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

High frequency EM scattering off an obstacle

March 9, 2016, 4:44 am
$
0
0
I would like to compute the scattered wave off a PEC obstacle (such as a polyhedron or ellipsoid) by an incoming time harmonic plane EM wave. The wavelength is much smaller than the scale of the obstacle (such as 1/100). If I choose the wave optics module -> em wave -> frequency domain then the mesh must be very fine and the computation can't be done in a PC. The wave optics module -> em wave -> beam envelopes model is also not suited since the tip says

The physics interface can be used efficiently for unidirectional and bidirectional propagation of electromagnetic beams. However, for optical scattering phenomena, where the field is scattered into many different directions, the Electromagnetic Waves, Frequency Domain interface is better suited.

In my problem the beams are split in more than two directions.

Are there any other better methods for this problem?
↧
Search

Centrifugal Force

March 9, 2016, 5:08 am
$
0
0
Hi,

I would like to include centrifugal force in my COMSOL (version 4.3a) model of a curved channel with laminar flow containing small particles.

Is it necessary for COMSOL to find the first and second time derivatives of the particles moving in x and y directions (to find the velocity and acceleration of these particles) at each point on the particle trajectory as the particles flow round the curve, and plug these values into the equation for the radius of curvature;

R= (x^'2+y^'2 )^(3/2)/(x^' y^''-y^' x^'' )

Since the radius of curvature is changing continuously for each particle?

Would COMSOL then use this R value to calculate the centrifugal force

F = mR(ω^2)

at each position the particle occupies?

Thanks in advance for any help that can be offered on this!
↧
↧

Positron trapping and release (charged particle tracing + electrostatics). How to optimise?

March 9, 2016, 11:53 am
$
0
0
I was wondering if someone could give any insight on what are the best solver settings for my problem, in terms of solving time vs accuracy. Currently I'm using Strict generalised alpha settings to be safe, but I'm wondering if this is necessary or if there are quicker methods?

I have a 2D-axisymmetric simulation in which the solution of a time-dependent electrostatic study (time-varying electric potentials) is used for a charged particle tracing study.

I am simulating a positron plasma inside a trap and then dumping them by lowering an electric potential gate.

The positrons bounce harmonically between the trap potentials with a period on the order of 100's of nanoseconds, and my simulation is over a period of 8000 nanoseconds.

However, using strict generalised alpha settings, the solver often has a reciprocal time step of 10E12, causing solving time to be very long.

I'm hoping to find information on optimizing the solving time of simulations and how this can affect the accuracy of the simulation.

Thanks in advance for any input on this matter.
↧

Failed to evaluate variable- comp1.ht.Tinit

March 11, 2016, 5:40 am
$
0
0
Hello,

Please I am working at simulating a 3D solar panel in COMSOL 5.1. After successfully fixing the errors on the mesh and the physics have been added, I proceed to compute the study, and I got the following error:

Failed to evaluate operator.
- Operator: mean
- Geometry: geom1
- Boundaries: 33, 34, 35, 36
Mean operator requires an adjacent entity of higher dimension.
Failed to evaluate expression.
- Expression: mean(comp1.ht.Tinit)
Failed to evaluate variable.
- Variable: comp1.ht.Tinit
- Geometry: geom1
- Boundaries: 33, 34, 35, 36
Failed to evaluate variable.
- Variable: comp1.ht.Jinit
- Defined as: (((comp1.ht.feb(comp1.ht.Tinit/unit_K_cf)*unit_W_cf)*comp1.ht.epsilon_rad)/(unit_m_cf^2))+((1-comp1.ht.epsilon_rad)*comp1.ht.e_bamb)
Failed to evaluate expression.
- Expression: comp1.ht.Jinit
- Feature: Dependent Variables 1 (sol1/v1)

Please I would appreciate if anyone out there can help me with 'Tips' to solve this problem.

Kind regards,
Samuel
↧

Eddy_Currents_In_PM

March 11, 2016, 10:46 am
$
0
0
Dear all COMSOL experts,

I am simulating a Synchronous Surface Mounted Permanent Magnet Machine. My aim is to estimate the eddy currents within the permanent magnets placed on the rotor surface, since my machine operates at high frequency and runs at high speed (17000 rpm). I have four different permanent magnets radially magnetized.

