Latest Discussions - COMSOL Forums

Inverted mesh elements during snap-back of a snap-through structure with stiff shutters

Sat, 03 Oct 2026 17:23:12 +0000

Hello, I am modeling a 2D plane-strain transient thermo-mechanical snap-through structure in COMSOL Multiphysics 6.3. The structure consists of a curved soft beam with a thermally contracting active layer. Heating causes the beam to snap through. Two long and relatively stiff shutters are attached to small soft silicone pads/protrusions on the beam. After snap-through, the shutters block the incident light, the structure cools, and eventually snaps back. The main problem occurs during snap-back. The computation becomes extremely slow, the time step decreases dramatically, and COMSOL reports inverted mesh elements. The inverted elements are consistently located near the connection between the stiff shutter base and the soft pad underneath it. The soft beam and pads are modeled as silicone with a Young's modulus of approximately 2 MPa. The shutter is much stiffer and relatively thin. Geometric nonlinearity is enabled and a time-dependent solver with Generalized-alpha time stepping is used. I have tried several approaches: 1. Assembly + identity pair + compliant thin layer between the pad and shutter. I also reduced the shutter Young's modulus to approximately 0.5 GPa. This improved convergence and prevented immediate mesh inversion, but the computation became extremely slow. About two snap-through/snap-back cycles required approximately 21 hours. 2. Different geometries, including attaching the shutter directly to the beam and attaching it through a small protrusion/pad. The problem still occurs near the shutter root. 3. Local mesh refinement and different mesh configurations near the shutter/pad interface. 4. Reducing the maximum time step around the snap-through and snap-back events. 5. Rigid/stiff shutter modeling and rigid-flexible connections. Large local deformation still develops in the soft material near the shutter root during snap-back. My main questions are: - Is this inverted-element problem more likely caused by the geometry/interface design, the mesh, or the nonlinear solver/time stepping? - What is the recommended way in COMSOL to connect a very stiff and thin shutter to a very soft silicone substrate undergoing large deformation? - Would a rigid domain with a compliant connection/thin elastic layer be more appropriate? - Would automatic remeshing or an ALE approach be suitable for this type of snap-through/snap-back problem? - Are quadrilateral elements or another local meshing strategy preferable to triangular elements in the highly deformed shutter-root region? - Are there solver settings that can handle this instability more efficiently without forcing the time step to become extremely small? I have attached figures showing the overall structure and the location of the inverted elements during snap-back. Any suggestions on the geometry, interface modeling, mesh, or nonlinear solver settings would be greatly appreciated. Thank you.

Piezoelectric Ultrasonic Cleaning Tank: Request for COMSOL Modeling Advice

Sat, 03 Oct 2026 09:27:16 +0000

I am currently studying a piezoelectric ultrasonic cleaning tank operating at approximately 100 kHz. I would like to use COMSOL Multiphysics to establish a fully coupled piezoelectric-structural-acoustic model. However, I have encountered some difficulties in setting up the model. So far, I have only found examples involving ultrasonic wave propagation in air, and I have not been able to find a case that closely matches my application. I would be very grateful for any help or suggestions. If anyone could share a similar COMSOL .mph example model, relevant paper, tutorial, or modeling recommendation, I would greatly appreciate it. The model mainly consists of the following components: （1）A circular PZT-4 piezoelectric ceramic disk with a diameter of 38 mm; （2）The PZT thickness is the design variable, and the current optimized value is approximately 4.130 mm; （3）The PZT disk is bonded to the outer surface of the bottom of a cylindrical 316L stainless-steel tank; （4）The thickness of the 316L stainless-steel tank bottom is approximately 0.5 mm; （5）The medium inside the tank is water; （6）The PZT is polarized through its thickness, and a harmonic voltage is applied across its top and bottom electrodes; （7）The target operating frequency is approximately 100 kHz.

Changing quadrature rules/settings

Tue, 29 Sep 2026 12:47:49 +0000

Hello,

I was wondering whether there was any update on this previous post. Is it possible to change the quadrature settings of elements, either for the entire geometry or specific domains, to e.g. Gauss-Lobatto or Newton-Cotes instead of standard Gauss quadrature?

After some searching, I found that there is an option 'Global Definitions/Equation Contribution/Integration Rule', where it is possible to manually set integration points and corresponding weights. However, it is unclear to me whether that now applies to my whole geometry automatically.

Any information or help is highly appreciated.

