Latest Discussions - COMSOL Forums

Convergence issue using Richard's equation

Tue, 02 Sep 2025 16:48:09 +0000

Hi everyone,

I am working on a hydraulic model. The geometry of my model is shown in the attached Figure. I am using Richard's equation. There are host rock that occupies most of the model, a fault shown as the narrow pipe, and other formations at the top right.

The general water flow process is that the inflow is through the top boundary with topography and outflow is through right boundary and top boundary when water table intercepts the surface, which means that pressure head at the top is higher than zero.

For boundary conditions, a constant precipitation was applied at the top boundary with topography, which is in green in the figure. A pressure head of 0 Pa was applied at the top of fault. For the rest top part, a previous layer was applied and it will be applied a big conductance when pressure head is higher than zero. Otherwise, a conductance of 0 will be applied.

The main object of these settings was to ensure that the location of the water table can be dynamic determined by the precipitation rate and permeability instead of applying a constant head at the top.

However, when I use Time-dependent solver for cases that don't have a steady state, the solver does not converge/converges slowly. The solving progress usually stays at 4% and does not move on. If anyone has any insights about how to make it converge or improve my model, i would really appreciate it! Please let me know if more information is needed. Thank you!

Best, Zihao

Implementing Direction-Dependent Etching Velocity in COMSOL Deformed Geometry

Sat, 30 Aug 2025 16:29:07 +0000

Hi everyone,

I am trying to reproduce the chemical etching tutorial from COMSOL (link: https://www.comsol.com/model/chemical-etching-44481 ). In the tutorial, they used Prescribed Normal Mesh Velocity in Deformed Geometry to describe the etching velocity, where the velocity value is constant. In my case, I would like to define a non-uniform etching velocity, meaning that the velocity in the x-direction is different from the velocity in the y-direction. I tried some approaches, but they didn’t work. Could you please give me some suggestions on how to implement this?

Thank you so much!

Color legend

Sat, 30 Aug 2025 03:36:53 +0000

I wanted to change the height and position of the color legend, but I couldn't find it

Undefined Value in Temperature Field (NaN/Inf in comp1.T)

Fri, 29 Aug 2025 10:01:39 +0000

I am currently working on a numerical simulation of a plasma torch using COMSOL Multiphysics. The goal of the simulation is to model the behavior of a plasma jet generated by an electric arc discharge between an anode and a cathode. As part of the setup, argon gas is injected through the torch, which becomes ionized due to the high-temperature environment created by the electric discharge. This ionized gas forms a plasma jet that exits the torch and interacts with an open boundary, representing ambient surroundings or downstream flow.

The simulation includes the following coupled physics:

Electric currents to model the arc discharge.

Laminar or turbulent fluid flow for argon gas transport.

Heat transfer to capture thermal effects and plasma temperature distribution.

Open boundary conditions to allow outflow of plasma without artificial reflections.

However, when I run the model using the Stationary Solver, I encounter the following error message:

Feature: Stationary Solver 1 (sol1/s1) Undefined value found. Detail: Undefined value found in the equation residual vector. There are 2590 degrees of freedom giving NaN/Inf in the vector for the variable comp1.T. At coordinates: (0.0176092, 0.298922), (0.0143478, 0.3), (0.0163046, 0.298921), ...

Extracting data in 3d format

Thu, 28 Aug 2025 15:59:56 +0000

So far, I have been extracting abs(emwHz) values for given parameter in a 2D cutplane in my loop antenna model. I would like to extract it in a 3D grid format.

However, most methods of trying a grid 3D dataset has resulted in errors, where Hz is an undefined variable, like as shown below.

"Undefined variable. - Variable: comp1.emw.Hz - Global scope Failed to evaluate expression. - Expression: comp1.emw.Hz"

Is there a way to overcome this error, or if not, to obtain Hz values in a 3d format?

LiveLink for Creo is very slow

Thu, 28 Aug 2025 09:24:57 +0000

It seems that the time required to import a model from PTC Creo to COMSOL increases exponentially with the number of components.

