Latest Discussions - COMSOL Forums

Difficulty obtaining initial values in cyclic charge–discharge simulations when current collectors are included

Wed, 28 Jan 2026 10:09:10 +0000

I am currently learning lithium-ion battery simulations in COMSOL and I encountered a confusing issue related to cyclic charge–discharge conditions and current collectors.

Specifically, when I include current collectors in the model and change the study from a single charge (or charge–discharge) step to a cyclic charge–discharge condition, the solver fails to find consistent initial values. The simulation diverges very quickly (around 0.001 s) with errors indicating that the initial values cannot be obtained.

However, if I remove the current collectors while keeping all other settings the same, the model runs successfully under the same cyclic charge–discharge conditions.

For example, in the official COMSOL model “Electrode Utilization in a Large-Format Lithium-Ion Battery Pouch Cell”, the simulation works well for a single charge step. But after modifying it to cyclic charge–discharge, the model diverges almost immediately when current collectors are present. Once the current collectors are removed, the simulation can run normally again.

I have observed similar behavior in other lithium-ion battery models as well: – single charge or discharge works – cyclic charge–discharge fails only when current collectors are included

I would like to ask:

Why does including current collectors make it difficult to obtain consistent initial values under cyclic conditions?

Is this related to potential continuity, current conservation, or the coupling between electronic and ionic potentials during the transition between cycles?

Are there recommended strategies (e.g., initialization steps, solver settings, or study sequence design) to make cyclic simulations with current collectors converge reliably?

Artefacts In Total Radiated Power Spectrum Of A Piezoelectric Transducer

Tue, 27 Jan 2026 16:33:54 +0000

Hello everyone!

I have the following question/problem:

I am currently modelling a piezoelectric transducer based on the COMSOL example "Piezoelectric Tonpilz Transducer with a Prestressed Bolt". I have made some modifications to the geometry and material, but the overall concept is still the same.

I have noticed that after making some geometric changes, artefacts/spikes appear in the Total Radiated Power Spectrum (see figures). Has anyone observed similar behavior, or can anyone explain what might be causing this?

Thanks!

Best, Ronja

Cross-section average as function of position

Tue, 27 Jan 2026 02:32:09 +0000

Hi everyone,

I have a simple system where I have a freely flowing tube depositing some concentration into a porous medium surrounding that tube. The tube is straight along a single axis and has regular dimensions.

I would like to calculate (and plot) the cross-sectional average of concentration as a function of length along the pipe (C_avg vs x). I only want data for the pipe, not the surrounding medium. However, I've had some trouble in locating the correct method/function to do this.

Any help would be greatly appreciated! Thank you in advance.

Best, Carl

Circular Dependency Error: Modeling T-dependent BH Saturation in Frequency-Transient Study

Mon, 26 Jan 2026 15:08:49 +0000

Hello everyone,

I am working on an Induction Heating simulation using the Frequency-Transient study step. My goal is to model a ferromagnetic workpiece heated above its Curie temperature (Tc), capturing two nonlinear effects simultaneously:

1. Magnetic Saturation: Dependent on ∣H∣ (B-H curve).
2. Phase Transition: The drop in permeability to unity as T>Tc.

I attempted to combine these effects by creating an analytic function for relative permeability:

mu_r_combined(T, H) = (mu_r_T(T) - 1) * mu_r_H(H) + 1

mu_r_T(T): Interpolation of relative permeability vs. Temperature.
mu_r_H(H): Interpolation of relative permeability derived from B-H curve data.
The logic is to scale the saturation curve by the temperature factor, ensuring μr→1 when T>Tc.

The Problem: When I define the material property using this function (passing the magnetic field norm, e.g., mf.normH or similar, as the argument), the solver fails with a Circular Dependency Error.

Is there a standard workaround for this error in the Frequency Domain? Or is there a better way to implement "Temperature-dependent Effective BH curve" to avoid manually defining μr?

Any advice on the correct setup to handle this nonlinearity would be appreciated.

Thank you!

Magnet-Spring system fails to vibrate along Y-axis in Solid Mechanics

Fri, 23 Jan 2026 02:34:39 +0000

Hi everyone, I am simulating a haptic motor (vibrator) mechanism. The setup consists of a magnet held between two springs, with a coil positioned at the bottom.

