Quantcast
Channel: Solving, solving\ - Latest Discussions - COMSOL Forums
Viewing all 3988 articles
Browse latest View live
↧

Free Convection - General Modeling

June 5, 2014, 1:49 am
$
0
0
Hi,

I'm new to Comsol and i'm trying to model free convection in a prety large domain.
I've realised the model in 2D and it has about 1 square meter and in 3D it has about 0.5 cubic meters.
I'm using 3.5a version and i have found a 2D example of free convection in it's library , example in which the dimension of the domain is a lot smaller by the order of centimeters thing that allows the mesh to be very small about 1e-4 on the heated plate. The example works perfect but when i tried to model my chase it had problems in realising the mesh (out of memory) so i had to build the model ten times smaller using the scale command. The problem is that the results are diferent from reality because of the scale.
My question is if is there a way to model your case at a smaller scale and somehow to obtain results for the real size. I'm thinking at a command or some option to specify that the model is at a certain scale.
And another thing, can you verify the volume force that simulates the force due to changes in density that is (rho0-rho_htgh)*g where rho0 is the density of air at 20 degC which is the initial temperature which gradualy increases (parametric solver).
The example which i was talking about is the heated vertical plate free convection from comsol team.
Another thing is that the heated plate in my case is at an angle (is not vertical like in the model)
The reason why i'm asking if the volume force relation is valid is because using the same conditons i rotated the domain so that the heated plate is horizontal and the air is flowing from left to right in stead of flow from the heated plate up (from down to up).


If someone has any ideea of how i could solve my model please help.
Any information is good. This model bugs me for some time.
Thanks for reading and sorry for all the long writing


All the best,
Viorel
↧
Search

Coupling Solid Mechanics and Optics

June 5, 2014, 9:25 am
$
0
0
Hello,
I am trying to analyse some optical data that changes when strained. I want to deteremine the values for specific strain which means that I need to do a parametric sweep of strain. However, I don't know how to set up the two physics (solid mechanics and optics). I defined the strain by calling a prescribed displacement in the solid mechanics physics. Then I run a frequency-domain study, but its does not work.

PLease help,

Have a great day,

Phil
↧
↧

eigenvalue shift with additional point, no mass assigned

June 5, 2014, 10:30 am
$
0
0
Hi, I've met some trouble when calculating the eigen-frequencies of notched plate. I've compared three cases, the first one is a simply notched plate, the second one is a notched plate with one more point in the bottom of the plate for future point load, the third cases is to calculate the dynamic response of a notched plate.

My current problem is the eigenvalues of the first two cases are quite different, i.e., simple notched plate and the notched plate with one additional point, I don't know what the difference comes from, is it due to the mesh? or it's due to the additional mass of this point(I don't assign additional mass to this point)? or due to some default damping that I'm ignoring?
↧

application of load on nodes

June 6, 2014, 12:54 am
$
0
0
is it possible to apply load directly on nodes
↧

Laminar flow without outlet (mixing with sliding wall) does not find solution

June 6, 2014, 6:39 am
$
0
0
Hey,

I have been trying to simulate a system where a sliding wall is causing convection in a closed box.

In brief, the system is a box with one wall sliding at a rate of 0.01 m/s. The other walls are No slip.

The solver seems to solve the problem quite fine, and the result looks reasonable. However, I get the following error message:

Failed to find a solution.
Maximum number of Newton iterations reached.
There was an error message from the linear solver.
The relative error (4e+002) is greater than the relative tolerance.
Returned solution is not converged.

The Solver log looks like this:

