Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Heating a Block of Water

Please login with a confirmed email address before reporting spam

Hi,

I'm ultimately trying to heat a gold sphere submerged in a sphere of water. I simply create two spheres and apply a boundary heat source to the inner boundaries. However I believe my results are incorrect because the water sphere does not heat up at all and the temperature of the gold sphere does not change much.

So now I am practicing heating a gold block and a water block individually to see what is going on. The gold block heats up with a seemingly correct temperature distribution when a boundary heat source is applied.

When I change the material to water, however. I get a funky temperature distribution which I believe to be incorrect. All the temperature is concentrated at the boundary and it looks like a very random distribution. Is there something extra I must do to heat fluids?

I am having trouble uploading the model. It is telling me "file size error". The file is 301 kB.


10 Replies Last Post 2011年4月12日 GMT-4 11:03

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月4日 GMT-4 01:43
I just tried changing the heating time to a very large value 100000s and got a better temperature distribution.

To give some values. The block of water is 1 m^3 and the heat flux I applied was 5000W/m^2.

I understand water takes a lot of energy to change 1 K so I figure that the errors from before were due to the very small temperature changes. (can i get confirmation on this conclusion?)

Would refining the mesh give Comsol a better temperature resolution? For 10 second simulation, I obtain random "spots" of high temperature mixed with spots of low temperature.
I just tried changing the heating time to a very large value 100000s and got a better temperature distribution. To give some values. The block of water is 1 m^3 and the heat flux I applied was 5000W/m^2. I understand water takes a lot of energy to change 1 K so I figure that the errors from before were due to the very small temperature changes. (can i get confirmation on this conclusion?) Would refining the mesh give Comsol a better temperature resolution? For 10 second simulation, I obtain random "spots" of high temperature mixed with spots of low temperature.

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月4日 GMT-4 04:54
Hi

I assume you can estimate this rather simply from the heat capacity of water: about 4.18 kJ/kg/K hence you need 4.18MJ per degree for your 1m^2 water block, no?

with the 6m^2 of lateral faces and your 5 kW/m^2 you are entering 30 kJ/s of power which means you need some 4180/30 seconds per deg temperature rise (assuming immediate heat conduction to homogenize your medium, what is certainly neither not the case). So indeed you need several minutes to heat significantly up your m^3 of water

Perhaps you have noticed that your electric fuses for the water boiler heats up nicely after you have taken a bath, because heating water is consuming a loooot of energy ;)

--
Good luck
Ivar
Hi I assume you can estimate this rather simply from the heat capacity of water: about 4.18 kJ/kg/K hence you need 4.18MJ per degree for your 1m^2 water block, no? with the 6m^2 of lateral faces and your 5 kW/m^2 you are entering 30 kJ/s of power which means you need some 4180/30 seconds per deg temperature rise (assuming immediate heat conduction to homogenize your medium, what is certainly neither not the case). So indeed you need several minutes to heat significantly up your m^3 of water Perhaps you have noticed that your electric fuses for the water boiler heats up nicely after you have taken a bath, because heating water is consuming a loooot of energy ;) -- Good luck Ivar

Nagi Elabbasi Facebook Reality Labs

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月4日 GMT-4 08:30
There are two factors causing this temperature distribution in the water. The first is that water has high heat capacity as Ivar mentioned. The second is that water has a low thermal conductivity. What that means is that it takes a lot of energy to heat water, and due to the low conductivity it is hard to maintain the high heat flux required to cause that temperature rise. Thermal diffusivity (ratio of thermal conductivity to specific heat capacity) is the parameter that affects this problem, and gold has a value that is about 1000 times bigger than that of water.

That explains why using much larger heating times works. Also, with the short heating time, the effective heated depth of water is so small that it is probably within the mesh resolution. That gives unreliable results and explains the strange temperature distribution you are getting. What you need is a boundary layer mesh with fine elements close to the boundaries. That is not enough however. Heat transfer to water will not be only due to conduction. Natural convection will, in general, be a much more significant source of heat transfer. For that you need a coupled CFD-heat transfer analysis and either fully account for the effect of temperature on density or use the Boussinesq approximation. COMSOL has examples of both types.

