Flow simulation of micro fluids

[Anonymous]

Hallo at everyone,

at first I want to try to give an overview and following my questions/problems that I couldn't solve jet.

To start with, I want to make a flow simulation for very small traps that should catch small spheres/cells. I want to analyse the pressure and the force on the sphere/cell. At the moment I have simulated a V shaped trap and have added a static sphere in the center of the trap. Now my questions:

- First, I thought that the boundary condition periodic should give me the possibility that the flow leaves one side and enters on the other side, but my results do not really show this

- Second, if I import my CAD model in CFD and try to create the meshing net, then cfd do not create a meshing net on the sphere. Therefore the question is the following. Is it possible toe create a sphere meshing net or does cfd create one, if i calculate the file? I can't click on the sphere surface to display the pressure and the force on the sphere, I just can just approximate them by the surface of the trap near the sphere.

-Third, i want to make an animation, where the trapping is visual. Therefore i have tried the particle flow, but this tool do not simulate the particle (the particle do not interact with the flow). It is just a presentation of the flow vektors. Furthermore, I have seen the possibility to enable motion to specific parts. I thought that I just try to construct a few spheres and let them move freely, but the solver crashes, or the moveable parts are not visual anymore. Like, if the spheres are to far away, even if I set the boundaries. Do I need something in addition to create the free motion, or is there another option?

I hope that someone can help me. Thanks for the help and have a nice day.

Puandra

[David.Short.]

Hi @Anonymous,

This sounds really interesting and hopefully we can help you get something going.

1. I am a bit lost regarding which surfaces / edges you would like to add periodic BCs. A simple inlet with flow and outlet with 0 gage pressure will be the easiest method.

2. Try splitting the spheres in CAD. CFD may need an edge on the sphere to be able to mesh it.

3. Have you tried using massed particle traces rather than normal traces? Simulating freely moving spheres will be tricky but maybe possible and could be fun! Please attach a share file (.cfz found under save as) of the free motion sphere setup and we can give some recommendations.

All the best,

David

[Anonymous]

Thanks @David.Short.,

-The uploaded pictures should be a very small extract of my whole component, therefore I wanted a BC, that shifts the flow that comes out of the right side on the left side and vice versa. But if I create the particle traces, it seems to be not the right BC. But i haven't worked with CFD a lot so I do not have the experience, if just the particle trace does not take the BC on. Furthermore the picture is the same as you have written "A simple inlet with flow and outlet with 0 gage pressure". Plus the inlet that is rotated to create a flow from a specific angle.

-The idea of splitting the sphere to create edges is a great possibility which I will try.

-Yeah i have tried the massed particle traces, but the traces just seem to be an added overlay and not an extra simulated particle, which interacts with the flow. (In addition I have to say, that the particles have a similar size as the traps. But I think this is obvious because i want to trap the particles)

- And because the massed particle traces do not producesof what i am looking for, I have tried to simulated free motion spheres. But as you mentioned this is a quite heavy paroblematic.

Thanks a lot for your ideas and maybe we can get a few more ideas on how I can try to fix my problems.

I attached a .cfz file that have the structue as an empty trap and with a static sphere.

Have a nice day

Puandra

[David.Short.]

Hi Puandra,

Thanks for the .cfz. 

Firstly i suggest first trying this in 2D and then we can step it up to 3D if necessary down the line. This will speed up the simulation work cycles significantly and we can get an idea of the settings we need without waiting for long simulations to fail!

For periodic BCs the boundaries should have exactly the same geometry with exactly the same mesh qpplied to each. Uniform or manual meshing will be required.

I think free motion solids is the way to go and should be fun. I will have a crack at getting a 2D sim going.

All the best,

David

[Anonymous]

Hello @David.Short.,

I was busy the past days and thus i only have achieved just a few thinks. On one hand I have splitted the sphere and now cfd can create a meshing net on the sphere. But now I am confronted with another problem, because when I import the CAD file in CFD, then CFD seems to add something inside the File like an additional edge, or it merges the trap and the sphere at a little space.

Furthermore I still have no option or clue, how i should or could calculate the pressure or force on the sphere? I thought I can just take the wallcalculator, if i can manage to creatte a meshing net on the sphere, but i still can only calculate them on the trap and not on the sphere.

I try to bring it in to the 2D area, and there I will try again. Plus I have to retry the solid motion.

Best regards

Puandra

[David.Short.]

Hi Puandra,

Jon Wilde and I have both been trying to get a 2D simulation going with your geometry and free motion particles. We have tried everything but we are just getting nonsense.

After discussing internally we have had to concede that simulating fluids at such small scale is not possible with our software. This is because at these tiny scales surface tension dominates fluid behavior. Surface tension is not considered within Navier Stokes based codes and thus at these tiny scales the mathematics falls apart and we get unusable results.

At a larger scale the simulation works fine. Would you be happy to work at a larger scale?

