## Fluids

#### convergence towards a good result

• abd fs
Subscriber

Hello dear experts,

I work with unsteady turbulent flow, with an interval between [0, 0,8]. Normally, I have to get good results when the simulation has reached 0.1 s, as indicated in a reference article, but in my simulation, I get these values when my simulation has reached 0.8 s.

Do you have any tips for solving the problem?

Does this problem come from time step, mesh or something else?

Best regards,

SER

• Rob
Ansys Employee

We need more information: chances are it's a mix of mesh, settings, initial values and lack of information in the literature (I've found published data & models to be somewhat lacking in crucial details as the author omits things they don't want to share).

• abd fs
Subscriber

Dear rwoolhou,

Turbulent flow with turbulence model K-W-SST.

At the Y + wall level used, in the other area, I used a triangular mesh

I have two velocity inputs, the first one is a tabular velocity profile, the second one is a sinusoidal function that depends on time, amplitude, frequency thanks to a C language code. I calculated the time step with:

F=45 Hz, Simulation time: 30 periods of the jet (T),  Number of time step / T: 200

T= 1/45 = 0.0222, time step= 0.0222/200 = 0.00111 for implicit

Is the way used to calculate the time step is right, mesh quality?

Best regards,

SER

• DrAmine
Ansys Employee
Tine step size might be okay but aspect ratio is quite high. Where are the cells with this high aspect ratio located? What about cell volume change?
• abd fs
Subscriber

Dear Amine,

Thank you for your quick response, in terms of ratio, their great value is located in the premises where there are circles in black, I'm not interested in their value in these premises, I'm interested in the circle in red where the value less than 1000.

Best regards,

SER

• Rob
Ansys Employee

What was the mesh like in the paper? What initial conditions did they use?

• abd fs
Subscriber

Dear rwoolhou,

Best regards,

SER

• Rob
Ansys Employee

How long does the flow take to get from one end of the domain to the other?  I'd use PRESTO!  but doubt it'll make much difference.

• abd fs
Subscriber

Dear rwoolhou,

He must take 0.1s

first case steady flow without jet (velocity inlet) the results are correct.

Second case transient flow add a jet (velocity inlet = sinusoidal function thanks to a UDF) in this case I find a lot of problems.

Do you have any recommendations for solving this problem? I would be very happy.

I am stuck for almost 4 months because of this problem

Best regards,

SER

• abd fs
Subscriber

Dear experts,

I have a doubt about how the average and the data sampling for the time statistics are activated.

My goal is to record the values by each period (T = 160 Number of time steps) and the averages it contains.

You will find attached files that represent the way I recorded and averaged.

Best regards,

SER

• abd fs
Subscriber

Please guys any help will be greatly appreciated

• abd fs
Subscriber

Dear experts,

I found this condition (immiscible phases) in an article because I have two velocity inputs. How can I insert it into my solver?

Best regards,

SER

• peteroznewman
Subscriber

Immiscible describes two fluids that don't mix into a homogeneous fluid, for example oil and water. Do the two inlets have different fluids that will separate if left standing?

If you do have two fluids that are immiscible, you would turn on the Discrete Phase Model in Fluent.

• abd fs
Subscriber

Dear experts,

I do not know why they indicated this condition despite the fact that both fluids are air?

Best regards,

SER

• peteroznewman
Subscriber