Results and visualization - Latest Discussions - COMSOL Forums

Does COMSOL treat the nodes at the boundary also as unknowns?

Tue, 22 May 2012 09:14:12 +0000

Hi Ivar,

But in COMSOL 3.5a, there is no option to clear solution, reset history etc.
The only option available is reset model, in which case everything including mesh will be deleted and the model will be saved as a .mfile.

So I have again created a new model with all the inputs, coefficients, B.C., solver settings (steady-state) and mesh.
But I have not run it as that will increase the file size. Please note that I have already entered the mesh element size and number for each subdomain, but I have not meshed it.
For subdomain 3, use mapped mesh (values already entered).
For subdomains 1 and 2, use free mesh (advancing front, values already entered in the dialogue box).

You just have to click mesh selected button for subdomain 3 in mapped mesh parameters. Click mesh selected separately for subdomains 1 and 2 in free mesh parameters. Then just click the solve button and see the results at z=0(boundary 10) using cross-section plot parameters dialogue box for subdomain 3.

Looking forward for your reply,

Regards,
Seetha

Does COMSOL treat the nodes at the boundary also as unknowns?

Tue, 22 May 2012 01:50:14 +0000

Hi Ivar,

I am using 3.5a.
The file I sent you earlier is .mfile and from 3.5a.
Isn't it not possible to open it in comsol GUI using 'comsol with matlab'?
I am attaching it again here.
I am not sure whether it has information about mesh. So I am mentioning here the mesh I used.

Domain 3: Mapped mesh- Boundary 9 & 12: no. of edge elements = 2000, element ratio =1
Boundary 10&11: no. of edge elements = 36, element ratio =1

Domain 2: Free mesh (advancing front) - Boundary 5: number of edge elements = 500, element growth rate=1
Boundary 6 & 7: element growth rate=1.1
Boundary 8: number of edge elements = 2000, element growth rate=1

Domain 1: Free mesh (advancing front)- Boundary 1: number of edge elements= 50, element growth rate =1
Boundary 2&3: element growth rate =1.2
Boundary 4: number of edge elements= 500, element growth rate =1

If you are not able to open this .mfile, how can I send my model .mph to you?

Regards,
Seetha

Does COMSOL treat the nodes at the boundary also as unknowns?

Mon, 21 May 2012 11:56:16 +0000

Hi Ivar,

Thank you very much for your reply.
I am attaching herewith my COMSOL model as .mfile (as .mph is very large) which is in v3.5a (I wrongly mentioned it as 4.2a)
It will be of great help if you can go through it and help me in this regard.

A short description about my model:
It is about transport of colloid in a tube. Very close to the walls of the tube there is a term containing 'phi' (potential energy) which is equivalent to advection in the direction normal to the flow. This advection varies with the distance from the wall. Due to the very small region in which this advection is acting, I have to use extremely fine mesh near the pipe wall. Hence I divided the pipe into 3 domains having different mesh. In domain 3, there is sharp gradient in the concentration between the boundary node and the node next to that in the flow direction. I use very fine mesh.
If I plot the concentration at the inlet boundary (boundary 10) ( for domain 3) using domain plot parameters, it is showing whatever the boundary condition I have given (Dirichlet, concentration=1) . But when I use cross-section plot parameters dialogue box to specify the line plot and plot the concentration at the inlet boundary for domain 3, it is showing wrong values of boundary condition. For the transient case, the plotted values at the boundary (boundary 10, domain 3 ) changes with time where as the original B.C. is constant. But while exporting the post-processing data also (format- nodes, elements data), it is showing different values at the boundary 10 (actual B.C. is dirichlet). Instead of the B.C. which is equal to one , it is showing a value of 95 at the boundary. The value of the variable inside the domain is 223. So is it because of diffusion or interpolation?

If it is an error only because of interpolation from inside the domain to boundary, can I assume that COMSOL is taking the B.C. which I had specified?

I tried using artificial diffusion, in that case the concentration inside the domain also gets diffused so much.

Also the boundary is the inlet boundary and it is not a common boundary between two domains.
Using boundary layer mesh also gives the same results.

Please help me.

Thanks in advance.

Regards,
Seetha

How to get the derivative of a force with respect to a self defined variable?

Mon, 21 May 2012 00:47:46 +0000

Hi All,

I am working on a simulation of a particle moving in water.

The particle moving velocity is V and its value is 0.01m/s, which we can define as a moving boundary condition.

Then COMSOL can calculate the hydrodynamic force (F) on the particle by integrating the hydrodynamic tensor (for example,T_z_chns and T_r_chns in a 2D axisymmetric geometry).

But, question is how I get the derivative of F with respect to V?

I guess coupled variable in comsol should be used as well as matlab.

Any experts can help me out?

Does COMSOL treat the nodes at the boundary also as unknowns?

Sun, 20 May 2012 05:53:18 +0000

Hi,

Does COMSOL treat the nodes at the boundary also as unknowns?

