Quantcast
Channel: Results and visualization - Latest Discussions - COMSOL Forums
Viewing all 3058 articles
Browse latest View live
↧

initial value set-up

January 24, 2013, 1:41 pm
$
0
0
Hi,


I have a 2D simulation running from T=0 to T=100. First, I run the simulation from T=0 to T= 50. Then I want to use the solution at T=50 plus some value changes in specific domain as the initial value for my simulation from T=50 to T=100.
Can anyone help me out?

Another simple question is: can I find the solution value of variable PIP at T= 50 by PIP(t=50)? if not, how can I reference the variable at specific time point and/or spatial points as well?

Thanks in advance,

John
↧
Search

Finding a mapping of Current Density (variable J) using the Electrostatics (es) Interface

January 24, 2013, 3:10 pm
$
0
0
Hi everyone,

I'm making a model using the Electrostatics Interface from within the limited AC/DC Module included with the base COMSOL Multiphysics package. Initially, I tried making my model using the Electric Currents (ec) Interface, but I was getting the 'wrong' answer (I will probably make a separate post for that problem, it would be interesting to see what you guys think about it). When I use the Electrostatics Interface, the answer I get is much better. The Electric Currents Interface includes the Current Density (J) variable from Ohm's Law, but the Electrostatics Interface does not.

Finding the Current Density (J) is relatively simple since the Electrostatics (es) Interface gives me the electric field (E). From Ohm's Law, I should easily be able to calculate J using: J = (sigma)(E). Where (sigma) is the electrical conductivity of the material (a vector/tensor quantity) and (E) is an electric field tensor.

How do I enter this into COMSOL? Is there a way to define this as a function somewhere? A variable? Or should I add this as another "physics" as some kind of PDE/ODE, even though it is only a matrix operation. Also, It may be hard to simply multiply an expression the sigma by E as sigma is a tensor/vector quantity and is unique to the material being modeled. The model is 2D and has two different materials in it, so the expression must take into account the E-field at that specific point and also the electrical conductivity associated with that point as well.

In case this post is too long for you to read or you don't understand the model after reading, I have attached it. It includes two different materials with assigned material electrical conductivities.

Thanks in advance for any help,

John

UPDATE: After some further investigation, I have found out that the Electrostatics Interface has an equation for x-direction current density as: es.Jx = es.Jdx. Where Jdx is the displacement current density, which happens to be zero for this modlue/model/interface. Investigating the Electric Currents Interface, the equation for equation for x-direction current density is: ec.Jx = ec.Jix + ec.Jdx. Where again, Jdx is displacement current density (again zero) and Jix is (ec.sigmaxx*ec.Ex+ec.sigmaxy*ec.Ey+ec.sigmaxz*ec.Ez) which is basically the Ohm's Law thing I was talking above where certain tensor quantities of of the electrical conductivity (sigma) is multiplied by the Electric Field (Ex, Ey, Ez). Why is there no induced current density in the Electostatics Interface? And when I try to add in the long expression for Jix in into the ex.Jx on the electrostatics I get an error saying that it doesn't know what es.sigmaxx, even though I have entered electrical conductivities for all materials I am studying.
↧
↧

computing accurate fluxes with periodic boundary conditions?

January 24, 2013, 3:42 pm
$
0
0
Hi everybody,

does anyone know if I can compute the fluxes accurately, as described in the user guide on page 308 ff. and file 0973 from the knowdledge database etc, when using periodic boundary conditions on the respective boundaries? In the descriptions, it is explained only for Dirichlet boundary conditions :( And if you think it should work, should I then use the weak constraint option? (and, how, if you have time to explain?) I model in the PDE mode of version 4.1. with matlab.

All best,
Sabine
↧

Manipulating data from tables

January 25, 2013, 9:05 am
$
0
0
Hi all,
I'm working on a model using piezodevice physics in the frequency domain, and I'm trying to plot admittance in dB (rather than siemen). I've found the admittance by taking the surface integral of the current density (pzd.JZ) over one of the electrodes, which results in a table with admittance values for each frequency. Is there any way to apply additional mathematical manipulation to the values in that table? To convert to dB, I need to take 10*log10(x) for each admittance value. I've tried doing the 10*log10(abs(pzd.JZ)) as the expression in the surface integral, but that doesn't work... though admittedly my math is a little fuzzy as to why not.