Nagi Elabbasi
Veryst Engineering
There are two factors causing this temperature distribution in the water. The first is that water has high heat capacity as Ivar mentioned. The second is that water has a low thermal conductivity. What that means is that it takes a lot of energy to heat water, and due to the low conductivity it is hard to maintain the high heat flux required to cause that temperature rise. Thermal diffusivity (ratio of thermal conductivity to specific heat capacity) is the parameter that affects this problem, and gold has a value that is about 1000 times bigger than that of water. That explains why using much larger heating times works. Also, with the short heating time, the effective heated depth of water is so small that it is probably within the mesh resolution. That gives unreliable results and explains the strange temperature distribution you are getting. What you need is a boundary layer mesh with fine elements close to the boundaries. That is not enough however. Heat transfer to water will not be only due to conduction. Natural convection will, in general, be a much more significant source of heat transfer. For that you need a coupled CFD-heat transfer analysis and either fully account for the effect of temperature on density or use the Boussinesq approximation. COMSOL has examples of both types. Nagi Elabbasi Veryst Engineering

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月10日 GMT-4 01:03
Hi,

Thanks a lot for your input, they really pushed me forward in my simulation progress. Unfortunately, I think I may have hit another bump in the road. I am currently trying to model a

21-nm-diameter gold particle in water
irradiated by a 6-nsec 0.13-J/cm2 laser pulse at 532 nm.
The laser has a 4 mm radius and the pulse is Gaussian.
The Gold particle has absorbance of k=4.01

My model consists of two concentric spheres. The outer sphere is water and the inner sphere is a gold particle.

I made the gold sphere boundaries the heat source and given the information above, I calculated an average incident intensity over the 6 ns to be

24.2 TW/m^2

I calculated the average value of the gaussian pulse over two standard deviations (95%) instead of multiplying the pulse by the laser intensity because I was having issues with the units when I tried

gp1*Intensity[TW/m^2]

Initial Temperatures: 293.15[K]
Surface radiosity: 0 (not sure what value to put for this one)

These are the only boundary conditions I have.

I ran the simulation and obtained results in the 350,000 K region as you can see in the picture I provided. The values obtained experimentally were 1500 K so I am way off.

Am I still missing something in my model? I have not yet added the convection process but would this be enough to drop the temperature by 2 orders of magnitude?

Sorry for asking such a large question. Heat transfer problems are not my specialty but this is a class assignment.
Hi, Thanks a lot for your input, they really pushed me forward in my simulation progress. Unfortunately, I think I may have hit another bump in the road. I am currently trying to model a 21-nm-diameter gold particle in water irradiated by a 6-nsec 0.13-J/cm2 laser pulse at 532 nm. The laser has a 4 mm radius and the pulse is Gaussian. The Gold particle has absorbance of k=4.01 My model consists of two concentric spheres. The outer sphere is water and the inner sphere is a gold particle. I made the gold sphere boundaries the heat source and given the information above, I calculated an average incident intensity over the 6 ns to be 24.2 TW/m^2 I calculated the average value of the gaussian pulse over two standard deviations (95%) instead of multiplying the pulse by the laser intensity because I was having issues with the units when I tried gp1*Intensity[TW/m^2] Initial Temperatures: 293.15[K] Surface radiosity: 0 (not sure what value to put for this one) These are the only boundary conditions I have. I ran the simulation and obtained results in the 350,000 K region as you can see in the picture I provided. The values obtained experimentally were 1500 K so I am way off. Am I still missing something in my model? I have not yet added the convection process but would this be enough to drop the temperature by 2 orders of magnitude? Sorry for asking such a large question. Heat transfer problems are not my specialty but this is a class assignment.


Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月10日 GMT-4 05:56
Hi

the radiosity is for radiative exchange, normally at a few hundred degrees of temperature difference the radiative emission of a body is relevant, so you might need to consider it, at least as an exercise. Because I'm not sure for your small sizes the scaling law would make radiative exchange so important. To be checked. Use the blackbody Stephan Boltzman law, i.e. as described here

en.wikipedia.org/wiki/Stefan–Boltzmann_law

Do not forget that it uses the difference in temperature in Kelvin ;) and with a T^4 relation it is very non linear

--
Good luck
Ivar
Hi the radiosity is for radiative exchange, normally at a few hundred degrees of temperature difference the radiative emission of a body is relevant, so you might need to consider it, at least as an exercise. Because I'm not sure for your small sizes the scaling law would make radiative exchange so important. To be checked. Use the blackbody Stephan Boltzman law, i.e. as described here http://en.wikipedia.org/wiki/Stefan–Boltzmann_law Do not forget that it uses the difference in temperature in Kelvin ;) and with a T^4 relation it is very non linear -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月10日 GMT-4 13:53
Once again, thanks for your input.

When I tried to add in BB Radiation, I got the following error message:

Failed to evaluate variable Jacobian.
- Variable: mat.epsilon rad
- Geometry: 1
- Boundary: 5 6 7 8 11 12 14 15

These are the boundaries of my gold sphere. From what I can interpret there's a missing material property but I filled in Surface Emissivity for gold already and added it and I still got the same error.
Once again, thanks for your input. When I tried to add in BB Radiation, I got the following error message: Failed to evaluate variable Jacobian. - Variable: mat.epsilon rad - Geometry: 1 - Boundary: 5 6 7 8 11 12 14 15 These are the boundaries of my gold sphere. From what I can interpret there's a missing material property but I filled in Surface Emissivity for gold already and added it and I still got the same error.

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月10日 GMT-4 17:44
Hi

it could be a version issue, try it on a newer v4.1 latest patch, it should work OK

--
Good luck
Ivar
Hi it could be a version issue, try it on a newer v4.1 latest patch, it should work OK -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月10日 GMT-4 18:19
Hi,

I was able to fix that issue by just changing to "user-defined" emissivity rather than changing the actual material property.

The drop in temperature was significant, it dropped by 2 fold. Now my max temperature is 1.4 x10^5. Still two orders or magnitude too high.

I will now try to add in convective cooling and see how that goes.

Thanks.
Hi, I was able to fix that issue by just changing to "user-defined" emissivity rather than changing the actual material property. The drop in temperature was significant, it dropped by 2 fold. Now my max temperature is 1.4 x10^5. Still two orders or magnitude too high. I will now try to add in convective cooling and see how that goes. Thanks.

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月11日 GMT-4 01:42
Hi

convection is normally a few W/m^2/K I'm not sure that will give you a big difference, because of your small scale, but always worth to try. Often its also useful to do some hand calculations to get the orders of magnitude

--
Good luck
Ivar
Hi convection is normally a few W/m^2/K I'm not sure that will give you a big difference, because of your small scale, but always worth to try. Often its also useful to do some hand calculations to get the orders of magnitude -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月12日 GMT-4 11:03
Hi,

Thanks a lot for your input. I was able to get the model very close to experimental data. You are right, adding in convection barely changes the temperature distributions.

It turns out I had made an error in the units, causing my input intensity to be 7 magnitudes off.

Still, your inputs were very informative and helped me learn a lot about the heat transfer process and what conditions I should be aware of.

Thanks
Hi, Thanks a lot for your input. I was able to get the model very close to experimental data. You are right, adding in convection barely changes the temperature distributions. It turns out I had made an error in the units, causing my input intensity to be 7 magnitudes off. Still, your inputs were very informative and helped me learn a lot about the heat transfer process and what conditions I should be aware of. Thanks

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.