipment to eliminate this. If I assume a worst case scenario of an emissivity of 0.1 and delta T of 100 C, the radiation leakage will only be 1.2 W. Everyone has to start somewhere, I get that my experiment is not extremely accurate but that's why I'm here.

Gerry Kan · January 13, 2022, 14:20

Quote:

Originally Posted by TinusPilot

Aluminum Commercial Sheet: 0.09
Aluminum Heavily Oxidized: 0.2 - 0.31
Aluminum Highly Polished: 0.039 - 0.057
Aluminum Anodized: 0.77
Aluminum Rough: 0.07

Since radiative heat loss is not the focus of your study, you can simply estimate the upper and lower bounds by taking reasonable values for emissivity and assert that this could account for the temperature difference you are seeing.

Having said that, 0.1 seems like a reasonable value, since it is not so finely polished, which still gives you 0.3 W (from the 0.6 you see now).

Gerry.

TinusPilot · January 14, 2022, 02:09

Quote:

Originally Posted by Gerry Kan

Since radiative heat loss is not the focus of your study, you can simply estimate the upper and lower bounds by taking reasonable values for emissivity and assert that this could account for the temperature difference you are seeing.

Having said that, 0.1 seems like a reasonable value, since it is not so finely polished, which still gives you 0.3 W (from the 0.6 you see now).

Gerry.

Thanks for your help, I really appreciate it. I have attached the results that I have generated with the feedback from the forum (especially the modeling of the pins). These are generated with the velocity inlet (measured with pitot) instead of mass flow. The dark green/blue bars represent the potential thermocouple error (1.5 C) and the lighter blue/green bars represent the 10% error bar. The error is larger with higher temperatures, this can be linked to the heat leakage due to radiation.

Gerry Kan · January 17, 2022, 09:57

Martijn:

Just out of curiosity, could you show the plots from your previous message with the normalized temperature $\theta\equiv(T-T_0)/(T_\infty-T_0)$ , where

is the initial temperature of the plate (also same as ambient, I presume), and $T_\infty$ is the steady-state plate surface temperature?

This way you have a more meaningful basis of comparison between the experimental and CFD data. Of course, you should use the same $T_\infty$ to normalize both the measurement and simulation data. I would use the observation temperature, but you could also choose the CFD temperature.

Gerry.

TinusPilot · January 18, 2022, 04:28

Quote:

Originally Posted by Gerry Kan

Martijn:

Just out of curiosity, could you show the plots from your previous message with the normalized temperature $\theta\equiv(T-T_0)/(T_\infty-T_0)$ , where

is the initial temperature of the plate (also same as ambient, I presume), and $T_\infty$ is the steady-state plate surface temperature?

This way you have a more meaningful basis of comparison between the experimental and CFD data. Of course, you should use the same $T_\infty$ to normalize both the measurement and simulation data. I would use the observation temperature, but you could also choose the CFD temperature.

Gerry.

I normalized the plot of the 5 m/s case, I used the Tinf of the observation as you recommended.

I have another question; STAR-CCM+ assumes a Turbulent Prandtl Number of 0.9, in my opinion it should be around 0.7-0.71 for dry air at 20C right?

LuckyTran · January 18, 2022, 07:39

0.7 is the molecular Prandtl number, which is a material property (of air).

The turbulent Prandtl number is a flow property and is for modeling of the turbulent heat flux. This number varies from flow to flow but you shouldn't ever touch this unless you really know what you are doing. It's the heat equivalent of turbulent viscosity.

TinusPilot · January 18, 2022, 07:54

Quote:

Originally Posted by LuckyTran

0.7 is the molecular Prandtl number, which is a material property (of air).

The turbulent Prandtl number is a flow property and is for modeling of the turbulent heat flux. This number varies from flow to flow but you shouldn't ever touch this unless you really know what you are doing. It's the heat equivalent of turbulent viscosity.

Thanks for your reply, shouldn't it be 'Prandtl number' then in the picture?Because it is a material property?

LuckyTran · January 18, 2022, 08:00

The turbulent Prandtl number is not a mislabel. It is indeed a turbulent Prandtl number. But it is here being listed as a material property, even though it is not a material property (in the thermodynamic sense). It shouldn't be there, but it is.

And this is one of my grievances of Star. This is what happens when you hire programmers to write a CFD code. They're really good at object oriented programming (emphasis on the object part) and doing things that makes sense to them and no one else.

TinusPilot · January 18, 2022, 08:17

Quote:

Originally Posted by LuckyTran

The turbulent Prandtl number is not a mislabel. It is indeed a turbulent Prandtl number. But it is here being listed as a material property, even though it is not a material property (in the thermodynamic sense). It shouldn't be there, but it is.

And this is one of my grievances of Star. This is what happens when you hire programmers to write a CFD code. They're really good at object oriented programming (emphasis on the object part) and doing things that makes sense to them and no one else.

Thanks for your help

, did not expect that it would be listed wrong. I am going to do some more research into why STAR put it there with the other material properties.

LuckyTran · January 18, 2022, 10:48

I don't want to add too much that it distracts from the original topic. But you will eventually find that turbulent Schmidt numbers, etc. are all also listed under material properties. It is a global Star philosophy to put them there and it's not a bug.

As for why this is allowed to happen? Well it probably doesn't help that the governing equations in laminar and (U)RANS are identical and we can use an effective viscosity (and effective conductivity, and so on) for turbulent flows and the rest of the code is the same. This convenience lasts only as long as we stick to eddy viscosity models. But if you weren't using an eddy viscosity model then you wouldn't have a turbulent viscosity, turbulent Prandtl number, turbulent Schmidt number, etc. to worry about anyway.