I am trying to solve colloid transport inside a pipe with constant concentration boundary condition at the inlet. The unknown is concentration. There is sharp gradient in the concentration between the boundary node and the node next to that in the flow direction. I use very fine mesh.
If I plot the concentration at the inlet boundary using domain plot parameters, it is showing whatever the boundary condition I have given (Dirichlet) . But when I use cross-section plot parameters dialogue box to specify the line plot and plot the concentration at the inlet boundary, it is showing wrong values of boundary condition. Also the plotted values at the boundary changes with time where as the original B.C. is constant. In the data exported also the values at the points in the inlet boundary are different from my boundary condition.

So does comsol interpolate from the results to find the value of variable at the boundary if we give the co-ordinates?

Does anybody know why is this problem happening?

Thanks in advance.

Regards,
Seetha

Particle tracing, how to release particles?

Sat, 19 May 2012 22:52:27 +0000

Hello,

I am using the particle tracing module in comsol 4.2 to track charged particles in an electric field in 3D geometry. I basically need to release particles from the top plane of a cube and tract their motion towards the bottom plane in an electric field.

I know how to release multiple particles simultaneously using the range function. If I use X=range(0.3,0.02,0.7) and Y=range(0.3,0.02,0.7) and Z=constant, then I get a line of particles released. (see image what_i_get)

How do I get a 2D grid of particles released simultaneously? (see image what_i_want)

Can someone please help me with this?

Thanks,
Aditya

Fluid Shear Stress in Thin Film Fluid Module

Thu, 17 May 2012 13:49:04 +0000

Hello,

I'm currently working with the journal bearing model (Comsol 4.2) from the model library and I'm trying to calculate the shear stress within the fluid film with is an essential for the torque calculation. Unfortunatly the shear stress is not one of the output values of this module.

I tried the surface integration and using the simple formula:
tau = eta*du/dh
with dh = tffs.dh
and eta = constant value

I used du = d(tffs.uy,y)+d(tffs.ux,x)

Unfortunately the results don't seem logical. I believe my expression for du or my usage of the integration (under results, derived values) is wrong.

Please help!

Thanks,
Toan

Mesh

Wed, 16 May 2012 06:26:13 +0000

Hi. I have a problem with moving mesh too. Please help me with materials. Where I can find something about it?

finding a specified flow rate in a '3D cut plane'

Tue, 15 May 2012 20:08:55 +0000

Hello,

I'm trying to do some postprocessing in a 3D fluid channel flow over a fractured hole.
The channel is filled with water and is laminar.
Here's what I'm trying to do:

1. cut the surface at an arbitrary channel location (i used the 'cut plane')
2. within that surface, I would like to start from one edge and integrate the velocity profile in a prescribed area.
eg) if the surface slice is a 2mmx2mm rectangle, i want to make another rectangle that is 1mmx1mm within that original 'cut plane' and measure the surface integrated velocity value (flow rate).

I was able to cut the surface where I wanted it, but I cannot find a way to resolve what I said in 2.

Any ideas?

Thanks!

TJ

Parametric sweep results

Tue, 15 May 2012 13:51:54 +0000

Hi,

I am launching a parametric sweep over one single parameter P. For each paremeter P, I am launching a modal analysis and get two eigenmodes. Then, I integrate my solution over four subsets of boundaries and get four scalar per eigenmode: int1_mode_i,int1_mode_i,int2_mode_i,int3_mode_i,int4_mode_i with i = 1 or 2 indexes the modes.

Therefore, if P is varying from p1 to p3, I would like to store in a table:

P1 EigenValue_1_P1 int1_mode_1_P1 int2_mode_1_P1 int2_mode_1_P1 int2_mode_1_P1
P1 EigenValue_2_P1 int1_mode_2_P1 int2_mode_2_P1 int2_mode_2_P1 int2_mode_2_P1
P2 EigenValue_1_P2 int1_mode_1_P2 int2_mode_1_P2 int2_mode_1_P2 int2_mode_1_P2
P2 EigenValue_2_P2 int1_mode_2_P2 int2_mode_2_P2 int2_mode_2_P2 int2_mode_2_P2
P3 EigenValue_1_P3 int1_mode_1_P3 int2_mode_1_P3 int2_mode_1_P3 int2_mode_1_P3
P3 EigenValue_2_P3 int1_mode_2_P3 int2_mode_2_P3 int2_mode_2_P3 int2_mode_2_P3

Is there simple way to do that without using the livelink MATLAB? Maybe with accumulated probe table. I am very struggling with this basic request. Any tutorial or example will be very helpful.

For your informations, I usually end up with error messages like: "Tabe input type 'None' does not match data input type 'Eigenfrequency'. Try evaluating in another table."

Thank you all for your answers,

Best Regards,

Denis

Metal-Insulator boundary

Thu, 10 May 2012 12:39:16 +0000

Hello everyone

I have a question about boundary conditions.
I have 3 layers (Insulator-metal-insulator) and this structure is placed in a solution that i want to sense its voltage with that metal layer.
Insulators are SIO2.
The problem is that when i choose dielectric shielding for insulator boundaries, the voltage on the metal, insulator and the solution will be all the same (and i was expected that it should not be the same on insulator layer!)
So i thought may be i should choose ground for insulator (!) and then the problem is solved. I have zero on insulator, different voltages on metal, and in solution.