/Timotheos

The below post is related to an archived discussion

Can Pressure Acoustics represent shear wave propagation in a fluid-filled cavity coupled to a poroelastic solid? (COMSOL 6.4)

Mon, 28 Sep 2026 12:34:58 +0000

Hi,

I'm modeling a 3D block of poroelastic material (Poroelastic Waves interface) with two rectangular cavities inside it, and I want to fill the cavities with a fluid that simulates water. The cavities are supposed to be inside the solid block.

Current setup Solid block and PML: Poroelastic Waves (pelw). Fluid cavities: Pressure Acoustics, Frequency Domain (acpr), coupled to the solid through an Acoustic–Porous Boundary multiphysics coupling on the cavity walls. Excitation: Prescribed Displacement on the top and bottom faces, in the x-direction only (u0). The faces are horizontal, so the motion is tangential to them, which makes it a shear-type excitation.

My concern My goal is to see how a shear wave propagates through the solid and into the fluid. Pressure Acoustics is an inviscid formulation with a scalar pressure field, so as far as I understand it only supports compressional waves and only responds to the normal component of boundary motion. Since my shaker motion is tangential to the top face, I suspect the fluid is only excited indirectly, through the cavity walls, and that Pressure Acoustics cannot carry shear waves inside the fluid at all.

Questions

Is Pressure Acoustics the right choice for the fluid if the physics I care about is shear-wave propagation, or is it fundamentally unable to represent it? If it isn't, what is the recommended approach for the fluid regions? For example: Thermoviscous Acoustics (to include viscous effects at the interfaces), Linear Elastic / Viscoelastic material with a very low shear modulus, if the fluid is gel-like, a fully coupled fluid–structure formulation. Whichever interface is used, which multiphysics coupling should connect it to the Poroelastic Waves domain? Is it physically reasonable to expect meaningful shear-wave transmission into a fluid, or should I expect it to be strongly attenuated or not transmitted at all?

I'd appreciate any guidance, or a pointer to an application in the Application Library that treats something similar. I can attach the .mph file and screenshots of the geometry and physics tree if that helps.

Thank you!

Unable to read or write out-of-core file

Tue, 22 Sep 2026 12:35:00 +0000

I have received the error:
" - Feature: Eigenvalue Solver 1 (sol1/e1) Unable to read or write out-of-core file."

Any idea what this means? I have TBs of disk space so I don't think that should be a problem.

Transient SST initialization fails on refined mesh / higher Reynolds number

Tue, 22 Sep 2026 08:03:54 +0000

Hello,

I am modelling 3D flow around an inclined circular cylinder in water. The cylinder axis is aligned with the z-direction and periodic conditions are used in the spanwise direction.

Instead of rotating the geometry, I decompose the towing velocity into:

Un = Usin(alpha) Ua = Ucos(alpha)

where alpha is the angle between the cylinder axis and the towing direction.

I impose Un as the cross-flow velocity, and I account for the axial component using a reference-frame/Galilean transformation by applying a cylinder moving-wall velocity of approximately -Ua.

The final goal is to run sweeps over approximately:

U = 1-25 knots alpha = 1-40 deg D = 20-25 mm

The attached model contains a stationary k-epsilon study followed by a transient SST study, with the SST initialized from the k-epsilon solution.

On a coarse mesh this works well and vortex shedding develops correctly. However, when I refine the mesh or increase the Reynolds number, the transient SST often fails right at startup.

Typical errors are:

"Failed to find consistent initial values"

or the segregated solver repeatedly reduces the first timestep and is not able to converge.

The problem seems to be related to the initialization/nonlinear coupling rather than the linear solver. I have also tested PARDISO.

A stationary SST solution is not a good workaround in this case because the flow is strongly unsteady and the stationary SST problem itself does not reliably converge.

I would appreciate advice on the most robust way to initialize the transient SST solution, especially regarding:

Starting SST from a stationary k-epsilon solution.
Transferring a developed transient SST solution from a coarse mesh to a finer mesh.
Correctly handling k, omega, and the SST wall-distance variable during mesh transfer.
Whether consistent initialization should be modified or disabled in this situation.
Recommended startup settings for timestep, BDF order, segregated damping, Jacobian updates, and maximum segregated iterations.
A robust continuation strategy for future sweeps in U, alpha, and D.
My main objective is to find an initialization procedure that remains stable when the mesh is refined and when the operating conditions change.

Any suggestions would be greatly appreciated.

Defining the area of a 2d domain with an ODE

Fri, 18 Sep 2026 16:54:10 +0000

Just as the title asks. I am trying to model the pressure waves caused by the expansion and contraction of an underwater bubble-like object through a net-like solid structure as a pressure wave passes over the entire area. It is not a domain that I can simply define as, for example, a sphere where the mesh boundary could be defined via velocity or simple geometric expansion around a point. It almost is a geometric expansion around a point besides the structure that is in the way.