Model Setup:

Physics: Solid Mechanics. Boundary Conditions: Fixed Constraints are applied to the ends of the springs where they connect to the frame. Loading: I applied a Y-axis force to the magnet and coil to simplify the electromagnetic interaction.

The Issue:

Despite the force applied, the magnet does not show any vibration or displacement along the Y-axis in the results.

My Questions:

I have two specific concerns: Constraint/Boundary Conditions: Is the "Fixed Constraint" at the spring ends sufficient? Spring Properties: Aside from assigning a Linear Elastic Material, do I need to define additional settings for 3D springs to behave elastically under dynamic loads?

How to select a singular point for fixed constraint boundary condition?

Tue, 20 Jan 2026 18:02:27 +0000

Hello,

I am simulating a battery model in 2D that involves multiple concentric circles and shells to build the geometry. I am using the Structural Mechanics module and seek to model the deformation in this 2D battery model.

I am trying to apply a fixed constraint boundary condition to prevent rigid body motion. Ideally, I would like to apply the fixed constraint boundary condition on a singular point on the outer shell of my 2D battery model, however I can't find a way to do so. At the minimum, I'm only able to have a quarter of my outermost shell undergo fixed constraint - not a singular point as I seek to do - and this conflicts with the underlying physics of my system.

Any input would be greatly appreciated.

Best, AU

How to define anisotropic thermal expansion coefficient tensor

Mon, 19 Jan 2026 06:51:12 +0000

Hello,

For a custom material I am defining, I wanted to know how to define the anisotropic thermal expansion coefficient tensor. More specifically, how to define the anisotropic thermal expansion coefficient tensor in cylindrical coordinates. For my project, I seek to have my material only expand radially upon an increase in temperature and I was wondering how to do that.

Any input would be greatly appreciated!

Best, AU

TCD Electrode Boundary Showing 'Not Applicable' in Aluminum–Air Battery Model

Sat, 17 Jan 2026 21:42:37 +0000

Hello everyone,

I am currently working on an Aluminum–air battery simulation using the tcd interface in COMSOL.

I am trying to define an Electrode Surface on a specific boundary (Boundary 11). However, when I select this boundary, the setting consistently shows “Not applicable”, even though I believe this boundary should physically represent the electrode–electrolyte interface.

Here is some additional context:

The model includes aluminum as the anode and an electrolyte domain with ionic transport.
Boundary 11 is located at the interface between the aluminum solid domain and the electrolyte domain.
The Electrode Surface node is added under the TCD interface, but Boundary 11 cannot be applied.

I would like to ask:

Under what conditions would an Electrode Surface boundary be marked as Not applicable in the TCD interface?
Are there any specific requirements for boundary eligibility when defining an Electrode Surface in aluminum–air battery models?

Any guidance or suggestions on how to properly diagnose and resolve this issue would be greatly appreciated.

Thank you very much for your help.

Best regards, Da-Wei Chen

Particle diameter larger than channel width still passes through in Particle Tracing – how to enforce size exclusion?

Sat, 17 Jan 2026 09:45:32 +0000

Particle diameter larger than channel width still passes through in Particle Tracing – how to enforce size exclusion?

I am using Particle Tracing for Fluid Flow in COMSOL (v5.6 and 6.3) to model transport of finite-size particles in narrow channels / porous-media-like geometries.

I have specified a particle diameter of 1 mm, while the channel (or pore throat) width is smaller than 1 mm. However, during the simulation the particle still passes through the constriction. Even when a Wall condition with Freeze is applied, the particle appears to slide through the narrow region instead of being geometrically excluded.

From what I can tell, COMSOL seems to track the particle as a point mass and does not automatically enforce hard-sphere size exclusion based on the specified diameter. I suspect the diameter is being used only for drag/forces and not for collision detection with walls.

My questions are:

Is there a built-in way in COMSOL to enforce particle–wall contact based on the particle radius, so that particles cannot pass through openings smaller than their diameter?

If not, what is the recommended approach to model size exclusion reliably (e.g. wall offsetting, modified geometry, near-wall capture forces, or other workarounds)?

Are there solver or time-stepping settings that can prevent particles from numerically “jumping” through narrow gaps?

Any guidance on best practice for modeling finite-size particle transport and exclusion in narrow channels or porous geometries would be greatly appreciated.

Thank you in advance.

