COMSOL simulation for heat and mass transfer across membrane for membrane distillation process

February 20, 2014, 12:32 am

≫ Next: ODE simple problem

≪ Previous: How to solve 6 Coupled Mode Non linear Partial Differential Equation

$

0

0

Hi,
For your information, I am developing a COMSOL simulation for the direct contact membrane distillation (DCMD) process. It is actually a process which involve the heat and mass transfer from the bottom cell (hot feed side) across the membrane (porous medium) and reach the upper cell (cold permeate side).

upper cell (hot feed side) --> heat convection
membrane (hydrophobic) --> heat convection (water evaporation) + heat conduction in membrane
lower cell (cold permeate side) --> heat convection (vapor condense back to water)

I am currently using 2 physics which are conjugate heat transfer + free and porous media flow.
I was able to simulate the convective heat transfer at the upper and lower cell + heat conduction in membrane matrix. However, no mass flux was flowing across the membrane, which means no heat convection was developed within the membrane.

Thus I would like to ask is it possible to simulate the heat convection through membrane together with the flows at the cells? How do I link the three layers together so that there is fluid flowing from the bottom to the upper side?

Thanks for your time and wish you have a good day.

Patricia

---

ODE simple problem

February 20, 2014, 7:15 am

≫ Next: 4th order pde

≪ Previous: COMSOL simulation for heat and mass transfer across membrane for membrane distillation process

$

0

0

Hi, I have a two problems.
I'm trying to solve simple differential equation in the form of:
dA/dt=-k*t*A
initial conditions:
k=1
A=5e-4
At=5e-4
This equation is describing reaction kinetics. To be more precise, first order reaction.
My problem is that obtained kinetic curve, is rather sigmoidal not expotential as expected. I was trying to change time step without success.
My second problem is more complex.
dA/dt=-k*t*(A^0.75)
initial conditions:
k=1
A=5e-4
At=5e-4
ERROR:
Nonlinear solver did not converge.
Time : 0.3011554385215172
Attempt to evaluate non-integral power of negative number.
Last time step is not converged.
How to fix those problems? Any ideas?
Best regards
Marek

4th order pde

February 20, 2014, 2:55 am

≫ Next: human head skull stimulation by current source electrods

≪ Previous: ODE simple problem

$

0

0

Hello everybody,

I have 4th order pde equation (Navier sokes lubrication aproximation). As I know I need to make intermediate depended variables of 2nd order. Maybe somebody could tel me where I need to fill in these intermediate dependent variable in comsol 4.2

Thank you

human head skull stimulation by current source electrods

December 4, 2011, 5:35 pm

≫ Next: Cluster Computing Convergence Plot

≪ Previous: 4th order pde

$

0

0

Hi

I am new to Comsol. I have tried to build a simple model resembling human head including three layers of skull, csf and brain using the provided conductivities. I then built a tri-pole electrode which the center one is grounded and the middle and the outer layer are given a current or current density.

My problem is that I fail to get any solution and I face an the following error:

Failed to find a solution.
Segregated group 1

Singular matrix.

There are 12414 void equations (empty rows in matrix) for the variable mod1.V
at coordinates: (-0.00104638,-0.000753554,0.10675), (-0.000753557,-0.00104637,0.10675), (-0.00119282,-0.000400008,0.10675), (0.0013,4.30826e-019,0.10675), (-2.29715e-020,0.00115,0.106625), ...
and similarly for the degrees of freedom (empty columns in matrix)
Returned solution is not converged.

Cluster Computing Convergence Plot

February 24, 2014, 1:17 pm

≫ Next: Unrealistic results in Elasto-Plastic analysis: Thermal Stress Problem

≪ Previous: human head skull stimulation by current source electrods

$

0

0

Is it possible to view a convergence plot when solving a problem on a HPCS2008 cluster? I know in the log file generated by the cluster that completion percentages are generated. If it is not possible on a HPCS2008 cluster Is this something possible on a linux cluster?

