nonsense coefficient of drag

tschultz · April 18, 2015, 11:37

Here is a picture of my new mesh and the mesh "check" readout. I do not know where to check the skewness however.

The 1.875s time step was taken from argument on the thread you directed me to. you approximated a strouhal number of 0.2 and found the shedding frequency to be 1/75s. 75s/40=1.875 so I just went with this number.

The diameter of the cylinder is 1 meter. I am going to quick run through a simulation to see if the changed mesh has any noticeable effect.

tschultz · April 18, 2015, 11:58

I made the changes you recommended as I stated in my last post and here are the results that I attained.

By the way, convergence was MUCH faster!! That was a sight for sore eyes.

The Cd value converges around 0.8 so that seems reasonable compared with experimental data.

I did a quick run with the velocity increased to 10 m/s to attain a higher Reynolds number but that simulation appeared to go unstable.

flotus1 · April 18, 2015, 12:07

That is because if you increase the inlet velocity, the frequency of the vortex shedding becomes lower.
You would have to decrease the time step size to account for this.
I recommend a time step size at least 20 times smaller than the estimated frequency of the vortex shedding.

tschultz · April 18, 2015, 17:20

I adjusted the time step accordingly as you suggested so that the timestep is 40 times smaller than the frequency of the vortex shedding and the solution becomes stable, with a believable coefficient of drag. I am doing this for numerous Reynolds numbers and seem to be able to effectively reporoduce wind tunnel data, which is outstanding.

I must say thank you to all that have helped me out so far, with special thanks to you Flotus1. I really appreciate your help.

flotus1 · April 19, 2015, 04:28

Just a few thoughts:
I posted a wrong formula in one of the first posts and corrected myself later. With an inlet velocity u of 1m/s and a diameter D of 1m the estimated vortex shedding frequency T is

$T= \frac{D}{Sr \cdot u} = \frac{1m}{0.2 \cdot 1m/s} = 5s$

Divided by 20 this gives you a time step size of 0.25s

Simulating the drag coefficient of bluff bodies over a large range of Reynolds numbers is challenging.
Even if you do everything right, That is just not the strong point of RANS or URANS modelling.
Dont be disappointed if the numbers dont match experimental results very closely.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Drag Coefficient Convergence Problem	John	FLUENT	18	June 24, 2023 09:22
wrong SU2 calculation for lift and drag coefficient for NAC4421	mechy	SU2	7	January 9, 2017 05:18
problem with saving drag coefficient	colopolo	FLUENT	5	April 12, 2013 10:59
Calculation of Drag Coefficient, Help Please	teek22	CFX	1	April 26, 2012 18:41
Automotive test case	vinz	OpenFOAM Running, Solving & CFD	98	October 27, 2008 08:43

April 18, 2015, 12:07		#23
flotus1 Super Moderator Alex Join Date: Jun 2012 Location: Germany Posts: 3,399 Rep Power: 46	That is because if you increase the inlet velocity, the frequency of the vortex shedding becomes lower. You would have to decrease the time step size to account for this. I recommend a time step size at least 20 times smaller than the estimated frequency of the vortex shedding.

April 18, 2015, 17:20		#24
tschultz New Member Tim Join Date: Apr 2015 Posts: 13 Rep Power: 11	I adjusted the time step accordingly as you suggested so that the timestep is 40 times smaller than the frequency of the vortex shedding and the solution becomes stable, with a believable coefficient of drag. I am doing this for numerous Reynolds numbers and seem to be able to effectively reporoduce wind tunnel data, which is outstanding. I must say thank you to all that have helped me out so far, with special thanks to you Flotus1. I really appreciate your help.

April 19, 2015, 04:28		#25
flotus1 Super Moderator Alex Join Date: Jun 2012 Location: Germany Posts: 3,399 Rep Power: 46	Just a few thoughts: I posted a wrong formula in one of the first posts and corrected myself later. With an inlet velocity u of 1m/s and a diameter D of 1m the estimated vortex shedding frequency T is $T= \frac{D}{Sr \cdot u} = \frac{1m}{0.2 \cdot 1m/s} = 5s$ Divided by 20 this gives you a time step size of 0.25s Simulating the drag coefficient of bluff bodies over a large range of Reynolds numbers is challenging. Even if you do everything right, That is just not the strong point of RANS or URANS modelling. Dont be disappointed if the numbers dont match experimental results very closely.