At some amount of surfaces, or solids, or boundaries, or whatever - I have not managed to figure out what exactly triggers the slowdown - the time required seems to go to infinity.

Looking at the status bar a ticker walks leisurely like "generating associativity for entity 7734 of 12995" and sometimes it just stops at "processing selections" and waits for half an hour. Or fails entirely after prolonged waiting. Which is odd since I have no selections defined in Creo.

Importing the same model as a STEP is massively faster, but then I lose associativity with all the rest of the features, boundary loads, shell thicknesses, et cetera - and have to rebuild it every time.

Can this be improved? I have no high hopes, since I suspect it is a consequence of the Java interface to Creo - which is why I post it here as a concern and suggestion rather than a support issue.

High aspect ratio

Thu, 28 Aug 2025 08:58:07 +0000

I have a fracture that is represented in terms of a layer in porous media of domain size 15001500500m and the fracture is situated in center. I'm not able to create proper mesh. Also, the mesh that i have created, the solution is not converging for it.

Particle tracing module

Wed, 27 Aug 2025 08:08:14 +0000

Can particle position be referenced as an independent variable in the electrostatic module?

Simulating Electrocapillary effect

Tue, 26 Aug 2025 15:24:28 +0000

I am trying to simulate droplet motion under Electrocapillary Force. I have two electrolytes, one as the droplet and the other as the continuous phase. I tried to couple the TDS and ES physics to form the EDL, but the droplet motion does not occur. Do I need to define surface tension based on the Lippmann equation as well, or is creating the EDL alone sufficient to change the surface tension?I am using a 2D geometry and applying a DC voltage. I have set velocity inlet and pressure outlet as boundary conditions, but at the inlet the velocity is 0.

I define space charge density in each species as : F_Const( C_cation-C_anion).

Trial Passcode

Tue, 26 Aug 2025 14:33:18 +0000

Hi guys,

I am trying to download a trial version but its asking the Passcode which I don't have, can any one please help me with this.

Unable to recognize the stress tensor component variable in the laminar flow module.

Tue, 26 Aug 2025 09:11:32 +0000

After adding laminar flow and 'Free and Porous Media Flow Coupling'modules, an error occurs during calculation. However, expressions and variables that mention errors exist in the laminar flow module

SAW Delay Line Simulation with Large Number of IDT Pairs

Mon, 25 Aug 2025 14:09:03 +0000

Hello,

I am trying to simulate a SAW delay line device in COMSOL with 100 IDT pairs on each side (input and output). Since modeling the full geometry with all these pairs would be very large and computationally expensive, I tried to simplify the model.

Currently, I built the geometry with only one IDT pair and applied periodic boundary conditions on both lateral sides of the device, assuming that this would mimic the effect of multiple pairs.

However, I am not sure if this approach is physically correct for a delay line with so many electrodes. Also, the results I am getting contain a lot of noise, which makes me wonder if my simplification is not valid.

My questions are:

Is it physically correct to model only one IDT pair with periodic conditions to represent many pairs in a SAW delay line?

Are there better approaches in COMSOL to simulate a large number of IDTs (e.g., 100 pairs per side) without building the full geometry explicitly?

Could the noise I observe be related to this simplification?

Any guidance or references would be very helpful.

Thank you!

Issue: Inconsistency Between Maxwell Stress Tensor and Lorentz Force Computations

Mon, 25 Aug 2025 07:57:38 +0000

I would like to confirm, theoretically, computing the force acting on a wire using the Lorentz force method, the surface integration of the Maxwell stress tensor (MST) over a control volume enclosing the object of interest, and the surface integration of the MST directly over the object’s surface should all yield the same result, correct?

I simulated two cases:

Solenoid

Under a non-uniform background magnetic field, the three methods (Lorentz force, MST over a cuboid control volume, and MST over the wire surface) produced consistent results.

However, under a uniform background magnetic field, the computed results from these three methods do not agree with each other.

Straight wire

For both non-uniform and uniform background magnetic fields, the Lorentz force method and the MST over the wire surface gave nearly identical results.

But the MST over a control volume surface did not match the other two methods.

