Discussion Forum

Posted: 1 decade ago 14 mars 2012, 21:48 UTC−4

I have used this material which was very useful: http://www.csc.kth.se/utbildning/kth/kurser/DN1240/numfcl08/labbar/IntroPDEFEMEng.pdf I hope the example in the above file illustrate your issue. all the best

Posted: 1 decade ago 15 mars 2012, 00:21 UTC−4

Hi, Thank you for response. But i think that this is for one PDE and don't talk about how to couple two PDES with two unknowns?

Posted: 1 decade ago 15 mars 2012, 20:40 UTC−4

The user guide said that u=(u1,u2) . So it will include 2 eqns. My question is how to define u=(u1,u2) Your generous help is appreciated. Thanks.

Posted: 1 decade ago 16 mars 2012, 05:04 UTC−4

Hello, Liang: The answer depends on what the equations are and how the two dependent variables are mathematically coupled. I think that, possibly, you could use "Model" / "Add Physics" / "PDE interfaces". Usually, the "Coefficient Form PDE" will suffice, if you can express your equation in that form. The best would be if you could express your two scalar equations in only 1 vector equation with two dependent variables {u1, u2} (for instance, they are components of the same vector field). In that case, in "PDE (c)", at bottom, in "Dependent Variables", you could change the name of the dependent variable to "u1", and add another one: "u2". Or else, you could choose two scalar PDE equations, each one in one dependent variable. The coupling would be done simply using the names of both dependent variables in both equations. If you give us the equations, perhaps the help could be more explicit. About the dimensions, if the problem reduces to 1D, it will be simpler and faster in 1D. Good luck.

Posted: 1 decade ago 16 mars 2012, 18:30 UTC−4

Thank you very much, Sir. This is helpful information. I am very frustrated by void information provided by the current PDE tutorial materials in user guide and all the available video about PDE in comsol I have a question. I 've tried many time to find options to change the name of dependent variables, but couldn't find. My license is 4.0 classkit license. THis feature is blocked? Could you show me the place you change the dependent variable. Thanks

Posted: 1 decade ago 21 mars 2012, 05:03 UTC−4

Hello, Liang: I am now with version 4.2a, but I think it was similar in 4.0. In Model * / Physics (in this case PDE), in the "settings tab" (central window), below, you will see "Dependent Variables". There you could change the names of the variables. If you want, you may write the equations and boundary conditions, and we will try to help you more explicitly. Bye. Jesus

Posted: 1 decade ago 24 mai 2012, 01:26 UTC−4

hello Jesus Lucio , I also have the same problem and my equations are u*∂u/∂x + v*∂u/∂y - (1/re)*(∂2u/∂x2 + ∂2u/∂y2) u*∂v/∂x + v*∂v/∂y - (1/re)*(∂2v/∂x2 + ∂2v/∂y2)

Posted: 1 decade ago 13 nov. 2012, 07:25 UTC−5

Hello Mr. Lucio, I made a similar implementation to the one you described above. I have a solid block which is the domain for a solid mechanics physics with variables u,v,w and i have an edge inside the block which is modelled by the beams physics with variables u,v,w,thx,thy,thz. I especially wanted the u,v,w of the sm physics to have the same name as the u,v,w of the beam. If I understood your suggestion this would suffice for coupling the beam and the sm variables. But looking at my results it looks like the coupling doesnt work. I would really appreciate a little comment whether i implemented the model right. Thank you very much kind regards Jan

Posted: 1 decade ago 13 nov. 2012, 09:38 UTC−5

Hi, Jan, I don't know the beam physics, and really I do not understand how you can solve simultaneously both the solid mechanics ph. in the block and the beam ph. in an edge, as I suppose they are not independent. But if the problem is correct as you enunciate it, you could access the block displacement simply by its variables names (u, v, w), and you could define a "prescribe displacement" condition for the beam physics in the edge, setting some of the displacement (let them be u2, v2, w2) simply equal to one of the corresponding solid displacements (for instance, u2 = u). But, sorry, I don't see how you can set a special physics for an edge of the block. If you could send the model, we will be able to help you better. Jesus.