But i beleive there is something wrong and i shouldn't work like this. Anyone has any idea?

Best
Bahar

Solving model with several ModelNodes

Thu, 10 May 2012 10:19:23 +0000

Hi,
I'm trying to build a model that contains several ModelNodes (sub-models). Each Node should have different properties (same geometry but I want to alter the mesh). The the Model should be solved at once.

My questions are:
How to set dependent variables (since the sub-models should not be coupled)?
How to access the solution for each ModelNode?
Are there some examples available (couldn't find anything about this in the Comsol 4.2a documentation)?

Best regards,
CS

Divergence of gradient of a variable calculation

Thu, 10 May 2012 09:04:52 +0000

Dear Support,
I need to calculate, in the post-processing stage, the divergence of the gradient (it is the laplacian) of a known variable, in a specified volume.
May you help me, please?

Hiding geometry boundaries in graphics window?

Tue, 08 May 2012 12:46:42 +0000

Hi,

I have a model geometry that includes some sub-domains. I want to show the results from these domains, but I don't want their boundaries to be visible in the graphics window, so visually the model looks like only one domian. Does anyone know if and how this can be done?

shear stress in microchannel

Tue, 08 May 2012 02:50:24 +0000

Hello,

I am clueless as to how to start the post-process data analysis to get the spatial distribution of shear stress of a micro channel. I have read the few discussion threads but it involves exporting the slice 3D plot of the velocity and then do the calculation of shear stress in Matlab. I would appreciated it if someone can either tell me (a) what Matlab code or equation you used to solve for shear stress from the velocity profile slice or (b) a way to get Comsol to compute shear stress for me and graph it (a much easier way for a newbie like me).

Thank you so much!

Enquiry on Livelinks for Matlab

Mon, 07 May 2012 11:26:32 +0000

Dear Mr Lars,

I'm using COMSOL 4.2a in my lab, and I have no problem with simple examples in extracting data from COMSOL to matlab.
The thing is, when I try to use a more complicated model, I just cannot get any values of any variable!!!
I posted it on this page: http://www.comsol.dk/community/forums/general/thread/28413/#p77051
The mph and m file is also attached there.
Could you please spend some time to help me out？
I'm looking forward to your reply.
Thank you very much!

Best regards,
Di Kai

No previous solutions after an error - parametric solver

Mon, 07 May 2012 06:20:04 +0000

Hi,

I am doing a quasi-static simulation with displacement BCs using solid mechanics module. I use the extension option in the stationary solver and use the parametric node to control my displacement steps. My model is in 3D and is big.

The problem is when there is an error-for whatever be the reason-the simulation exits and there are no previous solution steps available in the results section. The recovery file is not useful or not present at all in some cases. (I have the option for writing recovery files enabled.)

Irrespective of whether I choose to store solution at the "steps given" or "steps taken by the solver" in the "output" field of the "parametric" node and use the option "store solution out-of-core", I get no previous parametric solutions stored. I do not plot the results on the fly.

What is the way to keep the solutions of the previous parameters?

Regards,
Kodanda

Far field plotting needs a very long time

Fri, 04 May 2012 15:40:25 +0000

Hello,

I'm trying to simulate a antenna over a ground plane with far field condition around, like it is in the examples.

The attached model works fine. But when I try to show the emw.normEfar on a surface, my computer utilize all cores with 100% (HP xw6600, 2xXeonE5450, 16GB). This is fine for some time. But after half an hour I canceled the process, because there is no cancel button at the GUI.

So I have two question:
First: Is there a mistake in my model or preferences?

And second: I lost some models on this way, because the emw.normEfar plot was the last active view when I saved. And COMSOL tries to recalculate, when I open it. Is there a way to prevent the recalculating, when COMSOL gets opened?

Thanks in advance

--
stefan

voltage induction due to changing magnetic field

Tue, 24 May 2011 12:14:58 +0000

hello,

I'm a relative newbie to comsol. what I'm trying to do is (I think) relatively easy, but I have no idea how to tackle it. what I need to do is model voltage induction in a coil due to a changing magnetic field. I know how to how to create a coil in a fixed magnetic field, but I don't know how to set up time varying field. but, more importantly, I don't know how to measure the induced voltage. should the coil be open and then specify two points where the voltage would be measured? thanks for any suggestions.

r.

How to delete unwanted stored solutions?

Thu, 30 Sep 2010 11:50:35 +0000

Hello,
I am simulating large 3D RF problem in time domain in version3.5a. And I found my whole stored file is as large as 40GB, since I stored around 600 time steps. And I want to delete the last 100 time steps to save some memory of the server without recalculate the whole process.

I would be very happy if someone knows how to do it.

Best wishes,