Latest Discussions - COMSOL Forums Most recent forum discussions Thu, 24 Jan 2019 12:58:44 +0000 COMSOL Forum: Latest Discussions Degree of Freedom Error in Capacitive Pressure sensor design <p>Hello everyone, I have just started my PhD and I have no enough background regarding comsol. I am working on Capacitive Pressure sensor. Design is very simple there are two conductors of copper and there is a dielectric layer of PDMS between these two conductor plates. I am apply pressure on the top boundary and expecting there will be some deformation in top conductor plate because of pressure. I am trying to findout how the capacitance changing with applied pressure. I have also attached the design image here. I am getting this error when i try to compute the study of my design</p> <p>Undefined value found. - Detail: Undefined value found in the equation residual vector. There are 52015 degrees of freedom giving NaN/Inf in the vector for the variable comp1.spatial.u. at coordinates: (0,0,0.00014), (1.27451e-05,0,0.00014), (2.54902e-05,0,0.00014), (0,0,0.00015), (0,0,0.00016), ... There are 52015 degrees of freedom giving NaN/Inf in the vector for the variable comp1.spatial.v. at coordinates: (0,0,0.00014), (1.27451e-05,0,0.00014), (2.54902e-05,0,0.00014), (0,0,0.00015), (0,0,0.00016), ... There are 52015 degrees of freedom giving NaN/Inf in the vector for the variable comp1.spatial.w. at coordinates: (0,0,0.00014), (1.27451e-05,0,0.00014), (2.54902e-05,0,0.00014), (0,0,0.00015), (0,0,0.00016), ...</p> Thu, 24 Jan 2019 12:58:44 +0000 4.2019-01-24 12:58:44.226311 How to "Compute" keeping the previous result as it is <p>Hi everyone.</p> <p>Is it possible to run a new "Compute" without overwriting the results by the previous "Compute"?</p> <p>"Parametric Sweep" seems to give a solution, but at times I need to compute for one "Parameters" and check the result and then tune the parameters, again and again.</p> <p>I wan to avoid writing down the results, say the frequency, each time.</p> Thu, 24 Jan 2019 06:55:54 +0000 4.2019-01-24 06:55:54.226281 Couple Ray Optics and Structural mechanics modules <p>Hello, I have started with the 'Transparent Light Pipe' model in comsol, where the light transmittance can be found for different shapes. However, I would like to see the same effects by stretching the pipe. I believe the structural mechanics module (Boundary load and and a fixed constraint) should allow to stretch the pipe as needed. How should I add/couple the structural mechanics module to the aforementioned model.</p> <p>The link for the model can be found here:</p> <p>Any help is much appreciated.</p> <p>Best regards, GB</p> Thu, 24 Jan 2019 00:02:39 +0000 4.2019-01-24 00:02:39.226262 Questions about the thermal stress simulation <p>Now I am trying to do the thermal stress simulation, but comsol do the thermal and mechanical simulations together. I want to do the thermal and mechanical simulation separately, because the thermal one is too complecated. Both simulations are time dependent. How can I put the results of temperature into the mechanical simulation? Now I tried to buil 2 studies. In Study 1 I only do the thermal simulation. In Study 2 I set up a thermal simulation without boundary conditions and couple it with the mechanical simulation. Then I set the initial values and values of variables to the results of study 1, but the results seem to be wrong. Are there any other ways to do so?</p> Wed, 23 Jan 2019 22:16:37 +0000 4.2019-01-23 22:16:37.226261 I can't plot data from a stopped parametric sweep <p>I had a parametric sweep running for a <strong>long</strong> time and wanted to stop it mid-run and see what I have so far. I followed this thread:</p> <p></p> <p>But when I try to plot my data, nothing pops up except for my geometry. The legend outside the plot is there and changes depending on what parameter I pick from the sweep. The data looks like it's there. It takes a long time for COMSOL to "plot" when I click on a different parameter and then it just shows the geometry again.</p> <p>What can I do from here?</p> <p>EDIT: This only applies to data sets that are mirrored, apparently. I can plot the raw data. If I put it through a Mirror data set it just comes up blank.</p> Wed, 23 Jan 2019 21:16:12 +0000 4.2019-01-23 21:16:12.226242 Laser engraving - mesh solution <p>We are interested to re-create the example visible in the link : But we do not understand to implement the moving mesh to see the engraving effect. Can we have any suggestions ? We use Comsol 4.2a</p> <p>Thanks a lot</p> Wed, 23 Jan 2019 20:29:27 +0000 4.2019-01-23 20:29:27.226241 Brinkman Equations: Maximum number of newton iterations reached! <p>Hello everyone!