[SB123]

I'm looking to measure the wake width behind a bluff body in different configurations. Question: how would you expect to do it?

i was thinking about running 2 C/S planes behind the flow and measuring vorticity (potentially time averaged?) and see where i have a gradient and use that to define the wake width? problem is i don't know how to time average a visualization plane

i'm not sure, is there a better way to try to measure this ?

[SB123]

is this dumb or does no one know what i'm talking about ?

[lovecraft22]

There's no easy answer and also there's not only ONE answer… You can measure it in different ways and get different values.

One thing could to average your measurements in time (in case you are running transient simulations). Then you can plot the velocity magnitude divided by the free stream velocity in the direction perpendicular to the flow and along the width of the wake. Where you value is different than one… that's your wake. I've used this method for both wind tunnel tests and CFD.

Other methods include measuring the velocity standard deviation or the vorticity magnitude… You can get slightly different absolute results but relative results should be ok. I would do some of the above mentioned plots for all your geometries to make the comparison. Measuring the wake in terms of millimiters doesn't make much sense in my opinion.

[SB123]

thanks for the reply lovecraft. I'm in the process of trying to implement your suggestions. any chance you know how to do that in STar? I can't figure out how to time average a Derived part. ( ie plane measuring velocity) or even for that matter, a Time series of the velocity on that plane

[Totalsim]

Hi. We would do something similar.

Calculate a total pressure ceofficient on a slice behind the vehicle. Anywhere on the slice that this dips below 1 energy has been taken from the flow.

You can do a similar thing with isocontours of total pressure to show wake patterns.

Good luck.

[DaveyBaby]

Hi,
Just in case you are still stuck on this, you can create monitors that give you variances and/or co-variances of the velocity components. As part of this process, they update their calculated time-averaged mean velocity components on each time-step and then also record the difference between this and the current time-step's instantaneous velocity. Once the means have converged, you can access the mean velocity as well as the Re stress components! (field functions for them are automatically created).
1.) Right-click monitors, create Field covariances
2.) Select the velocities you want ie to get U,V,W and all six Re stresses make all six ui'uj' covariance monitors
3.) Edit the newly-created monitors and tick "Publish mean" for the mean velocity components you want.
4.) You can select these calculated values as scalars, ie XY plot, choose part and selecting "Mean of Velocity[i] from co-variance monitor" or something will only give you the mean values on the derived parts you want.

Hope this helps, I don't ever run a time-dependent sim without doing this now.

[DaveyBaby]

Don't start the monitors recording until you have a good idea that you have reached a statistically steady state or it will take much longer to reach the correct mean! If you have been recording and realise that it hadn't yet reached such a state you can reset the monitors. To see if you have yet you can just do a time-dependent plot of e.g centre-line velocity or something else that will converge towards a solution that is visibly fluctuating around the then established mean.

[SB123]

Originally Posted by DaveyBaby

Hi,
Just in case you are still stuck on this, you can create monitors that give you variances and/or co-variances of the velocity components. As part of this process, they update their calculated time-averaged mean velocity components on each time-step and then also record the difference between this and the current time-step's instantaneous velocity. Once the means have converged, you can access the mean velocity as well as the Re stress components! (field functions for them are automatically created).
1.) Right-click monitors, create Field covariances
2.) Select the velocities you want ie to get U,V,W and all six Re stresses make all six ui'uj' covariance monitors
3.) Edit the newly-created monitors and tick "Publish mean" for the mean velocity components you want.
4.) You can select these calculated values as scalars, ie XY plot, choose part and selecting "Mean of Velocity[i] from co-variance monitor" or something will only give you the mean values on the derived parts you want.

Hope this helps, I don't ever run a time-dependent sim without doing this now.

this is spot on, thanks Davey