Quantcast
Channel: Latest Discussions - COMSOL Forums
Viewing all 26512 articles
Browse latest View live
↧

Radiative de-excitation in a plasma

March 25, 2014, 6:36 am
$
0
0
Hello,

I am using the plasma module and wanted to include radiative de-excitation reactions. For example, I want to include the transition from N2C to N2B with a wavelength of 337.1 nm. I know that COMSOL has Plank's constant. If I give it a wavelength I get a Token error, Position: 11, Reaction: 0. Has anyone done this before? Is there a way to include radiative de-excitation?

Thank you.
↧
Search

Set Stefan condition manually

March 31, 2014, 6:12 pm
$
0
0
Hello,

I'm currently struggling with stefan problem and want to implement the stefan condition L(V.n)=Kl grad(T).n-Ks grad(T).n manually on the Solid/Liquid interface. L is latent heat, V is interface velocity.

I figured out there could be 2 possible ways:
(1) HT in liquid with solid subnode defined in this physics;
(2) 2 HTs, one is HT in solid and the other is HT in liquid and set;

for (1), since the default continuity condition assigns Kl grad(T).n-Ks grad(T).n=0, how to override this to add the latent heat term?

for (2) Do I need to define temperature and flux for both HTs in their boundary settings?

Thanks in advance for any suggestion!
↧
↧

Pressure change with time

April 1, 2014, 2:43 am
$
0
0
Hi everyone,

I have a problem like this:
Two big container are linked with a thin tube, there is a valve at the middle of the tube.
At time t<0, The containers are filled with air but at different pressure. At time t=0, the valve is opened.
I want to know the time dependence of pressure distribution of the containers and tube.
A simple drawing is given in the attached picture.
Can anyone tell me which model should I choose and how to define the boundary of valve?

Thanks a lot

Wei
↧

Fixing constant mass - fluid in closed cavity

April 1, 2014, 3:26 am
$
0
0
Hello everyone,

I am doing a model of a gas in a closed cavity with heated and cooled walls. I fixed the initial conditions (pressure, temperature, density). If I use the time dependent solver, everything is OK; the program gives me the expected solution, which after a long time is stationary. I am using the CFD

To solve this kind of problems with a stationary solver, a point wit ha fixed pressure is needed; otherwise, there are infinite solutions and there are no convergence. My problem is, I don't know the final pressure but only the initial one; this condition must be replaced by some other one.

This other condition is the constant mass: since the cavity is closed, the total mass in the initial and final state must be the same. I can compute the total mass and then force the final one to be the same.

My question is, how to do it properly? I have included an integral coupling over the whole and only domain , and then add a global constrain in which I integrate the density and subtract the total (initial) mass. However, doing it this way the model doesn't converge either.

I don't know what I am doing wrong; if the problem is in the coupling integral, in the global constrain, or in that I shouldn't use this nodes to enforce the condition at all!

Thanks in advance for any help.

Álvaro

(file attached. Only the stationary version for reasons of space, the time-dependent works fine)
↧

How to have a 'tolerance plot' with COMSOL 4.4

April 1, 2014, 4:06 am
$
0
0
In order to have a better resemblence to reality, getting to see the effect of tolerance is very crucial. In these regards, I would be interested to know if there is a possibility with COMSOL 4.4 to have the tolerance plotted in the graph (an example of such a tolerance plot may be seen in the attached Picture. In the Picture, the parts highlighted in red are the tolerance plots over the Graphs. This is how it is seen in theory to understand the influence of tolerance.) This is what I intend to have.

Thank you

Regards
Tanumay
↧
↧

what is the best method (iterative or direct) for solving 3D problems?

April 1, 2014, 4:42 am
$
0
0
hi
i'm new in comsol
i try to do a very simple example conjugate heat tranfer
i simulate a simple flow in a channel characterized between 2 parallel planes in 3D geomtry, i have do a finer mesh
what is the best parameter for solving my example with the method iterative multigrid in Presmoother, Postmoother and Coarse Solver (SOR, SCGS, Jacobi, Vanka, SOR line, SOR Gauge, SOR Vector, direct) and what is the differnce between this parametrs
i search an explication for the iterative method multigrid
Thank you.
↧

How apply a sinusoidal voltage?

