Learning Forum

[maksay]

[RK]

Hello, nYou will need to specify the injection direction in the injection properties window: https://ansyshelp.ansys.com/account/secured?returnurl=/Views/Secured/corp/v202/en/flu_ug/flu_ug_sec_discrete_set_inject.htmlnnI hope this is helpful. n

[maksay]

Hello, You will need to specify the injection direction in the injection properties window: https://ansyshelp.ansys.com/account/secured?returnurl=/Views/Secured/corp/v202/en/flu_ug/flu_ug_sec_discrete_set_inject.htmlI hope this is helpful.https://forum.ansys.com/discussion/comment/105761#Comment_105761

Dear rahkumar, thank you for your answer. I am not doing a discrete phase analysis. There is continuous gas phase injection of air as a gas into the flowfield. I checked the manual and if I am not mistaken discrete phase simulation can be done with injected material being either solid or liquid.nSincerely,n

[YasserSelima]

Assuming you are using density based solver and your mesh is good, the problem you are facing means that you don't get deeply converged solution. Could be large time step ... or not enough iterations per time step.nn

[maksay]

Dear Yasser,nThank you for your answer,nAbout the problem you are talking about, normally the boundary condition should not change through iterations much I think as it is basically how the flow field is calculated based on. So if time step is not enough or convergence is not enough, the solution may be wrong, but still the boundary condition should be kept as it is I think.n

[Rob]

Remember that if you use two pressure boundaries the mass flow through that boundary is then part of the solution. The solution isn't changing the flow vector, it may be reducing the amount of material. What happens if you add a mass flow boundary to that location? n

[YasserSelima]

ArrayIf you are using inlet pressure, you might not get flow into the system depending on the system pressure. You can even get backflow in many cases if the solution is not converged. You might see the residuals 10^-3 but the pressure distribution in the flow field is still far from reality. Conversion means that the residuals reaches an asymptote .. now the solution doesn't change. If you never get constant value, at least, make sure you reach very low values 10^-6 for example.nAlso remember that the boundary conditions are set on the boundary nodes only. The adjacent cells will have different conditions that change with the solution progress. So, you will only find the temperature and pressure constant on the face ... only the face. nAs I recommended earlier, more iterations, smaller timesteps and finer mesh. This gets you closer to reality. If you have difficulty converging, decrease the URF and run few time steps then you can increase them back. Also ignoring the higher order terms gets faster convergence, but less accurate solution. So, you can ignore them at the beginning.nFrom my experience, you might get completely different transient behavior with different time steps or different convergence criteria. n

[DrAmine]

Can you add screenshots of pressure inlet settings?n

[maksay]

Remember that if you use two pressure boundaries the mass flow through that boundary is then part of the solution. The solution isn't changing the flow vector, it may be reducing the amount of material. What happens if you add a mass flow boundary to that location?https://forum.ansys.com/discussion/comment/106123#Comment_106123

I tried mass flow inlet, but then the other BC parameters defined in the inlet BC did not hold again. Still the injected gas should start the iteration and continue with the pressure coming from the injection face BC I think. I don't know what happens in FLUENT exactly but theoretically this should be the case and this case holds pretty well for the normal injection...n

[maksay]

Can you add screenshots of pressure inlet settings?https://forum.ansys.com/discussion/comment/106135#Comment_106135

Of course, thank you.nn

[maksay]

@maksay If you are using inlet pressure, you might not get flow into the system depending on the system pressure. You can even get backflow in many cases if the solution is not converged. You might see the residuals 10^-3 but the pressure distribution in the flow field is still far from reality. Conversion means that the residuals reaches an asymptote .. now the solution doesn't change. If you never get constant value, at least, make sure you reach very low values 10^-6 for example.Also remember that the boundary conditions are set on the boundary nodes only. The adjacent cells will have different conditions that change with the solution progress. So, you will only find the temperature and pressure constant on the "face" ... only the face. As I recommended earlier, more iterations, smaller timesteps and finer mesh. This gets you closer to reality. If you have difficulty converging, decrease the URF and run few time steps then you can increase them back. Also ignoring the higher order terms gets faster convergence, but less accurate solution. So, you can ignore them at the beginning.From my experience, you might get completely different transient behavior with different time steps or different convergence criteria.https://forum.ansys.com/discussion/comment/106129#Comment_106129

I only check boundary face values not the adjacent node values. In addition I think it is not quite possible to have a backward flow because injection pressure is a lot larger than the freestream pressure. I can understand the convergence criteria, but still I don't have the same problem with normal injection, it only exists when there is an angle.n

[DrAmine]

From my experience I prefer working with stagnation pressure boundary conditions. I usually provide pressure inlet Bcs and at the outlet zone I start with high pressure and then provide a function to ramp that pressure down to the value I require.n

[DrAmine]

Supersonic or Static Pressure won't be used if the flow at inlet is subsonic. For that reason pressure Inlet or at least mass flow inlet for one injection face + reasonable static pressure at inlet (2nd option not as good as 1st Option) are the way to go.n