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.

problem with convection in dilute species

Please login with a confirmed email address before reporting spam

Hi,

I have a time dependent cylindrical symmetric model with dilute species. The diffusion of the species c is always limited to half the outer concentration whenever I implement a convection therm. Only with diffusing the profile looks like expected.

When reproducing exactly the same calculation in 1D the concentration c can increase all the way to the outer concentration as expected.

I tried it in Version 4.3b and 5.

Would be great if anyone has an idea how to understand this.

Thx
Tobi


12 Replies Last Post Jul 2, 2015, 1:44 a.m. EDT

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 25, 2015, 5:30 a.m. EDT
First, please define the units of your variables, it makes understanding easier. Second, t is a reserved variable for time in Comsol. Your time stepping is "range(0,t,t)", t defined as 600[s]. I changed that to "range(0,tf/50,tf)" with tf as 600[s].

What I see that concentration increases from the bottom as time passes. So, what is exactly the issue you wished to raise?

br
Lasse
First, please define the units of your variables, it makes understanding easier. Second, t is a reserved variable for time in Comsol. Your time stepping is "range(0,t,t)", t defined as 600[s]. I changed that to "range(0,tf/50,tf)" with tf as 600[s]. What I see that concentration increases from the bottom as time passes. So, what is exactly the issue you wished to raise? br Lasse

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 25, 2015, 8:57 p.m. EDT
Dear Lasse,

thank you very much for your reply.

I removed the unnecessary parameters and added units to the surface exchange and diffusion coefficient to make it easier to read.

Now the surface exchange coefficient is increased by one order of magnitude to see it more drastically. The concentration within the model only goes up to exactly half of the outer concentration (“TracerKonz”/2 = 0.971/2=0.4855). As I mentioned, the 1D model gives the correct concentration within the model of 0.971.
This problem appears only when adding the velocity field term “u” in convection and diffusion. If you set this term to 0 the concentration goes up to 0.971 as it should.

If you look into the “1D Plot Group 3” you see that the concentration only goes up to 0.4855 and not to 0.971.

Thanks again for helping!
Tobi
Dear Lasse, thank you very much for your reply. I removed the unnecessary parameters and added units to the surface exchange and diffusion coefficient to make it easier to read. Now the surface exchange coefficient is increased by one order of magnitude to see it more drastically. The concentration within the model only goes up to exactly half of the outer concentration (“TracerKonz”/2 = 0.971/2=0.4855). As I mentioned, the 1D model gives the correct concentration within the model of 0.971. This problem appears only when adding the velocity field term “u” in convection and diffusion. If you set this term to 0 the concentration goes up to 0.971 as it should. If you look into the “1D Plot Group 3” you see that the concentration only goes up to 0.4855 and not to 0.971. Thanks again for helping! Tobi


Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 26, 2015, 6:10 a.m. EDT
What are the units of TracerKonz and Nat?
What are the units of TracerKonz and Nat?

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 26, 2015, 6:16 a.m. EDT
And you have convective flow from the bottom, but where is its outflow? What the flow is? Water, air?

Is the upper part of the capillary diffusion in solid and the lower part diffusion in gas? In gas Fick's law cannot be used.
And you have convective flow from the bottom, but where is its outflow? What the flow is? Water, air? Is the upper part of the capillary diffusion in solid and the lower part diffusion in gas? In gas Fick's law cannot be used.

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 28, 2015, 8:12 p.m. EDT
TracerKonz and Nat are both fractions. TracerKonz is the tracer concentration of the Oxygen 17 (mass 17) isotope (97,1 %) we use for our experiments. Nat is the natural abundance of oxygen 17 in the atmosphere (0.0205%).
The model is tracer diffusion in solids. The lower part represents a Lanthanum strontium manganite (LSM) thin film. The upper part is Yttria Stabilized Zirconia (YSZ). I use this model because it is solving the Laplace equation I need for this problem.
The model I posted is a small part of my data fit routine I use for data evaluation of isotope depth profiles. Everything works fine without activating the convection therm. For really simple geometries, the simulated data can also be fitted by Crank’s solutions of Ficks Laws.
The physics is correct for diffusion without the convection therm. If I activate the convection term in a 1D model the results looks like expected with a surface near concentration up to outside concentration (TracerKonz). However if I activate the convection the concentration inside the model is limited to exactly 1/2 of the outside concentration.
The convection here represents a bias enhanced term of the tracer diffusion and is no a gas diffusion. It is simply using the same partial differential equation. The derivation for this kind of problem can be found here (equation 24): pubs.rsc.org/en/content/articlehtml/2009/cp/b822415c
I don’t understand why it is working on 1D but not in this geometry.
TracerKonz and Nat are both fractions. TracerKonz is the tracer concentration of the Oxygen 17 (mass 17) isotope (97,1 %) we use for our experiments. Nat is the natural abundance of oxygen 17 in the atmosphere (0.0205%). The model is tracer diffusion in solids. The lower part represents a Lanthanum strontium manganite (LSM) thin film. The upper part is Yttria Stabilized Zirconia (YSZ). I use this model because it is solving the Laplace equation I need for this problem. The model I posted is a small part of my data fit routine I use for data evaluation of isotope depth profiles. Everything works fine without activating the convection therm. For really simple geometries, the simulated data can also be fitted by Crank’s solutions of Ficks Laws. The physics is correct for diffusion without the convection therm. If I activate the convection term in a 1D model the results looks like expected with a surface near concentration up to outside concentration (TracerKonz). However if I activate the convection the concentration inside the model is limited to exactly 1/2 of the outside concentration. The convection here represents a bias enhanced term of the tracer diffusion and is no a gas diffusion. It is simply using the same partial differential equation. The derivation for this kind of problem can be found here (equation 24): http://pubs.rsc.org/en/content/articlehtml/2009/cp/b822415c I don’t understand why it is working on 1D but not in this geometry.