Magnet Modeling Assistance

Sat, 17 Jan 2026 01:53:25 +0000

Hey y'all,

I am trying to create a simulation measuring the torques of cubic magnets in an external field, very similarly to the "Force Calculation 3 — Magnetic Torque BEM FEM" COMSOL guide. However, COMSOL uses a series of pre-defined variables in this guide, which I am not sure can be applied to mine. I have attached below my file, and am wondering 1) whether the same variables can be used (accounting for differences in surface area etc.), and 2) if not, how should I go about simulating these torques? Additionally, COMSOL does not define any specific materials in the guide, just a blank with a permeability of 1. Am I able to specify a neodymium material for the magnets or do I have to go that route? Thanks in advance!

Error while using time-dependent study

Thu, 15 Jan 2026 11:51:17 +0000

Hello, I get the error

" Complex value found in discontinuous Galerkin evaluation. -Expression:(comp1.cee.gamma_rho)*1-(0)" while using explicit Runga-Kutta method

and "BDF solver does not support discontinuous Galerkin." while using Implicit BDF solver

I am working on High Mach number flow, Laminar and Compressible Euler equations for supersonic expansion in different types of nozzles . I encounter these errors only in time dependent study. The stationary study works perfectly fine.

Is there anyway to solve these errors?

Thank you.

Periodicity (Floquet?) conditions on a nanowire-based device

Thu, 15 Jan 2026 10:33:00 +0000

Hello,

I am developing a device composed of nanowires (vertical metal cylinders) embedded in a polymer matrix, topped with metal electrodes (shown in gold in the illustration).

To facilitate calculations, I am simulating on a 2D model.

I would like to: - draw half-spheres to represent the cylinders in a top view (I will not draw the electrodes in this case) - use periodicity conditions (why not Floquet conditions?) at the half-spheres to duplicate them in one direction.

-> What periodicity conditions could I use? How are they configured?

Thank you in advance, have a great day!

Help Request: Fluid-Structure Coupling Simulation of a Silicone Soft Tentacle

Wed, 14 Jan 2026 15:15:52 +0000

Hello everyone,

For the soft tentacles of a jellyfish robot, I need to conduct fluid-structure coupling simulation using COMSOL. The fluid is water, located in the cavity of the tentacles, and the solid is silicone with a Shore hardness of 30. I am a beginner and hope someone can guide me on how to operate specifically. Thank you all!

Need help while reproducing a paper

Mon, 12 Jan 2026 13:58:20 +0000

Hello everyone, I'm reproducing a TEG (Multiphysics Modelling and Multilevel Optimization of Thermoelectric Generator for Waste Heat Recovery.) and can't seem to get past a particular checkpoint of the electric circuits part, because my voltage always shows 0, and I have to reproduce the graphs of that paper, any help would be really appreciated

Frequency study not converging

Sun, 11 Jan 2026 18:06:48 +0000

These two files have been uploaded, one is the planar layer only and other is planar with a dielectric layer.

The coil geometry analysis + stationary study works fine in both the cases, but while doing freq study with freq sweep as 0.1,0.5,1,10,25,50,75,100 KHz , the one with dielectric is not converging as the error is not going 0.

Is there any prob with the geometry pr some issue with the study.

Modeling radiative loss to the ambient environment for a system operating in vacuum conditions

Sun, 11 Jan 2026 17:48:54 +0000

Hi,

I am modeling the temperature distribution in a Hall Thruster, to be operated in vacuum conditions (either in space or in a vac chamber). I am using Surface-to-Surface radiation to model radiative transfer between various thruster interfaces. However, the thruster has a radiator that is meant to radiate heat into the ambient environment. My question is, does S2S radiation account for this or do I need to separately define Surface-to-Ambient radiation at the external boundaries of the thruster?

Thank you in advance :-)

Thermal expansion for cryostat model

Sun, 11 Jan 2026 17:44:09 +0000

Hi there,

I am new to using COMSOL and I am setting up a simulation to test the cooling capabilities of my cryostat. I have built up the model slowly, starting with a basic simulation to model the cooling power of my cryostat then making it more advanced to incorporate radiation and other sources of loss.

The final step that I am trying to incorporate into the model is thermal expansion specifically to test whether our sample will contract away from some surfaces due to thermal expansion. I have run an initial simulation however the results showed that every part of my model was connected together so they all contracted together, rather then showing whether the sample contracts away from other parts. From looking online, I think the issue is that I used "form union" rather than "form assembly". I have tried again to run the simulation using form assembly however it is not progressing past t = 0 in the time dependant study.

