Hi,

I do not work with Fluent, but I am somewhat familiar with VOF methods. VOF tracks the volume resp. the volume fraction of the droplets, and not each droplet as a Lagrangian particle. Therefore, with VOF you may track the zone occupied by a single droplet provided that the droplet diameter is large compared to the size of a typical mesh cell. If the diameter of a typical droplet is smaller, then only the volume fraction of the droplets within a cell can be tracked.

Hope this answers your question.

Regards

Hi Dr. Daniel,

Thank you for your answer. I visualized the droplets (there are more than one and each are about same size and much larger than cell sizes) by ploting isosurfaces where the volume fraction is equal to 0.5. Fluent calculates a thin band around the interface between different fluids (around droplets for my case), where the volume fraction differs from 0 to 1. However I want to mark each droplets as separate bodies and keep track of their volume center positions.

I probably should start a loop which scans each cell and mark the ones with volume fraction is greater than 0.5. After that I must separete them to groups by an elegant algorithm which I couldn't find yet. There are some issues which makes this harder; the droplets collide and unite or moves very close to one another. There may also be a tool in fluent which is suitable for this purpose. I need either a tool or a method which allows me to group the marked cells as droplets.

If possible I want to solve this issue by including an algorithm to Fluent using UDF and not by visually detecting the droplets using a post-process tool.

Kind Regards

Hi Umut,

Here is my suggestion (I do not know if this is possible with Fluent):

If there are not too many droplets and the droplets do not break up or coalesce, then I would give each droplet its own material ID for easily addressing it. To track the average motion of each droplet you may run the following algorithm after every time step:

i) select all cells having a particular material ID;

ii) re-select only those cells with volume fraction 1.0 (this gives a rough estimate of droplet size and shape);

iii) compute the center of mass of all the selected cells, i.e. the droplet center of mass X. This is an easy task because the cell geometry is known;

iv) store droplet center of mass X it for post-processing;

v) the average Lagrangian (material) velocity of a droplet is a simple finite difference approximation (Xn+1-Xn) / (tn+1-tn).

Regards

Dear Dr. Daniel,

I was thinking the same steps from second to end however I couldn’t define the volume fraction 1 zones as different droplets at each timesteps. The cells are stationary and only the volume fraction property of cells changes at each timestep. I cant mark the cells of volume fraction 1, give id and tract the positions at each timestep. I can easily mark the cells of volume fraction 1 as single zone, but my problem is to recognize the enclosed volumes of volume fraction 1 as separate droplets.

Lets assume there is only one droplet and if I try to average center of mass of each cell where the volume fraction equals to 1 then it gives me the position of this single droplet. If I repeat this process at every timestep then I can get the trajectory and the velocity of the droplet using finite difference approximation as you suggested. Because whole volume fraction 1 zone belongs to a single droplet.

Assume there are two droplets which are quite far from each other, if I try to id the cells with volume fraction 1 then I get both droplets and cant recognize them as different droplets. I can try to sort the marked cells by their position in x-axis (they move dominantly in x-axis and this will work for short amount of time) however at some cases the droplets are very close to one another or one droplet can pass the previously created one.

What I need is an algorithm to recognize the cloud of nodes (positions of vf 1 cells at each timestep) as separate closed volumes since there are gaps between them. I prefer a numerically cheap method if possible since I must repeat the process at each timestep.

Thank you.

Sincerely

Hi Umut,

this is why I suggested step i) and the preprocessing step of assigning an individual material ID to each droplet BEFORE calculation starts. The Fluent preprocessor should have this capability. Indeed, this only works if the number of droplets does not change between time steps and if all droplets are already there in the initial configuration. If this is not the case, talking about droplet trajectories does not make much sense.

Regards

Hi Dr. Daniel,

I don’t know if Fluent assigns values for different patch zones and keep track their movement. Are you suggesting that I should assign different fluids for each droplets? If thats the case it may be very interesting to try. Even if the droplets coalesce I can still easily track them.

Normally in my case, the droplets are same fluid (water) and the carrying fluid is oil. I enter the surface tension between them and contact angle between both fluids and the wall material. Instead, I can create say 5 materials (like water1, water2, …) for 5 droplets with same surface tension with oil, density and contact angle with walls. But then I also have to enter some value for surface tension between each droplet fluid and not sure what to use for ST between these (same) fluids. (Shall I use zero or near zero value) I’m also not sure how does it affects the convergence.

There are no droplets in the initial field however I’m sending oil and water from the same inlet periodically and I can choose different fluid at each period.

I will give it a try, thanks once again for the advice.

Kind regards

Hi Umut,

the problem of droplet coalescence and the influence of surface tension has to be modeled independent of the strategy by which droplets are regarded as either a single fluid or multiple fluids having the same properties. The only difference might be that Fluent assigns default values for a single droplet fluid.

Please give us feedback if you solved your problem.

Regards

Dear Dr. Daniel,

I have tried creating two fluids (materials) with same specifications as water. For the first droplet I choose inlet vf of the first fluid as 1 and second as 0. I did the same thing for the second droplet except I change the vf order to 0 and 1. By this, I created two droplets about two diameter size apart from each other and moving at same direction and velocity. They interact with oil and walls identically. It worked quite well.

If the droplets don’t interact with each other it works and half of my cases they don’t. I also wanted to see what happens if they do, so I reduced the velocity of the first droplet by reducing it’s contact angle with walls. I was half expecting that they will unite and act as miscible fluids (as two water droplets do) however the second droplet penetrated the first one and didn’t create a mixed droplet. They created an interface just as they did with oil. Maybe if I change the surface tension between the droplet fluids, I can overcome this problem, too.

Thank you for your help. I think I can use this method at most of my cases.

Best regards