Hello Elmerers,
is there a canonical way to model the volumetric absorption of light in a partially translucent (isotropic) material?
I'd like to model the transient heat input (as a transmission path dependent body force) and the resulting temperature rise if pulsed light is shone into a spatially structured medium (to be represented as a set of distinct volumes) of differently absorbing materials.
Regards,
Peter
Volumetric absorption of light?
-
raback
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Volumetric absorption of light?
Hi Peter
I don't know about canonical way in general except that none of them exist in Elmer. The best that I can suggest is using the "Rosseland Approximation" (google gives a lot of references) which results to T^3 dependent thermal conductivity and write a MATC/UDF for that. There are some heavy assumptions made here, for example that the optical distance is smaller than the geometric dimensions, and that scattering of light is isotropic. If the optical distance is large then the interaction with the walls becomes more important and some more advanced methods are needed, I guess.
-Peter
I don't know about canonical way in general except that none of them exist in Elmer. The best that I can suggest is using the "Rosseland Approximation" (google gives a lot of references) which results to T^3 dependent thermal conductivity and write a MATC/UDF for that. There are some heavy assumptions made here, for example that the optical distance is smaller than the geometric dimensions, and that scattering of light is isotropic. If the optical distance is large then the interaction with the walls becomes more important and some more advanced methods are needed, I guess.
-Peter
Re: Volumetric absorption of light?
Hello Peter,
thanks for your quick pointer to the Rosseland Approximation! I scanned some of the googled results, but judging from those texts this would by far overdo or not meet what I need, coupling the radiation intensity to the thermic re-radiation effect in a given volume. In my problem only temperature rises of some (to some ten) degrees Centigrade are to be considered. The main objective of the work is to calculate the volumetric redistribution of absorption-generated heat in the vicinity of the more absorbing volume bodies.
Let me therefore re-formulate my problem in a potentially better understandable way: I'd like to perform a (node-wise) anteceding calculation of laser beam weakening due to the absorption of the light on its path to a given node. The irradiated node does not emit radiation by itself but only swallows a bit of it by absorption (and thereby generates a node-dependent body force) and creates a bit of scattering to "illuminate" parts of volumes that lie behind more absorbing ones in the direct direction of the laser beam. In a first approximation one might even forget about this scattering effect and only consider the direct-path weakening.
Since the volumetrically varying heat production is only dependent on the given geometric setting the workflow could be redesigned in a way that would include an anteceding raytracing, yielding a node-wise/position-dependent body force that would not change for a whole transient simulation. But since this calculation would have to be performed for any position on the mesh I would have thought it to be better integrated in the node-wise calculation that Elmer does perform anyhow.
And, to be honest, I am not aware of a raytracer that would yield a volumetrically resolved absorption value that could be fed into Elmer as body force ...
Regards,
Peter
thanks for your quick pointer to the Rosseland Approximation! I scanned some of the googled results, but judging from those texts this would by far overdo or not meet what I need, coupling the radiation intensity to the thermic re-radiation effect in a given volume. In my problem only temperature rises of some (to some ten) degrees Centigrade are to be considered. The main objective of the work is to calculate the volumetric redistribution of absorption-generated heat in the vicinity of the more absorbing volume bodies.
Let me therefore re-formulate my problem in a potentially better understandable way: I'd like to perform a (node-wise) anteceding calculation of laser beam weakening due to the absorption of the light on its path to a given node. The irradiated node does not emit radiation by itself but only swallows a bit of it by absorption (and thereby generates a node-dependent body force) and creates a bit of scattering to "illuminate" parts of volumes that lie behind more absorbing ones in the direct direction of the laser beam. In a first approximation one might even forget about this scattering effect and only consider the direct-path weakening.
Since the volumetrically varying heat production is only dependent on the given geometric setting the workflow could be redesigned in a way that would include an anteceding raytracing, yielding a node-wise/position-dependent body force that would not change for a whole transient simulation. But since this calculation would have to be performed for any position on the mesh I would have thought it to be better integrated in the node-wise calculation that Elmer does perform anyhow.
And, to be honest, I am not aware of a raytracer that would yield a volumetrically resolved absorption value that could be fed into Elmer as body force ...
Regards,
Peter
-
raback
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Volumetric absorption of light?
