Latest Discussions - COMSOL Forums

Need Help with Suction-Based Climbing Robot Simulation

Thu, 07 Aug 2025 20:38:32 +0000

Hi everyone,

I'm working on simulating a suction-based climbing robot using COMSOL Multiphysics. My objective is to check whether the robot will stick to a vertical surface or slip based on applied suction.

Project Details:

Suction pressure: –15,750 kPa applied to each suction pad

Robot weight: 117 N

I want to simulate how the robot behaves under gravity and suction force—whether it stays adhered or begins to slide.

I'm confused about contact pairs and setting up proper interaction between the robot body and the wall.

Also, deciding between stationary and time-dependent studies is a bit unclear.

Pipe flow and pump

Thu, 07 Aug 2025 13:12:27 +0000

Hi there, can someone please share how to set a pump with inlet and outlet of a pipe flow model (gas, not liquid) with the goal of simulating the pressure losses due to elbows and pipe length, please. The flow rate and ultimate vacuum of the pump are set, trying to work out how to achieve this flow rate reduction / pressure difference drop when piping gets very long. thanks for you help. I found the liquid example in pipe, but not anything with pump and gas. any reference to some documentation is appreciated.

tcd.kappaxx-zz in the Tertiary Current Distribution governing Equations

Thu, 07 Aug 2025 12:43:27 +0000

Hi All,

I took a deep dive on porous electrode reaction node equation view and i saw electrolyte conductivity term tcd.kappaxx/xy/xz/yx/.../zz. I tried find and replace (ctrl + F) for this variable but i could not find what this variable do and how does this contribute to the overall tertiary current distribution physics.

does anyone know how COMSOL uses this term? or is it just a post-processing variable?

How to stimulate the electrode on the semiconductor module?

Wed, 06 Aug 2025 05:57:05 +0000

Hi guys,

I want to stimulate a ITO-ETL-PVSK-HTL-FTO Perovskite solar cells by using the semiconductor module of COMSOL to investigate quantities like electrical potential, concentration of holes and electrons, and etcs, and I want all layers are included. Now I encounter a problem, which is I couldn't include FTO in my model although I can include ITO.

The first thing came into my mind for this issue is that maybe it is due to the PN effects on HTL(P)-FTO(N), such that there exists an large energy level gap on their interface, so the holes in HTL could hardly move into FTO, which then fail to solve the continuity equation on that boundary. However, after some adjustments on the electric affinity of FTO and HTL, I was almost close that gap(left a <0.5eV energy gap), the stimulation failed again.

As for the other b.c.s and i.c.s, I set 0V on the top of ITO, and a scanned volt Vapp on the bottom of FTO, and doping ITO/FTO with 1e20[1/cm^3] in N-type, HTL with 1e18[1/cm^3] in P-type, and ETL with 1e18[1/cm^3] in N-type.

Thanks Pals

Worstation choice

Tue, 05 Aug 2025 00:19:55 +0000

Dear support,

I'm thinking of upgrading one of my workstations, with running Comsol as its primary goal. I'm considering two options (mainly): both are single CPU computers 1) with an AMD Ryzen™ Threadripper™ PRO 7975WX (4.0 GHz, up to 5.3 GHz w/Boost, 128 MB L3 cache, 32 cores, 350W) 2) with an Intel® Xeon® w7-3565X Processor (2.5 GHz, up to 4.8 GHz w/Boost, 32 core, 335 W)

In both cases I aim to 8x16Gb, following Comsol's recommendation of having all memory channels populated.

Would you have any suggestions? Should I go for a different architecture? My workstations today have two CPU, 8 and 16 cores each respectively, and relatively slow Xeon processors. I lean towards the Ryzen because of the higher clock, would you concur? I would expect any of the computers above to be significantly faster/more capable than the best I currently use (2 Xeon Gold 5218 processors, 16 cores each, 2.3 GHz speed). Would that be your expectation too?

Bests, and thank you for your help...

Javier

PS. Comsol does not make much use of GPUs yet, right? And it does not make use of hyperthreading, as I seem to remember?

How to define the contact angle in a model

Mon, 04 Aug 2025 11:29:20 +0000

Hello there, I am trying to simulate droplet impact on solid stationary particle. I am finding it difficult to define contact angle. I used wetting wall BC to give values of contact angle but it doesn't matter whatever value i give in contact angle, I am not able to see it in output, it feels like the contact angle value which i give has no effect upon simulation. Please guide me where i am doing wrong. Thanks

PML Wavelength

Sun, 03 Aug 2025 22:02:27 +0000

Hello I'm doing a structural analysis in Comsol using PML. The analysis is in the frequency domain, and I'm unsure which wavelength to use in PML.