Since my simulation is a 2D one I have to force the following condition in order to reflect fairly the reality: the net current, i.e. the integral over the magnet cross section, must be zero for each time step. I tried to enforce it establishing a scalar global equation defining the current computed as an integral over the magnets cross section. After this step I impose a surface current density exactl equal to the opposite of the current divided by the magnet area.

I performed two simulations: firstly the permanent magnets are demagnetized (just a conductive material attached to the rotor) and the winding is excited with a current (just checking the eddy currents generated solely by the armature reaction on the stator); secondly I magnetize the magnets and "switch off" the windings (no load test).

It turns out that in the first case the simulation is working fine, i.e. it converges without problems, whereas in the second case there is no way the time stepping converges. If I choose an adaptive time step (free) the solver computes several time steps (very small compared with the actual time step subdivision I impose in the range visualization of the time stepping settings) and after some hundreds it stops and signal a convergence error. If I force a manual time step it is not even capable to compute the first time step, even if it is able to converge in the first stationary simulation which is imposed before the time stepping one.

I also tried the Single Turn Coil imposing zero current instead of defining a scalar equation for the current and imposing an external current density. No way it works.
In the time stepping solver configuration I tried to relax some error conditions, like posing Exclude algebraic instead of include. In this way, utilizing the Single Turn coil it is able to compute the time stepping. Nonetheless the results are not reliable since the current density integral over the airgap is not null anymore.

Basically I can see that when I magnetize the magnets (the machine is "magnetically charged") the time stepping is not able to go on in the computation. I tried several non linear materials iron curves, even with hundreds of points for definition. No way it works.

Has anyone else encountered the same issue before?

If you have any suggestions about this point I will be immensely grateful.

Sincerely,
Nicola
↧
↧

Multi-turn coil into simple SPICE circuit

March 14, 2016, 7:25 am
$
0
0
Hi,

I have recently downloaded and played around with the COMSOL model "Inductor in an amplifier circuit" (www.comsol.no/model/inductor-in-an-amplifier-circuit-990). This worked well, but when adding my own NETLIST from a simple circuit (picture attached), I get various error messages from the time dependent solution. (Reach singularity, maximum number of iterations reached, failed to find solutions for all parameters and so forth).

I have attached my model file and hope that someone with more experience than me can point me in the right direction and increase my knowledge about this extremely powerful tool.

NETLIST below:

*
R1 N002 0 10MEG
*L1 N001 N002 244µ (this is the inductor i removed from the circuit and replaced with the COMSOL multi-turn coil)
R2 N002 N003 59.4
C1 N003 0 138n
V1 N001 0 PULSE(0 813 2u 0.1u 0.1u 10u 30u 10) Rser=1u Cpar=100n
.tran 0 100u 0 5n
.backanno
.end

--
:)
↧

How to continue particle tracing from a certain time point?

March 15, 2016, 8:10 am
$
0
0
Dear all,

I am now working on the simulation of particle tracing with fluid flow.

I have finished the staionary fluid flow and time dependent particle tracing calculation and obtain the particle tracing results.

Now I want to apply a certain calculated time point particle to continue with another turbulent flow mode. Are there anyone knowing how to do this?

Thank you very much.
↧

Parametric solver does not support parameters in geometry or mesh.

March 15, 2016, 9:24 am
$
0
0
Hello guys,

When I change the size of a geometry with a parameter, parametric don't work as I wish. And there is a warning:Parametric solver does not support parameters in geometry or mesh.

is there anyone who has an idea about this issue?

Thanks
↧
Search

Error: "Attempt to evaluate negative power of zero."

March 15, 2016, 12:22 pm
$
0
0
I'm building a model using the Reaching Flow in Porous Media in a 2D axis-symmetric component, and my goal is to remove some of the dilute species from the system as a function of concentration and fluid velocity (e.g. the reaction slows as concentration decreases, and is increased when fluid velocity is high).

