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.

Solving nonlinear PDEs with higher-order gradients using COMSOL

Please login with a confirmed email address before reporting spam

Hi,

I'd like to use COMSOL to solve for the nonconserved dynamics of the phase field crystal (PFC) model:

du/dt = - ( \epsilon + ( 1 + \nabla^2 )^2 ) u - \tau u/2 - u/3,

where u = u(t,x,y) is the dependent variable, \epsilon and \tau are constants and \nabla^2 is the Laplacian.

Using the coefficient form for PDEs the time derivative and the linear term are easy, the second- and third-order terms I can specify in the source term and there's also a coefficient for a term with the Laplacian.

But the PFC model incorporates also fourth-order gradient terms:

d^4 u / dx^4
d^4 u / dy^4
d^4 u / ( dx^2 dy^2 ) (if I'm not mistaken...)

How can I implement these in COMSOL? It doesn't recognize variables and functions such as uxxxx, uxxyy, grad(), div(), del(), nabla() or Laplacian(). I googled extensively, watched tutorials and read documentation, but haven't yet figured this out.

My primary motivation is to be able to solve the model on curved surfaces (I really don't need COMSOL for solving flat systems, but I guess I have to figure out how to use COMSOL for that first), so tips regarding gradients in this context are also welcome.

A big thanks!

Edit: I think that I'll also have to set the shape functions of the elements to fourth order at least for COMSOL to be able to evaluate the fourth-order gradients, right?

5 Replies Last Post 10 août 2017, 15:48 UTC−4
Jeff Hiller COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 7 years ago 10 août 2017, 09:22 UTC−4
Hello Petri,
The Support Knowledge Base has an entry on how to solve a PDE with space derivatives of order higher than two. Here is the link: www.comsol.com/support/knowledgebase/816/ .
On the topic of solving custom equations on curved surfaces, that is often best achieved by using tangential derivatives. An example model is here: www.comsol.com/model/shell-diffusion-in-a-tank-222
Best,
Jeff
Hello Petri, The Support Knowledge Base has an entry on how to solve a PDE with space derivatives of order higher than two. Here is the link: https://www.comsol.com/support/knowledgebase/816/ . On the topic of solving custom equations on curved surfaces, that is often best achieved by using tangential derivatives. An example model is here: https://www.comsol.com/model/shell-diffusion-in-a-tank-222 Best, Jeff

Please login with a confirmed email address before reporting spam

Posted: 7 years ago 10 août 2017, 09:52 UTC−4
Thanks for the reply!

It appears my account is lacking a valid licence (I run COMSOL on my work computer at the university, but I guess I have no personal licence) so I can't access your first link right now. I'll figure out if it's possible to connect my university's licence to my account next week when I'm back at the office. I'll report back if I was successful or not (with the licence and in implementing the model in COMSOL).
Thanks for the reply! It appears my account is lacking a valid licence (I run COMSOL on my work computer at the university, but I guess I have no personal licence) so I can't access your first link right now. I'll figure out if it's possible to connect my university's licence to my account next week when I'm back at the office. I'll report back if I was successful or not (with the licence and in implementing the model in COMSOL).

Please login with a confirmed email address before reporting spam

Posted: 7 years ago 10 août 2017, 14:03 UTC−4
Okay, I think I solved it. I did some googling based on your comments for the first link and figured out that I had to reformulate the 4th-order PDE as two 2nd-order PDEs. I did that and finally everything seems to be working smoothly. Thanks again!
Okay, I think I solved it. I did some googling based on your comments for the first link and figured out that I had to reformulate the 4th-order PDE as two 2nd-order PDEs. I did that and finally everything seems to be working smoothly. Thanks again!

Jeff Hiller COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 7 years ago 10 août 2017, 14:37 UTC−4
Glad I could help, Petri.
Glad I could help, Petri.

Please login with a confirmed email address before reporting spam

Posted: 7 years ago 10 août 2017, 15:48 UTC−4
In case someone else ever wants to do something similar, I could add it seems that quadratic elements work fine for qualitative results at least - haven't tried doing any quantitative analysis of the solutions yet.
In case someone else ever wants to do something similar, I could add it seems that quadratic elements work fine for qualitative results at least - haven't tried doing any quantitative analysis of the solutions yet.

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.