Posted: 1 decade ago 13 nov. 2012, 14:48 UTC−5

Hi the knitting of the 3 DOF of SOLID to the 6 DOF of Shell or beams is delicate, you need also to link the rotation dependent variables. There are different ways, if you forgetyou will get a "perfect hinge" this is often called a "mechanism" in FEM jargon, and does not behaves as expected (in most cases). Then a shell or a beam connecs to a line in a Solid, this makes often singularities, so you should ideally knit togeher a boundary of a Solid to a boundary of a shell or a beam, and as these have at elast 1 DoF less than the Solid this requires some thoughts. There was a good but rapid explaantion aot this during the last COMSOL "solid" mini-course in Milan, proably the same were given at the other conferences too. Its rather model dependent -- Good luck Ivar

Posted: 1 decade ago 19 avr. 2013, 02:28 UTC−4

[QUOTE] Hello, Liang: The answer depends on what the equations are and how the two dependent variables are mathematically coupled. I think that, possibly, you could use "Model" / "Add Physics" / "PDE interfaces". Usually, the "Coefficient Form PDE" will suffice, if you can express your equation in that form. The best would be if you could express your two scalar equations in only 1 vector equation with two dependent variables {u1, u2} (for instance, they are components of the same vector field). In that case, in "PDE (c)", at bottom, in "Dependent Variables", you could change the name of the dependent variable to "u1", and add another one: "u2". [/QUOTE] Hi everyone, This is my first time with Comsol but I'm already working on a similar problem. I have to solve the thermoelectric problem, so I have two coupled equations and my u vector became u=(T,V). In the first equation the "c" coefficient in the "coefficient form PDE" is: lambda + (sigma)*((alfa)^2)*T, when u=T; (sigma)*(alfa)*(T), when u=V. In the second equation the "c" coefficient in the "coefficient form PDE" is: (sigma)*(alfa), when u=T; (sigma), when u=V. Suppose now I want to impose thermal isulation [n• (lambda*∇T) = 0] on δΩ using Neumann boundary condition so I set q=g=0 (a and Y were set to 0 on Ω ). But how I can set c=lambda on δΩ so that thermal insulation holds? Obviously a similar problem arises if i want to impose electric insulation. Thanks a lot, dario

Posted: 1 decade ago 20 août 2014, 04:21 UTC−4

hello jesus!! i look forward to solve two coupled equations with variable coefficients. how can i proceed please guide.

Posted: 1 decade ago 22 août 2014, 14:49 UTC−4

Hello I have the same problem, but I have five equations and five dependent variables, I couldn't solve it. -- Omar S. Lateef

Posted: 10 years ago 12 déc. 2014, 07:46 UTC−5

dear all, i have some problems with PDE, i have 5 dependant variables, u1,u2,........and u5, and i have some questions, can i defined in same equation- i mean the general PDE- or can i use the other way, such as defined five equations -- Omar S. Lateef

Posted: 10 years ago 12 déc. 2014, 07:58 UTC−5

Hello, You can do any of both things: or define five scalar (i.e., each one with only one dependent variable) PDEs, or define only one (vector) PDE (with all your five dependent variables as components of the unique vector dependent variable). If the equations are strongly coupled, I recommend the second option. But surely the first option is quite simpler to write. If you use option one, just take into account the names of those variables, usually u1, ..., u5. Regards, Jesus.

Posted: 10 years ago 16 févr. 2015, 03:33 UTC−5

Dear all, Can i solve coupled PDEs of Two Temperature Model equation for ultrashort pulse laser heating ? How can i coupled these two equations ? Ce*(d/dt)*Te= d/dz(k*(d/dz)*Te)- G(Te-Tl)+S(z,t) Cl*(d/dt)*Ti=G(Te-Ti)

Posted: 9 years ago 12 août 2015, 08:24 UTC−4

Hello Ivar my problem is a little different, but I would be grateful if you could help me. I am defining my system of equation using Coefficient Form PDE . and as you may find from the attached photo, I have 3 dependent variables: n and p (ion species), E (electric field) which are functions of x (position). could you please guide me through building these equations? I am confused with the terms, since for each equation the diffusion term only contains either the variable p or n , but the convection term contains also the variable E. Regards

Attachments:

• equations.png

Posted: 9 years ago 23 mars 2016, 06:37 UTC−4

Hello, please my question is how I can coupled two equations PDEs , for example (schrodinger-poisson) as the 1 st depends on two variable u1 and V . or V is a variable in the 2nd equation (poisson) thank you.