[Numerics]

Hello,

before i start to explain my problem, i'll explain my case:
This is my geometry:


A hot air flow goes through the tube (1.1m) (it's symmetrical. That's why there is only a half of it in the picture) at the inlet (blue points). And they shall escape at the outlet (red points). The small tube at the top is where cool water will be injected as a spray. The geometry of the small tube is only for turbulence. The water will be injected where the small tube ends.

The cool water shall flow through the outlet as well and do the physics stuff (heating, evaporating ..).
Inlet to Outlet direction has the same direction as gravity.

This are my specifications for the DPM-Model:
-interaction with continuous phase
-unsteady particle tracking
-10 Continuousphase iterations per DPM iteration
-Particle Time Step size 0.0005, Number of time steps 10

The injection is a cone droplet injetion with a number of 600 streams of water liquid which can evaporate.

Point Properties:
Diameter of particles 6.2e-05
Temeprature 320K - 300K
Cone angle 40 deg - 30 deg
Radius 0.0002 m -0.000275 m
Total Flow Rate 0.00122 kg/s - 0.00225kg/s

The Boundary Conditions at the velocity inlet:
T=873K-763K
v=1.5m/s-0.3183 m/s
Turbelent Intensity 5%
Hydraulic Diameter 0.2m

Pressure Outlet:
p=110000 Pa
Turbelent Intensity 5%
Hydraulic Diameter 0.2m
T=873K-763K

My problem is the following:
I gave two sets of parameters. The first set is the case where the problem is solved correctly from ansys and everything looks perfectly fine. But if i use the second set of parameters which is needed, there will be a backflow/reversed flow which is not appropriate for my case. Droplet particles do not escape through the outlet.The velocity parameters seem to be the problem, but i have to keep them exactly this way.

I searched this forum for solutions but couldn't find anything that helped. The Outflow boundary condition doesn't seem to be allowed for my case. I smoothed the mesh and made it polyhedra. That reduced the problem a bit but did not remove it.

I gladly appreciate any help you can give me.

Best kind regards, Numerics

[Large Epic Simulations]

I'm not sure, but it may depends on your turbulence model, your solver and the schemes you're using.
Regarding the particles trajectories, maybe the particles are deviating their trajectories because of the choice of the turbulent dispersion model.

[Numerics]

Thank you for your answer, Large epic simulations.

I am using the k-epsilon realizable model with enhanced wall-treatment,
a coupled pressure velocity scheme.
And the solutions methods for turbulent kinetic energy and dissipation rate are first order upwind.
The solver is pressure based, steady and absolute velocity formulation.

Is there something to change?

Best regards, Numerics

[Large Epic Simulations]

I would suggest you to try a steady particle tracking to see if there are significative changes in the outcome of the trajectories.
Are you using a turbulent dispersion model (you can find it between the injection properties).

[Numerics]

Thanks again for your answer Large Epic Simulations.
I have "Discrete Random Walk Model" and "Random Eddy Lifetime" enabled with a time scale constant of 0.15 .
For physical Models the Drag Law is dynamic drag and the TAB Breakup MOdel is enabled with y0=0 and Breakup Parcels 2.

I will start another calculation now with steady particles.

[Large Epic Simulations]

You're welcome,
since I'm performing roughly a similar simulation, it is also helpful to me .
Anyway, known the fact that these calculations implies an high number of modelization choices in the modelization, you have to find the effect given by each of these "ingredient".

[Numerics]

So i did some calculations without the injection and the reverse flow is there nonetheless. Increasing the velocity removes the problem but i can't do that for my calculation.

Something else i can do?

[Large Epic Simulations]

Well at this point the "problem" seems to be circumscripted to the gas-flow.
Do you have any gravity or buoyancy terms in your model?
Maybe they play a role.
In the mean time you can try to shift all the schemes to the second order upwind and use a pseudo-transient formulation on your solver. Let me know!

[Numerics]

Hi, i did a pseudo transient calculation and it lowered the reverse flow. Reducing the back flow temperature also lowered the reverse flow. But both did not remove the reverse flow.

This is a tempearature profile of the solution, if it helps (blue=593K, red = 763K)


Gravity is activated but did not change much. Also it should accelerate the normal flow at max. I don't think i have a Buoyancy method (it is not activated in the viscous tab) or model activated. At least i don't know where to activate that.

I see roatation axis direction is z=1 for the cell zone conditions is activated (that's the dircetion of the normal gas flow)

Also it is a problem if most of the properties of my materials are "constant"? It seems to use ideal gas law nevertheless.

Thanks again, for your help!