Quantcast
Channel: Solving, solving\ - Latest Discussions - COMSOL Forums
Viewing all 3988 articles
Browse latest View live
↧

Function

April 2, 2016, 3:02 pm
$
0
0
How to enter summation of a series ?

I am trying to enter prony series(generalized maxwell model) in which there is summation of series for a viscoelastic model. SO if anyone can help me.
↧
Search

Problem when using "Remesh Deformed Configuration"

March 26, 2012, 4:37 am
$
0
0
Hi,

I'm running into problems when using the "Remesh Deformed Configuration" on a solution data set that contains results from a simulation including "moving mesh" and "solid mechanics" physics. A new mesh configuration is created but this is the original geometry and not the deformed one. As suggested eariler in the forum I've tried switching the order of the two physics in the model tree, although with the same result.

I've created a simple file to illustrate this problem and would be grateful with any ideas what causes this. I've seen in the forum that other people as well have encountered this problem.

Thanks in advance,
Wiktor
↧
↧

Solve 2D Euler equations by using general PDE (g)

January 16, 2016, 3:10 am
$
0
0
Hi everyone,

I want to solve a 2D Euler equations using Comsol to save programming and debugging time. Here the Euler equation is used to describe gas expansion, thus it includes conservation of mass, momentum and energy, and four independent variable exist (rho, u, v, and p).

I treated this problem following the way in the blog about shock tube: www.comsol.com/model/shock-tube-100, where the 1D Euler equation is modified to a 2D example (t was represented by the y) to save time. In my module as attached, I use a 3D module to represent the 2D problem (using z to stand for t). A weak contribution is also added as in the shock tube blog. The 2D Euler equation were derived and added in my attached module.

The initial condition is read from a 2D Riemann problem (6.1 2D Riemann problem) from: www.oxygen8.org/HTML/numerical...oject/MUSCL_for_2D_Euler.html.
And this initial condition is expressed by Matlab function. For example, density is expressed by function initial_rho, as attached.

My question is, I cannot get a solution from this module, the comsol error informing that "Failed to find a solution. Maximum number of Newton iterations reached." I tried to increase the iteration number, but it seems it cannot figure out this problem perfectly.

I want discuss following problems:

1). For CFD problem, is there special solver or configuration or special meshing scheme if we want to use FEM to solve these equations? I noticed that in the shock tube example, the Adaptive Mesh Refinement is used. I tried it, but failed.

2). I don't think this 4 PDEs problem is more difficult than solving Navier STokes equation. If Comsol can handle the CFD problem properly, there should be a proper way for me to carry out this module calculation.

3). I also tried to calculate this problem in 2D using Time dependent, but it is too slow.

Thank you very much for any suggestion about this module.
Hao


↧

Global eqn. Help

April 4, 2016, 7:25 am
$
0
0
Hi guys,

I'm having some issues with a global equation, specifically with the initial value assignment.

I have a global equation governing a prescribed mesh displacement (rotation). The equation contains a cos(u) term and a sin(u) term so it is 0 at integer multiples of pi/2. The initial value for u, u_0, is set at 30deg (0.8ish rad).

When I ran the simulation, no rotation occurred, so I added a global variable probe on u. At t=0 the table shows u = 1.0916E-4 rad, which is very much not the value I've entered as the initial value for u. The rest of the table is made up of values tending even closer to 0; as far as I can tell, no rotation is observed because the equation is basically 0 for the duration. What I don't understand is why the value for u_0 is not the value I've entered...
↧

Current conservation issue when integrating charge density

April 4, 2016, 2:30 pm
$
0
0
I am working with an AC/DC, 3D, Stationary, Electric Currents model.

I am currently studying charge density between deep brain stimulation electrodes of different surface areas and geometries. To represent an oversimplified electrode, we created a cylinder with high conductivity (1e10 S/m) and a point current source at it's center. The cylinder was in a conductive medium similar to the brain, 0.2 S/m.

When evaluating this simplified problem, I encountered some error that seems violate current conservation. Specifically, I am finding 47% error between the point current source (0.001 A) and the integral of current density (ec.normJ) over the surface area of the cylinder.

Note: I did conduct a mesh convergence study - below the current mesh size (0.00002), the error does not notably improve.

I then explored this issue with different geometries (spheres, ellipsoids, cubes, rectangular boxes, etc). See page 1 of the attached TroubleshootingComsol.pdf for the results. To quickly summarize, symmetric geometries - spheres and cubes - integrate to a current within 2% error. However, non symmetric geometries, ellipsoids, cylinders, rectangular boxes - have unacceptable integration error (11-63%).

*****Question: Any ideas of the possible sources of error in the simple problem? Any suggestions to further troubleshoot?

An an alternative method, I also explored representing the electrode as a cylindrical shell with floating potential of 0.001 A. In this case, the integral of normJ is 0.0004- less than half than expected. The integral of nJ was closer - 0.0009. To troubleshoot this, I went back to a sphere, with similar results - integral of normJ was 0.00048 and integral of nJ was 0.00097. See page 2 of the TroubleshootingComsol.pdf for more details.