Unrealistic results in Elasto-Plastic analysis: Thermal Stress Problem

February 24, 2014, 8:41 pm

≫ Next: Generalized alpha settings

≪ Previous: Cluster Computing Convergence Plot

$

0

0

I am modeling a Thermal stress analysis of 3D wire wherein Temperature field (x,y,z) was imposed by using Interpolation function (imported a .txt file). It is simply an Elastic-Plastic analysis. Because of higher values of imposed temperatures, specimen goes into plastic region. I set the Linear elastic-Perfectly Plastic model (i.e. The plastic region starts when induced stress exceeds the Yield strength of the material).
However, the result are unrealistic i.e. Elastic strain are very low, a zero plastic strain and only thermal strain is very high. I think the total strain is mainly showing up as Thermal strain and contribution of other strains are almost zero.

I have attached the report file. Need help. Any comments or suggestions are appreciated.

Generalized alpha settings

February 24, 2014, 11:58 pm

≫ Next: Convergence in convective heat transfer

≪ Previous: Unrealistic results in Elasto-Plastic analysis: Thermal Stress Problem

$

0

0

Hello everyone

Can someone indicate me the different settings for the "maximum time step" and "time step increase delay" when choosing the generalized alpha tims solver?

For example, I'd like to have a maximum time step of 1e-3s for t<0.1s and then the maximum can go up to 1e-2...

I thought it was possible to do so with previous version of comsol...

Thanks!

Convergence in convective heat transfer

February 25, 2014, 1:57 am

≫ Next: What would be your ideal COMSOL workstation?

≪ Previous: Generalized alpha settings

$

0

0

Hi there,
I am trying to replicate the model of a published paper about the performances of horizontal micro-fins in natural convection (Mahmoud et al. (2011). Effect of micro fin geometry on natural convection heat transfer of horizontal microstructures. Applied Thermal Engineering, 31(5), 627–633). The essential part is that:
" The air cylinder was modeled as a smooth adiabatic wall with an opening boundary condition at its top to allow the air flow to go in and out of the computational domain. The bottom and side surfaces of the insulation material were modeled as adiabatic boundaries. The temperature of inflow from the top was 293K. The simulation started with an initial temperature throughout the computational domain of 293K. The heat was added to the domain through a source term in the energy equation of the heating element sub domain. The buoyant flow above the copper plate and inside the cylinder was modeled as laminar natural convective heat transfer condition with radiation effects included."

I create the same geometry and tried to replicate the functions as well. In order to improve the convergence, I used some tips found on the forum and on the examples: I am using a Highly Nonlinear solving method and a continuation parametric solver for the air density.

Since the model is not yet giving a solution, I would like to know if anyone can suggest some tips or feedback. I hope I have been clear enough: if not, please let me know if you need any other information. Many thanks in advance.
Leo

---

What would be your ideal COMSOL workstation?

August 13, 2010, 6:01 am

≫ Next: COMSOL Error for Initial Values(RF Module)

≪ Previous: Convergence in convective heat transfer

$

0

0

Hello all,

I am in the process of building my own COMSOL workstation, and am looking for advice. If you were building your own computer from scratch, what parts would you be sure to get (besides a supercomputer, of course)?

COMSOL's list of system requirements is pretty basic -- 1 GB of memory, 1-4 GB of hard disk space. I'd like to run 3D simulations with multiple physics, and also use parametric sweeps for optimization. Mainly, I'm looking to solve simulations as fast as I can.

Does anyone have any hardware recommendations?

Is there a noticeable difference between 64 bit and 32 bit performance?

Any speed differences between Windows (XP or 7) and Linux?

Optimal graphics card and memory size?

Number of cores?

I'm currently looking at an Intel quad-core processor and 8 GB of memory running 64-bit Windows 7. What do you all think?

COMSOL Error for Initial Values(RF Module)

June 16, 2011, 12:23 am

≫ Next: No species released when using Transport of Species model

≪ Previous: What would be your ideal COMSOL workstation?

$

0

0

I am defining a model with different concentrations of oil and water and trying to measure the heat distribution by using the microwaves. But I obtained the error mentioning Initial values:

THE FOLLOWING FEATURE HAS ENCOUNTERED A PROBLEM
-FEATURE:TIME DEPENDENT SOLVER
-ERROR:FAILED TO FIND CONSISTENT INITIAL VALUES

Failed to find consistent initial values.
Undefined value found.
NaN or Inf found when solving linear system using SOR.

