Learning Forum

[Abdelrahman92]

Hello

I am new at Ansys fluent. I am trying to simulate flow of liquids and solid inside Annulus between two concentric cylinders. For that i am using Eulerian multiphase model with laminar flow. I obtained the below residual (I guess it is a converged solution). However, when I compared the results with the experimental study, I found around 20% difference between simulations and experiments. For the validation I used to compare the average fluid and average particle velocities in the in annulus (I obtained theses values from Report -volume average .

[Karthik Remella]

There could be many reasons for the model results being different from expts. Please read the experimental conditions carefully and make sure that you are capturing the physics correctly in your model (especially, boundary conditions, initial conditions). Also, make sure that your results are grid-independent. nSince this is a transient analysis and because you are comparing with experiments, it is important to make sure that you are comparing data at the correct instant in time. If that ends up making the simulation run longer, that would be the direction you will need to take, unfortunately.nDepends on where you are computing your averages. If this is on a surface, you might want to estimate the area-weighted-average (instead of volume weight). Again, please make sure you are comparing 'like' results from the expt. paper.nAgain, this is a decision you need to take based on the expt. test conditions. Please try and engage with the team who performed these experiments and brainstorm with them to understand if you are on the right track with your modeling.nThank you.nKarthiknn

[Rob]

Assuming you set the inlet velocity of the second phase (as I suggested /forum/discussion/23502/boundary-condition-for-e-e-multiphase-model#latest) check the mass into the domain for each phase balances. Depending on the domain size it's quite possible solids are building up. nAlso check the experiment: both may be correct but running at different conditions. n

[YasserSelima]

Why are you considering the solids Euleria? Can you use Eulerian-Lagrangian Model? This would decrease your simulation time by an order of magnitudenWhat is the nature of the solid? Are they fine particles?

[Amine Ben Hadj Ali]

Mixture model might be first thing to check heren

[Abdelrahman92]

There could be many reasons for the model results being different from expts. Please read the experimental conditions carefully and make sure that you are capturing the physics correctly in your model (especially, boundary conditions, initial conditions). Also, make sure that your results are grid-independent. Since this is a transient analysis and because you are comparing with experiments, it is important to make sure that you are comparing data at the correct instant in time. If that ends up making the simulation run longer, that would be the direction you will need to take, unfortunately.Depends on where you are computing your averages. If this is on a surface, you might want to estimate the area-weighted-average (instead of volume weight). Again, please make sure you are comparing 'like' results from the expt. paper.Again, this is a decision you need to take based on the expt. test conditions. Please try and engage with the team who performed these experiments and brainstorm with them to understand if you are on the right track with your modeling.Thank you.Karthik/forum/discussion/comment/103182#Comment_103182

Thank you very much for your reply. Regarding the physics of the problem, I am not sure about my approach of setting the inlet B.C (and this is one of the questions that I have posted' em here). Let take your suggestions point by point:n(1) The comparison with the Experiments is done based on the average V of the phases. These velocities were measured experimentally when the system reached a steady sate (defined when the inlet mixture flowrate equals the outlet flowrate). In my simulation, I have run the case for 5 s at 0.005 timestep. You could be right may need to run the simulation for longer time. However, I have check the inlet and outlet flowrates and they are almost the same which means i reached the SS condition according to the experimental procedure. Honestly I was thinking about doing the simulation longer. But I don't want to go for longer time unless I have solid evidence that the time could be the problem, since a single simulation takes more than 50 h.n(2) I want to compute the average velocities values at the fluid domain (that is the annulus), I tried Mass-weighted average at fluid domain, and It gave me similar results to volume average. BUT, I am not sure if these methods are correct. I hope someone can help here? n(3) Actually, this question is about using Fluent models, and its degree of freedom. If I specify the volume fraction of one phase and velocity of the second phase, will the model translate that into volumetric flowrates and then compute the velocity of the second phase. OR,I have to specify the velocities of both phases, as well as volume fraction of the second phase?( In that case, i don't understand why someone would need to define the volume fraction, since the concentration is already identified by the volumetric flowrates (velocities).Thank you again for your help!n

[Abdelrahman92]

Assuming you set the inlet velocity of the second phase (as I suggested /forum/discussion/23502/boundary-condition-for-e-e-multiphase-model#latest) check the mass into the domain for each phase balances. Depending on the domain size it's quite possible solids are building up. Also check the experiment: both may be correct but running at different conditions./forum/discussion/comment/103186#Comment_103186

Thank you again Rob. I have tried setting both phases at the same velocity. However, I obtained significant overpredictions. n

[Abdelrahman92]

Why are you considering the solids "Eulerian"? Can you use Eulerian-Lagrangian Model? This would decrease your simulation time by an order of magnitudeWhat is the nature of the solid? Are they fine particles?/forum/discussion/comment/103513#Comment_103513

Thanks Yasser for the reply. Actually, E-L can not work here, because the volume fraction of the secondary phase may reach 0.6 inside the annuals due to particle settling (or bed formation). I want also to track the volume fraction of the particle inside the annuals. n

[Abdelrahman92]

Mixture model might be first thing to check here/forum/discussion/comment/103535#Comment_103535

Hi DrAmine, nI think the problem with my simulation is the accuracy of the results. And as far as I know from Fluent manuals, Eulerian can provide ore accurate results than Mixture model. n

[Amine Ben Hadj Ali]

Not always it is always a matter of closure. Mixture does not consider all physics but it is a very good simplification.n

[YasserSelima]

Now I see your point. nBy the way, if you have solids settling, the steady state could be reached in minutes, not 5 seconds. Initialising with the expected results and volume fraction would decrease the simulation time by a lot. Good Luck! n