Latest Discussions - COMSOL Forums

How to calculate the co and cross polarized transmittance/reflectance in Comsol?

Thu, 09 May 2024 04:27:11 +0000

In Comsol, I want to calculate the Txx(transmittance to x polarized for incident x polarized), Txy(transmittance to x polarized for incident y polarized), Rxx, Rxy, and so on in comsol. S21/S11 never gives info about the particular polarization, it gives the overall transmittance and reflectance. But how to find these components in Comsol?

Disabling variables in study step

Wed, 08 May 2024 20:18:16 +0000

I am working on a heat transfer problem that requires me to calculate a variable (let's call it X, defined in "Variables 1") as the first step. X is defined as 2Y - 10, where Y is a state variable (defined in "State Variables 1") that is being updated after each timestep. In the second study step, I need to use the last computed value of the variable X in the physics and solve again. Which means that the variables should not be solved for in the second step.

I am implementing a two study scheme, where in study 1, I run the physics while computing X and Y. In study 2, I am disabling "Variables 1" and "State Variables 1", and setting the values of the variables not computed for through the previous solution. When I run this, the first study runs successfully, but the second study throws an error saying variable X is not defined. I understand that this might be because I am disabling the variables in the study, but I am not sure how else to not solve for the variable and state variable in the second study, while still using the last stored values.

Would be of great help if anyone can guide me to implement it properly.

Thanks!

Failed to evaluate AWE expression. Not all parameter steps returned.

Wed, 08 May 2024 20:05:33 +0000

Hello,

I am getting the error "Failed to evaluate AWE expression. Not all parameter steps returned." when running an Adaptive Frequency Sweep study for an Electromagnetic Waves, Frequency Domain physics. The model is a waveguide antenna with two lumped ports in a coaxial configuration. I run a "Frequency Domain Study" and it converges with no problem. But when I run "Adaptive Frequency Sweep" from 3 GHz to 4 GHz, in order to plot S11 parameters vs frequency, solution does not converge. I tried both Direct solving with MUMPs and Sugessted Iterative Solver, but they both did not converge. I am confused why this is the case since the "Frequency Domain" problem converges without a problem. Is it possible adaptive frequency sweep doesn't work properly when you have more than one excited lumped port?

I can't attach my file as it is too large but I have attached a screen shot.

Any information would be greatly appreciated.

Rapid solidification

Wed, 08 May 2024 19:16:08 +0000

Apologies, as I am fairly new to COMSOL.

I am trying to simulate using a pulsed laser to melt and recrystallize silicon. Right now, I am just trying to model the rapid solidification after the melting has been done. I am doing a 2D simulation with a 10 um x 10 um piece of Si using heat transfer in solids and liquids with a time-dependent study. I am starting with a 100 nm thick layer of liquid Si at the top of the sample. For the sides, I have periodic boundary conditions, and the bottom is held at constant temperature. I also have surface-to-ambient radiation on the top surface. I give it an initial temperature distribution where the interface is at the melting point of Si, and then there is a linear temp distribution in the solid that goes from the melting temp to room temp. In the liquid layer, I set the top surface to be at a temperature higher than the melting temp and again use a linear temp distribution in that layer.

What I am ultimately trying to get is the position of the liquid/solid interface as a function of time. Right now, it is working to the point that I can see the interface move towards the top surface with time, but the simulation gets an error at one point that is due to the mesh. Then it stops the simulation at that point, but it does give me all the data for the time points before that happens. I believe that when the liquid part is completely solidified, it has zero thickness, so when it tries to mesh, it can't make an appropriate mesh element, which is why it gives the error.

I have two questions:

Is there a way to tell the simulation to stop once the top layer is fully solidified so that I don't get this error?
How can I extract the position of the interface as a function of time?

How to input a no rotation constraint

Wed, 08 May 2024 17:47:45 +0000

Im working on the same system I need to put a no rotation constrant on the top face of the lower flexure.

I added the image of the location of the no rotation constrant and a pdf of the system. In addtion I attach the mph file of the system.

What i did was I use a rigid connector at the location and constrant it rotation however I suspect it does not work correctly because in the first 2 mode the flexure is not bended correctly and the system was not rigid.

If there an another way to put a no rotation constrant on the solid mechanics interface that will be perfect.

Any help will be very useful.

Problem with magnetic field and solid mechanics

Wed, 08 May 2024 17:09:19 +0000

I’m trying to build a three-phase transformer to predict the sound in the core. But Im having problems computing my electromagnetic model so Im wondering if anyone could take time and take a look at my solid mechanics and the magnetic field? Thanks.

