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

error test failures

January 2, 2014, 2:22 am
$
0
0
hi
i use ccp module for simulation,my time range is range(0,5e-5,0.001) but when i compute model times go 7.579247e-019 to this and report this error
Repeated error test failures. May have reached a singularity.
Time : 7.579247088737066e-019
Last time step is not converged.

how can i solve it?
thanks
↧
Search

finite volume, volume of fluid, boundary integral method

August 13, 2012, 5:00 pm
$
0
0
Hello,

I'm trying to model a cavitation air-bubble in water under ultrasound waves. For that I need to use a Boundary integral method to solve the Laplace equation. The finite element method I try to use until now is not stable for my problem.

Is there any boundary integral method (using Greens function) module for Laplace equation in Comsol 4.2 or 4.3 available?

Thanks for your answer!

Karl
↧
↧

minimum time step

September 5, 2014, 8:51 am
$
0
0
Hi all,

I am stuck at trying to choose correct solver options/properties for my model. I am working on H-formulation in Time-dependent case. I cannot find an option where I can set minimum time step because now my model would take forever to find a solution.

Any help is appreciated.

Rokas
↧

deformed geometry for Electromagnets

September 6, 2014, 2:52 pm
$
0
0
hi every one ,
i defined a moving armature by using deformed geometry in AD\DC module .also auto Remeshing defined .when the distortion exceed from 1 ,comsol remesh geometry again .
but after three remeshing ,comsol give an error(A problem occurred when building mesh feature 'Reference 1')also inverted mesh element can be seen .
does any one know how can i solve this problem ?
thanks with regards
↧

"Undefined value found in the equation residual vector" error in "bipolar_transistor.mph" model (Semiconductor Modul)

September 7, 2014, 6:31 pm
$
0
0
Hi everyone,

I try to calculate bipolar_transistor.mph example from COMSOL library models.
After sucsesful finishing study "V_B sweep, V_CE=0.5 V, V_E=0 V" I run Compute command for study "V_CE sweep, V_E=0 V, I_B=1 uA" and recieve Error message "Undefined value found in the equation residual vector".

I have the same errors on both Windows 8, x64 and Fedora 20, x64 computers.
Can somebody explaine me, what's wrong?

The log see below:

COMSOL 4.4.0.195

Opened file: bipolar_transistor.mph
Complete mesh consists of 8047 domain elements and 519 boundary elements.
Solution time (Solver 1): 3 s.
Number of degrees of freedom solved for: 24141.
Solution time (V_B sweep, V_CE=0.5 V, V_E=0 V): 78 s. (1 minute, 18 seconds)
Number of degrees of freedom solved for: 24142.

Error:
Undefined value found.
- Detail: Undefined value found in the equation residual vector.
There are 438 degrees of freedom giving NaN/Inf in the vector for the variable comp1.Ne.
at coordinates: (6.46671e-008,-1.47054e-007), (7.47439e-008,-1.47035e-007), (6.9664e-008,-1.44089e-007), (7.47295e-008,-1.52949e-007), (6.97037e-008,-1.5588e-007), ...
There are 363 degrees of freedom giving NaN/Inf in the vector for the variable comp1.Ph.
at coordinates: (5.46959e-008,-1.47084e-007), (4.99044e-008,-1.44442e-007), (5.9625e-008,-1.44138e-007), (8.96945e-008,-1.43933e-007), (9.46986e-008,-1.46974e-007), ...
↧
↧

Editing Parametric Sweeps

September 7, 2014, 7:23 pm
$
0
0
Hi all,

I created this model to test the effectiveness of AntiReflective Coatings. I wanted to modify this so that it can check results over a wide variety of wavelengths as well as angles. Can anyone help me fix this (creating two parametric sweeps does not seem to create the correct data).
↧

BAW Resonator Mesh

September 7, 2014, 9:04 pm
$
0
0
Hai

I am designing a AlN BAW resonator operating at 3GHz. It is being observed that the frequency response changes widely with the mesh density.

The mesh criteria of the mesh size being between lambda/4 and lambda/6 seems kind of very difficult as the lambda values at 3Ghz is very low and my resonator active area is large.So the computational time is really too long to serve as a good solution.

Please do help me with suggestions as to how to build a reliable mesh for such devices at GHz range.

Regards

Gayathri Pillai
↧

3D Model of "Voltage Induced in a Coil by a Moving Magnet"

May 16, 2013, 10:39 am
$
0
0
Hi

I wanted to do the 3D version of the Model "Voltage Induced in a Coil by a Moving Magnet" which is in the Comsol Library.