I have the derived the ODE that allows me to define this object via area/volume. But I am having a hard time finding a way in COMSOL to define the explicit volume of a domain directly.

https://www.comsol.com/blogs/computing-controlling-volume-cavity is the closest I can find but for my example the pressure change will not need to be calculated, it is included in my ODE. This is a problem that is fairly simple when there is no object in the way as I have read other examples of them defining the expansion of bubble like objects in terms of the wall-velocity calculated via the rayleigh-plesset equation, or even simpler as defining the radius of the domain etc. But as the expansion I want to study is by virtue of the solid object unable to act as a simple radially symmetric breathing mode for my results I would be best served by defining the volume programatically.

If anyone has any experience in defining the volume of a domain using an ODE, a PDE or just directly using another means of defining the changing volume of a domain please offer advice.

Best, Luke

swept mesh problem in branched tube

Tue, 15 Sep 2026 16:53:17 +0000

i have branched tube y shape and has hexahedral mesh done by swept mesh in horizontal tube but i have problem in creating hexahedral mesh in branched tube it is showing unsuported topology . how to mesh branched tube branched tube is at 25 cm from inlet and length of horizontal tube is 50 cm and branched tube is 25 cm

Nonsymmetric matrix

Tue, 15 Sep 2026 13:59:39 +0000

In a simulation I am running, two coils wound on separate cores separated by an air gap are connected in series (Coils 1 and 3 are in series, and Coils 2 and 4 in series as well). When I run, I see the following in my log with almost no progress:

Nonsymmetric matrix found. Scales for dependent variables: Magnetic vector potential (comp1.A): 1.7e-05 Divergence condition variable (comp1.mf.psi): 1

The physics I use is similar to that in the ecore_transformer tutorial.

Mismatch Between Bandgap Calculated by Eigenfrequency Study and Transmission Loss Spectrum from Frequency-Domain Study in the Solid Mechanics Module

Mon, 14 Sep 2026 16:12:15 +0000

I am using the Solid Mechanics module in COMSOL Multiphysics 6.3 to study in-plane elastic wave propagation in a periodic acrylic plate with holes. From the eigenfrequency study with Floquet periodic boundary conditions, the dispersion curve shows a bandgap between the 3rd and 4th bands, approximately from 133 kHz to 154 kHz.

To verify this bandgap, I performed a frequency-domain study on a finite structure with PML layers, I gave prescribed displacement of 1mm in x direction and calculated the transmission loss spectrum. However, the transmission results do not fully agree with the eigenfrequency analysis. The transmission loss is not consistently high throughout the predicted bandgap range, and noticeable attenuation also appears near 118 kHz, where no bandgap is observed in the dispersion curve.

Could you please explain the possible reason for this discrepancy? How can I determine whether the bandgap given by the eigenfrequency study is a true complete bandgap for in-plane elastic waves, and what is the recommended COMSOL procedure for validating it using a frequency-domain transmission loss analysis?

Flow electrodes modeling

Sat, 12 Sep 2026 10:39:21 +0000

I want to model a flow electrode where the fluid is a mixture of water and activated carbon with 1M Na2SO4. Can any one help me ? The two electrodes are separated with a nafion 114 membrane.

Level set variable no longer maintaining values between 0 and 1 after ~3 s

Fri, 11 Sep 2026 09:43:21 +0000

Hello,

I am using the Two-Phase Flow, Level Set approach in COMSOL for a transient simulation. The model converges successfully until the end of the simulation time. However, after approximately 3 s, the level set variable no longer maintains the expected values close to 0 and 1.

In particular, the maximum value of the level set variable gradually decreases to around 0.7–0.6, instead of remaining close to 1.

Initially, I suspected that this was related to the mesh, so I refined the mesh, but the problem still persisted. I also adjusted the parameter controlling interface thickness and the reinitialization parameter. These changes improved the behavior to some extent, but did not completely resolve the issue.

Has anyone experienced a similar problem with the Level Set method? I would appreciate any suggestions regarding mesh resolution, reinitialization, interface thickness, solver settings, or any other parameters that may help maintain the level set variable closer to its expected range.

Thanks in advance.

Comsol 6.4 on Linux w/ multiple GPUs

Tue, 08 Sep 2026 08:49:54 +0000

Hi,

I've recently been trying to make Comsol use multiple GPUs on a Linux-based supercomputer. Both CPU-only and single GPU configurations work fine.