Gerry Kan · January 19, 2022, 05:39

Hi Martijn:

Sorry for not getting back to you sooner. I was stuck in a writer's block for 3 days and now it's over!

Thank you for plotting this in normalized form. You can now really the level of discrepancy between observation and simulation results, which is about 20% in the steady state.

For heat transfer problems I would say it is not unreasonable. Of course, the various contributing factors can be the starting point of your discussion. I am interested to see this.

For the record, the turbulent Prandtl number is usually 1 or close to 1 to reflect the effects of mechanical mixing due to turbulence in the flow. I doubt you need to change this value because you don't know what additional calculations are done under the hood of Star-CCM+, in addition to what we were taught from text books.

Gerry.

TinusPilot · January 20, 2022, 05:26

Quote:

Originally Posted by Gerry Kan

Hi Martijn:

Sorry for not getting back to you sooner. I was stuck in a writer's block for 3 days and now it's over!

Thank you for plotting this in normalized form. You can now really the level of discrepancy between observation and simulation results, which is about 20% in the steady state.

For heat transfer problems I would say it is not unreasonable. Of course, the various contributing factors can be the starting point of your discussion. I am interested to see this.

For the record, the turbulent Prandtl number is usually 1 or close to 1 to reflect the effects of mechanical mixing due to turbulence in the flow. I doubt you need to change this value because you don't know what additional calculations are done under the hood of Star-CCM+, in addition to what we were taught from text books.

Gerry.

Thanks for the feedback, turbulent prandtl number is clear for me now. This project is now shortly on pause, I first have to do some propeller simulating.

January 18, 2022, 07:39		#28
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,728 Rep Power: 66	0.7 is the molecular Prandtl number, which is a material property (of air). The turbulent Prandtl number is a flow property and is for modeling of the turbulent heat flux. This number varies from flow to flow but you shouldn't ever touch this unless you really know what you are doing. It's the heat equivalent of turbulent viscosity. TinusPilot likes this.

January 18, 2022, 08:00		#30
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,728 Rep Power: 66	The turbulent Prandtl number is not a mislabel. It is indeed a turbulent Prandtl number. But it is here being listed as a material property, even though it is not a material property (in the thermodynamic sense). It shouldn't be there, but it is. And this is one of my grievances of Star. This is what happens when you hire programmers to write a CFD code. They're really good at object oriented programming (emphasis on the object part) and doing things that makes sense to them and no one else. TinusPilot likes this.

January 18, 2022, 10:48		#32
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,728 Rep Power: 66	I don't want to add too much that it distracts from the original topic. But you will eventually find that turbulent Schmidt numbers, etc. are all also listed under material properties. It is a global Star philosophy to put them there and it's not a bug. As for why this is allowed to happen? Well it probably doesn't help that the governing equations in laminar and (U)RANS are identical and we can use an effective viscosity (and effective conductivity, and so on) for turbulent flows and the rest of the code is the same. This convenience lasts only as long as we stick to eddy viscosity models. But if you weren't using an eddy viscosity model then you wouldn't have a turbulent viscosity, turbulent Prandtl number, turbulent Schmidt number, etc. to worry about anyway. TinusPilot likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[OpenFOAM.org] Compile OF 2.3 on Mac OS X .... the patch	gschaider	OpenFOAM Installation	225	August 25, 2015 19:43
[swak4Foam] GroovyBC the dynamic cousin of funkySetFields that lives on the suburb of the mesh	gschaider	OpenFOAM Community Contributions	300	October 29, 2014 18:00
ParaView for OF-1.6-ext	Chrisi1984	OpenFOAM Installation	0	December 31, 2010 06:42
checking the system setup and Qt version	vivek070176	OpenFOAM Installation	22	June 1, 2010 12:34
DecomposePar links against liblamso0 with OpenMPI	jens_klostermann	OpenFOAM Bugs	11	June 28, 2007 17:51

January 13, 2022, 06:26		#22
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,728 Rep Power: 66	Yeah this is exactly why you run heat leakage experiments. It's really not cool having an uncertainty of 100% in a quantity that you've assumed until now to be 0.

January 17, 2022, 09:57		#26
Gerry Kan Senior Member Gerry Kan Join Date: May 2016 Posts: 360 Rep Power: 11	Martijn: Just out of curiosity, could you show the plots from your previous message with the normalized temperature $\theta\equiv(T-T_0)/(T_\infty-T_0)$ , where is the initial temperature of the plate (also same as ambient, I presume), and $T_\infty$ is the steady-state plate surface temperature? This way you have a more meaningful basis of comparison between the experimental and CFD data. Of course, you should use the same $T_\infty$ to normalize both the measurement and simulation data. I would use the observation temperature, but you could also choose the CFD temperature. Gerry.

January 19, 2022, 05:39		#33
Gerry Kan Senior Member Gerry Kan Join Date: May 2016 Posts: 360 Rep Power: 11	Hi Martijn: Sorry for not getting back to you sooner. I was stuck in a writer's block for 3 days and now it's over! Thank you for plotting this in normalized form. You can now really the level of discrepancy between observation and simulation results, which is about 20% in the steady state. For heat transfer problems I would say it is not unreasonable. Of course, the various contributing factors can be the starting point of your discussion. I am interested to see this. For the record, the turbulent Prandtl number is usually 1 or close to 1 to reflect the effects of mechanical mixing due to turbulence in the flow. I doubt you need to change this value because you don't know what additional calculations are done under the hood of Star-CCM+, in addition to what we were taught from text books. Gerry.