[e101533]

Hi all,
I am running a turbulent simulation with regions of swirling, this is causing reverse flow at my outlet. My solution however is converging. Wil the reverse flow affect my results?

thanks in advance

[TomK]

It depends on where you are looking for data. If you are measuring something far from the exit, then no it will not affect it. Try using a pressure boundary for the exit. That should fix it.

[purdue1]

Hi. I am solving a steady state laminar problem (flow through microchannel) with velocity inlet and pressure outlet. Although my solution converges (and the mass fluxes are well balanced at convergence); I get reverse flow on pressure outlet (in almost 50% of the total number of faces). That is definitely not acceptable as a solution even though it is converged !! What options do I have?
Note: If I give the outlet as "outflow" instead of "pressure-outlet" ; the number of reverse flow faces significantly decreases!! Am I doing something wrong here ?

[ara1362]

if you are getting reverse flow, you should change the temperature on the outlet to the local temperature, to reduce instability. Initial set up is 300k.
what this means is, if you get reverse flow the flow comes back to the region through the outlet boundary with temp value of 300K

if you dont want to see any reverse flow, change turbulence specification to "Intensity + length scare" and change the length scale to be 10% of your outlet diameter.

modeling outlets with flow split outlet is not an accurate way of modeling.

sometimes, for numerical reasons, you need to extent the outlet to be further downstream. in ccm+ you can choose the extruder in mesh models and then select the boundary to be extruded and by how much.

Hope this is all helpful.

[purdue1]

Hi,

Thanks for the response. The thing is, so far I am just solving the flow equations, and getting reverse flow. I havent even turned on the energy equation yet. So, Temperature may not be an issue. My flow is laminar, so turbulent specifications should not really matter, if I'm not wrong.

I will probably try extending the domain though. that might wokr. I did not understand the last part of your post.. about split model not being a good way of modeling. Could you please explain what is meant by that?

Thanks!

[tH3f0rC3]

Originally Posted by ara1362

if you are getting reverse flow, you should change the temperature on the outlet to the local temperature, to reduce instability. Initial set up is 300k.
what this means is, if you get reverse flow the flow comes back to the region through the outlet boundary with temp value of 300K

if you dont want to see any reverse flow, change turbulence specification to "Intensity + length scare" and change the length scale to be 10% of your outlet diameter.

modeling outlets with flow split outlet is not an accurate way of modeling.

sometimes, for numerical reasons, you need to extent the outlet to be further downstream. in ccm+ you can choose the extruder in mesh models and then select the boundary to be extruded and by how much.

Hope this is all helpful.

Hi,

I'm wondering which outlet boundary to use.
I have set up two similar cases (a flow streaming through a pipes). One is set up with "pressure outlet" and the other with "flow-split outlet". The results show slight differences in the calculated velocity values.

What could be the reason for that or rather which boundary is the best to use? I'm afraid I have no data to validate yet.

Best Regards,
tH3f0rC3

[djack]

Hi,
I may have an answer for you, if you're still alive...
Regarding my system (jet in cavity) it seems that the "pressure outlet" boundary condition give a symetric flat velovity profile.
Using the "Flow-split" boundary condition give an unsymetric velovity profile which is strongly influenced by the air motion in the cavity.
Using this second boundary conditions the solution need quite less iterations to converge (x6 in my case).

Regards,
Djack

[irenefong92]

Originally Posted by djack

Hi,
I may have an answer for you, if you're still alive...
Regarding my system (jet in cavity) it seems that the "pressure outlet" boundary condition give a symetric flat velovity profile.
Using the "Flow-split" boundary condition give an unsymetric velovity profile which is strongly influenced by the air motion in the cavity.
Using this second boundary conditions the solution need quite less iterations to converge (x6 in my case).

Regards,
Djack

Hi Jack, I hope u can help me. I am facing a problem now.. I am simulating water and oil flow in a valve like element, and keep facing the problem of reverse flow at the outlet. isit normal to have reverse flow when the flow is split due to hindrances?
p/s I had extent the outlet to be longer then the 7-8x longer then the inlet. but still it shows reverse flow.

Please help.

[Mohammad80]

Hi.
I have a problem with the reverse flow at the outlet of my geometry. I am doing a simulation to determine the water drag force on a pier inside a channel using FLUENT. I want to get the results for a Laminar flow (v=0.0001 m/sec). the simulation is going well but the drag force I got is negative, that means the reverse flow affecting on the results. I increased the elements density and I changed the geometry to be as shown in the attached picture and made the outlet far from the pier and I decreased the time step size but I still get the reverse flow warning messages. I read all the comments in this site about this problem and i did all what were suggested but unfortunately, the problem stills unsolved. Any tips please?

[tooran]

Hi all,
I created viscosity model base on the changing viscosity by y component of cell and HershelBulkley model as below :

// if y < e then calculate HershelBulkley1
// else calculate HershelBulkley2


// * * * * * * * * * * * * Private Member Functions * * * * * * * * * * * * //

Foam::tmp<Foam::volScalarField>
Foam::viscosityModels::triv::calcNu() const
{
dimensionedScalar tone("tone", dimTime, 1.0);
dimensionedScalar rtone("rtone", dimless/dimTime, 1.0);
const fvMesh& mesh = U_.mesh();

tmp<volScalarField> sr(strainRate());

// Define a tmp refering to this min thing
tmp<volScalarField> tmpEtat = min(
nu0_,
(tau0_ + k_*rtone*pow(tone*sr(), n_))
/(max(sr(), dimensionedScalar ("vSmall", dimless/dimTime, VSMALL)))
);


forAll (tmpEtat.ref(), cellI) //loop through cell centres
{
tmpEtat.ref()[cellI] = mesh.C()[cellI].y();

if ( tmpEtat.ref()[cellI] < e_.value())

{
//HershelBulkley1
tmpEtat.ref()[cellI] =
min
(
nu0_.value(),
(tau0_.value() + k_.value()*rtone.value()*pow(tone.value()*sr()[cellI], n_.value()))
/(max(sr()[cellI], 0.0002))
);
}
// Hershelbulkley2
else {
tmpEtat.ref()[cellI] =
min
(
nu0_.value(),
(tau0ll_.value() + kll_.value()*rtone.value()*pow(tone.value()*sr()[cellI], n_.value()))
/(max(sr()[cellI],0.0002))
);
}
}




return tmpEtat;
}



It compiled and simulated well . I applied this model for 2-D geomtry (simple block with length of 0.5 meter and diameter of 0.05)

I enter e = 0.048 it simulated well. However for the amount of e smaller than 0.047 it shows reverse flow. I deleted reverse flow by applying inletoutflow boundary condition but after that the solution does not converge.

Could you please help me?

Thanks