Although the initial values defined are by default and I did not change any thing.Please see the attached file.

Can anybody suggest me some solution of these problems?

Thanks in advance.

No species released when using Transport of Species model

February 26, 2014, 12:50 pm

≫ Next: To demonstrate a "sieve" at the particle tracing module

≪ Previous: COMSOL Error for Initial Values(RF Module)

$

0

0

Hi,

I am trying to add some tracer to the bulk flow. In Study 1, I am running a Steady air flow, and then in Study 2, I add a tracer to the flow. The tracer is released from a point (a source using weak contribution ) and concentration changes with time with a Gaussian Pulse (gp1). To set up this problem, I am pretty much following the steps taken in the Turbulent_mixing.mph in the Model Library model. Although the solution does not diverge, but I am not getting any released tracer in the domain.

I would appreciate it if someone could look at the attached file and let me know what has caused zero tracer release.

Mazyar

To demonstrate a "sieve" at the particle tracing module

February 26, 2014, 7:14 pm

≫ Next: How to transmit solution data between two models?

≪ Previous: No species released when using Transport of Species model

$

0

0

Hi all, now I'm trying to demonstrate a particle sieve using the particle tracing module.
I expected that the particles bigger than the pore size of the sieve could be filtered and stuck on the sieve surface.
But it seems that the particle tracing module cannot demonstrate such effect because every particle can pass through the pores even which are much smaller than those particles!

Do you think that I have to set up a sort of force equation or "if-condition" which can describe the sieve effect by the size difference??

Thanks!!

Dukwoo

How to transmit solution data between two models?

June 16, 2012, 12:23 am

≫ Next: coupled PDEs

≪ Previous: To demonstrate a "sieve" at the particle tracing module

$

0

0

Hello!
How to transmit solution data between two models?
I have set up two models as follows.
1. To solve u1 satisfied the PDE, D1(u1)=f1, in the first model;
2. to solve u2 satisfied the PDE, D2(u2)=u1, in the second model.
I try to transmit the data of u1 by interpolation function by the following steps.
1. Solve u1 in the first model and save u1 in text file, u1.txt, in the formate described in the help document for interpolation function;
2. create interpolation function, u1, in the second model from the text file, u1.txt, and then solve u2.
This solution does work, but the problem is time consuming by writing and reading data of u1, because in 3D model, there are many many interpolation points.
In my opinion, the solution data of u1 interpolated by regular interpolation points may lead to accurracy lossing. Another question is how to transmit the solution between model in the form of mesh nodes.
--
Computational Electromagnetics, Finite Element Method, Computer Aided Engineering.

coupled PDEs

February 27, 2014, 6:25 am

≫ Next: FSI Module for blood vessel simulation

≪ Previous: How to transmit solution data between two models?

$

0

0

Hello,

I want to find the profiles of two fields (u1(x,y) and u2(x,y)) on a 2d geometry with known boundary conditions that satisfy the following equation:

[(u2y*u1x-u1y*u2x)^2]x + [(u2y*u1x-u1y*u2x)^2]y = 0

The geometry is a square box with the following boundary conditions:
left edge: u1=1, u2=no flux
right edge: u1=0.01, u2=no flux
top: u1=no flux, u2=1
bottom: u1=no flux, u2=0.01

I know (at least one) solution which is u1(x,y)=1-0.99*x and u2(x,y)=0.01+0.99*y. If I give COMSOL this solution as initial condition (or actually even exponential profiles satisfying the boundary conditions) it returns it as expected. However when I start e.g. with a random distribution it doesn't find it. What can be the problem? And what solver configurations make sense? I am quite new to COMSOL, so I have no idea what to change...

And the future idea is to extend the problem to different shapes, so trying it on the square is trivial but I need to make it work there before going to a geometry of which I don't know the answer.