*****Question: I originally believed that normJ, the total current density norm, was the output of interest when studying charge density on a surface, but I am confused by the fact that it is substantially lower than nJ in the floating potential situation.

I greatly appreciate in advance any tips, explanations or corrections.
Sincerely,

Ashley
↧
↧

setting variable for stationary vs time-dependent study in same project

April 5, 2016, 6:28 pm
$
0
0
I have a component definition for an electrostatic potential constraint:

Vgate = 100 * (1 - exp(-t/3e-9))

This ramps up the gate bias to 100 volts with a time constant of 3 nsec.

However, I'd also like to do a stationary study in addition to the time-dependent study. When I run the stationary study I get:

"Undefined variable: t"

Ideally I'd like to be able to do two studies: one time-dependent, another stationary with the time also ramped as a parameter. Alternately I could explicitly ramp Vgate in the stationary study with the exponential time dependence applied to the time-dependent study.

As it stands I need to redefine the boundary condition depending on which study I am using.

Any help to this seemingly super-basic question?

thanks,
Dan
↧

Non-Local Modelling

April 6, 2016, 3:53 pm
$
0
0
Hello all,

I just started using comsol, moving from ABAQUS. I was hoping to gain advice on how to procceed with non-local modelling.

I would like to know if comsol can for every point compute an integral over a finite volume surrounding that point and use it in the solution.

e.g. every point can sense and use variables a finite distance away, like a sphere (3D) smaller than the modelled part but some multiple of the mesh size.

I appreciate I should bury myself in the documentation and figure this out myself but I hope there is some speed up by asking the community while searching.

Kind Regards and thank you in advance,

Alex
↧

How to add inertial lift force and Dean force

October 8, 2015, 10:54 pm
$
0
0
Hi everybody,

I m doing simulation for inertial focusing. I need to see the effect of inertial lift force and dean force on particle. I m just wanted to know where these forces can be defined. I tried to add them in particle tracing-> forces. still no use.

Appreciate help from someone.

Thanks in Advance.

↧
Search

TIME to FREQUENCY FFT

March 20, 2016, 9:36 pm
$
0
0
Dear all,

I am simulating pulse echo technique using ultrasound. I have my results in time domain. Now I need frequency spectrum of my time dependent study result. In study 1 "time dependent " time range is = range( 0, 5.0e-8, 9.0e-5 )s and generated ultrasound frequency is 1 MHz.

So I add study 2 "Time to frequency FFT" in my COMSOL model. Now I dont know what to do. I tried once and get 0 Hz frequency result.

Thank you
↧
↧

Simulation fails when I add heat transfer physics

April 7, 2016, 8:33 pm
$
0
0
Good morning,
I'm a student and I'm trying to model on Comsol an implantable fetal pacemaker (PM). I modeled the system with a corkskrew (helix) that represents the working electrode of the PM; then, I modeled the body of the PM with two cylinders linked to the helix mentioned above (one of the two is used as the ground of the PM). Finally, to simulate the environement of the PM (heart tissue, blood), I added two larger cylinders who enfold the PM.
When I run the simulation (stationary study) using just the physics "electric current", it works without any problem: I succeed to simulate the voltage and the current density reached in the system.
As I try to add a second physics (Heat transfer in solids or Bioheat transfer) the simulation fails and I always got the same message "Failed to find a solution. Segregated group 2. Out of memory during LU factorization".
Now, even if I don't have such a great experience with comsol, I don't think that my model is too complex to run a simulation with two physics; however, I still can't get the solution of this problem.
If anybody could give me any tip, I'd appreciate it.

Thank you very much!
P.s. I attach some photo of the model in the case somebody was so kind to help me.
↧

Number of degrees of freedom solved for: 0.

March 23, 2011, 12:33 pm
$
0
0
Hello all,

When I try to use 4.0, I met with this problem when solving.

Does anyone know how to fix this? I think it might be that I did not add the mesh to the solver, which is different from 3.5a.

Thank you.

↧

schottky contact

April 12, 2016, 3:16 am
$
0
0
Hello,

I am trying to make Schottky contact with the help of ZnO/Pt in semiconductor module. During solving i am getting this error(segregated step, undefined value found,
Details-Undefined value found in the equation residual vector.
there are 3600 degree of freedom giving NaN/inf in the vector for the variable conp1.Ne
there are 3300 degree of freedom giving NaN/inf in the vector for the variable conp1.Ph)

Could anyone please tell me what is this and why is this happening?
↧

Schottky contact

April 12, 2016, 3:44 am
$
0
0
Hi,
I am sovling for schottky contact of ZnO/Pt in semiconductor module.During solving i am getting the error like Segragated step
undefined value found
Details-Undefined value found in the equation residual vector.
there are 3600 degree of freedom giving NaN/inf in the vector for the variable comp1.Ne
there are 3300 degree of freedom giving NaN/inf in the vector for the variable comp1.Ph
Could anyone please tell me what is and why is this coming?
↧
↧

Undefined value found in the equation residual vector.

