Learning Forum

[Ghinna Nur Akram]

Hi, I'm currently doing a simulation regarding dissolved oxygen in a pond containing liquid water. I want to validate the graph of the increase in dissolved oxygen concentration against time from experiments conducted by Du (https://www.sciencedirect.com/science/article/pii/S0144860919301815).
So, on the inlet side, air in the form of bubbles (oxygen) enters from the bottom of the pool and the outlet is at the surface of the pool. I set water (liquid water + oxygen) as the primary phase and oxygen as the secondary phase. I used the multiphase mixture model for this case after reading various references. Then, I turned on species mass transfer and set it from air bubbles to water with species in the form of o2. The results I found were only after 6 iterations, the simulation became divergent and "floating point exceptions" appeared. Below I show the setup that I made. My question is:
1. Is my setup correct? If there is something wrong, which one should be corrected?
2. Does the mixture material only exist as a gas phase? Because when I look at the properties of the mixture that I made it shows that it is only for the gas phase? How do you set up the liquid phase as a mixture?
3. Do I have to input the o2 source terms value in the cell zone conditions? If I must, I do not have any references from the paper I refer to because the research conducted was experimental.
I'm still very new to this. I've been stuck for almost two months, so I'm asking for your help. Thank you very much for all your help..

[Rob]

I'd tend to use nitrogen & oxygen for an air mixture, but otherwise it looks OK.  Turn off the species/phase transfer. Does the model run then?

[Ghinna Nur Akram]

Thanks for your advice
Can we control the nitrogen and oxygen composition? How?
I tried turning off species mass transfer, and the mixture material in the multiphase model changed to air in both the primary and secondary phases.
I'm wondering how to determine the properties of oxygen-water mixture because the available properties are still default to the mixture-template.
I looked for some references and found that I have to enter the mass source in the cell zone conditions, is that correct? But I didn't find any references regarding the mass source of the paper that I wanted to validate

[Ghinna Nur Akram]

The divergence occurred after I changed the mixture material from h2o(g) to h2o(l) in the water-oxygen material

[Rob]

You will need to change the contents of the template, and make a second one for the liquid phase.  Re the phase composition, it's controlled by boundary conditions/patch and phase mass exchange. Not sure why you'd want a mass source unless you're coding up phase exchange yourself, in which case you may need some care with the mass transfer rates. 

[Ghinna Nur Akram]

Hi, Rob. Thanks for your advice.

I wanted to change the properties of the water-oxygen material and this is what I did, I wanted to change the density to constant but there was no option to change it. How to?

Can this mixture also be used for other phases besides the gas phase, such as the liquid phase? Because I looked at the database, the only phase available was the gas phase.

Also I want to show the geometry I created where the inlet velocity is 0.0166 m/s.

[Rob]

Do NOT use ideal gas for liquid properties! It won't end well as there's nothing in the model to assign phase as there's nothing linked to latent heat in the model. It's usually safe to assume liquids are incompressible unless you need buoyancy, in which case read the manual and explain what you think you need to include. 

[Ghinna Nur Akram]

Hi Rob, I want to set the density of the liquid properties but there is no option for the liquid phase in the mixture like the following picture, what should I do? Thanks

[Rob]

Use volume weighted and then set the two liquid species to a sensible value (ie around 998kg/m3 for water). 

[Ghinna Nur Akram]

So, the oxygen in water-oxygen must be in liquid phase? Ok, I’ll try it, thanks Rob

[Rob]

Yes, it's dissolved so isn't a gas. 

[Ghinna Nur Akram]

 

Hi Rob, I tried it, but “floating point exceptions” still shows up, what should I do?
Do boundary conditions matter? Because I just set the boundary conditions for the inlet velocity of the bubble. Do I have to adjust the volume fraction and species mass fraction at the boundary conditions? And what is meant by backflow at outlet boundary conditions?

 

[Rob]

Yes, boundary conditions matter and need to reflect what's going on at that position. 

[Ghinna Nur Akram]

Hi, Rob. I have tried setting the boundary conditions but there is no change. Then, I tried to remove the h2o (g) species in the mixture-template, and the divergence no longer occurred although there is a possibility that it will occur again. However, there was no increase in o2 concentration in the water. Apart from that, in my opinion the residual is not that good. What should I do?

[Ghinna Nur Akram]

Hi Rob, what do you think? I'm really stuck here and the results always diverge

[Ghinna Nur Akram]

 

Hi Rob, why doesn't mass transfer work when I use liquid phase water? Meanwhile, it works on gas phase water. Where should I set it?

 

[Rob]

Phase transfer will depend on how you set up the mass transfer. If you're seeing divergence what is going on in the flow? Too high a mass transfer rate may cause problems, but that's mostly a case of checking the maths and reducing the time step. 