It seems like there must be an easy way to manipulate data in a table, but I can't seem to find it. Any advice would be greatly appreciated!

Thanks,
Jillian
↧

Free & Porous Fluid Flow -- Measure of mass flux

January 28, 2013, 3:39 pm
$
0
0
Hi All,

I am attempting to model a stream bottom seepage collection device, but I getting some strange results.

I am starting out fairly simple with no lateral flow in the free fluid, and setting the BC's so that I have a vertical flow field. But my results show the velocity contours have local maxima along the stream bottom under the device. I was expecting the maxima to be near the outlet of the device as it is a considerable constriction.

Also, is there a way to measure the mass flux of fluid through a plane? I am trying to evaluate how much water is being captured by the device and how much fluid is going around.

Thanks,
John
↧
↧

Integration Coupling Variables

January 29, 2013, 12:07 am
$
0
0
I am trying to find the volume of a spherical shell using Comsol 4.3a in a 2D axisymmetric coordinate system. I attempted to do this by creating a integration coupling variable along the boundary of the shell with the expression pi*r^2, but it returns a value that is significantly different from the expected 4/3*pi*r^3. Can anybody help me find the volume contained within the boundary of a spherical shell?

Incidentally, integrating along the boundary with the expression 2*pi*r for the surface area returns the expected value, as does integrating the contained subdomain with the expression 2*pi*r for the volume, but I would like to be able to calculate the volume using boundary integration.

Thanks in advance,
Sean MacGahan
↧

Undefined Value Found

January 30, 2013, 4:46 am
$
0
0
I was runing a 2D model and came up with this error..
anyone can help me??


Undefined value found.
- Detail: Undefined value found in the equation residual vector.
.There are 912 degrees of freedom giving NaN/Inf in the vector for the variable mod1.E10
. at coordinates: (-2.83333e-007,-8.5e-007), (-2.91667e-007,-8.5e-007), (-3e-007,-8.41126e-007), (-3e-007,-8.32252e-007), (-2.94182e-007,-8.26412e-007), ...
There are 284 degrees of freedom giving NaN/Inf in the vector for the variable mod1.E20
. at coordinates: (-2.94182e-007,-8.35286e-007), (-2.85849e-007,-8.4416e-007), (-2.62218e-007,-8.43432e-007), (-2.73067e-007,-8.37592e-007), (-2.92567e-007,-8.08802e-007), ...
There are 284 degrees of freedom giving NaN/Inf in the vector for the variable mod1.E21
. at coordinates: (-2.94182e-007,-8.35286e-007), (-2.85849e-007,-8.4416e-007), (-2.62218e-007,-8.43432e-007), (-2.73067e-007,-8.37592e-007), (-2.92567e-007,-8.08802e-007), ...
There are 630 degrees of freedom giving NaN/Inf in the vector for the variable mod1.Ez
. at coordinates: (-3e-007,-8.5e-007), (-2.75e-007,-8.5e-007), (-2.875e-007,-8.5e-007), (-3e-007,-8.36689e-007), (-3e-007,-8.23379e-007), ...
↧

Problem Modeling Capacitive Cantilever Sensor

March 16, 2012, 11:57 am
$
0
0
Hi,

I want to find out the pull-in voltage of a capacitive cantilever sensor. To do so I did the following:

I chose to do the 2D model first to get an idea, then I added Moving Mesh (ale), solid mechanics, and electrostatics (in order) for my physics and chose a stationary solver.

In geometry I made 3 rectangles (300um * 2 um). I chose the top and bottom rectangles to be gold; the top one is going to be the cantilever and the bottom one is going to be a fixed ground plane.

I chose only the top and bottom rectangles for the solid mechanics module. I chose the middle rectangle, representing the gap to be air. Then is solid mechanics module I fixed the left, right, and the bottom boundaries of the bottom rectangle (the ground plane) and fixed the left boundary of the top one, using the fixed boundary constraint.

I chose air for the electrostatics module. In the electrostatics module, I chose the bottom face of the middle rectangle to be ground and the top face to have a fixed potential.

At the end, I chose the middle rectangle only for the ALE, chose it to have free deformation (free deformation domain), and added a prescribed mesh displacement (after the free deformation) overriding the default mesh displacement. I chose the top and bottom face of the rectangle, and chose u and v for dx and dy.

I used a physics-controlled mesh.