I can get the model to behave as expected, but I cannot successfully implement a velocity dependent Reaction. I am able to use a constant (e.g. -0.01 [mol/m^3/s]) or concentration dependent reaction rate, but if I try to use average velocity I get the error:

Attempt to evaluate negative power of zero.
- Function: ^
Failed to evaluate temporary symbolic derivative variable.
- Variable: comp1.rfds.U@VDN$comp1.w2
- Defined as: (0.5*(((comp1.u2^2)+(comp1.w2^2))^(-0.5)))*(2*comp1.w2)
Failed to evaluate expression.
- Expression: d(comp1.rfds.U,comp1.w2)

I attempted to incorporate the velocity in the porous domain (or any domain.. just trying to get anything to work for starters) by setting up an Average of the porous domain (Component --> Definitions --> Average), then defining a variable as the velocity magnitude of that domain (comp1.aveop1(rfds.U)).

I have consulted other threads here and saw recommendations to use 'eps' when defining my variable to prevent my initial expression from being zero, but this doesn't seem to help. Any guidance is appreciated, and I can provide further details of the model as necessary.


↧
↧

Error Index:2, Size 2, Feature: Stationary Solver 1

March 15, 2016, 12:23 pm
$
0
0
Hi

I have a Frequency Domain study looking at the dynamic response of an un-deformed structure immersed in water. I am using Solid Mechanics, Pressure Acoustics - Frequency Domain (acpr) and Multi-physics defined in the model. This is a problem type I have solved many times (recently upgraded to v5.2).

Initially looking to solve the model at zero pressure (un-deformed) but I unfortunately get an error message "Index: 2, Size: 2" and despite looking through the model numerous times, resetting to defaults etc. I can not resolve this!

Can anyone help give an indication of what this error may relate to ...its so brief its unhelpful!

Unfortunately I am unable to share the model - sorry

Thanks in advance Colin
↧

Solving air-cups inner pressure and its inner air leakage with giving heavy weight in time dependent

March 15, 2016, 1:00 pm
$
0
0
Hello everyone
I want to solving air-cups inner pressure and its inner air leakage with giving heavy weight in time dependent
But i don't how to set air leakage with giving heavy weight on air-cups!!
For now, i can solve air-cups inner pressure with giving heavy weight in steady state , so i want to know the result if my air-cups leakage just like in my real case.
And i want it to leak with time , so i use time dependent to simulate it, but time has failed to converge.

PS. My model refer to 〝Hyperelastic Seal〞in COMSOL model example!

Thanks everyone help!!
↧

How to get a solution for EDL

March 31, 2012, 5:16 am
$
0
0
Hi.

I'm trying to solve the electric double layer structure using Nersnt-Planck eqn and Poisson eqn rather than Poisson-Boltzmann eqn.

My domain is pallel plate of which height is nanometric, typically ~10nm.

However, my model failed to converge.

Sometimes, converge is succed. But that result is not coincided with analytical solution.

What is the problem?

Is there anyone solved this problem?
↧

Custom Linear Solver

March 17, 2016, 5:24 am
$
0
0
Hi,
I'm wondering whether it is possible to use custom linear solver for my model on comsol.
For example, can I use my original solver implementation instead of MUMPS or PARDISO at Study ->Solver configuration -> Solution -> Direct Solver -> Solver?
↧
↧

Water Electrolysis through SOEC

March 17, 2016, 11:19 am
$
0
0
Hello everybody,
I'm trying to simulate water electrolysis in an SOEC. In my stationary model I use SIEC physic, two TCS respectively for cathode and anode and two FP in the same manner. In my model I have defined equilibrium potentials described by Nernst equation for anode and cathode.
I'm simulating three different operative conditions at a constant temperature of 850°C. Endothermic with Phis=1.1V, termoneutral voltage with Phis=1.28V and exothermic with Phis=1.4V. I started using Butler Volmer equation for calculating the current. In the first condition when I define Nernst potential using the concentration of the reactants [mol/m^3], I have no problem with the solution, at least with the distribution of mole fraction, while the electrolyte current density vector in z direction, which is normal to the plane of reaction, is constant, and I would that it decreases from the inlet to the outlet of the cell according with the decrease of water concentration and the increase of hydrogen. I suppose it is because I have to use the concentrate dependent kinetic to calculate the current, but in this case the solution gives me a constant distribution of mole fraction, as if the physics are not coupled, but they are.
If I use mole fraction defining the Nernst potential, the solution don't converge.
Another problem is if I try to simulate the conditions with Phis=1.28V and 1.4V, in these cases the solutions never converge.
Can you help me to understand where I'm wrong?