Hi Peter
Well, there is no volumetric ray tracing in Elmer. But we're just working an a particle tracker module which might be used for this. The particles should just have a decay of intensity in the material and might also randomly reflect. As one application I've tested it for is to trace a number of particles which while moving at each point they act as gaussian integration points and thereby accumulate a finete element equation. As said this is under work and will hopefully open many interesting avanues for non-orthodox coupled models. The hope is to model different kinds of particles, either tracer or newtonian dynamics. Close-range interaction between particles will be supported (granular flow, collisions, etc.).
-Peter
Well, there is no volumetric ray tracing in Elmer. But we're just working an a particle tracker module which might be used for this. The particles should just have a decay of intensity in the material and might also randomly reflect. As one application I've tested it for is to trace a number of particles which while moving at each point they act as gaussian integration points and thereby accumulate a finete element equation. As said this is under work and will hopefully open many interesting avanues for non-orthodox coupled models. The hope is to model different kinds of particles, either tracer or newtonian dynamics. Close-range interaction between particles will be supported (granular flow, collisions, etc.).
-Peter
Re: Volumetric absorption of light?
Hello Peter,
Kind regards,
Peter
This does sound interesting indeed! Could you drop me a note if there is some code to give it a try? It is not very urgent for me, as the objective I am after is somewhat playground-like and mostly satisfying my curiosity ...raback wrote:... But we're just working on a particle tracker module which might be used for this. ...
Kind regards,
Peter
Re: Volumetric absorption of light?
Hi Elmer team,
just as a hint: Years ago, I have used the raytracing tool radiance (http://radsite.lbl.gov/radiance/) and found it very reliable and efficient. It is Open Source. I don't know how far you are in your work, but in case you still need expertise in that field, you could certainly contact those guys and/or use their software as inspiration or maybe even directly as a library or module. The raytracer is a standalone program which you can launch from the command line, so I imagine that indeed it could be integrated into the Elmer framework if this would help.
Matthias
just as a hint: Years ago, I have used the raytracing tool radiance (http://radsite.lbl.gov/radiance/) and found it very reliable and efficient. It is Open Source. I don't know how far you are in your work, but in case you still need expertise in that field, you could certainly contact those guys and/or use their software as inspiration or maybe even directly as a library or module. The raytracer is a standalone program which you can launch from the command line, so I imagine that indeed it could be integrated into the Elmer framework if this would help.
Matthias
Re: Volumetric absorption of light?
Hello Matthias,
) need: Raytracing means you are only interested in "what comes out" in the direction of the viewport. What is happening inside the scene is only import to the extent it changes the final view.
I just threw an eye on Radiance's longer description of functionality, and it confirmed my view: There doesn't seem to be a means to retrieve volumetric information on intensities - or at least not without any very dirty tricks.
Regards,
Peter
To use a raytracer was my first thought as well, but: In my understanding the very method those programs use is contradictory for what we (or let's better say: Imzenker wrote:... I have used the raytracing tool radiance (http://radsite.lbl.gov/radiance/) and found it very reliable and efficient.
) need: Raytracing means you are only interested in "what comes out" in the direction of the viewport. What is happening inside the scene is only import to the extent it changes the final view.I just threw an eye on Radiance's longer description of functionality, and it confirmed my view: There doesn't seem to be a means to retrieve volumetric information on intensities - or at least not without any very dirty tricks.
Regards,
Peter
Re: Volumetric absorption of light?
Hi Peter,
you are right, the radiance package is intended for generation of photorealistic view, so the rays are traced backwards from the irradiated surface to the light source in order to save computation time.
However, using the standalone raytracing module of radiance (called "rtrace"), you can follow any ray through your geometry. At each interface, it splits up according to the laws of optics, with intensity distributed among the daughter rays. So you are not limited to the viewport if you proceed this way.
I do not know how volumetric absorption is treated, however. You could try to ask the authors.
The drawback of using rtrace alone is that you have to generate your rays on your own. At least that is what I did (actually 10 years ago) when I used rtrace for a raytracing application, where I wanted to calculate energy flows between a radiating surface and other surfaces (mirrors, absorbers, filters). This necessitated some programming work indeed. If I remember right, I wrote a C program to determine starting point and angle of the rays, call rtrace to do the actual tracing, and read back the results.
I don't know how radiance has evolved since that time.
I just thought it might help the the Elmer team to have had a look on radiance/rtrace in order to avoid reinventing the wheel. But the work on the Elmer particle tracker is probably already in an advanced state. They certainly will do it the right way...
Oh yes, and the question is also whether we want to do raytracing in the optical sense, which is what radiance does, or if we want to track particles using the Monte Carlo approach, meaning that when hitting an interface, the particle is either reflected, absorbed or transmitted according to probabilities.