At the beginning when I tried to solve the problem it gave me a Newtonian n iteration number error so I increased the number of iterations; however, the y displacement result that I'm interested in, still does not make physical sense (it is ~3E-308, which is zero).

Would someone be kind to tell me where I am making the mistake? Eventually I am going to need to model the #D version; I tried the 3D version using a similar approach and keep getting 0 displacement as answer. I don't get any errors or warnings from the solver. I have also tried playing with the fixed boundary conditions, and the voltage level to no avail.

Thanks.
↧
Search

Change x-axis units in table plot

January 31, 2013, 1:06 am
$
0
0
Hi all,

I have a 3d model of diffusion from a source and am calculating surface integrals of flux over particular features in my geometry. This outputs a table of total flux vs. time point which can then be plotted in a 1d plot group table plot. However, the default output of this is with respect to seconds as used in the time dependent solver and I cannot figure out any way to change the x-axis units to something else for display. When my changes of interest occur on the order of many hours, it is inconvenient to look at things in seconds. Thanks!
↧
↧

How to integrate a table result

January 31, 2013, 1:09 am
$
0
0
Hello,

I have calculated surface integrals for total flux over particular features in my geometry. This outputs a table of total flux in (mol/s) vs. time points. I would like to then be able to integrate this vs. time to output a table of total mol vs. s but I cannot figure out how to integrate table data. Thanks!
↧

Colors For 2D Plots

January 31, 2013, 3:46 pm
$
0
0
Hello,

I have a parametric Sweep for 10 data point. Then I need to plot them in a 2D Plot with a legend. I have done that but some colors recycle, so it can be hard to distinguish them. Is there a way to add more colors to the table or get a different style of line, like dashed??
↧

Show parameteric value on plot

January 31, 2013, 4:05 pm
$
0
0
I've made a model of a hi-pot test setup to visualize how the electric field changes at a discontinuity as the conductivity changes (fixed potential).
I'ts working great, but now I want to make an animation of the results. I can create the series of 2D plots without a problem, but I don't know how to show the value for the conductivity on the plot, making it difficult to distinguish between them.
↧

piezoelectric potential due to a deformation

February 1, 2013, 10:00 am
$
0
0
hi,

an electric potential applied to a piezoelectric material will cause a deformation of the material. On the other hand a deformation of a piezoelectric material will end up in an electric potential. I'm interested in the latter case, which I will investigate in detail by using COMSOL.

Therefore I have taken a piezoelectric material (pieM) and on that material I put a bimetallic structure (bimS). First, I added an Structural Mechanics Module (SOLID), which is used to set up a thermal change. That will cause the deformation of the bimS and therefore a deformation of the pieM. But now I want to see the induced electric potential inside the pieM. So I added another physics module for piezoelectric devices (PZD). Inside of Piezoelectric Material I defined the "Initial Stress and Strain" by inserting the values for "initial stress" from SOLID in the following form:

S0=((solid.sx,solid.sxy,0)(solid.sxy,solid.sy,0)(0,0,solid.sz)) N/m²

Inside "Linear Elastic Material" I selected the non-piezoelectric bimS.

After computing I can take a look to the electric potential inside the pieM by using the expression "V". But it gives me only one potential for the whole structure. But I found the reason for that: Comsol takes the calculated induced electric potential which leads into an additional deformation. So, what should I do in order to make the induced potential visible and to forbid that this potential ends up in an additional deformation, which is shown up by COMSOL .

I hope that my description won't confuse too much, so that you can give me some hints.

Thanks a lot in advance!
↧
↧

Problem Modeling Capacitive Cantilever Sensor

March 16, 2012, 6:57 am
$
0
0
Hi,

I want to find out the pull-in voltage of a capacitive cantilever sensor. To do so I did the following:

I chose to do the 2D model first to get an idea, then I added Moving Mesh (ale), solid mechanics, and electrostatics (in order) for my physics and chose a stationary solver.

In geometry I made 3 rectangles (300um * 2 um). I chose the top and bottom rectangles to be gold; the top one is going to be the cantilever and the bottom one is going to be a fixed ground plane.

I chose only the top and bottom rectangles for the solid mechanics module. I chose the middle rectangle, representing the gap to be air. Then is solid mechanics module I fixed the left, right, and the bottom boundaries of the bottom rectangle (the ground plane) and fixed the left boundary of the top one, using the fixed boundary constraint.