Thank you - I am really happy for any hints/ideas/... :)
Kathrin

FSI Module for blood vessel simulation

February 27, 2014, 6:39 am

≫ Next: studies with multiples steps

≪ Previous: coupled PDEs

$

0

0

Hi,

I'm trying to implement a very basic 2D axisymmetric model of a blood vessel (straight tube) using the FSI module as I'm interested in getting a wall deformation profile. The problem is that my model always fails with the following error:

"Failed to find consistent initial values.
Segregated group 2

Attempt to evaluate real square root of negative number."

I have tried finer meshes and get the same error message. My suspicion is that the BC are not correct. For example I am unsure as to how I should define the Boundary Load on the FSI Boundary. At the inlet I would really only like to specify the pressure using the piecewise function that creates a pressure pulse.

Would someone with more experience with the FSI module be able to help me?

Thank you
Alexandra

---

studies with multiples steps

February 17, 2014, 6:14 am

≫ Next: Probe plots and plot while solving in Comsol 4

≪ Previous: FSI Module for blood vessel simulation

$

0

0

Hi all,
there is something so weird with my COMSOL 4.3, when I create a time dependent study with different steps the solution only represents the last step and It is not posible to change the time interval and impose the one that I want to have (from the beginning to the end), what can I do to solve it?
Thanks a lot.

Probe plots and plot while solving in Comsol 4

June 1, 2010, 8:56 am

≫ Next: magnetic field gradient in 3D return zero!

≪ Previous: studies with multiples steps

$

0

0

How to have "plot while solving" and probe plots in different windows when solving a simulation? When I simply enable them both they show up alternately in the "graphics" tab, making them useless.


Comsol v4.0.0.937

magnetic field gradient in 3D return zero!

March 19, 2013, 8:35 pm

≫ Next: How to use solution as initial value for 2nd study with changed geometry?

≪ Previous: Probe plots and plot while solving in Comsol 4

$

0

0

Hi everybody

If anybody knows that why the gradient of magnetic field in 3D models is always zero? I checked the different methods. for example I wrote d(Bx,x) or different component of B and H. Even d(d(Ax,x),x) is always zero! I am using the quadratic discretization. It is OK when I simulating in 2D or 2D axisymmetric, However the similar geometry in 3D give the zero for gradient of magnetic field.

could somebody help me?

How to use solution as initial value for 2nd study with changed geometry?

March 23, 2013, 7:37 am

≫ Next: ALE moving mesh + electrostatics force calculation are in conflict

≪ Previous: magnetic field gradient in 3D return zero!

$

0

0

Hello Everybody. I’m using comsol 4.3a.

I try to use the solution from study 1 (stationary heat transfer) as the initial temperature value for a second time dependent study. Further a new geometry interval with new initial value and BC has to be added to the model for to the second study.

Can anybody give me some help to realize that? Thanks a lot in advance.

ALE moving mesh + electrostatics force calculation are in conflict

June 20, 2013, 7:42 am

≫ Next: calculate the electric field near a pylon

≪ Previous: How to use solution as initial value for 2nd study with changed geometry?

$

0

0

Hello Everyone,
thanks for all the support or tips you may give on the following problem.
I am trying to couple three different modules, to monitor the displacement of a bubble in a channel in 3D under the effect of an electric field.
I solve the displacement of the bubble through a differential equation in the ODE solver, for which one term is given by the electromagnetic force computed by the electrostatic module at each position of the bubble.
The force computation is done on the domain of the bubble. The bubble boundaries move with the prescribed mesh displacement given by the solution of the ODE solver. I use a ALE moving mesh to move the bubble.
I have verified the single physics/modules together and coupled two by two. The ODE and ALE work fine together with an arbitrary force not given by the electrostatic module. The ODE and electrostatic work well together and now that I am coupling the ALE with the electrostatic module, Comsol does not recognize the Maxwell stress tensor anymore (the variable mod1.es.unTz evaluated on the boundaries of the bubble.
I don't understand why with adding the moving mesh ALE, the simulation just doesn't start anymore and says it fails to evaluate the variable. I attach my model file for clarifications. I hope anybody out there can point some way around this.
Thanks a lot.

Stefano