I would like to confirm whether this discrepancy is expected (e.g., due to numerical or formulation differences) or other factors ? I have reduced the size of mesh but still does not seem to work. Thank you

Lorentz Force Contribution Feature Issue

Mon, 25 Aug 2025 04:22:13 +0000

Hi, I would like to ask a question regarding the Lorentz force contribution feature in COMSOL. Theoretically, for a solenoid placed in a uniform background magnetic field, the net force acting on the wire should be zero. However, in my simulation, the computed force is not exactly zero; it typically ranges from about 10^-8 to 10^-10 N. Even after refining the mesh (by reducing the maximum element size), the force does not converge to zero. Furthermore, when I increase the number of turns and the current, the magnitude of the force also increases rather than converging to zero.

I would like to confirm whether this behavior is expected (e.g., due to numerical errors, discretization, or modeling limitations), or if it suggests a possible issue with my setup. Thank you

Grounding / Earthing system

Sun, 24 Aug 2025 19:06:01 +0000

Hello everyone.

I want to simulate a grounding/earthing system. I started with a simple grid. As usual to grounding/earthing simulation softwares I applied a current to a line segment. The problem is that the current doesn't flow from one segment to another. All line segments are apparently isolated. Line segments were imported from autocad, but also tried to draw the lines in COMSOL or to draw a square polygon. The result was the same. I tried "union" geometry and "convert to curve". Nothing works. I appreciate suggestions to work around this problem.

About "Generate New Turbulence Model Interface"

Sun, 24 Aug 2025 06:25:17 +0000

In the User's Guide doc, the "Generate New Turbulence Model Interface" is designed to "create a new turbulent flow interface based on the current laminar or turbulent flow interface." And "Note that the parameters and expressions in text fields in all the features, except for Initial Values and the Name column in Global Equation, get exactly the same expressions as in the original model. " What is the difference if someone adds a new turbulent interface and inherent the old solution in the study node? What is the purpose for this interface? If I want to add an source term in the k-epsilon equation, what should be done?

Calculation of force on only a magnet in a air domain

Sat, 23 Aug 2025 05:04:05 +0000

I’m setting up a simple simulation where a permanent magnet is attached to the tip of a non-magnetic beam. I want to measure the beam’s deflection due to magnetic force. Later, I’ll add a coil so the force comes from the coil–magnet interaction. For now, since there’s no other magnetic material or external field in the model, the net magnetic force on the magnet should be ~zero, correct

Issue with components disappearing after running simulation

Fri, 22 Aug 2025 18:08:54 +0000

Hello. I am relatively new to COMSOL. I am trying to model a heat pipe simulation.

When I click 'build all,' I can see all of my components. When I compute the model however, a few components are missing. Does anyone know what could be the case?

Everything is assigned to a material. Attached are screenshots and my .mph file.

Exporting Node Coordinates, Data, Node_id, and Connectivity from a Target Surface

Fri, 22 Aug 2025 17:26:45 +0000

Hi,

I would like to ask whether it is possible to simultaneously export the node coordinates, associated data, node IDs, and the connectivity information of nodes on a specific surface in COMSOL. Could you please advise if this can be achieved, and if so, how?

I already know COMSOL can output the node coordinates, associated data at the same time.

Thank you very much for your kind assistance.

WCL

How can I get only scattered field without background field in Electromagnetic Waves, Transient

Thu, 21 Aug 2025 21:53:36 +0000

Hello, everyone! I tried to reproduce a FDTD article in COMSOL. I would like to introduce the model briefly: I couple temw with general form PDE to describe the dynamic in material. I set a Gaussian incident pulse in SBC and make it illuminate on a metasurface.

I set a probe to get the output signal. There are two signals caused by the nonlinear PDE. I want to observe: (1) THz signal(2)400THz. I can get 400THz signal now, but I can't get THz signal. It may be covered by the background noise from Gauss pulse, and I want to eliminate it now.

However, TS/SF technique used in the FDTD article which can help them get scattered field directly without background field doesn't exist in COMSOL. And background/scattered field technique only exists in Electromagnetic Waves, Frequency domain. So what I can do in Electromagnetic Waves, Transient now?