I am wanting help to set up this simulation properly to incorporate thermal expansion of the parts, and allow them to contract away from each other. Since the simulation is working properly when I use "form union" I am wanting to learn how to change the model to the "form assembly" situation. I am also curious to know if there is a way I can see what is causing my simulation to stall at t = 0 without waiting for hours to get an error message.

I am using COMSOL 6.3 on Ubuntu 24.04. I am happy to provide as much detail about the system as needed.

Material data ordering in Solid Mechanics simulations: Voigt vs. Standard

Fri, 09 Jan 2026 20:56:29 +0000

I'm simulating mechanical resonators made of anisotropic materials and am struggling to understand COMSOL's expected input for the material's elasticity tensor.

Under the Linear Elastic Material node, there's a dropdown to choose "Material data ordering" and can be toggled to select either "Standard" or "Voigt". Under the Materials node, you can either specify a material's "Elasticity matrix", or its "Elasticity matrix, Voigt notation."

I ran the same study on the same component 4 times with all possible permutations of the toggles above, and I'm not understanding how to interpret the results.

Study 1: "Standard" ordering and "Elasticity matrix." Study 2: "Standard" ordering and "Elasticity matrix, Voigt notation." Study 3: "Voigt" ordering and "Elasticity matrix." Study 4: "Voigt" ordering and "Elasticity matrix, Voigt notation."

Studies 1 and 3 gave identical results, while studies 2 and 4 also gave identical results. This makes it seem like the "Material data ordering" toggle has no effect on simulations, so what does it do? I understand that changing between "Elasticity matrix" and "Elasticity matrix, Voigt notation" while inputting the same tensor values changes the order of the shear terms, so seeing different results between those studies was expected.

In each case, I set the elasticity matrix to be assigned from the material rather than choosing "User defined." Is the toggle really only necessary when inputting "User defined" material properties?

I compared these 4 studies with a study where the "Material symmetry" was set to "Crystal" and inputted the c_ij values for the elasticity tensor (in Voigt notation); and this study gave identical results to studies 2 and 4.

What is the "proper" way to set the "Material data ordering" and input elasticity tensors in these simulations?

Thanks!

Running COMSOL server on Linux cluster, LiveLink for MATLAB on local Windows machine

Fri, 09 Jan 2026 10:08:10 +0000

Dear all, my system is configured as follow:

I work on a windows laptop, I access a UNIX workstation via native windows Remote Desktop. I do not have access to the folders there from windows (I tried to map the workstation to windows with popd and net, with no success, maybe it would work installing Samba on the workstation, but I cannot). Can I access the files on the workstation with a Comsol Livelink Matlab command?

I tried as follows:

on the workstation I start the comsol server

 ./comsol mphserver -user myself -passwd mypwd

on the laptop I start matlab&Livelink and I connect to the comsol server
```
 mphstart("the.correct.ip", thecorrectport, "myself", "mypwd")
```

I can mphload a model from the local laptop, for example with

model=mphload('myModelwithResults.mph');

and for example re run solutions on the comsol server, but I would like to have the model located on the workstation and loaded it from there, for example

model=mphload('//the.correct.ip/home/myself/myModelwithResults.mph');

but it does not work. What is the correct syntax for mphload (and then I assume is the same for mphsave, I would like to be able to save the model on the workstation, too).

I can run matlab directly on the workstation, but I am interested in the specific configuration I mention here.

Thanks,

How to couple the built-in PID controller with the induction heating?

Thu, 08 Jan 2026 22:30:53 +0000

Hi All,

I am trying to add the built-in PID controller to my induction heating system. I chose a domain point probe and monitored this probe's temperature as my PID controller's input. Then, I got an output whose unit is "volt" as the input of the induction coil. However, no matter what the target temperature that I set, the model always cut off the output at the very first beginning no matter I set "big" or "small" time step.

I also checked with the convergence plot and found that it didn't seem to be converged over time. I also downloaded the COMSOL's official induction heating demo and tried to add the PID controller into it and replace the input of the coil with the output of the controller, but it failed too.

I really have no idea how to address it and really appreciate any suggestions you give to me.

Thanks.