============================================================
Number of vertex elements: 4
Number of boundary elements: 462
Number of elements: 10532
Minimum element quality: 0.8176
Number of vertex elements: 4
Minimum element quality: 0.1194
Stationary Solver 1 in Solver 1 started at 6-juin-2014 17:26:14.
Nonlinear solver
Number of degrees of freedom solved for: 19266.
Nonsymmetric matrix found.
Scales for dependent variables:
Pressure (comp2.p8): 0.27
Velocity field (comp2.u8): 1
Iter ErrEst Damping Stepsize #Res #Jac #Sol LinErr LinRes
1 0.034 0.0100000 0.034 2 1 2 0.053 1.2e-015
2 0.032 0.1000000 0.036 3 2 4 0.0055 3e-016
3 0.031 0.2297586 0.038 4 3 6 0.0031 2.4e-016
4 0.035 0.3012911 0.047 5 4 8 0.0029 3.2e-016
5 0.026 0.5186614 0.047 6 5 10 0.0054 5.1e-016
6 0.0033 1.0000000 0.033 7 6 12 0.025 4.4e-014
7 0.00033 1.0000000 0.0037 8 7 14 0.54 1.7e-012
8 9.7e-006 1.0000000 0.00035 9 8 16 9.2 1.6e-011
9 9.2e-007 1.0000000 1e-005 10 9 18 1.8e+002 3.3e-010
10 1.1e-007 1.0000000 7e-007 11 10 20 4.2e+002 9e-010
11 3.3e-007 0.7500000 1.4e-007 20 11 30 3.9e+002 8.1e-009
12 2.3e-007 0.1128516 3.5e-007 23 12 34 3.9e+002 5.8e-009
13 2.6e-007 0.7500000 2.3e-007 32 13 44 4e+002 2e-008
14 2e-007 0.0049796 2.7e-007 37 14 50 4e+002 1.5e-008
15 3.5e-007 0.7500000 2e-007 44 15 58 4e+002 7.6e-008
16 3.5e-007 1.0000000 3.7e-007 45 16 60 4e+002 5.6e-007
17 2.1e-007 0.0798715 3.7e-007 48 17 64 4e+002 3.4e-007
18 1.3e-007 0.2229211 2.2e-007 50 18 67 4e+002 2.5e-007
19 4.7e-007 0.7500000 1.3e-007 59 19 77 4e+002 2.6e-006
20 2.3e-007 0.1053436 4.9e-007 60 20 79 4e+002 1.3e-006
21 1.2e-007 0.0112598 2.4e-007 62 21 82 4e+002 6.6e-007
22 5.6e-007 0.7500000 1.2e-007 70 22 91 4e+002 8.8e-006
23 3.6e-007 1.0000000 5.9e-007 71 23 93 4e+002 4e-005
24 3.3e-007 0.7500000 3.6e-007 80 24 103 4e+002 0.00011
25 2.7e-007 0.0751680 3.5e-007 83 25 107 4e+002 8.9e-005
Stationary Solver 1 in Solver 1: Solution time: 20 s
Physical memory: 1.18 GB
Virtual memory: 1.23 GB

So for some reason ErrEst goes down but LinErr goes up.

Is this just a problem of COMSOL even though the solution is reached, and more importantly, can I trust the result?

Thanks for any comments and suggestions!

↧
↧

Time dependent Navier Stokes does not converge - inlet boundary condition

June 7, 2014, 12:24 am
$
0
0
Hi all

I am modelling a laminar flow problem in a micro channel in a transient mode. My model works well
but when I change the inlet boundary condition to velocity instead of pressure it does not
converge! the outlet is pressure boundary condiiton

I get below message

Nonlinear solver did not converge.
Maximum number of Newton iterations reached.
Time : 0
Last time step is not converged.

Also when I change the viscosity to non newtonian, my model only works with velocity
inlet boundary condition and not the pressure?


Can you please comment?

Regards
Abe
↧

Asymmetrical & one-directional torque on rotating particle about center point

June 8, 2014, 1:28 pm
$
0
0
My colleague and I have been struggling with COMSOL modeling recently. Specifically, COMSOL seems to calculate forces that should be symmetrical to be assymetrical about a zero point.

In my case, I am calculating the torque on a 2D triangular particle suspended in a liquid medium over a gap between biased electrode digits (on the x-y plane running parallel to the y axis) as a function of the rotation of the particle in the x-z plane using DC model (please see the attached .gif). The particle is centered over the biased gap, and lying parallel to the electrode surface is defined as having 0 rotation. As the particle is rotated about its axis parallel to the electrode edges ( the y-axis), the torque on the particle about this axis should be symmetrical about its original position (flat), but in opposite directions (opposite signs).

However, Comsol outputs a torque that is asymmetrical about zero rotation and is always positive (torque is only in one direction). I believe these are artifacts of the modeling because the torque profile did not change when I switched the bias on the electrode digits. In addition, I tested the symmetry using a flat square particle with obvious geometrical symmetry, and obtained similarly asymmetrical results.

I have been using custom meshing with small elements and low growth rate, so I don't believe that to be the issue, but any help is appreciated.

Thank you in advance,

Donna
↧

Please help - Maxwell Boundary element integral

June 8, 2014, 10:20 pm
$
0
0
Dear all,

I'm trying to solve Maxwell electrostatic surface integral equation,

lambda*U(r) = 1/(2*pi)* Integrate( U(r') (r-r')/(abs(r-r'))^3 ds)

for a sphere. Lambda is an eigenvalue and the integration is over the surface of the sphere. I used weak form boundary (wb) module as mentioned in (www.comsol.com/community/forums/general/thread/3389 ) by Marc Jouan. I used boundary free mesh and solved for eigenvalues near 0.