April 1, 2014, 4:49 am
$
0
0
I'm using joule heating, time dependent domain.
↧

rigid plate supported by injected gas beneath it

April 1, 2014, 5:00 am
$
0
0
Dear all,

I have a question, maybe a very stubid one, in which a rigid plate is supported by injected N2 beneath it. Hence the plate is vibrating up and down due to gravity and gas pressure. There is no other constraint in this problem.

Can I solve this problem in FSI module or other modules?

Thank you in advance.

↧
Search

Modeling A Dove Tail Joint Between Two Plates

April 1, 2014, 5:09 am
$
0
0
I want to model two long plates that fit together using a dove tail joint (similar to a puzzle piece) and are glued. The geometry I have created is a 1 x 6 in plate with loads on either end. In the center, a line sequence creates a geometry which would indicate the actual boundary of the two plates and how they are connected. I have set the line to a thin elastic boundary condition as there will be glue as well.

I am running into an issue with modeling the plates as separate materials. When I run the study, the plates separate and seem to be overlapping even though this is a two dimensional model. This is obviously impossible, so I tried using contact pairs but can't find out how they work and don't know if this is the best way to proceed.

Any advice about modeling this situation would be appreciated.
↧
↧

Use the result of 2D simulation into 3D simulation

April 1, 2014, 5:48 am
$
0
0
hi everyone!

i want to increease the speed of my 3D simulation, so i have a idea: the result of a 2D simulation is used as the initial value of a 3D simulation, then maybe the 3D simulation can find a suitable result quickly. Does my idea work? how can i use the result of 2D into the 3D simulation?

thank you very much!
↧

Mulitple Modes on Numerical Port

April 1, 2014, 9:42 am
$
0
0
Hello Comsol community,

I am trying to get a rather simple simulation working so that I can use it in a more complex case, but I am running into problems with computing S-parameters between multiple modes on a port. So the setup is as follows: circular PEC waveguide with ports on each end. Each port is set to numerical which are solved using the boundary mode analysis with two modes being set. I can compute the study step for the boundary mode analysis and I observe two separate modes at each port.

Now I want to use these two solved for modes in a frequency domain study, in which scattering parameters between each port and each mode will be computed. For example, getting S_{21} between mode 1 on port 1 and mode 2 on port 2. I can get it working for a single mode on each port, but I have no clue to get it for multiple modes. Can anyone lend me a hand?

↧

Profiling/Optimizing Comsol Simulations

April 1, 2014, 10:09 am
$
0
0
Hi everyone,
I was wondering if anyone has tips/suggestions for optimizing solution times in Comsol 4.4? I have produced a model, which is taking far longer to solver than I would like, and I'm trying to isolate the cause. Many programing/development environments can produce breakdowns of code execution time/performance, is it possible to do this in Comsol? Perhaps via matlab or JAVA? I would like to know how much time is being spent on various tasks so I can intelligently scale back the expensive parts. For example, my model requires many function evaluations (Bessels, interpolation, ramps). I'd like to be sure that the combined execution time of these features is not the problem.

Depending on my input parameters, my model can take several minutes to run to several days. My current largest is approximately 1M DOF. The model is a 2D time-dependent simulation with coupled heat-transfer, solid-mechanics, contact, 5 domain ODE, two geometries, coupling operators and a deformed geometry. For example, one time-step of this model required six iterations of the non-linear solver (Paradiso) and took about 3 minutes on a 12-core Xeon server (with significantly more ram than required).

Is there a simple way of visualizing the Jacobian to check the sparsity? I've tried using the nojac operator to improve the sparsity of the Jacobian matrix with the coupling operators.
↧

floating potential node and potential distribution

April 1, 2014, 12:12 pm
$
0
0
Hi,

I'm simulating a piezoelectric block that is depressed by a uniform load. I have set the top plate as floating potential and the bottom as a ground.

I'll like to ask what does setting Qo=0 mean. does it set the initial charge on this plate to be zero and any extra charges generated due to the load will then cause the top plate to have a voltage? Also, since I'm able to get a voltage, why is it that I can't get the charge density on the plate.