Another question is, when I select the Comsol option to use the wavelength based on the physics of the problem, is there any way I can access that wavelength?

Schur complement of the global stiffness matrix

Sun, 03 Aug 2025 07:46:19 +0000

I am trying to calculate the Schur complement of the global stiffness matrix, I have already exported the matrix through the livelink. To calculate the Schur complement of the matrix, I have to figure out which part of the matrix corresponds to which node, therefore, are there any way for me to get how comsol assemble the matrix? Besides, I know that comsol has a schur complement solver, are there any way for me to perform this on the global matrix? (e.g. I got two same cascaded units, and I only need to calculate one stiffness matrix and cascaded them through the schur complement)

Silica Precipitation

Sun, 03 Aug 2025 02:50:44 +0000

I'd like to know how to achieve discontinuous precipitation of silica. My idea is to use the dissolution and precipitation of silica to change the porosity and permeability of the fractures, thereby achieving a discontinuous precipitation effect. I'd like to ask if this can be achieved.

How to integrate over infinite domain or scaling coordinate system

Sat, 02 Aug 2025 19:03:24 +0000

Hello!

Every integration I tried didn't recognize or consider the stretching of an infinite or scaling domain. What is the right or appropriate function or way to integrate over stretched scales? Whether in Results or in Component.

Thank you.

P.S. Datasets->Emped 1D/2D seemed a good way or workaround, but the 'Output space expression' doesn't recognize used component functions or expression operators, and I don't want to use redundancy or a workaround.

Nice greetings

Non Hermitian Acoustic Model

Sat, 02 Aug 2025 00:09:48 +0000

I'm summarizing my whole work.

With COMSOL 6.2. Non-Hermitian Study with Balanced Loss and gain with only COMSOL Weak Form PDE. I have an Hamiltonian, H=[0 (v+gamma)+w.exp(-ik);(v-gamma)+w.exp(-ik) 0]. I tried weak form of equation.

Model Details: 1. Parameter: L=60[mm], kx=pi/L, h1=5[mm], r=5[mm] g=1[1/m], c_air=343[m/s], rho_air=1[kg/m^3], k=1 [1/m].

Geometry: Rectangle1 (Width L/2, Height h1, Position, x=0, y=0); Rectangle2 (Width L/2, Height h1, Position, x=L/2, y=0); Circle1(Radius r, Position, x=L/4 y=h1/2), Circle2(Radius r, Position, x=3L/4 y=h1/2). Then Union. Then delete Entities to delete internal boundaries inside the circle.
Materials: Now we have 5 domains. Domain 2 is the channel between between circles A (domain 4) and B (domain 5). We used user defined materials. Material Type: Non-Solid, Speed of sound, c=c_air and density, rho=rho_air. All the domains have same material properties.
In Component 1> Definitions>Identity Boundary Pair 1, I selected Source boundary as Left (1) boundary and Destination boundary as Right (10) Boundary.
Weak form of PDE: Applied to Domain all Domain. Weak Form PDE 1< Weak Expressions: - (test(px)px + test(py)py)+ ((2pifreq/c_air)^2)test(p)p.

Then took Weak Form PDE 2 and selected only Domain 2 and put expression: igtest(p)*px. This is non-reciprocity term.

Next, for Periodic Condition, I used Weak Form PDE (w) >Pair> Weak Contribution. Then at Pairs section selected the predefined "Identity Boundary Pair 1". And at Weak expression field, wrote expression as "test(p)(p - dest(p)exp(ikxL))".

Mesh: Free Triangular with Predefined size 'fine'.
Study 1: Parametric Sweep: g=0:0.01:1 and kx=0:0.1*pi/L:pi/L. Step1: Eigenvalue, Desired number of eigenvalues 6 and Search for eigenfrequencies around 1+1i[rad/s].

I run the simulation. There was no error. But, the periodic condition "test(p)(p - dest(p)exp(ikxL))" is not working. How to solve ?