Thanks

P.s. I have used the COMSOL Current Density Distribution in a Solid Oxide Fuel Cell tutor as support
↧

Comsol Paralel Computing Problem !

March 17, 2016, 4:39 pm
$
0
0
I want to use Comsol Paralel computing with Windows HPC 2008 R2. There are five Nodes one of which is Head-Node (Windows SERVER HPC). Among the workstations one Head-Node + quantity of four Workstation Node(Windows 7 ) only A workstation Node works actively with Parallel computing and when I try to make the others WORK, it says " Scheduling" and it freezes like that gives no other error.
All the settings in A workstations exist also in B-C-D Workstation Nodes. What might be the reason for Comsol being not able to make parallel computing between Head-Node and A Node while it makes it with the others? If that was a license problem, should not it suppose to give an error between A and Head-Node? Which parts I should check on Windows HPC and Comsol?


Thankyou.
↧
Search

In Segregated Step 1: There was an error message from the linear solver. The relative error (2.2e+002) is greater than the relative tolerance.

March 17, 2016, 7:17 pm
$
0
0
Hi all,

Why does this error happen and how do I solve it?

This is the log file:

============================================================
Minimum element quality: 0.03333
============================================================
Stationary Solver 1 in Study 1/Solution 3 (sol3) started at 18-Mar-2016 08:12:59.
Nonlinear solver
Number of degrees of freedom solved for: 86118.
Nonsymmetric matrix found.
Scales for dependent variables:
Electric potential (mod1.V): 29
Displacement field (Material) (mod1.u): 6.5e-008
Pressure (mod1.p): 1e+005
Displacement field (Material) (mod1.u2): 1e-008
Orthonormal null-space function used.
Iter SolEst ResEst Damping Stepsize #Res #Jac #Sol LinErr LinRes
1 0.065 3.2 1.0000000 7.2 2 1 2 2.1e-010 1.9e-011
2 3e-006 0.00061 1.0000000 0.031 3 2 4 1.7e-010 8.3e-012
Stationary Solver 1 in Study 1/Solution 3 (sol3): Solution time: 3 s
Physical memory: 1.28 GB
Virtual memory: 1.41 GB
Time-Dependent Solver 1 in Study 1/Solution 3 (sol3) started at 18-Mar-2016 08:13:04.
Time-dependent solver (BDF)
Number of degrees of freedom solved for: 73170.
Nonsymmetric matrix found.
Scales for dependent variables:
Pressure (mod1.p2): 1e+005
Velocity field, continuous phase (Spatial) (mod1.uc): 3.8e+002
Velocity field, dispersed phase (Spatial) (mod1.ud): 3.4e+002
Nonsymmetric matrix found.
Scales for dependent variables:
Volume fraction, dispersed phase (mod1.phid): 0.0093
Nonsymmetric matrix found.
Step Time Stepsize Res Jac Sol Order Tfail NLfail LinErr LinRes
0 0 - out 22 15 22 0
Group #1: 11 3 11 0.005 1.2e-013
Group #2: 11 12 11 1.7e-013 1.5e-016
1 2.5e-005 2.5e-005 94 65 93 1 0 1
Group #1: 34 5 34 0.00032 5.8e-014
Group #2: 60 60 59 1.4e-013 2.2e-016
2 7.5e-005 5e-005 100 69 99 1 0 1
Group #1: 37 6 37 0.0094 3.9e-014
Group #2: 63 63 62 1.4e-013 2.2e-016
3 0.000175 0.0001 102 71 101 1 0 1
Group #1: 38 7 38 0.0024 5.5e-014
Group #2: 64 64 63 1.4e-013 1.8e-016
4 0.000375 0.0002 104 73 103 1 0 1
Group #1: 39 8 39 0.0047 2.8e-014
Group #2: 65 65 64 1.4e-013 1.7e-016
5 0.000775 0.0004 107 76 106 1 0 1
Group #1: 40 9 40 0.0042 1.6e-014
Group #2: 67 67 66 1.6e-013 1.8e-016
6 0.001575 0.0008 110 79 109 1 0 1
Group #1: 41 10 41 0.00086 2.2e-014
Group #2: 69 69 68 1.9e-013 1.6e-016
7 0.003175 0.0016 122 88 121 1 0 1
Group #1: 45 11 45 0.0078 3.5e-014
Group #2: 77 77 76 1.6e-013 1.6e-016
8 0.006375 0.0032 125 91 124 1 0 1
Group #1: 46 12 46 0.0054 7.2e-014
Group #2: 79 79 78 1.4e-013 1.5e-016
- 0.01 - out
9 0.012775 0.0064 150 109 149 1 0 1
Group #1: 54 13 54 0.00069 3.9e-014
Group #2: 96 96 95 1.6e-013 1.6e-016
- 0.02 - out
10 0.022775 0.01 190 133 189 1 0 1
Group #1: 71 14 71 0.0098 1.4e-014
Group #2: 119 119 118 2.4e-013 2.1e-016
- 0.03 - out
11 0.032775 0.01 195 137 194 1 0 1
Group #1: 73 15 73 0.004 1.1e-014
Group #2: 122 122 121 1.6e-013 2.1e-016
- 0.04 - out
12 0.042775 0.01 204 144 203 1 0 1
Group #1: 76 16 76 0.0082 5.6e-014
Group #2: 128 128 127 1.8e-013 1.6e-016
- 0.05 - out
13 0.052775 0.01 235 161 234 1 0 1
Group #1: 91 17 91 0.00085 8.7e-015
Group #2: 144 144 143 1.9e-013 1.6e-016
Warning: In Segregated Step 1:
Warning: There was an error message from the linear solver.
Warning: The relative error (2.2e+002) is greater than the relative tolerance.
Stopped
Time-Dependent Solver 1 in Study 1/Solution 3 (sol3): Solution time: 3143 s (52 minutes, 23 seconds)
Physical memory: 1.67 GB
Virtual memory: 1.74 GB
↧

