I have attached a schematic of the geometry and inlet boundaries. How difficult is it to model the formation of droplets? Inlet vapor velocity is 1m/s and water inlet velocity of 0.001m/s. It is very similar to a spray nozzle. The first step is to model a cold flow simulation and then in future will study heat transfer between two phases. So the main point is to model the flow when gas (compressible and high velocity) and water mix in a T- junction. From, my initial studies i have used VOF method of FLUENT, but not able to generate droplets. Is there something am missing? Do you think it is not possible to resolve droplet formation?
The VOF model is used to simulate the phase boundary explicitly. In your case, when we expect a lot of small droplets, we will need an extremely fine mesh in order to simulate each droplet's shape.
On the other hand, when we simulate the multiphase flow using the mixture model (or the full Euler-Euler model), we simulate the content of two phases averaged in each element of the mesh. We don't "see" individual doplets, but we see a volume fraction field for every phase, so we can treat these phases as quasi-continuous, even if the secondary phase consists of droplets. The positive effects is that we don't need such as fine mesh as for VOF.
As you noticed, we need the diameter of the secondary phase in order to desribe the behaviour of the system properly. There are generally two ways of doing this:
a) constant diameter of secondary phase elements - you take this value from experimental results obtained for the same conditions and geometry, or you calculate it from correlations available in the literature.
b) dynamic change of diameter of secondary phase elements - you have to extend your model to take into account droplet breakage and coalescence, which depend on local process conditions (such as phase fraction, droplet diameter, interfacial tension, viscosity, energy dissipation rate, relative velocity of phases). This way will lead to better results, but requires complex modeling.
Dear Shailesh Konda,
1) I would like to suggest you to use mixture model rather than VOF model.
2) Then, you can define inlet velocity of each phase at their inlets respectively. Then you have to define volumetric gas fraction of each phase at their inlets. It could be 0 for liquid phase inlet and 1 for vapor phase inlet.
3) However, please note that I am not sure for that you will see the droplet (circle or spherical shape) in the results. It will show just simple contour of localized volumetric vapor fraction. From that vapor fraction contour, you must imagine it like droplet formation if I am not wrong.
Hi!
Any specific reason to switch from VOF model to mixture model? For the case of mixture model, we have to specify the diameter for the secondary phase, which is water in my case. How do we get that data?
The VOF model is used to simulate the phase boundary explicitly. In your case, when we expect a lot of small droplets, we will need an extremely fine mesh in order to simulate each droplet's shape.
On the other hand, when we simulate the multiphase flow using the mixture model (or the full Euler-Euler model), we simulate the content of two phases averaged in each element of the mesh. We don't "see" individual doplets, but we see a volume fraction field for every phase, so we can treat these phases as quasi-continuous, even if the secondary phase consists of droplets. The positive effects is that we don't need such as fine mesh as for VOF.
As you noticed, we need the diameter of the secondary phase in order to desribe the behaviour of the system properly. There are generally two ways of doing this:
a) constant diameter of secondary phase elements - you take this value from experimental results obtained for the same conditions and geometry, or you calculate it from correlations available in the literature.
b) dynamic change of diameter of secondary phase elements - you have to extend your model to take into account droplet breakage and coalescence, which depend on local process conditions (such as phase fraction, droplet diameter, interfacial tension, viscosity, energy dissipation rate, relative velocity of phases). This way will lead to better results, but requires complex modeling.
That's a very nice explanation. I have few unanswered questions running in my mind.
For the mixture model, getting the diameter of the secondary phase is the problem. What do you mean by dynamic change of diameter? Do i have to make a UDF to incorporate all the process conditions. I am interested in building a robust model. The problem with constant diameter case is am not sure how will i validate my results from the data i got it from empirical or literature studies. So, if i go for a bottom up approach, how should i proceed with the dynamic change of diameter case.
Do you think quality of the mesh would be the only limitation in the case VOF modeling?
What do you think is the reason for droplet formation in such nozzle problems? This is more of a fundamental question. I believe it is due to surface tension between the phases and pressure drop.
Yes, Shailesh Konda
(1) It is true. Knowing the diameter of droplets for simulation is essential and tricky task.
(2) There are two options. The first option is possible if you have experimental observation. We can estimate the average diameter of droplets from qualitative results. You can approximate the average diameter of droplets in your qualitative results by comparing with nozzle diameter in your test section. Then you can use this information
(3) Then, the second option should be adopted if you have no experimental data. You can search a few works that concerned with mist jet flow. In their works, they may provide the information such as gas flow velocity, liquid flow velocity and the diameter of the nozzle. Then you can approximate the average diameter of droplets in their work. Then you can evaluate non-dimensional parameter between droplet diameters and nozzle diameter at each liquid velocity because dimensions in their case may not be the same with your case. Then you can apply these non-dimensional parameters with respect to your case.
(4) These are just my suggestions. It may not be match with your opinion. Also, there can be several ways to get approximate average diameters from literature. Then you can perform simulation by setting different approximate average diameters of droplets until you are satisfied with the results. Since the diameters of droplets are obtained by approximation, please don’t expect your CFD results must be exactly match with your experimental data. Acceptable level is enough.
(5) The nature is droplet formation in high gas stream is close to nature of gas-liquid mixture. Thus, we should use mixture model. VOF is mostly suitable for stratified flow and slug flow.
Shailesh,
Take a look at population balance model (PBM) in Fluent. Once you are able to predict the formation of droplets using Eulerian-Eulerian model, population balance can predict the spatial and temporal evolution of droplet size distribution. I am not entirely sure if PBM is applicable in your case (since I don't fully understand the physics of the problem) but you can look at the literature and find out if anyone has used it in the past for similar problems. However, I agree with the previous answers about either using mixture or full Eulerian model.
Aung and Manjunath,
Thanks for your input. I will proceed forward with mixture model. I guess the high speed gas stream as explained by Aung is more of a gas- liquid mixture and would be more difficult to actually separate the interface of gas-liquid during the formation of droplet. So what do you think is the flow regime called for the case of high speed jet stream? Since VOF models are well developed for slug and stratified flows, cases were velocity is comparable and slow.
I am trying to find technical reasons behind droplet formations in high speed mist formation.
I will start working with the mixture model and get some initial results, where i can see the volume fraction of liquid in the high speed gas stream. I will keep you guys updated.
Shailesh,
The high speed jet stream with droplet formation could be thought of as atomization process where focus is on maximizing the interfacial area between the droplets and gas phase. Take a look at the attached paper and see if it of any relevance to your problem.
Hope this is useful.
shailesh
you can use discrete phase method , there you can inject gas bubbles ...
and for simulation you can use volume of fluid method...
Sridhar,
I am interested in studying the formation of droplets, Discrete phase model is used when the droplet is formed.
Aung and Manjunath,
I seem to have a hit road block in the model. I did some literature studies and came up with the initial diameter as 38microns for secondary phase(water). I used mixture model for the simulation, and selected slip velocity since both the phases have different model in my study. I did incorporate the surface tension between the two phases from the phase interaction dialog box. I am facing divergence issues. The model just does not seem to run. Surface tension, drag and slip- velocity are the sub- models that i have considered in my phase interaction dialog box. Is there anything that i am missing in the two phase model? The primary phase is just air.
- Badges
- Science topic