Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Fluid Structure Interaction with Transport of diluted species

Please login with a confirmed email address before reporting spam

Hello,

I'm new to COMSOL and i'm trying to model the following: First of all i'm dealing with a microfluidic device capable of microfluid mixing. I start with a 2D Model with a rectangular body representing the fluid. Within the center of the fluid i placed a circle which represents a PMMA Particle. For the FSI i first started with the time dependent FSI model. For the particle i added a "Solid Mechanics -> Prediscribed Displacement" property with the following function for the displacement in the x-direction: 5[µm]*sin(1[1/s]*t) and 0 for the y-displacement.

These Settings worked fine for me but i got problems when i wanted to place the particle at the proximity of the fluid wall, lets say at a gap of about 1 nm or something like that. I guess, that there are some problems during the solving progress since the movement of the particle in vicinity to the wall induces a very strong buoyancy, i.e. the particle wants to move in the y-direction, but i'm not really sure about it at all.

As a result, i observed the typical behaviour of the lamellae velocity, which is exponentially decreasing along the y-direction und modulated with a periodic function, i.e. a sine.

In the next spet, i wanted to use the calculated velocity profile of the time dependent FSI simulation within a Transport of Diluted Species Simulation because i want to know, how the movement of the particle effects the mixing efficiency which is correlated to the mass transport within the device.

So first of all, i made a Transport of Diluted Species simulation without FSI within the same geometry and the same Materials. I defined a step profile for the concentration, centered at x=0. The time dependent model worked well and i observed the diffusive progress.

Now i come to my Problem: As soon as i add the FSI to my Transport of diluted species model, where i explicitly deactivate the FSI in my time dependent solver model, i got a nother solution for the Transport model, i.e. the concentration is 0.5 for all times and positions. When i disable FSI, than the model works fine.

I guess that these two models might interfere with each other, but i'm not sure at all.

Do you have any suggestions for me, y i'm not able to run both models simultaneously?


Thanks for your answers and best regards!

4 Replies Last Post 1 févr. 2012, 15:36 UTC−5
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 31 janv. 2012, 14:02 UTC−5
Hi

A few things, if you have a "solid" sphere of 5 um, and you want to model it with FSI in a fluid you need some mesh elements around it, so when you say you approach a solid wall by 1 nm, I'm afraid there is not really much place for a mesh between the sphere and the wall, you will have some issues on how to handle the mesh with such dimensions

Also when you add or remove multiple physics models such as FSI which have many dependent variables, you must ensure that these variables do not interact, or if so that you control how they interact. often simple physics uses p and T and ... with given default values, then when you add another physics you redefine some of these, hence change their values, or they become dependent variables and will influence the other physics, sometimes in an unexpected way

--
Good luck
Ivar
Hi A few things, if you have a "solid" sphere of 5 um, and you want to model it with FSI in a fluid you need some mesh elements around it, so when you say you approach a solid wall by 1 nm, I'm afraid there is not really much place for a mesh between the sphere and the wall, you will have some issues on how to handle the mesh with such dimensions Also when you add or remove multiple physics models such as FSI which have many dependent variables, you must ensure that these variables do not interact, or if so that you control how they interact. often simple physics uses p and T and ... with given default values, then when you add another physics you redefine some of these, hence change their values, or they become dependent variables and will influence the other physics, sometimes in an unexpected way -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 1 févr. 2012, 13:57 UTC−5
Thx Ivar for your reply!

Yes, indeed, i also spended a lot of time for the gap between my particle and the substrate surface. To my mind, the easiest way without falsifying the real physics to much would be to place the particle in center (y-direction) of the fluid container and then cut the solution below the particle. Do you agree, that this would be a good approximation of the problem?

Conerning FSI and Transport of diluted Species: Based on your reply, i checked all dependent variables of the two models and tried a lot, but nothing changed until i set the boundary "fixed mesh" to my FSI. When i set "fixed mesh" to the FSI model for the fluid volume, the mesh stays constant during the solver progress but the time dependent behaviour of the fluid lamellae velocity is quite simular to my results when the bead is moving within the mesh. When the mesh is fixed, the transport of diluted species model works quite well together with the FSI.

Pls correct me if i'm wrong, but to my mind, when i set "fixed mesh" it seems to me that COMSOL changes then reference system for the solver, i.e. we take a look from the particle reference system and the fluid starts moving based on the displacement of the particle which is zero within the particle reference system?

Tomorrow i will have a look on how COMSOL deals with the problem, when i approach the particle near the substrates surface while keeping the mesh fixed because i observed that within the ALE moving mesh simulation, the structure of my mesh is destroyed at some time step and than my simulation breaks down. Maybe i can fix the problem when keeping my mesh fixed!

Best regards,

Stuart
Thx Ivar for your reply! Yes, indeed, i also spended a lot of time for the gap between my particle and the substrate surface. To my mind, the easiest way without falsifying the real physics to much would be to place the particle in center (y-direction) of the fluid container and then cut the solution below the particle. Do you agree, that this would be a good approximation of the problem? Conerning FSI and Transport of diluted Species: Based on your reply, i checked all dependent variables of the two models and tried a lot, but nothing changed until i set the boundary "fixed mesh" to my FSI. When i set "fixed mesh" to the FSI model for the fluid volume, the mesh stays constant during the solver progress but the time dependent behaviour of the fluid lamellae velocity is quite simular to my results when the bead is moving within the mesh. When the mesh is fixed, the transport of diluted species model works quite well together with the FSI. Pls correct me if i'm wrong, but to my mind, when i set "fixed mesh" it seems to me that COMSOL changes then reference system for the solver, i.e. we take a look from the particle reference system and the fluid starts moving based on the displacement of the particle which is zero within the particle reference system? Tomorrow i will have a look on how COMSOL deals with the problem, when i approach the particle near the substrates surface while keeping the mesh fixed because i observed that within the ALE moving mesh simulation, the structure of my mesh is destroyed at some time step and than my simulation breaks down. Maybe i can fix the problem when keeping my mesh fixed! Best regards, Stuart