The machine has 4 GPUs, and given the same CUDA-related variable values, Python manages to use all 4 while Comsol only uses 1. My study uses the cuDSS solver, and I've ticked the option to use multiple GPUs in the solver config menu, to no avail...

Anyone facing the same issue or having thoughts on the matter ?

Cheers

Request a Trial Passcode

Mon, 07 Sep 2026 16:59:27 +0000

Hello Team, need a trial passcode to evaluate the software for my research work in wearable sensors on Human Body.

trial passcode

Mon, 07 Sep 2026 13:28:36 +0000

Hello COMSOL team i need a trial passcode please,

Thank you

Manel A

Phase specific turbulence on Euler Euler Interphase

Fri, 04 Sep 2026 12:03:59 +0000

How can I use k epsilon turbulence model only on the continuous phase on the Euler Euler Turbulence interface.

How to extract data from Layered Shell dataset to matlab

Sat, 29 Aug 2026 06:13:05 +0000

The mpheval and mphinterp functions fail to extract localized variables (such as r_llmat1_xdim) when querying standard evaluation datasets directly in memory. I need help on how to export this data into MATLAB. lshl1_eval = mpheval(model, {'T'}, 'dataset', 'lshl1'); 错误使用 mpheval mpheval (0): Java exception occurred: Exception: com.comsol.util.exceptions.FlException: 数据集类型无效。 Messages: 数据集类型无效。 - 标记: lshl1

lshl1_eval = mpheval(model, {'T','r','r_llmat1_xdim'}, 'dataset', 'dset1', 'selection', 3, 'edim', 1);

错误使用 mpheval mpheval (0): Java exception occurred: Exception: com.comsol.util.exceptions.FlException: 无法计算表达式。 Messages: 无法计算表达式。

未定义变量。
- 变量: r_llmat1_xdim
- 几何: geom1
- 边界: 3

无法计算表达式。
- 表达式: 1000*r_llmat1_xdim
- 特征: 计算

Difference in Results Depending on the Journal Velocity Setting in Multibody Dynamics–Hydrodynamic Bearing Coupling

Thu, 20 Aug 2026 05:38:13 +0000

I am modeling a rotating shaft using Multibody Dynamics (MBD) and a Hydrodynamic Bearing (HDB).

In MBD, I applied Prescribed Rotational Motion to the shaft and specified a constant angular velocity, omega.

I then defined the journal bearing surface using the HDB interface. I tested two different settings for the Journal Velocity:

Multiphysics-coupled velocity: The Journal Velocity is automatically obtained from the MBD interface through the Multiphysics coupling.
Manually specified velocity: The Journal Velocity is directly set to the same constant angular velocity, omega.

Since the shaft is prescribed to rotate at the same angular velocity omega in both cases, I expected the HDB results to be identical. However, I am obtaining different results between the two cases, such as differences in pressure distribution and other bearing characteristics.

Why can these two Journal Velocity settings produce different HDB results even though the shaft is rotating at the same prescribed angular velocity omega in both cases?

I would appreciate any explanation of what velocity is actually being passed to the HDB interface through the Multiphysics coupling.

COMSOL 6.4 "Operation not supported" Error

Wed, 19 Aug 2026 16:42:31 +0000

My workplace just migrated to COMSOL 6.4. All of the models I developed in 6.3 are not working because of an error in several Electrical Circuit (cir) elements.

I have five elements in my model: Ground Node 1, Voltage Source 1, Resistor 1, Inductor 1 and External I-Terminal 1. The Ground Node 1, and External I-Terminal 1 work as normal, when I click on them the GUI displays their settings. Clicking on Voltage Source 1, Resistor 1, or Inductor 1 produces a pop-up Error box with the text "Operation not supported" and a single OK button. Clicking OK closes the box but the Settings window is empty.

I have recreated the Error in multiple old model files.

I created a new model in COMSOL 6.4, added Electrical Circuit (cir) physics, and then added a Voltage Source. The Error was present in the new model as well so its not simply back compatibility, something is not working with the Electrical Circuit (cir) elements themselves.

Please help,

David

Invalid XML character (Unicode: 0x1)

Sat, 15 Aug 2026 13:42:03 +0000

Hello,

I ran a COMSOL simulation and saved .mph file with all the data preserved to be able to use it again in future. I opened the same file again and got this error.

"An invalid XML character (Unicode: 0x1) was found in the element content of the document."

Is this error solvable? How can I recover my .mph file?

Regards,

Zain