CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FloEFD, FloWorks & FloTHERM (https://www.cfd-online.com/Forums/floefd-floworks-flotherm/)
-   -   Solidworks Flow Simulation 2011: Thermal Modeling of light sources (https://www.cfd-online.com/Forums/floefd-floworks-flotherm/99535-solidworks-flow-simulation-2011-thermal-modeling-light-sources.html)

EMS1 April 5, 2012 11:56

Solidworks Flow Simulation 2011: Thermal Modeling of light sources
 
Hello,

I am a Mechanical engineer and a trying to use solidworks to model the thermal output and effects of light sources.

Things that need to be considered:

1. Radiation from the light source
2. Spectral output of light source
3. Effects on objects and environment temperature.
4. Light source in a reflector to create a collimated beam. (point the light in a desired direction)

I have been attempting this for some time now however I am not satisfied with my results. Rather than showing what I have done I was wondering if anyone else has tried to do this also. Maybe give me some steps to how you were able to achieve this task.

Any help would be greatly appreciated.

Thank you.

Boris_M May 4, 2012 15:53

Yes, I've worked with this quite some time and received very good results compared to measurments.

In general there are two ways to define a light source. For LEDs you usually don't need any spectrum definition as the LEDs radiate so little, but for incadecent bulbs it is necessary.
You can define the filament of the bulb (ideally modelled as a cylindrical shape, not a helical shape) with a heat source and apply a radiative surface. With that you will radiate with the grey body accoring to the radiative surface properties.
If you define the surface temperature as a solid temperature for the filament and use white body wall as a radiative surface property on the filament you cause the applied temperature load to only use the convection in the bulb and not radiate at all. Then apply a surface goal for the heat transfere rate on the surface of the filament to see what the convective part of the heat is. Then you apply a radiation source on the surface of the filament with either the temperature of the filament surface just like in the heat source and a black body spectrum or you create a spectrum and use that. For the heat source of the radiation source you select a formula dependency and enter for the equation your total electrical power of the bulb minus the goal you created. That way you are saying that the radiation power is the rest of the power you put into the bulb, without the convective part, as your bulb has convection on the filament and radiation and both together are the overall power.

For the rest you need to define the appropriate materials for solid and radiative surface such as for the reflector.
This data you should know.
For the solid material in transparent solids please consider that the transmissivity is "only" of the material. That means in case you measur such a materials transmissivity you usually place a bulb behind it and a sensor on the other side. Now the difference is the absorbed power. This is not quite correct, because there is part of the radiation reflected on the surface of the material which is not absorbed by the material and therefore not heating the material. If you consider this part too, you will get too hot temperatures on the glass. The ammount is accoring to the optical laws such as lambert-beer law.

I hope this helps,
Boris

EMS1 May 17, 2012 16:51

Boris,

Thank you for the reply. I have modeled using your suggestions and feel I am getting good results. Question about using the diffusive radiation source. I model the filament and apply the spectrum to the surface like you had suggested. Then I am prompted to enter the power for this. For example I modeled a 300[W] light source that only transmitted 36.75[W] of flux. What value should I be putting in this field?
I am also modeling a reflector using symmetry as a radiation surface, seems to work well. Is that correct?

Thanks again,
Eric

Boris_M May 22, 2012 10:23

Hi Eric,
Are you sure it only has 36,75W flux? That would be just 8% radiation?
I would rather use the constant temperature on the filament accoring to it's surface temperature which is usually pretty close to the light temperature for most incandescent bulbs and then use a equation definition for the radiation power with

300W-(heat transfer rate goal on filament surface)=(radiated power over spectrum that is defined)

This should give the most accurate results.

For the reflector I would use a user defined radiation source where you can define what percentage is specular or diffusive reflection and define the absorbtion on this surface. With the symmetry surface everything is specular and there is no absorption at all which is not quite as in reality.

Boris

ethankellough December 22, 2016 09:48

Hi Boris,

I have found your advice on this thread very helpful. However, I am trying to model a fluorescent lamp undergoing forced convection. For me, the constant surface temperature condition does not work as you described it because the lamp should become cooler as flow increases (that is the point of the forced convection).

Do you have any pointers for modelling fluorescent lamps?

Thanks!
Ethan

Boris_M December 24, 2016 06:00

Hi Ethan,

try this, but it might need some calibration with an experiment as the filament definition does as well.

1. Extract the inside volume of the fluorscent bulb as a fluid volume solid model and build it into your assembly again. FloEFD can help you with that when you use only that volume as a fluid volume in a simulation project. You can show if only that volume is shown in the check geometry feature and here you can also select the export the fluid volume and it will create a separate part for it which you then can add to the assembly again.

2. deactivate that volume with the component control and specify the heat generation rate that the fluorescent lamp has on that volume and specify also the gas properties in the lamp for the fluid in it. This should then account for the inside heat generation similar to the filament. This will then also heat up the glass body of the lamp for any external convection.

3. since there is no radiation source inside the lamp like the filament (gas in FloEFD does not radiate), you have to specify the radiation portion of the lamp to the glass body with the corresponding spectrum or black body spectrum if you don't have a spectrum for the lamp that you can specify.

This should now account for the heat generated in the gas that make it generate light as well as for the radiation emitted. You don't have to specify the glass body as transparent or semi-transparent as there is nothing radiating from within it. The radiation source is probably best defined as a dependency (function) of the overall fluorescent lamp power consumption minus the heat generation inside the lamp that was specified.
The calibration part then has to be done to adjust the heat generation portion to the point where the measurements of the lamp fits the simulation as it is not an exact model as you might see. This should be done with filament bulbs and any other bulb the same way and is noting wrong as it is done with any simulation modelling where no details physics or models of the component are available. It is done for chip modelling in electronics cooling and other areas as well.
Since there is no reaction in the gas that generates the losses etc. the exact amount and the physical modelling are not that well known for the simulation and therefore needs to be calibrated.

This calibration can be done with the parametric study method of finding a goal where you specify to alter the heat generation rate until you meet a goal that you can specify from the measurement. Take for example an IR camera or thermocouple measurement of the bulb at a specific point and alter the heat geneation rate until that value is approximately reached. then compare it again with the measurements in maybe a full assembly with housing or if you used IR with the full IR image and the surface plot of the lamp.

I hope this helps,
Boris

Merry Christmas!


All times are GMT -4. The time now is 00:20.