hello,

i've managed to get some answers for my flow past square cylinder simulation. although the velocity plot seems correct, i need to verify against some values.

my code gives pressure and velocities at all pts. how am i going to use these data to get the shedding frequency?

i've tried reading some papers and search for answers in this forum but to no avail....

hope someone can help...

thanks!

In case of the steady flow, using the velicities and pressure fields you can calculate the drag coefficients, wake length etc.

In case of unsteady flow, drag & lift coefficients and Strouhal number. These can be benchmarked with the literature.

You can also compare the pressure and vorticity varition over the surface of the cylinder.

... the main point being: You have to run a time-accurate unsteady simulation to obtain the vortex shedding frequency. Even for the time-averaged drag, a steady-state solution will not be accurate.

Under to steady flow regime, the drag will not be dependent of the steady/unsteady simulation. I think the accuarcy will also not be dependent on it. The unsteady computation is required only for the vortex shedding characteristics.

>Under to steady flow regime, the drag will not be dependent >of the steady/unsteady simulation. I think the accuarcy will >also not be dependent on it. The unsteady computation is >required only for the vortex shedding characteristics.

Unfortunately, it's not that simple. The time-average drag of a blunt body has has a lot to do with vortex shedding. The unsteady generation and shedding of vortices (stirring up the flow over time) is what imposes a drag on the body. That's a little like the induced drag you observe on wings, although with vortex shedding, this process is inherently unsteady. If you don't run a time-accurate solution, you will get no vortex-shedding, the work that the body performs on the flow is significantly lower, and therefore the drag is going to be underpredicted by the steady-state equations.

There have been smart efforts (by Li He in the UK, and maybe others) to implement a model for the additional unsteady flow stresses into a steady simulation, kind of like a very large eddy simulation, but I don't recall how successful that was...

Don't forget that solving the steady-state equations is not the same as taking the time-average of an unsteady solution, in the case of a non-linear problem like vortex shedding!

Zonexo

Before all that was told by Ramp and Mani, do not forget to consider the following things before trying to compare your results:

- Is your computational model (geometry, Reynolds number, time step, CFL, constraints, etc...) equal or equivalent with the model of your reference? The validation process turns easier when you replicate the problem from the reference.

- Is the cylinder fixed or not?

- How and where are you applying the no-slip bnd conditions?

- For 3D problems: are you solving a problem like a wind tunnel (with no-slip at the bottom, or top, surface) or an idealized problem (without top and bottom walls)?

At first, I would not be worried about the accuracy of my solution. I would be engaged to reproduce the solution from some reference, being it high or low accurate (I am just comparing results...). For a first compare, just extract the velocities components from some point in the shedding street and plot against the time. It will give you some idea about your Strouhal number (some papers also have these plots with the velocities). After matching some results you will feel how are you going to try a well elaborated solution and study other parameters (drag, lift, etc...) and the accuracy of your solver. I used to do it: try some relaxed comparison before studying more elaborated parameters.

Cheers