I chose air for the electrostatics module. In the electrostatics module, I chose the bottom face of the middle rectangle to be ground and the top face to have a fixed potential.

At the end, I chose the middle rectangle only for the ALE, chose it to have free deformation (free deformation domain), and added a prescribed mesh displacement (after the free deformation) overriding the default mesh displacement. I chose the top and bottom face of the rectangle, and chose u and v for dx and dy.

I used a physics-controlled mesh.

At the beginning when I tried to solve the problem it gave me a Newtonian n iteration number error so I increased the number of iterations; however, the y displacement result that I'm interested in, still does not make physical sense (it is ~3E-308, which is zero).

Would someone be kind to tell me where I am making the mistake? Eventually I am going to need to model the #D version; I tried the 3D version using a similar approach and keep getting 0 displacement as answer. I don't get any errors or warnings from the solver. I have also tried playing with the fixed boundary conditions, and the voltage level to no avail.

Thanks.
↧

Oblique incidence and floquet condition problem

January 15, 2013, 6:16 pm
$
0
0
In the RF module, I found that there are 2 problems.

First,
When the thickness of the layer which uses scattering boundary condition gets thin (below ~200nm), the direction of power flow [sqrt(emw.Poavx^2+emw.Poavy^2+emw.Poavz^2)] doesn't go 90 degrees even if I set the incident angle as 90 degrees.

Second,
The power flow distribution reaches below some part of port condition area even if I set the incident angle as 90 degrees.

I also tested it in the plasmonic wire grating of model library which uses port condition instead of scattering boundary condition. However same problems occurred.

I attached figures of my question.

please help me.
↧
Search

electric current induced

January 16, 2013, 6:13 am
$
0
0
hello,


i am beginer with Comsol, i want to determine the induced current in coil surrounding a moving magnet


thanks for your help
↧

complex current density

January 16, 2013, 6:51 am
$
0
0
Hi all,
I'm working on getting the frequency dependent conductance and susceptance (real and imaginary admittance) of a simple resonating PZT block using the Piezoelectric Devices physics. I seem to be getting a reasonable admittance by integrating the normal current density over the top electrode and dividing by the applied voltage.

My problem comes in when I try to separate the admittance into conductance and susceptance. It seems like this should be done by integrating the real and imaginary parts of the normal current density, but the current density doesn't seem to have an imaginary part - giving a susceptance of zero. Is there a setting I need to change to find the imaginary part of the current density? I've checked the boxes "Allow complex-valued output from functions with real input" and "split complex variables in real and imaginary parts" in the solver, but left everything else as the default option.

Please help! The model is attached.

Thanks,
Jillian
↧
↧

Rotate Graphics Window 90deg in 2D Axisymmetric

January 16, 2013, 6:59 am
$
0
0
Hi,

Is it possible to rotate (90deg.) the graphics window in 2D axi. so the z-axis is horizontal and the r-axis is vertical.

Thanks!
↧

solvers in 4.3a/4.3 version

January 21, 2013, 3:16 am
$
0
0
Hello

I am interested in accesing solver information for ODE/PDE in Comsol.

So far I was using the Matlab - COMSOL connection in 3.5 version which was showing a sequence like:

fem.sol=femtime(fem, ...
'solcomp',{' ',' ',' ',' ' }, ...
'outcomp',{' ' , ' ',' ',' '}, ...
'blocksize','auto', ...
'odesolver','genalpha', ...
'tlist',t:dt/4:t+dt, ...
'tout','tlist', ...
'incrdelay','off', ...
'nlsolver','manual', ...
'ntolfact',1, ...
'maxiter',4, ...}

I would like to know if and where this information is available in newer version.

What I could understand untill now id that all the solving is unde (for example):

model.sol('sol1').runAll

Thank you in advance for your support.

Dana Ofiteru
↧

Two interdependent studies - but I cannot use results from frequency domain study as input for stationary study

January 21, 2013, 10:35 am
$
0
0
Dear all,

How can I use the result from a parametric (parameter 'b') sweep-frequency domain study (e.g. emw.Ez(b) ) in another function defined under Definitions (e.g. f(emw.Ez(b)) ) which is used as an input for a stationary study?

According to the results that I am getting, what is actually passed to the second study is ZERO.

Could you please send me a reference?
Or tell me how to shall I input the results from a study to another study?

Thanks a lot!

↧
Viewing all 3058 articles
Browse latest View live
Search
<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>