</p> <p>I have been trying to model fluid flow through a tube that has a porous media section. I have been using the Brinkman equations module, but I always get the error: Maximum number of Newton iterations reached! My simulation is very similar to COMSOL's "Porous Reactor with Injection Needle" example that can be found on COMSOL Application Gallery. The only difference is that I am only interested in the flow through the porous media and my model does not have an injector (doesn't have two inlets). I am sure if I can figure out the right way to mesh my model, it will work. Please, could anyone help me with this problem? <strong>(I added some captures of my model)</strong> The tube is composed of three cylinders. The middle cylinder (#3) has the porous media, #1 has the inlet and #3 has the outlet.</p> Wed, 23 Jan 2019 17:31:13 +0000 4.2019-01-23 17:31:13.226232 How to calculate the Scattered Cross Section. <p>Hi everyone,</p> <p>I am trying to calculate the scattering cross section of a silver cylinder. However, the scattering cross section seams to depend on the radius of the integration circle around the cylinder even when I use emw.relPoavx for the time average of the scattered Poyting vector. I have tried with the integration being exactly on the cylinder surface, on a cylinder with 2<em>radius and with 5</em>radius and in some cases I have both positive and negative values for the cross section. Can someone help me please?</p> <p>Thank you very much</p> <p>Ruben</p> Wed, 23 Jan 2019 17:20:29 +0000 4.2019-01-23 17:20:29.226231 How to execute conditioned interuption of an iteration of simulation using matlab livelink? <p>Hello, evebody. I am using matlab livelink to sort out the best parameters with genetic algorithm. Surely some sets of parameters are impractical so the COMSOL simulation iteration for such a parameter set will carry on forever and never converge. How do I use matlab command to interupt such non-converging simulation if, say, this simulation takes over 1 minute, and then start the next iteration?</p> Wed, 23 Jan 2019 15:50:05 +0000 4.2019-01-23 15:50:05.226213 LMSTAT - finding number of licences in use <p>The below post is related to an <a href="//">archived discussion</a></p> <hr /> <p>Hi,</p> <p>I've recently tried the advice given in the linked article to find out how many licences are in use on our system but I get this:</p> <p>glnxa64]$ lmstat -c /usr/local/software/comsol/5.4/license/license.dat -a -bash: ./lmstat: /lib64/ bad ELF interpreter: No such file or directory</p> <p>The system is running Centos 7, and the licence file at this location contains the following text (serial number and server name changed!)</p> <h1>SN=aabbccdd</h1> <p>SERVER ANY 1718 USE_SERVER</p> <p>Grateful for any input!</p> <p>Thank you.</p> Wed, 23 Jan 2019 10:58:11 +0000 4.2019-01-23 10:58:11.226201 arrhythmia model <p>I have developed a 2d cardiac model of the heart.I need to simulate an arrhythmia case study. What are the points to be noted?</p> Tue, 22 Jan 2019 15:19:37 +0000 4.2019-01-22 15:19:37.226172 What is the model for solving surface tension in COMSOL? <p>What is the model for solving surface tension in COMSOL?</p> Tue, 22 Jan 2019 13:13:59 +0000 4.2019-01-22 13:13:59.226163 Modeling cloth <p>Hi there,</p> <p>I have a problem modelling a piece of cloth in COMSOL. The key feature of cloth is that <strong>it has no bending stiffness or out-of-plane shear stiffness</strong>. Could you tell me how to set the right material for the cloth?</p> <p>Thanks so much, Ke Wu</p> Mon, 21 Jan 2019 15:37:27 +0000 4.2019-01-21 15:37:27.226121 Viscoplastic Creep Test <p>Hello,</p> <p>I am new to COMSOL and I am trying to simulate a VISCOPLASTIC creep/relaxation test on a simple 2D rectangular block. I would like to either have a constant boundary load or prescribed displacement for sometime and then the load is removed. The required results are attached below. I don't have a problem with the BC. I need help in setting up the physics and material properties to get the type of results required.</p> <p>NOTE - Any material can be used.</p> Mon, 21 Jan 2019 15:36:33 +0000 4.2019-01-21 15:36:33.226112 Fixed constraints properties <p>Dear all, I am doing a modal analysis of an unstable rock block to study its eigenfrequencies. One face of the limestome block was attached to the vertical cliff and therefore I set on it the FIXED CONTRAINT. However, this boundary condition is conceptually wrong since the "rock bridges" have their own stiffness. I would like therefore to set a boundary condition which take into account the Young's and Poisson's modulus of the rock compartment.