How to place a Point mass on COMSOL

Wed, 08 May 2024 16:56:23 +0000

Good morning everyone,

Im currently working on a Eigenfrequency Prestressed study.

I want to put a point mass at the top face of the lower flexure (see the image below)

I attached another image that show where i want the point mass (see the image with the blue arrow).

For referance I attach the mph.file for the system that is used for the simulation and a image of what it look like.

Before I put a point mass I use a cube with a very high density with interia and placed it where arrow is at but it won't be useful for long term research, and before I use a point load but it have no mass of interia.

An idea that I have is to use a added mass load but it only for boundary but I am not sure it a correct substitute for the point mass.

Any help or advice much appreciated thank you.

In the solid mechanics module, what is the difference between boundary load and external stress?

Wed, 08 May 2024 15:39:01 +0000

I am exploring the solid mechanics module. My geometry is a simple square in 2D axisymmetry. I am using the nonlinear materials node.

The equations of the boundary load node says

Where, as far as I understand, S is the second Piola-Kirschoff stress and n is the column array containing the unit direction vectors. So the total stress tensor holds the contribution from the boundary load.

Now, looking at the equations for nonlinear elastic material, firstly, inelastic stress is calculated as

where the tensor can be defined in the external stress subnode. So the inelastic stress tensor holds the contribution from the external stress subnode. But again, total stress is calculated as

Then the total stress tensor also holds the contribution from the external stress subnode.

So the total stress tensor can hold the contributions from BOTH the boundary load node AND the external stress subnode. So it seems to me like the boundary load node and the external stress subnode are just two different approaches to achieving the same effect mathematically. I can add a certain amount of stress to my geometry in both ways.

But surely, there must be a reason why they are implemented separately. Either I am wrong in thinking that they can both achieve the same numerical effect on the problem and they are actually totally different. Or even if the same numerical effect can be achieved in both ways, the two approaches might be conceptually different.

I want to know how they are different and when one should use the external stress subnode rather than the boundary load node to apply stress. Can someone please explain? Thank you.

Change Thickness in 2D models

Wed, 08 May 2024 15:19:10 +0000

See my attached drawing for a visual description of my question: I have a simple 2D model of two rectangles. Lets say I set a thickness in Solid Mechanics (1m) but then use Change Thickness to change the blue rectangle to 1.5m. Will this add 0.25m to the thickness into the plane and 0.25m to the thickness out of the plane, thus keeping the orange rectangle centered in the out-of-plane dimension (figure A)? or will it add 0.5m to the thickness out of the plane so that the backside of both rectangles are aligned (figure B)?

Boundary condition for material layering in IGBT module to analyse thermal stress

Wed, 08 May 2024 14:16:12 +0000

Hi, I am making a model of an IGBT module. Currently I have used the Heat transfer in solids, but I want to add thermal stress, which automatically adds Solid mechanics interface. Here the default boudary is "free", but I am unsure wheter or not I have to change this to something else in between the layers. Does COMSOL automatically account for the stress between two types of different materials?

Coil problem

Wed, 08 May 2024 11:12:08 +0000

Good afternoon, I would like to know why when I use the magnetic and electric field, it does not compute and also, why when I use the magnetic field, my field is so big. Thank you in advance.

BR, Juan Carlos Server

Complex 2D geometry from rigid and elastic beams with hinges in multibody dynamics

Wed, 08 May 2024 09:15:24 +0000

I am trying to set up a 2D system of maybe more than 100 individual beams, some of them rigid some of them elastic. The indidivual beams should be interconnected by hinges. And some simple load cases are to be studied. I am creating the geometry with lines as beams and also everything else in matlab.

The problem is, to keep track which parts of the geometry are supposed to be rigid and which are elastic. If I only create one geometry, It is quite hard to track later which edge number belongs to which kind of beam as the sequence of edge numbers of the assembly is not connected to the geometry parts. Same for the vertices. If I create multiple geometries, the coupling in the multibody dynamics does not seem to work (I get a random Java error when trying to run a simulation). In the GUI I'm not even able to create multiple geometries.

Is there any way of setting up a simulation like this and keep an overview? I'm just not sure how to divide the different kinds of beams and also if I can bundle several beams in one physics beam part.

Low quality mesh is generated when Meshing a Thin film on a thick metal

Wed, 08 May 2024 06:49:23 +0000

My model contains a block of 8mm*8mm*0.5mm copper block and coated with a 8mm*8mm*300nm Niobium. I am unable to mesh the model with air boundary. This model was meshed properly with no boundary conditions. Can you help me with a solution to mesh the block with air boundary condition.