All the best,

David

[Anonymous]

Hello @David.Short.,

thanks for this information, unfortunately my area of interest is about this small rates. But then i do not have to try to find out, if it is possible. But even if I can not make this solid motions, is there a way that I can calculate the pressure- and force vector on the sphere? Or do i have to estimate this by calculating the pressure- and forcevector near by the sphere for example on the surface of the trap?

Another question, the solid motion, you have said that this is going well in higher scales. Higher flowrates? Or everything in a higher scale?

Thanks alotfor your help im very grateful

Have a nice day

Puandra

[David.Short.]

Hi Puandra,

Sadly it is not the solid motion that is the problem, it is the flow itself that is unrealistic at such small scales.

By scales I am referring to the physical size of the domain, traps and particles. 

At these tiny scales the flow, pressure and hence forces will be wrong.

The simulation works if you scale everything up. I think Jon got it going with a domain of 1 meter width. Obviously this is not really what you are looking for.

In terms of calculating pressure forces on a sphere in a flow the wall calculator will work fine.  Have a look below.

All the best,

David

[Anonymous]

Hello @David.Short.,

I have the idea that i could scale up everything, just to see how the motion looks, and perhaps it is possible to have multiple solid motions. Furthermore this could be an illustration of the trapping process.

How have you or Jon be doing it?  I can not select the sphere... Do i need something in addition?

I describe my workflow and perhaps there you can find the missing or failure... Ok at First i import the .sat CAD file into CFD, then i change the geometrie from cm to mm. After looking if the import have any issues (like added edges or angular shapes or something else) I add the material and the boundary conditions. The next step is to suppress everything except the medium (water). Then I define the meshing net and after that I define the calculation start ( 500 Iterations, ADV5, Physics hydraulic pressure and then Komponents) but i am not really sure about the last option.

I hope that there is just a simple misstake on my side and thanks

Have a nice day,

Puandra

[David.Short.]

Hi Puandra,

Your process sounds good but suppressing the spheres will make it very hard to select the appropriate surface. Also the spheres cannot be suppressed if you want to add motion. Try without suppressing the sphere's. Give them a course mesh as there internal mesh is not relevant for the simulation.

Have you scaled up the model yet? If so please send me the .cfz and i will do my best to get it going and then explain settings afterwards.

Al the best,

David

[Anonymous]

Hello @David.Short.,

I have splitted the sphere to get a meshing net on it, and even if i do not surpass the sphere ican not select the surface with the wallcalculator....

But I haven't tried to scale everything up. Therefore I have to decide, if I want to get an animation of the motion and if i can relate the scaled up to my very small settings. I do not think that everthing is linear or am I wrong?

Best regards,

Puandra

[Anonymous]

Edit:

I was able to select the sphere 😉 I just have to hide the rest of the structure to click on it.

How can I show the coloured gradient on the surface? my Sphere just have one colour...

Reedit:

You have said that the force and pressure can be wrong at this low scales? so the Force and pressure onside the sphere are not significant?

Sry that i ask so many questions, I just want to understand everything.

[David.Short.]

Hi Puandra,

You are correct, things are definitely not linear and the scaled up results would only give you a rough impression of the behaviour at small scales. 

This is because the physics is different at small scales. Surface tension is a crucial factor at small scales where as its effects are negligible at larger scales. Navier Stokes based CFD does not take surface tension into account and therefore it is not accurate at small scales. 

Therefore it is not valid to apply the large scale results as a representation of your small scale model.

You can probably only see one color on your sphere as you are visualizing velocity which will be zero on the sphere surface due to the no slip condition on fluid boundaries.

Not sure i understand your last point about the force and pressure. Pressure and force will of course be significant on the surface of the sphere however there is no point having a fine mesh on the inside of the sphere as we do not care what is happening inside for this simulation.

All the best,

David

[Anonymous]

Hello @David.Short.,

thank you for the explanation of why cfd is not accurate at small scales. If this is right then I am not well-advised if i want to make any significant statements out of the force and the pressure, which is calculated by the wall-calculator. Or do you just mean the solid motion in small scales. 

As I have read through your reply I have recognized why the sphere has only one colour. But my question was about how to create this gradient on the sphere, as I can see it on your uploaded picture. Is it just the pressure? Because i only know how to select pressure at planes.

best regards,

Puandra

[Anonymous]

Set the Global Results to Pressure, then right-click on the colored legend bar, select Set to Part ... and then select the sphere.   Chances are you have Global Results set to Vmag, so your sphere would be blue since the velocities at the wall are zero.

Navier-Stokes is based on a fluid continuum methodology so it starts to become invalid for very small flow channels (individual molecules or predominant surface tension influence) or very high vaccuums (molecules are spread too far apart).

[David.Short.]

Hi Puandra,

I mean everything is wrong at such small scales (pressure, force and motion). You would need to utilize a software that includes surface tension.

In my screenshot I simple selected pressure for global results and made the outer volume transparent.

I hope you can find a way to keep moving forward with your project as it is a really interesting one!

All the best,

David