[Ghinna Nur Akram]

Thanks for your response, Rob.
If I look at the divergence, the increase in o2 concentration is very drastic.
I have a question, how do I determine the backflow species mass fraction and backflow volume fraction at the pressure-outlet in this simulation? Because I don't have any references for backflow species mass fraction and backflow volume fraction.

[Rob]

Backflow information needs to be known, or backflow avoided. However, if the solver is diverging chances are the backflow isn't helping but is also being caused by the diverging solution. A starting point will be the phase you'd expect at the boundary at what you'd estimate the species fractions to be if the device worked. Ie use the design calculations to estimate back flow. Ideally, extend the domain to stop it occurring. 

[Ghinna Nur Akram]

Hi Rob, thanks for your response.
I didn't change it from the inlet setting because if I changed it divergence would occur. I set the O2 concentration in bubbles to 0.21, while in water I set it to 0. Meanwhile, for the volume fraction of bubbles, I set it to 1. What do you think?
I also want to ask, how do we regulate the saturation O2 concentration in the water in the Ansys fluent?

[Rob]

Saturation on the liquid side will be limited by whatever mass transfer you set up. Worst case you may need to code that: most of the models we see are bubble column/sparger systems so saturation is rare. 

Sketch out the domain, and write onto the sketch all the boundary conditions, post processing points etc. Post that image here. I'm after a clear representation of what you want to do, not a detailed CAD drawing or van Gogh!   I assume the "sky" will be all gas phase, but your earlier image looks like something flows in the bottom of the domain and out of the top. 

[Ghinna Nur Akram]

Hi Rob, thanks for your response
This is a sketch that I made. Is this possible to carry out a simulation? Because most of the available references do it by experiment. If available, the case is simpler.

[Rob]

Water is moving? Is air just at the pressure boundary? 

[Ghinna Nur Akram]

The water doesn't move. The fluid that flows from the velocity-inlet is only air.

[Rob]

Sparged? Ie the water is in a tank and bubbles are added into various locations? 

[Ghinna Nur Akram]

Yes, that's right

[Rob]

OK, and how does it look with just the phase flow?

[Ghinna Nur Akram]

This is the result I found

and this is the result I want

Should I add air domain on top of water domain? Because Henry’s Law works at the interface of the two fluids so Henry’s Law mass transfer can work, I guess

 

[Rob]

If you need oxygen from the free surface you'll need to model it, I don't think you can get mass transfer like that from the boundary. You also need to review the mass transfer models to check which laws are considered for saturation. Check Fluent's Beta features too. 

[Ghinna Nur Akram]

OK Rob, what do you think about the simulation results? What should I fix?

[Rob]

As I don't know what you're seeing I can't say. We can only give general advice, so are limited where things become very detailed or technical. I've also not see a volume fraction plot for phase. 

[Ghinna Nur Akram]

OK Rob, Why is the initial water volume fraction 0? Even though I have set water as the primary phase and the air volume fraction at the inlet is 1.

[Rob]

How did you initialise the model? 

[Ghinna Nur Akram]

 

Oh I see, I haven’t set the air volume fraction to 0 in initialization

 

[Ghinna Nur Akram]

 

Hi Rob, I have a question, how do I define the porous medium inlet? should I create another geometry and then place it on the inlet? This is what my paper says.

 

[Rob]

You can't. Are there any pictures or more details? It's not uncommon to model a set of holes (eg a sparger) as a longer strip/slot inlet but I've not seen it termed a porous medium before. 

[Ghinna Nur Akram]

This is what is written in the paper I refer to

[Rob]

I think the last paragraph may be more helpful. What does that tell you about the inlet area in the CFD model compared to the real design? 

[Ghinna Nur Akram]

The actual design is fine pore aeration tubes like this. I modeled the velocity inlets on the sides of the fluid domain. I still don't understand how to model porous medium as written in the paper.

[Rob]

OK. Modelling all of the holes isn't sensible so it's normal to add the gas from a slot or larger patch on the bars. That may be what the paper means, but that's not something I can answer: if you really need to know you'll need to contact the author(s). 

[Ghinna Nur Akram]

OK Rob, I have a question, Will it be a problem if I don't set the inlet velocity for the water (primary phase)? Because the only phase that enters the domain and leaves the domain is air (secondary phase)

[Rob]

Assuming the inlet is purely the second phase, it shouldn't as there isn't any primary phase entering. 

[Ghinna Nur Akram]

OK Rob, I have a question, why doesn't the air move upwards?

[Ghinna Nur Akram]

Hi Rob, does this kind of case require UDF or not? I'm really stuck right now

[Rob]

I'll relock after commenting.

No UDF needed. Check gravity, and depending on the multiphase model how the flow & volume fraction are behaving.