Another question is about plotting the surface plot of potential. When I have a curved piezo-layer and I depress it, there will be a terminal voltage generated. However, the potential distribution in the surface plot is not a smooth gradient along the whole curved surface. As in there will be max potential at the middle instead of having the max potential at the top (floating potential node). Is this due to how potential is calculated in comsol? Thanks.
↧
↧

determine domain number

October 18, 2011, 8:16 am
$
0
0
Comsol says I need to assign a material to domain 6. How do I figure out which piece of my model is domain 6?
↧

FGM Material properties

March 30, 2012, 9:06 pm
$
0
0
I am new to comsol. Is it possible to include FGM properties in COMSOL
↧
Search

Accessing solution of one study into another one ?

May 14, 2012, 6:29 am
$
0
0
Hi,

I been modeling hyperelastic material models in comsol which revolves around three different studies on the same geometry, with different boundary conditions. I would like to access results from the second study to give input to some of the variables in the third study. Well, i could use the second solution as initialization to third study, but i want to access, for example, shear stress at every point in the material domain, which in turn affects inelastic deformation gradient matrix in third study.

Any one has any idea how to do this ? Any help is greatly appreciated !!

Thanks !

/Krishna
↧

Coupling Physics (Piezoelectric + Laminar Flow)

September 9, 2013, 6:43 am
$
0
0
Good day,

I am currently running a simulation of a SAW device that interacts with a fluid domain.
I have managed to simulate the SAW device with the piezoelectric physics but have yet to be able to couple it with the laminar flow physics to be able to see the interaction between the SAW produced and how it affects the fluid domain on it; as well as if the viscosity of the fluid dampens the SAW.

I have tried defining the wall velocity of the fluid as the time derivative of the displacement from the piezoelectric physics in the Global Definitions > Functions > External. But it proved futile.

Can someone teach me the right way to be able to couple these two physics?

Thanks!
↧
↧

Problems in Electromechanics in contact analysis.

March 25, 2014, 12:39 am
$
0
0
Hi,
I want to model a capacitor whose top plate is to be pulled down upon the bottom plate under a large bias voltage(above the pull-in voltage). Electromechanics physics with stationary study is employed for simulation. Contact pairs are defined on the bottom of top plate and the surface of the bottom plate. Identity pairs are also defined between the plate and the air gap.
At the first stage, only an external force is applied as the boundary load without any bias voltage for simulating the contact phenomenon. The top plate contacts to the bottom plate as expected.
Then a DC voltage is applied as the terminal condition with a ground condition, but the Segregated solver cannot converge! Because the gap is air, in the emi physics the Identity pairs could not be set asContinuity condition.
I guess that the problem is caused by the geometry of assembly. For I change to "Form union" to build the geometry, the top plate is deflected as expected. But the "assembly" geometry is necessary for contact analysis.
I don't know how to set the condition for applying electrical voltage in contact analysis. It has confused me for some days.
Would you please give me some advices? Thanks a lot!
↧

how to convert Lagrange elements to Vector(Curl) elements

April 1, 2014, 5:47 pm
$
0
0
how to convert Lagrange elements to Vector(Curl) elements


thanks a lot.
↧

File import/export procedure: Not reading whole data

April 1, 2014, 9:39 pm
$
0
0
I am importing a notepad .txt file that contains the temperature values with respect to position i.e. T(x,y,z) in 3D geometry.

I am using:
1- Global Definitions --- Interpolation --- import file
2- Structural module ---- Thermal expansion--- int1(x,y,z) as Temperature load ---- Reference Temperature= 0

However, after solving the Time-Dependent problem, when I plot the Temperature in" Results": It shows only first entry / value of temperature of the imported file i.e. no gradient at all or not matching with interpolated values plot.. When I plot the interpolation: it shows the right plotting of T(x,y,z)

Can someone tell me why it is not being imposing as gradient? Because, I am not getting the expected stresses. The result of stress value is very low than expected.
↧
Viewing all 26512 articles
Browse latest View live
Search