[amin.z]

Well, first I must modify the St. Nu.! I was wrong about it, I calculated it again and checked it with some available data, it’s actually St=1.78! I’m afraid it’s completely different from previous one,
About the mesh dependency, I’ve not checked the solution in term of grid effect, I just checked it with finer grid, but both of them have same results,
About the checking results of single phase with exp. data, unfortunately there’s no information available about turbulence or velocity distribution, it’s just some data about 2nd phase distribution, the flow is split into 3 smaller pipes, and the phases distribution is available inside the pipes, I also know the 2nd phase pathline visually, not precise!
I’m not familiar enough to Schmit Nu. And have no idea to how I can calculate it at the moment,

2 question,
Are you thinking SST k-omega could be the reason of the issue?
And, the results I’m giving form steady solution fit better to exp. comparing with transient, do you still suggest to pursue transient solver?

[CeesH]

turbulent Schmidt number gives the balance between diffusion of momentum due to turbulence, and diffusion of mass. In some well-controlled situations it can be measured, and it is found to range between say, 0.1 and 2. The default value in FLUENT is 0.7, as this is a value (approximately) observed in many cases. It is however very difficult to observe experimentally for many realistic setups, and quite often is also used as a fitting parameter to reconcile CFD work with experimental predictions on mixing.

Whether the SST model correctly predicts the flow inside your geometry is difficult to say lacking experimental data - it is one of the more versatile models for the type of situation you are describing, but some validation should be done.

Whether you should do steady or transient; don't know - what is it you want to study? That's what determines what the most suitable solver is, in the end.

[amin.z]

Actually I'm studying fluid flow inside a long pipe, it's including several bends and finally the flow passes through a splitter and divided in three smaller pipes,
and I want to extract phase distribution inside each smaller pipes to validate CFD results, at first step.
note that the domain length is more than 15 m , doesn't the length too long? can it make a problem for convergence?

[gush]

hi,

i am experiencing exactly the same problem. my study is a ventilating enclosure with monodispersed contaminant coming through inlet. 3D, isothermal.

for only steady continuos phase solution eveything is ok. velocity results are correct. i can get particle pathways with one way coupling easily.

but when i interact continuos phase with DPM there is no convergence feasible it seems. especially continuity behave like a mad!

i searched this probelm on the web and it seems everyone with different study areas stuck exactly in the same point. and i see no solution yet!

did you manage it somehow??

thank you!

[ari003]

Originally Posted by gush

hi,

i am experiencing exactly the same problem. my study is a ventilating enclosure with monodispersed contaminant coming through inlet. 3D, isothermal.

for only steady continuos phase solution eveything is ok. velocity results are correct. i can get particle pathways with one way coupling easily.

but when i interact continuos phase with DPM there is no convergence feasible it seems. especially continuity behave like a mad!

i searched this probelm on the web and it seems everyone with different study areas stuck exactly in the same point. and i see no solution yet!

did you manage it somehow??

thank you!

I know it has been long time since this thread was initiated but I would like to state something based on my understanding. In Two-way coupling simulation, the jump in residual is pretty obvious as it creates a source term in continuous phase after each DPM iterations. After every DPM iteration the momentum, energy etc of the particles are exchanged to the continuous phase as the respective source terms and this creates a sudden jump in the residual. So far I've managed to understand this but if anyone finds it contradictory or has any better explanation please share it as well.
Also, it is better to monitor a parameter like Volume average velocity to check the convergence and not solely remain dependent on residuals.