</p> <p>Thanks a lot for your help, Marco</p> Mon, 21 Jan 2019 14:39:34 +0000 4.2019-01-21 14:39:34.226111 Modelling of thin layer <p>Hi;</p> <p>I am studying on thermo-mechanical simulation in Comsol 5.4 (licensed version). But I have some difficulty related to thin layer meshing and assigning solid boundary conditions especially. To make a true simulation, I have started with basic blocks. For example; assuming that there are three domains. The thicknesses of these three domains are 0.4 mm , 20 um and 0.4 mm, respectively. The domain with 20 um thickness is an adhesive layer and this adhesive layer bonds the other two domains to each other. Actually; I want to analyze mechanical behaviour of thin adhesive layer under temperature variations. The thin domain is a kind of adhesive (photopolymer). My study is time dependent, physics is thermal stress (multiphysics).</p> <p>My questions are firstly related to thin layer meshing and solid boundary conditions. Do you suggest any method to mesh thin layer effectively? I have made free tetrahedral and after that refined thin layer. It is true?</p> <p>For solid boundary conditions: the two boundaries of thin layer are “thin elastic layer”. In the settings, I have calculated only total spring constant. And I have made the bottom boundaries of other two domains “fixed constraint” to fix on a surface. Is that true?</p> <p>My model is actually more complex as I said, but firstly I want to start with basic blocks.</p> <p>Regards,</p> Mon, 21 Jan 2019 14:23:34 +0000 4.2019-01-21 14:23:34.226101 contour problem <p>my probleme is with the contour ploting problem* there is no triangulate from the data set in tecplot 2018 as in the older version to make our problem so much difficult</p> Mon, 21 Jan 2019 08:56:59 +0000 4.2019-01-21 08:56:59.226071 Skin effect simulation in tinned/covered coil <p>Dear Comsol Comunnity,</p> <p>I have been using Comsol at my University a few times so my experience in this software is only scientific and quite general. Now I have a problem which is quite complex. Here is the potential problem abstract. I want to simulate the skin effect in tinned coil or coil covered by other material like kanthal, lead, nickel and constantan. This is the air coil 20mm diameter with 14 turns with 4mm separation between them covered by around 100 um to 300um different material. It should have estimated inducted around 1 or 2uH (the exact value is doesn't matter for now). I want to do frequency sweeping from single MHz to tens of GHz for chosen material and its thickness. For example copper cover by tin 100umthickness -> frequency sweeping, same 150um, kanthal 100um then 150um etc etc. My goal is to chose the configuration where the surface current density norm (A/m) is highest. Thing is that I'm not sure how to cover the copper with other material. What I have done is STP file with example coil model I wanted to examine (attachment). I have add the boundaries for all of the coil excluding the boundries that I want to define as Coil input and output in Coil geometry analysis. My question are: 1. Whether my thinking is correct? 2. If yes (1) is this the only way to make the material coverage? 3. If my thinking is correct: a) Whether all of the boundaries which covers the copper should be defined in Coil explicit additionally or whether they should be defined as second coil and mutual inductance between them ? b) How to generate the Mesh for that kind of problem ? 4. Maybe there are special material library for that kind of problems ?</p> <p>Best regards, Krzysztof Szybinski</p> Sun, 20 Jan 2019 12:39:26 +0000 4.2019-01-20 12:39:26.226041 Infinite Domain (Half-Space) and Anisotropic Conductive Sheet (Electromagnetic Frequency Domain) <p>Hi,</p> <p>I am trying to study a scenario of infinite domain half-space that is separated by infinite anisotropic conductive sheet. The excitation is coming from a dipole. My problem is in defining the anisotropic conductive sheet. Can you suggest what is the best method for implmenting this?</p> <p>A picture is available in this link <a href="">image</a></p> <p>Thanks</p> Sun, 20 Jan 2019 06:29:10 +0000 4.2019-01-20 06:29:10.226022 Why solution is different when using 1 or 3 Pressure Acoustics interfaces? <p>Hello everyone!</p> <p>I'm working on a transient model consisting of one piezoelectric emitter transducer, one receiver and a cylindrical short sample. My first approach was to use one single <em>Pressure Acoustics, Transient</em> interface branch to obtain the acoustic pressure field of the entire model (emitter + sample + receiver). My second approach was to use 3 simultaneous <em>Pressure Acoustics, Transient</em> interface branches, one for each component, and connect them with appropriate Pressure and Impedance boundary conditions on the contact faces.</p> <p>In the end, both methods provided different solutions. The first method (denoted as 1-P in the attached images) gave me a lossless periodic signal and the second method (denoted as 3-P in the attached images) gave me a periodic signal that asymptotically goes to zero as if I had inserted losses, which I didn't. Why both methods provide different solutions?</p> Sun, 20 Jan 2019 02:51:44 +0000 4.2019-01-20 02:51:44.226012