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 29, 2015, 2:57 a.m. EDT
Hi

I think I got it.

When you deactivate the convection term, you still have the boundary flux boundary condition activated, i.e. Flux = k_k*(TracerKonz-c) which brings material to the lower end of the capillary. Yet, it is not transferred further in the capillary by convection but diffusion only, which makes the concentration to build up close to the boundary.

When you activate the convection term, it carries material further in the capillary and the build up of concentration at the boundary does not take place. If I put k_k to zero, then hardly anything appears in the capillary, but the initial Nat remains all over.

There must be some other difference between your 1D and 2D models, otherwise I cannot understand it.

BR
Lasse
Hi I think I got it. When you deactivate the convection term, you still have the boundary flux boundary condition activated, i.e. Flux = k_k*(TracerKonz-c) which brings material to the lower end of the capillary. Yet, it is not transferred further in the capillary by convection but diffusion only, which makes the concentration to build up close to the boundary. When you activate the convection term, it carries material further in the capillary and the build up of concentration at the boundary does not take place. If I put k_k to zero, then hardly anything appears in the capillary, but the initial Nat remains all over. There must be some other difference between your 1D and 2D models, otherwise I cannot understand it. BR Lasse

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 29, 2015, 3:22 a.m. EDT
May I also add that with convection activated the total amount of c in the lower domain is 6.18e-23 mol and with convection deactivated 2.22e-24 mol, hence 96% less. This makes perfect sense.
May I also add that with convection activated the total amount of c in the lower domain is 6.18e-23 mol and with convection deactivated 2.22e-24 mol, hence 96% less. This makes perfect sense.

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 29, 2015, 3:37 a.m. EDT
k_k is the surface exchange therm. if this is 0 nothing enters to the material.
if convection is activated the concentration should build up at the surface exactly the same like without convection. Only the transport should be faster. I don’t want to change the incorporation...

I did exactly the same for the 1D and this model and the result is different. Does this mean there is a bug in comsol?

I attached the plots for 1D and 2D and 1D without convection...
k_k is the surface exchange therm. if this is 0 nothing enters to the material. if convection is activated the concentration should build up at the surface exactly the same like without convection. Only the transport should be faster. I don’t want to change the incorporation... I did exactly the same for the 1D and this model and the result is different. Does this mean there is a bug in comsol? I attached the plots for 1D and 2D and 1D without convection...


Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 29, 2015, 9:10 p.m. EDT
Is there anyone who can help me with this problem?
Is there anyone who can help me with this problem?

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 30, 2015, 2:04 a.m. EDT


if convection is activated the concentration should build up at the surface exactly the same like without convection. Only the transport should be faster. I don’t want to change the incorporation...


I must disagree because convection carries away the material which has passed across the boundary. Why this does not happen in the 1D model, beats me.

br
Lasse
[QUOTE] if convection is activated the concentration should build up at the surface exactly the same like without convection. Only the transport should be faster. I don’t want to change the incorporation... [/QUOTE] I must disagree because convection carries away the material which has passed across the boundary. Why this does not happen in the 1D model, beats me. br Lasse

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jun 30, 2015, 3:30 a.m. EDT
There question cannot be answered so easy. Surface concentration is always a result of the proportion between surface exchange and diffusion.

If the surface is limiting the profile is manly determine by k_k. But when we decrease the diffusion coefficient D_k, at some point the diffusion well be limiting. Than the concentration in the vicinity of the surface should increase because tracer is incorporated but cannot go into the material.
The strange thing in the 2D model is, that irrespectively of the diffusion coefficient D_k the surface concentration is limited to exactly half of the outside concentration.

- First exactly half of the outside concentration looks like an arbitrary value.
- And second, in case of a super-fast surface exchange and super-slow diffusion what would be the physical explanation that the surface concentration is not rising, even the transport in the material is so slow?
There question cannot be answered so easy. Surface concentration is always a result of the proportion between surface exchange and diffusion. If the surface is limiting the profile is manly determine by k_k. But when we decrease the diffusion coefficient D_k, at some point the diffusion well be limiting. Than the concentration in the vicinity of the surface should increase because tracer is incorporated but cannot go into the material. The strange thing in the 2D model is, that irrespectively of the diffusion coefficient D_k the surface concentration is limited to exactly half of the outside concentration. - First exactly half of the outside concentration looks like an arbitrary value. - And second, in case of a super-fast surface exchange and super-slow diffusion what would be the physical explanation that the surface concentration is not rising, even the transport in the material is so slow?

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jul 2, 2015, 1:44 a.m. EDT
Nobody any idea how to explain this behavior?
Nobody any idea how to explain this behavior?

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.