However, the results I'm getting is quite different from the known Mie theory results. ( The lambda for lowest dipole mode should be around 0.333). But here result doesn't even give a dipole.

It seems this is a very simple simulation but, even after so many efforts I still cannot figure out what goes wrong here. Please can any one be kind enough to look at my model and give some feed back. I have also attached the example given in above link.

Thanks in advance
Ind Atan
↧
Search

Parameter list gives empty sweep

June 9, 2014, 10:29 am
$
0
0
Hello,
I am modeling a system intended to display light scattering off of a particle suspended in a liquid medium. I used Electromagnetic Waves, Frequency Domain (emw) and keep getting the error "The following feature has encountered a problem: Parameter list gives empty sweep - Feature: Parametric 1 (sol1/s1/p1).
What does this mean? Am I required to include a parametric sweep in order to use the pre-made study option? I only intend to model under one set of conditions.

Thanks,
Daniel Frantz
↧
↧

What is the difference between solving in one step and solving separately

June 9, 2014, 12:08 pm
$
0
0
Hi,

I have built a modeling including two modules, one is laminar flow and the other is transport of diluted species. Strangely, I found there is a little difference for the results between solving these two modules together using one step and separately solving the two modules using two linked steps. Could you explain why this difference happened.

If so, when can I separately solve modeling? And when I need to solve them together? Thank you.


↧

Stationary solver will not converge

June 10, 2014, 4:44 pm
$
0
0
Hi,

I have a cylinder, and I'm applying a Gaussian load to one of its faces and trying to find the displacement field. I am trying to hold the center of mass stationary by applying a body force in the opposite direction with the same magnitude. When I try to solve this with the stationary solver, it gives:

Failed to find a solution.
The relative error (2e+003) is greater than the relative tolerance.
Returned solution is not converged.

I have tried putting a constrained wall behind the cylinder, and it converges in this case. I've attached the file. Please let me know if you see anything I am doing wrong.

Thanks!
↧

Leasts-squares gradient-free optimization

June 11, 2014, 7:31 am
$
0
0
Hello all,

I've been trying to fit parameters of my model to experimental data (9 parameters), I can set up a least squares problem and solve it with the SNOPT algorithm. The only thing is that some parameters aren't changed, whereas I can go in by hand and change them and there is a significant difference. Is there some way to do an optimization that doesn't rely on gradients with least squares, or am I missing something with the solver?

Thanks,
Aaron R. Shifman
↧

Computing two physics models Simultaneously

June 11, 2014, 12:18 pm
$
0
0
Hello,

I am trying to combine MAGNETIC FIELD physics with Electromagnetic Waves, Frequency Domain physics for a Helmholtz coil application. In middle of coils is a FR4 board with transmission line on it. I need to encase this board in a perfect electric conductor block. This where I need the Electromagnetic Waves, Frequency Domain physics.

However, I can never get both the models to work together. Infact, Electromagnetic Waves, Frequency Domain never gets computed in the presence of Magnetic field physics.

Can anyone suggest anything?
↧
↧

Multigrid failure during solver

February 1, 2011, 1:56 pm
$
0
0
Hey all,

I'm working on a 3D model of a microfluidic system using the laminar flow (spf) and transport of diluted species (chds) physics models. I've got everything modeled with all the physics settings done, and the mesh applies quickly, but when I set it to solve, it errors out saying:

"Error: Problem setting up multigrid"

Looking through the model the error appears to be when the solver creates a coarser mesh in order to get a simpler solution, the "coarse" mesh has elements too large for the geometry and so it errors out. To give an idea of what I'm working with, the microfluidic channel is 301um x 6000um x 110um with microcantilevers that are 300um x 45um x 1um in the middle.

I know this will be a very intensive model; I have a Core i7 system with 16GB of RAM dedicated to this (as well as about 200GB of virtual memory). The problem is it never even gets to needing any of that RAM (gets to about 500 MB working set and 500 MB virtual) and just errors out trying to simplify the mesh. (Or at least that's my understanding.)

Is there anything I can do to get it to stop with the multigrid preconditioner, or get it to push through it? I'm using Comsol 4.0a, and have a copy of Comsol 4.1 that I haven't upgraded to yet, so if there's a solution that needs 4.1 I can update my system.