Resource for COMSOL Tutorials

Wed, 08 May 2024 06:05:30 +0000

Hello Everyone,

I have previously worked in COMSOL for 3 years. I have recently started a Youtube channel on COMSOL tutorials to share what I have learnt, with others.

The name of my channel is "COMSOL Multiphysics for Dummies"

I would love to know your interests and possibly make tutorials on them. Let me know your feedback.

I would also like to know the thoughts of my former COMSOL colleagues on the kind of tutorials which would be useful to the users.

Regards, Rahul Bhat

How to perturb the solution of previous step when using parametric sweep?

Wed, 08 May 2024 04:08:34 +0000

Hi there,

I am trying to combine the time-dependent solver and parametric sweep to model the evolution of a system from trivial configuration to buckling and post-buckling configuration. The overall framework is like discussed in this thread.

Because the trivial configuration is an equlibrium one though becomes unstable beyond the critical point, the solver is stuck to the trivial solution when the 'reuse solution from previous step' under the parametric sweep node is checked. See another tread for the example, where only the trivial solution can be obtained even far beyond the critical point.

I think a perturbation should be added to the pervious solution before proceeding with the current step. So, how to implement this in COMSOL?

Best regards,

HC L

Continue solution in multistep study

Wed, 08 May 2024 02:38:13 +0000

When a solution is stopped by the user one can continue the solution from where it stopped. However in a multistep study, this approach seems to be wrong in that it restarts from the last solution of the last study step, rather than where it actually stopped in the middle of the current study step?

Fermi Level Determination

Tue, 07 May 2024 16:18:24 +0000

Is there a semiconductor physics module in COMSOL which correctly describes the Fermi level position for any bandgap system?? I mean is it true to say that COMSOL does not explicitly account for the complex evolution of the Fermi level??

How to draw a coaxial closed pipe

Tue, 07 May 2024 14:05:09 +0000

The below post is related to an archived discussion

Hello,

I am trying to draw a 1D coaxial closed pipe geometry that consists of lines and arcs (portion of a circle). From my understanding, using the Bézier Polygon would be appropriate for this.

However, I am having a hard time figuring out how to define the arcs. The cubic Bézier node in COMSOL requires four control points and four weights as inputs. For the control points I put in the two endpoints 1 and 4 coordinates values (0,0,0) and (0,0,L), as for the two control points 2 and 3 corrdinate values (0,0,1) and (0,0,-1). For the weight point i left the defaults values of 1. Can anyone help validate if correct. Thank you for your help.

Initial Values for Magnetic Vector Potential

Tue, 07 May 2024 13:05:58 +0000

Hi,

I want to perform a time-domain study of wave propagation in a WG (input wave frequency is sweeped from a Fmin to a Fmax in a given time interval). I followed what reported on this discussion:

https://www.comsol.com/forum/thread/121532/Plane-Wave-and-Excitation-in-Time-Domain-Analysis-

My problem is that I know the Electric field of my incident wave (TE10 field), but also the magnetic vector potential of the incident wave is requested. How can I solve this problem?

Modeling a three phase transformer

Tue, 07 May 2024 08:35:21 +0000

I'm thinking when a magnetic sheet is magnetized, it generates a volume force within it. This phenomenon is known as magnetostriction. The force stimulates a strain in the solid, causing it to vibrate at the same frequency as the input signal. If my settings are correct which i don't know and cant find anywhere how to do, I would be able to observe the displacement field in the output which i don't and cant fix myself.

In the final stage, I wish to simulate noise, then I'm thinking to convert the transient solution of displacement to the frequency domain using an FFT solver. Then, i would use the acoustics modulus and harmonic source of displacement from FFT to simulate the acoustics noise in the frequency domain. Like surface normal acceleration as the source of the noise.

Also Trying to write an expression such as: u2 - solid.disp. That will store the diplacement variable from Solid Mechanics (sm) into u2. That will solve the (bode) along with the sm in Time dependent study. Then run the Time to frequency FFT, which uses the values stored in u2 and computes the FFT. So I can use the result as Surface normal acceleration input to Acoustics interface. But I'm getting inconsistent unit, why? Thanks in advance!

Also thinking something is wrong with my magnetic field which i cant understand eaither becouse when i remove magnetostriction it works instantly, but i have only choosen the core in amperes law so it should work right?

Placing my model here so if someone could take a look thru it, it would be so appreciated.