I used the same physics, materials and boundary conditions, but unfortunately the 3D Model does not converge.

Do I have to adapt something in order to make the 3D model work??

I would be really glad if you can help me out!!!

Thanks
↧
Search

Parametric Sweep versus Parameter

September 8, 2014, 6:07 am
$
0
0
I want to make a sequence of plots and/or movie, with a variable changing, but my model is not a function of this variable. For example I want surface plots of emw.Ex*cos(theta)+emw.Ey*sin(theta) for theta 0 to 360 degrees in 45 degree increments where the model isn't a function of theta.

I can use "Parametric Sweep" but I worry that this will cause Comsol to recalculate and/or store the
same solution multiple times. [ed: my testing indicates that Comsol _does_ store multiple copies of the solution even though the model isn't a function of a parametric sweep parameter]

Also, it would be good to make this sort of "not-parametric" sweep without needing to recalculate the solution.

Is there an alternative to parametric sweep for this situation?
↧
↧

Imposing pumping well in 2D model

September 9, 2014, 2:54 am
$
0
0
Hi comsolers,

I want to impose pumping well in 2D model NOT 2D axisymmetric .
Is there any way to impose well, if yes which boundary condition should I define that comsol knows that in my model I have a well?

and I know that I should also consider the thickness of the well, radius of the well,....

Any help or answer is appreciated.
Thanks
Sepideh
↧

Calculation time

January 8, 2013, 5:55 am
$
0
0
Hello,

I developed a model which simulate the trajectories of atmospheric particles in an electrostatic precipitator. In order to optimize the collection efficiency, I plan to run the model by changing the voltage and the flow rate using the "parameter sweep" option. Before to run these simulations, I would like to estimate the calculation time needed to solve each COMSOL simulation. Does anyone knows how to do that?

Thank you in advance for your help.
Best regards,
↧

MWP combined with DC

August 5, 2013, 1:15 am
$
0
0
Dear Daniel,

I am geting tired of comsoling... because I do not know anymore what I can do to make my models converge.

i started with the waveguide microwave plasma from the model-library. I put some substrate under bias within the plasma region.. so far it is working, even if it is already a tough calculation that needs 2 days.
whenever some e-temperature spikes occurs, then the timestepping gets very small, that is OK, it needs smoothing. so increasing the mesh density seems to be a solution to prevent this.

but my design is even more complicated since the microwave has to move (propagate) in the DC sheath..

the models run up to 10e-5 sec or so when the plasma reaches cut-off density and the fields get expelled from the plasma region, the ion currents make a big jump... but then it tells me after some days that the convergence fails, and the time increased some 10e-7 sec within the last 2 days.. the convergence plots has a plateau around 10e-08/10e-9.. I have really the suspiscion that the solver starts to look at each GHz oscillation !!!

what I can I do to have different solver conditions for the 2 segregated steps: the one for the plasma (heavy species and electrons) and the other for the E-fields..
the E-field solver should be very fast.. 1 or 2 iteration should be enough.. how to set this?

when I used the mwp models only, it is possible to have a solution within a day.. but not when I include the electrical potential..

one strange thing happened when I had a circuit with the ballast resistor and the bolcking capacitor (like in the pdf library model of the 2D-DC column).
my voltage at the capacitor was increased beyond the DC-voltage level? how to explain that?

Mr. Tong at Kesco (Japan) is trying to support me, but he tells me to keep the bias voltage low (50V) as well as the pressure (beyond 30 Pa).. but that is not the real world.. 400 to 500V are needed and 60 to 100 Pa.
(at the low pressures the cutoff may not be reached, and at low voltages the running of the microwave in the dark-space doesn't work)

do I have the go to the model-builder a have my own module? is the mwp module missing some important physics my set-up requires? (or do we miss the AC/DC modul for that problem? I run RF and PLASMA only)

thank You very much for the support.

Lukas Alberts
Nagoya University
↧

Expanding the Capacity Fade tutorial model - Convergence issues

September 10, 2014, 12:10 pm
$
0
0
Hello,

I've been working with the Capacity Fade lithium ion tutorial model, and I've been trying to expand the parametric study from the default 1-10 cycles to a maximum of 30 cycles.

I usually end up with a convergence error around cycle 17.

I've tried a few things to help convergence, but had limited success with using a finer mesh and loosening the error tolerances.

