[yousefaz]

Hello fellow CFD Onliners!

I am using Fluent to simulate 3D flow around a bluff body (rectangle) using steady state realizable k-e RANS with a standard wall function. The solution converges relatively quickly at low inlet velocities. When increasing the velocity, however, I am facing issues with oscillatory convergence, which I believe is due to the inherently unstable flow generated by the bluff body in the wake region, in particular the generation shedding vortices.

For this problem, a transient simulation is clearly the way to go. However, I believe there is no option in Fluent for automatically generating the time-averaged solution. Apparently, the only way to do this as suggested by other posts is to combine the exported instantaneous solutions using coding, which I am seriously trying to avoid!

The other option I've been experimenting with is the steady-state pseudo transient method, which addressed the problem of convergence at high velocities. However, the pseudo transient method produces a clearly visible Karman vortex street pattern, which would not otherwise be present in a time-averaged solution based on a fully transient simulation. This maybe because of the small time step (0.01s) and total solution time of 100s (10,000 iterations).

So what exactly does the pseudo transient solution represent and when should this method be applied? Can it be used to replace a time-averaged solution from a fully transient simulation?
And how can one get the time-averaged solution in Fluent?

Thanks in advance!

[shereez234]

Originally Posted by yousefaz

Hello fellow CFD Onliners!
So what exactly does the pseudo transient solution represent and when should this method be applied? Can it be used to replace a time-averaged solution from a fully transient simulation?
And how can one get the time-averaged solution in Fluent?

Thanks in advance!

Hi Yousef.

Unsteady simulations Fall in to two categories in time advancement.
1) Non-Iterative Time advancement
2) Pseudo Time advancement.

Number 1 mentioned above i.e. NITA scheme is specific to Flow Courant Number <= 1 (For RANS), therefore they are within the Stability region and Accuracy is ultimate given other conditions of flow are well defined. This method represents a real physical time approach and therefore does not need more than 1-2 iterations within the time step. It is well agreed that algorithms like PISO (Pressure-Implicit Split Operator) is effective for such simulations. The disadvantage is that these simulations take a very long time to converge

Number 2 - i.e. Pseudo Transient methods use large time Steps i.e. Courant number > 1 to may be upto 200 or more. They advance in time using much larger time difference and therefore to capture an accurate representation of the Unsteadiness and time averaged values you will need to put Iterations (10 - 30) or more depending on the courant number). It has been said in this forum and in literature that the amount of iterations to put in a time step should be sufficient to reach a satisfactory convergence within the time step.

Personal comments: Pseudo Methods are quicker and if carefully done can reach the same results as NTIA methods at the end of the solution (converged resuts). But this is really upto you to decide and investigate. A comparison of these two methods for your specific case might be very helpful for you to understand.

Regards

Shereez

[LuckyTran]

Why can't you just enable data sampling for time statistics and get the mean velocities?

[shereez234]

That is a more appropriate answer. I think I got carried away

[vineet_k_t]

Can we obtain Temperature time variation also using Pseudo transient method.
In many literature i have found that people mention steady sate simulation and then come up with time variation of temperature.
someone mentioned that in Fluent use pseudo transient setting in solver options and calculate (volume or surface) average of temperature in your mesh in solution monitor, then you can plot the temp vs time step.

is it possible ?

[vinerm]

Time averaged data required transient simulation. Pseudo-transient is just a numerical method to solve coupling of velocity and pressure field. But the results are steady-state. There is no transient data extraction possible. However, time-averaged and steady-state results would be same if the system is statistically steady.

[melvinardan]

Can we obtain Temperature time variation also using Pseudo transient method google account manager apk
In many literature i have found that people mention steady sate simulation and then come up with time variation of temperature.
someone mentioned that in Fluent use pseudo transient setting in solver options and calculate (volume or surface) average of temperature in your mesh in solution monitor, then you can plot the temp vs time step.

is it possible ?

Yes, it is possible to obtain the time variation of temperature using the pseudo transient method in Fluent. The approach mentioned by the person you quoted is one way to do this.

To use the pseudo transient method to obtain the time variation of temperature, you can follow the steps below:

1. Set up a steady-state simulation using the realizable k-epsilon RANS model with standard wall function.
2. In the Fluent solver settings, enable the pseudo transient option and set a small time step size (e.g. 0.01s).
3. Run the simulation for a specified duration (e.g. 100s).
4. Once the simulation is complete, you can obtain the time variation of temperature by calculating the (volume or surface) average of temperature at each time step using the Solution Monitor feature in Fluent.
5. Plot the time variation of temperature against the time step to obtain a graph of temperature as a function of time.

It is important to note that the pseudo transient method is not a true transient simulation, and therefore the temperature variation obtained using this method may not be as accurate as that obtained using a fully transient simulation. However, if the flow is relatively steady, the results obtained using the pseudo transient method may be reasonable approximations. Additionally, the accuracy of the results will depend on the time step size used and the total simulation time.