.lock file

March 18, 2016, 1:37 am
$
0
0
I am trying to do a dynamic simulation using particle tracing module.

However, at a certain time, COMSOL got frozen with creating .lock file.
I don't think that there are any busy computation to cause crash.
Can anyone give me any clue?
↧
↧

stop condition error: stop condition fulfilled for initial values

March 18, 2016, 7:01 am
$
0
0
I am trying to build a model of a 2D li-air battery based on the 1D li-air batt tutorial. I have made it most of the way through the model until I compute, and the error appears:

stop condition fulfilled for initial values.


I have checked all the initial conditions which as correct as for the 1D model, as I understand it the stop condition stops the sweep when the electric potential drops from the initial 2.95V to 2.5V, what the error suggests is that the electric potential is 2.5V or less to start with...

Can anyone see how this has happened? I have only been using comsol for a week or so, so would appreciate the help.
thanks
↧

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
↧

Nonlinear Dynamic Response of a Clamped-Clamped Euler-Bernoulli Beam

March 21, 2016, 1:49 am
$
0
0
I am currently modelling the nonlinear time response of a clamped-clamped Euler-Bernoulli beam using the geometric nonlinearity tool in COMSOL. I intend on doing so for a wide frequency range in order to show the dynamic hardening response of the beam. Currently, I am having some problems with the quality of my solution and the physical interpretation.

From this link, wiki.csiamerica.com/display/kb/Damping+coefficients, and my understanding of Rayleigh Damping parameters, I believe that the damping values used for my COMSOL simulation plays a big role with the accuracy of my solution.

My main question are:
How can I accurately represent viscous damping with the Rayleigh mass and stiffness damping parameters for a wide range of frequency values?
How does the solver method affect the accuracy and speed of geometric nonlinear problems on COMSOL?

Thanks in advance,
--
FA
↧
Viewing all 3988 articles
Browse latest View live
Search
<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>