Thanks for any advice.
↧

PML error

March 11, 2012, 10:20 am
$
0
0
Hi all,

Can someone point me what I do wrong?
I tried to repeat Scattering from plasmonic nanosphere model, so I have Au sphere + Air domain (sphere) +PML/ABC domain (sphere).

When I defining PML as spherical type I'm getting error that Comsol could not find any spherical or cylindrical regions.
↧
Search

Scattering Model Doesn't Compute

June 12, 2014, 9:11 am
$
0
0
Hello,
I'm trying to model the scattering effects of light off of a thin box. I used the model example "Optical Scattering Off a Gold Nanosphere" as a reference model. " I've tried a few approaches when modeling the system, and I keep getting the error "Undefined value found - Detail: NaN or Inf found when solving linear system using SOR. - Feature: Stationary Solver 1 (sol3/s1)". I don't know where exactly the undefined value is coming from. How can I avoid/fix this error?

Thanks,
Daniel Frantz
↧

Trouble coupling heat transfer physics for both fluids and solids

April 24, 2012, 8:19 pm
$
0
0
Hey all,

So I'm getting stuck with my model where I try to couple three different physics. The three physics are: Electric Currents, Heat transfer in Solids, and Heat transfer in Fluids. Basically what I'm trying to model is a silicon microbridge which is doped, which makes it so that it has a temperature gradient accross it when heated. I am able to couple and run the electric currents and heat transfer in solids physics, where the power dissipated from the ec physics is inputed into the ht physics as the heat source, and then the temperature that is calculated is used to calculate a new resistivity value for the silicon, until the model converges to a value for the temperature at the given voltage.

The problem arises when I try to incorporate the ht2 physics to mimic the effects of natural convection from the sorrounding gases. I set a temperature boundary condition within the ht2 physics which tells the program that the temperature of the microbridge is given by T, which is calculated in the ht physics. However, when I run this, I get the following error:

"Constraing found for variables in different Segregated groups. Try to merge these groups.
-Feature: Stationary Solver 1 (sol1/s1)
-Error: Constraint found for variables in different Segragated groups. Try to merge these groups."

Any ideas? Attached you will find my model. Thanks.
↧
↧

Thermal expansion in a core wire

June 13, 2014, 6:41 am
$
0
0
Hello, We are trying to simulate the thermal expansion of a guide wire from the medical industry, when it is exposed to 4 different temperatures in 4 different ovens. We are using the solid mechanics module, add the thermal expansion node to add the reference temperature and the temperature in the oven, and the initial stress and strain node. Also we used 4 different physics of solid mechanics to model the 4 ovens. The initial stress and strain node was used to stablish the initial strain and stress of the second oven due to the first one and so on.
However, We are having problems to define the boundary conditions to model this problem, and we are not sure if the model it´s well define. We are pretty new at comsol. Can you help us please?
↧

How to simulate 2 objects in free rotation ?

June 13, 2014, 11:59 pm
$
0
0
Hello,

I try to simulate 2 objects A and B in free rotation, but I don't find the solution, Comsol say problem with solid.mises but I don't know how to resolve the problem, could you help me ?

imageshack.com/a/img834/3528/fz99.png

Object A (blue) have an initial rotation at 6.2 rd/s
Object B (brown) have an initial rotation at 6 rd/s

No fixed rotation, just initial value. I give the file in case you can help me.

After I would like to measure the sum of energy evoluate in the time.

Regards
↧

3d laminar flow - solver problem: Segregated group 1.

January 1, 2012, 4:52 am
$
0
0
Hi

I´m simulating the convection heat transfer in a pipe. The external surface of the pipe receive solar radiation to heat the fluid that is flowing.

I start my model with the laminar flow and after I can validate this part of my model I'll put the heat transfer in the pipe to the fluid. But some problems ocurred when i started to comput my model.
When I selected the inlet boundary condition to "velocity" I obtained successful simulation with initial velocity field "0". When change the initial velocity field the simulation dont converge. I dont understand this.

I have changed the inlet boundary condition to "laminar inflow - flow rate", because in the real model I have just the values of fluid flow. At the first time that I tried to simulate occured the following error : "Segregated solver steps do not involve all components." So I added the variables in the segregated step and now the following error is occuring in the simulation:

Failed to find consistent initial values.
Segregated group 1

Matrix has zero on diagonal.
Last time step is not converged.

I dont understand why this error is occuring. I tried to change the initial values but the error persists.
Can someone help me? I put my file in the attach.

Another thing I would ask is how to plot the velocity profile along the pipe?

thanks.

Gustavo
↧
Viewing all 3988 articles
Browse latest View live
Search
<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>