Reverse flow at inlet

- January 24, 2019 at 5:28 pm
  
  TingWangUNO
  Subscriber
  
  Hi,
  
  I have a flow reversal in whole domain. In fact, I have put velocity inlet and if I monitor my vector plot at the inlet itself, they are reversed. It is a transient simulation that is initialized with a steady state case. The original steady state case has the correct direction of velocity vectors at inlet.
  
  Question: Why does Fluent overwrite my velocity inlet boundary condition and creates reverse flow!
- January 24, 2019 at 5:53 pm
  
  DrAmine
  Ansys Employee
  
  Screenshot of Mesh and Geo as well as boundary condition definition for inlet.
- January 24, 2019 at 7:12 pm
  
  Raef.Kobeissi
  Subscriber
  
  Hello,
  
  It must be something to do with your boundary conditions or your mesh. Can you please provide what Amine has requested.
  
  Thanks
- January 25, 2019 at 4:47 am
  
  TingWangUNO
  Subscriber
  
  The geometry is a siphon as shown in the figure above. It has velocity inlet boundary condition on the left horizontal face...and a pressure outlet boundary condition on the right horizontal face. Gravity is turned on.
  
  For a steady-state case the results are perfect. However, I used the steady case results to initialize an unsteady case...this involves thermal phase change model as liquid water enters the siphon it flashes to produce vapor.
- January 25, 2019 at 8:53 am
  
  DrAmine
  Ansys Employee
  
  Mesh is okay and at velocity boundary condition I won't expect reversal flow (not possible from the implementation side). Please provide a zero operating density under Operating conditions and provide the right gauge pressure at your outlet (as it opens to atmosphere I guess ambient). Please add the plot of the velocity contours as well as the inlet boundary condition. Do not use the results of single phase flow as initialization for two-phase flow runs. I would rather start with two-phase flows, switch off VOF equation from the equations which will be solved and then switch back.
- January 25, 2019 at 3:50 pm
  
  TingWangUNO
  Subscriber
  
  Dear Amine,
  
  1. I'm using the Eulerian model.
  
  2. Yes, I have provided zero operating density.
  
  3. When you say correct outlet pressure, I used Bernoulli's equation to calculate it for the single phase steady-state simulation. However, for the two-phase (the one with flow reversal) I don't know what value it will have!
  
  Question: Shouldn't Fluent increase the inlet pressure (velocity inlet BC) and adjust according to the outlet pressure boundary.
  
  4. If I don't initialize with the steady-state case...what should be the appropriate initial condition! Initialize from the inlet BC then?
  
  5. I will attach the velocity vector plots and pressure contours soon!
  
  Thank you.
- January 25, 2019 at 4:08 pm
  
  DrAmine
  Ansys Employee
  
  Yes Fluent would provide correct total pressure at inlet. The correct outlet pressure is the pressure outside: Example your pipe is opening into the ambient room then the pressure will be 1 atm. Or your pipe is opening into an ocean under 1000 meters then the pressure outside will be 1000*g*water density.
  
  Unsteady problems are Initial and boundary value problems so provide the most realistic appropriate initial field for the iterative process.
- January 25, 2019 at 4:40 pm
  
  TingWangUNO
  Subscriber
  
  I see that my simulation has a lot of mass imbalance...like if I plot mass flow rate difference between inlet and outlet...its huge.
  
  Question: In transient simulation should the mass flow difference between inlets and outlets be equal to zero? This is a phase change problem so it involves condensation and evaporation!! Would this be considered diverged solution then!!
- January 25, 2019 at 5:31 pm
  
  DrAmine
  Ansys Employee
  
  No as In transient you will have transient mass accumulation. But the mass out of one phase has to be the one into the other phase for phase change.
- January 25, 2019 at 6:10 pm
  
  TingWangUNO
  Subscriber
  
  So this is the Mass imbalance I plotted! It seems fine, right?
- January 25, 2019 at 6:12 pm
  
  TingWangUNO
  Subscriber
  
  I have adiabatic walls in this case, is there a way I could plot thermal energy imbalance as well?? Thank you.
- January 25, 2019 at 6:14 pm
  
  DrAmine
  Ansys Employee
  
  I don't know as it is just non scaled residual. Judge quality of results by monitoring temperature void and mass transfer rates so that the behavior remains constant towards the end of the time step.
- January 30, 2019 at 3:54 pm
  
  TingWangUNO
  Subscriber
  
  Hi,
  
  I realize there is a huge mass imbalance. The mass flow rate of mixture at inlet and outlet have a difference as high as -19.986084 kg/s. The mass flow rate at the inlet is around 6.2980032 kg/s. This is a huge mass flow rate difference!!
  
  In transient simulations the mass flow rate difference between inlet and outlet should be low at each time-step? Thank you.
  
  Best regards
  
  This is for time-step t = 0.3s
  
  mixture
  
  Mass Flow Rate (kg/s)
  
  report-def-0 -19.986084
  
  mixture
  
  Mass Flow Rate (kg/s)
  
  inlet 6.2983917
  
  mixture
  
  Mass Flow Rate (kg/s)
  
  outlet -26.284476
  
  water-liquid
  
  Mass Flow Rate (kg/s)
  
  inlet 6.2980032
  
  water-liquid
  
  Mass Flow Rate (kg/s)
  
  outlet -26.284476
  
  water-vapor
  
  Mass Flow Rate (kg/s)
  
  inlet 0.00038851637
  
  water-vapor
  
  Mass Flow Rate (kg/s)
  
  outlet 0