Simulating Grain Evolution with Surface Heating in COMSOL

Fri, 01 Aug 2025 14:47:17 +0000

Hello,

I'm new to using COMSOL. I’ve already defined the geometry and assigned materials to a cubic sample. Now, I would like to simulate how the grain structure evolves when heat is applied to the outer surfaces of the cube.

Could you please guide me on how to set up this simulation?

Thank you. Nafish (PhD. Student)

NACA 0012 transient

Fri, 01 Aug 2025 12:55:32 +0000

hello dear community

i was trying to do time dependent study but unfortunately always show inconsistent initial values even though i followed the steps of the model that is provided by the comsol community i would like to ask for guide and what is the exact issue here

Heat transfer from housing to fluid

Fri, 01 Aug 2025 09:43:07 +0000

I've been struggling with what would seem a fairly simple heat transfer model.

I have a domain with a power loss. Above that is another domain which represents a housing, within which is a cooling channel. I've been trying to calculate the power dissipated to the fluid to correlate to another FEA package.

When I calculate dissipated power from temperature rise of the fluid, I get more power than what is introduced to the model from the power loss domain.

All walls are insulated, so all power should be going into the fluid minus the power to increase each component temperature.

I've increased mesh density but the overall result hasn't changed which leads me to believe the problem is more fundamental.

Can anyone share some insight?

What is the meaning of (nitf1.qwf_u)?

Thu, 31 Jul 2025 01:00:13 +0000

I just know this formula for calculate the power of exchange but i do not know the exact formula contained in it. I will appreciate it, please help me.

Suppress recovery files writing from Matlab

Wed, 30 Jul 2025 12:28:23 +0000

Dear Support, I'm running Comsol 6.1, and doing simulations from Matlab. I am running a different Matlab/Comsol instance for each of the 32 cores of my workstation, and each is running a loop. The problem I have is that each is loading the hard drive with writing recovery files. Besides of the wasted space, I worry this is going to reduce its life, and I want to suppress all unnecessary writing. How can it be done? Looking at the names of the files, there seem to be geometry files, mesh files, etc. I got the impression that each Comsol server instance opened its own folder in the AppData/Local/Temp folder, and wrote recovery information for each loop step, which quickly amounted more than 100Gb of unneeded data.

Best regards, Javier

Quantification of Loss Due to Evanescent Phenomenon in Underground-to-Aboveground RF Communication

Tue, 29 Jul 2025 21:45:05 +0000

Hi everyone,

I am simulating underground-to-aboveground (UG-to-AG) RF communication using the RF Module in COMSOL. Specifically, I placed a dipole antenna underground and evaluated the power both just below and above the soil-air interface to study the evanescent wave phenomenon and quantify related losses.

As we know, evanescent waves occur when the incident angle exceeds the critical angle at a dielectric boundary (in this case, from soil to air), preventing the wave from propagating into the upper medium.

However, in the case of a dipole antenna, which emits a spherical radiation pattern, it is unclear how to define or compute a unique incident angle. I am wondering:

What is the correct way to verify the presence of evanescent waves in this scenario?

Can the incident angle be inferred from the phase of the field (e.g., arg(emw.Ez)) or perhaps by tracking the direction of the Poynting vector near the interface?

How do we quantify evanescent loss, and how can we distinguish it from reflection loss in COMSOL simulations?

Any guidance or references on how to confirm evanescent behavior and quantify associated losses in full-wave simulations like COMSOL would be highly appreciated.

Thanks in advance!

Simple model not converging

Tue, 29 Jul 2025 15:33:35 +0000

Hi all,

I am trying to model electric field shielding with a copper box that has some holes in it. I feel as though this should be a very simple simulation for COMSOL to run, but for some reason my convergence plot plateaus before the specified tolerance of the solver. I have tried doing very fine meshing, and this did not help.

Does anyone know what I can do about this?

Thanks

Importing "STEP file" into geometry of COMSOL showing no geometry?

Tue, 29 Jul 2025 09:30:04 +0000

Please if someone can help with this matter?

Error in Fluid Structure Interaction

Tue, 29 Jul 2025 09:14:53 +0000

Dear all, I am trying to simulate a closed rotating cylinder with a solid object like cube or ball which will be displaced due to the rotation of the cylinder (The cylinder is filled with water). Maximum no of newton iteration reached, i also tried direct solver which normally worked when i had convergence errors in other models i had done. Also what type of boundary condition should be given for the object?. I had given moving mesh to the fluid domain and also prescribed displacement for the solid body.

Thank you for your time