Can anyone give more insight on what I might need to do? Right now I am toying with the idea of completely changing the load cycle to get more charge-discharge events within one iteration of the parametric study.
↧
↧

Parameter as function of time in time dependent simulation

September 11, 2014, 2:53 am
$
0
0
How do i specify a parameter which is function of time in a time dependent simulation ??
in my case one of the parameter is function of time
↧

Convergence versus meshing granularity

September 12, 2014, 12:47 pm
$
0
0
I have noticed interesting behaviour in my models.

If the model will solve using a coarser mesh, it may not solve well using a finer mesh.

And although use of the boundary layer tool can reduce the wall lift-off, it does not necessarily reduce convergence time, and often can lead to convergence instabilities. Sometimes I can only add 2 boundary layers.

Also, if I use an initial field from a coarsely meshed solution, the convergence behaviour for the finer mesh will only begin to act erratically earlier.

I have noticed these effects using the k-epsilon and k-omega turbulence models.
↧
Search

how to write stop condition in time dependent study?

September 12, 2014, 10:57 pm
$
0
0
Hi
i am doing particle tracing, in time dependent study, how to write stop condition, like i need the study stops when the number of particle passes through the outlet is over 100 , thanks!
↧

Comsol ignores second step in the same Study and solves only first step

September 13, 2014, 1:40 pm
$
0
0
Hi all!

I want to solve my model using several steps in the same study. First step is supposed to solve from t=0 to t =1s and second step is supposed solve from t=1 to t=100s. Both steps are using the same physics and second step starts where first step finished. However, Comsol only solves first step and that is it. It seems that Comsol doesn't see the second step. What I am doing wrong here?

Thanks for helping.
↧
↧

Lorentz force calculation and implementation

September 15, 2014, 3:15 am
$
0
0
I have got a clamped-clamped beam, with a DC current flowing through.
If the beam is surrounded by a DC magnetic flux density (orthogonal to the direction of the current flow), a Lorentz Force arises and the beam deflects.
If next to the beam there is another similar structure (not moving) and if the two blocks are conductive, the two form a capacitor.
The capacitance value depends on the magnetic flux density magnitude, as it depends of the distance between the two.
I would like to implement this system in COMSOL.

I was thinking of three steps:
1) Magnetic Fields (mf)
I can define the external magnetic flux density and the current density flowing through the spring. The Lorentz force x,y,z components are available as simulation outputs (mf.FLtzx, mf.FLtzy and mf.FLtzz). This step gives almost no troubles.

2) Solid Mechanics (solid)
I define a body load on the beam, defined as force per unit volume, using the "Lorentz force contribution (mf/al1)" option (COMSOL itself suggests this), and the simulation should return the corresponding displacement. But, if I do this, the solution gives 0 displacement as a result. No displacement is shown.
On the other hand, if I apply an "user defined" body load as [B*I*L/V], where B is the magnetic flux density, I is the current, L is the length of the spring and V is the volume of the spring expressed in [N/m^3], the simulation works.
Any idea of why this happens? It's annoying because the user should "manually" input the Lorentz force and I would like to avoid this.

3) Electrostatics (es)
I define two terminals (the two structures, the beam and the fixed one) and I can calculate the capacitance.
And here's the other problem: how can I "tell COMSOL" that the beam is not "at rest" but it's deformed (due to the Lorentz force application)?

Thank you!

Paolo
↧

How to couple shear stress and moving velocity on the same wall?

September 15, 2014, 9:14 am
$
0
0
Hi everyone!

I am trying to add a shear stress boundary condition on a growing bubble wall, which should be a moving wall boundary condition. I used to add shear stress on a fix wall by slip wall boundary condition and weak contribution. However, in this case, it is hard to couple these both slip wall and moving wall (I tried to transform moving wall into weak contribution, but it did not work out). Does anyone know the method to deal with this kind of situation?

Thanks,
YFC
↧

2D-axis symmetry, 3D no symmetry not possible to calculate due to limited fixed points

September 16, 2014, 9:38 pm
$
0
0
Dear all,

I have a situation were I need to simulate a kind of tube with a pressure inside, the target is to calculate the displacement of the tube due to the pressure. Reinforcements are needed and these will be evaluated.

In the case of a round tube we can work with a 2D-axis symmetry study and put a boundary load on the inside of the tube and use 1 roller on the bottom, this is calculating fast and accurate.

Now I want to this in 3D (because my final tube is a none axis-symmetric shape (round rectangular)).
But in the 3D version I can not work without a fixed constraint or with too many rollers.

In fact this is the same just that the 3D version is not thinking that the applied force is completely symmetric.
I tried with splitting the rectangular tube in 2 and work with a symmetry but that did not work out neither.

Any suggestions on how to deal with this kind of problems?

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