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 1 févr. 2012, 15:04 UTC−5
Hi

Could you pls remind me with the physics you use and the order they appear in your model tree (top down), just use the identifier name FSI or ALE ...
Just to see if I manage to understand the main interactions

--
Good luck
Ivar
Hi Could you pls remind me with the physics you use and the order they appear in your model tree (top down), just use the identifier name FSI or ALE ... Just to see if I manage to understand the main interactions -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 1 févr. 2012, 15:36 UTC−5
Hello Ivar,

i could do this. My 2D Model (mod1) consists of the two time dependent physics: Fluid Structure Interaction (1.) and Transport of diluted Species (2.). The geometry is simply a rectangle centered at (0,0) with a width of 100 µm and a height of 1000 µm. At the center (0,0) i place a circle with a Radius of 5 µm. The two materials are water for the rectangle and PMMA for the particle. In the Fluid Structure Interaction Study, i use the mesh smoothing type "Hyperelastic" and the option "incompressible flow". At the "linear elastic material" node i selected the sphere and activate "nearly incompressible material". Than i added a node "solid mechanics -> prediscribed displacement" for the particle, where the v(x) is 5[µm]*sin(1[1/s]*t) and v(y)=0. Than i added a last node "Fixed mesh" for the rectangle fluid domain because then the interaction with the "Transport of dilued Species" physics worked for me.

Afterwards i added a time dependent "Transport of diluted Species" to my model with convection and i only activate this physics for the fluid domaine, i.e. the rectangle. At the "convection and diffusion" node i added velocity field "fsi/fpi" at the velocity field option and set the diffusion coefficient to "1e-11[m^2/s]. At the "initial values" node i selected the rectangular fluid domain and added a concentration named "conc" which is defined as follows:

In the "Global definitions" i added a step function "step1" centered at x=0 with a step from 0 to 1 and a smoothing transition of 0.01. At the "Definitions" in model1 i added the variable "conc" which is defined as "1[mol/m^3]*step1(x[1/µm])".

For my mesh i added a free triangular mesh for the entire geometry, where the size node is set to "fluid dynamics" with a "finer" meshing option. I than added boundary layers for the partcle boundaries, where i set the number of boundary layers to 5.

At the solver node, i activated both physics to be solved simultaneously for a time of 5 seconds. These settings worked for me.

But when i deactivate the "fixed mesh" option for the rectangular fluid domain, the particle moves due to the displacement boundary and my mesh is deformed, which gives me a concentration of 0.5 for every time steps when i keep the other settings konstant.

I also tried to add some additional boundaries for the density, temperature and pressure of the fluid but the multiphysics only works when i set the mesh of the fluid to be fixed.

I hope you can follow me and reconstruct the model in your mind.

Thx for your help and best regards!
Hello Ivar, i could do this. My 2D Model (mod1) consists of the two time dependent physics: Fluid Structure Interaction (1.) and Transport of diluted Species (2.). The geometry is simply a rectangle centered at (0,0) with a width of 100 µm and a height of 1000 µm. At the center (0,0) i place a circle with a Radius of 5 µm. The two materials are water for the rectangle and PMMA for the particle. In the Fluid Structure Interaction Study, i use the mesh smoothing type "Hyperelastic" and the option "incompressible flow". At the "linear elastic material" node i selected the sphere and activate "nearly incompressible material". Than i added a node "solid mechanics -> prediscribed displacement" for the particle, where the v(x) is 5[µm]*sin(1[1/s]*t) and v(y)=0. Than i added a last node "Fixed mesh" for the rectangle fluid domain because then the interaction with the "Transport of dilued Species" physics worked for me. Afterwards i added a time dependent "Transport of diluted Species" to my model with convection and i only activate this physics for the fluid domaine, i.e. the rectangle. At the "convection and diffusion" node i added velocity field "fsi/fpi" at the velocity field option and set the diffusion coefficient to "1e-11[m^2/s]. At the "initial values" node i selected the rectangular fluid domain and added a concentration named "conc" which is defined as follows: In the "Global definitions" i added a step function "step1" centered at x=0 with a step from 0 to 1 and a smoothing transition of 0.01. At the "Definitions" in model1 i added the variable "conc" which is defined as "1[mol/m^3]*step1(x[1/µm])". For my mesh i added a free triangular mesh for the entire geometry, where the size node is set to "fluid dynamics" with a "finer" meshing option. I than added boundary layers for the partcle boundaries, where i set the number of boundary layers to 5. At the solver node, i activated both physics to be solved simultaneously for a time of 5 seconds. These settings worked for me. But when i deactivate the "fixed mesh" option for the rectangular fluid domain, the particle moves due to the displacement boundary and my mesh is deformed, which gives me a concentration of 0.5 for every time steps when i keep the other settings konstant. I also tried to add some additional boundaries for the density, temperature and pressure of the fluid but the multiphysics only works when i set the mesh of the fluid to be fixed. I hope you can follow me and reconstruct the model in your mind. Thx for your help and best regards!

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.