April 14, 2016, 9:46 am
$
0
0
I'm solving for lid driven cavity with stratified flow.

My error message looks:
Undefined value found.
- Detail: Undefined value found in the equation residual vector.
There are 3 degrees of freedom giving NaN/Inf in the vector for the variable comp1.T.
at coordinates: (0.080597,0.2), (0.0835821,0.2), (0.0820896,0.202422), .

Thus the error would belong to the heat transfer domain. As I understand, residual vector is the vector that is formed to solve my equation, which in this case would be the temperature equation. It would consist of my thermal properties, Temperature and velocity as a variable along with boundary conditions. But I’m unable to figure what is actually missing.

According to manual I followed “Check for undefined numerical values after each operation” in the assembly settings. After that it display this error :

Function evaluation returns undefined numerical value.
- Function: ^2
- Expression: (comp1.spf.xi_m*comp1.spf.res_v)^2
Failed to evaluate variable.
- Variable: comp1.spf.vhNS_lin
Failed to evaluate expression.
- Expression: nojac((1.4142135623730951^(1-shapeorder(comp1.u)))*max(comp1.spf.vhNS_lin-comp1.spf.vhNS_quad,0))
Failed to evaluate variable.
- Variable: comp1.spf.vhNS
Failed to evaluate temporary symbolic derivative variable.
- Variable: comp1.spf.crosswindns@VDN${test@85}
Failed to evaluate expression.
- Expression: d((comp1.spf.crosswindns)*(dvol),{test@85})
Failed to evaluate Jacobian of expression.
- Expression: (comp1.spf.crosswindns)*(dvol)

But here it seems that it is referring to components of velocities. My guess would be that it is unable to resolve velocities at the boundary or getting division by zero there.. I’m lost on how to fix this.

I don't think it is due boundary conditions or material properties are the culprits, due to fact that there are 200 iterations. Although the error plot looks really bad i.e no convergence is achieved. My initial guess is good enough since it gives an error of magnitude ~ 10^0 .

Kindly help!!


Regards
Parth Swaroop
↧

Varying capacitance using electromechanics

April 14, 2016, 10:59 am
$
0
0
Dear Comsol User,

I am trying to find variation in capacitance between two plates with applied body load. I defined an air box around the plates. Now I'm trying with electromechanics physics for applying body load and finding capacitance. I defined the two plates as linear elastic material. Applied body load to the moving plate. I applied 1V to moving plate and fixed plate is grounded. I also gave parametric sweep for body load. But capacitance is not varying with body load.
What changes should I make to get a varying capacitance? Please Help.
Thank You,
Aswathi M Madhu
↧
Search

Solidworks assembly imported into comsol - simulation difficulties

April 15, 2016, 11:31 am
$
0
0
I am a student trying to improve my personal knowledge and skill with Comsol and I have run into a problem while trying to run simulation on an assembly that I've designed. I built the assembly (a small number of small parts) using Solidworks and have imported the parts into comsol. I'm sure everything about the assembly is dimensioned and should fit together correctly (adding/activating pairs in comsol where necessary). However when I try to apply a load, fix the bottom, and compute the simulation, it works on it for ages every time before then coming up with a list of error messages, usually "maximum number of iterations reached" and "the relative error is greater than the relative tolerance". I've played around with the settings of the study to see if I could work it out but no luck yet.

Any advice will be appreciated!
↧

how to do half model analysis

April 18, 2016, 3:17 am
$
0
0
Dear all,
i am doing a simulation where no. of elements and nodes are very large. that is why time taken is very large. my restriction is i can't take larger time step as well as coarser mesh.
so all i want to know if i can use the half model FEM analysis in COMSOL. if YES then HOW?
thanks
↧
↧

Ordinary differential equation

April 18, 2016, 2:17 pm
$
0
0
I have to enter two ordinary differential equation for single variable "a" such that for 0<a<.5 first equation will be governed and for .5<a<1 second equation.
How to enter it in comsol.
↧

chirp signal as a prescribed displacement

April 19, 2016, 4:40 am
$
0
0
Hi everyone

Can anyone help me to add a prescribed displacement of a cantilever beam as a chirp signal ?.
I'm using 2D structural mechanics (plate) physics and time dependent study.
Thanks in advance...
↧

unknown varibles

April 19, 2016, 11:48 am
$
0
0
Hi all,

When I using the free and porous media flow module, a strange variable automatically appeared in the segregated step settings, usually named comp1.fp.Pmfinl11, together with the velocity variable and pressure variable. This variable can be set to be solved in any steps defined by me like other normal variables.

What is that variable used for?

Regards
Bin
↧
Viewing all 3988 articles
Browse latest View live
Search
<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>