Matthias
you are right, the radiance package is intended for generation of photorealistic view, so the rays are traced backwards from the irradiated surface to the light source in order to save computation time.
However, using the standalone raytracing module of radiance (called "rtrace"), you can follow any ray through your geometry. At each interface, it splits up according to the laws of optics, with intensity distributed among the daughter rays. So you are not limited to the viewport if you proceed this way.
I do not know how volumetric absorption is treated, however. You could try to ask the authors.
The drawback of using rtrace alone is that you have to generate your rays on your own. At least that is what I did (actually 10 years ago) when I used rtrace for a raytracing application, where I wanted to calculate energy flows between a radiating surface and other surfaces (mirrors, absorbers, filters). This necessitated some programming work indeed. If I remember right, I wrote a C program to determine starting point and angle of the rays, call rtrace to do the actual tracing, and read back the results.
I don't know how radiance has evolved since that time.
I just thought it might help the the Elmer team to have had a look on radiance/rtrace in order to avoid reinventing the wheel. But the work on the Elmer particle tracker is probably already in an advanced state. They certainly will do it the right way...
Oh yes, and the question is also whether we want to do raytracing in the optical sense, which is what radiance does, or if we want to track particles using the Monte Carlo approach, meaning that when hitting an interface, the particle is either reflected, absorbed or transmitted according to probabilities.
Matthias
-
raback
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Volumetric absorption of light?
Hi
Well, I think we are so far that we have more or less fixed the approach. We want the tracer to be coupled to the fields, if needed, and this means that
A lot has already been done and even parallel performace looks promising. There should be some examples out hopefully within a few weeks. All support for testing will then be welcome.
-Peter
Well, I think we are so far that we have more or less fixed the approach. We want the tracer to be coupled to the fields, if needed, and this means that
- each particle must know in which element it sits
- FE fields can be evaluated at the particle position
- Particle properties may be averaged to obtain FE fields
A lot has already been done and even parallel performace looks promising. There should be some examples out hopefully within a few weeks. All support for testing will then be welcome.
-Peter
Re: Volumetric absorption of light?
Hi Peter,
Just one question w/r/to your envisioned development in the direction of multi-physical applications: Will it be possible to have two different species of particles at the same time? I envision the following scenario for a high-temperature solar receiver I hopefully will have to deal with next year: Concentrated (visible) sun light is entering a geometrically slightly complex hollow structure (cavity), being partly reflected and absorbed by the volume's walls. As the tempeatures rise to some 800+ deg Centigrade in some parts of the receiver (some of) the walls start to emit infra-red radiation as well that is directionally not coupled to the sunlight, of course. At the same time various fluids are to be heated inside thermally coupled tubings to different extents by absorbing the wall-absorbed heat in the receiver themselves. The cavity will also be heated by an adaptable fossil combustion for cloudy shading and night operation, but with the aim of minimizing heat losses during that operation mode.
To summarize: A heat transport problem coupled to a convective flow, consisting of radiative transfer for two different raytracing domains, rigid body heat transport, convection of flue gasses, and thermal fluid-wall interaction. I think that may prove multi-physical enough as a test case ...
Regards,
Peter
Sounds interesting indeed! If you need another DAU tester, just say a word as soon as you feel ready for it, and I'll give it a try!raback wrote:Well, I think we are so far that we have more or less fixed the approach. ...
A lot has already been done and even parallel performace looks promising. There should be some examples out hopefully within a few weeks. All support for testing will then be welcome.
Just one question w/r/to your envisioned development in the direction of multi-physical applications: Will it be possible to have two different species of particles at the same time? I envision the following scenario for a high-temperature solar receiver I hopefully will have to deal with next year: Concentrated (visible) sun light is entering a geometrically slightly complex hollow structure (cavity), being partly reflected and absorbed by the volume's walls. As the tempeatures rise to some 800+ deg Centigrade in some parts of the receiver (some of) the walls start to emit infra-red radiation as well that is directionally not coupled to the sunlight, of course. At the same time various fluids are to be heated inside thermally coupled tubings to different extents by absorbing the wall-absorbed heat in the receiver themselves. The cavity will also be heated by an adaptable fossil combustion for cloudy shading and night operation, but with the aim of minimizing heat losses during that operation mode.
To summarize: A heat transport problem coupled to a convective flow, consisting of radiative transfer for two different raytracing domains, rigid body heat transport, convection of flue gasses, and thermal fluid-wall interaction. I think that may prove multi-physical enough as a test case ...
Regards,
Peter