# CoP simulation comparison, including CFD

### Help Support The Rocketry Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Wow @Buckeye !

That is BEAUTIFUL !

Could the CFD model somehow be factoring in the stall angle for the fins ?

Both the NACA 0012 and a flat plate stall at about the same AoA as your 'anomolous data' ...

-- kjh

Wow @Buckeye !

That is BEAUTIFUL !

Could the CFD model somehow be factoring in the stall angle for the fins ?

Both the NACA 0012 and a flat plate stall at about the same AoA as your 'anomolous data' ...

-- kjh

Oh, wow! I did not consider that. CFD is picking up the stall. That is a good explanation!

<<snip>>

The CFD, minus the 12 degree anomaly, matches the linear trend of the Galejs equations. The y-axis scale purposely runs out to 1.0 caliber to illustrate the absurdly- small movement of CP with angle of attack. This is why short, stubby rockets are very stable with low CP-CG margins.

View attachment 635340
Hmmm ... so maybe you could write a paper like, say ... "What Barrowman and Galejs left out"

-- kjh

Oh, wow! I did not consider that. CFD is picking up the stall. That is a good explanation!
A look at streamlines at 12 degrees and a little bit on either side should be illuminating. It seems odd that the CP would shift back to the original spot if the fins lost lift due to stall, though.

I think the real takeaway here is that OR sims accurately reflect CP movement, so we likely should go to stability as a percentage of length rather than a strict number of calibers.

Would surface pressure plots show the stall on the fins? This is really juicy stuff!

Would surface pressure plots show the stall on the fins? This is really juicy stuff!

Working on that...

Yep, 12 degrees is near the onset of full flow separation off the fins.

Drag and lift coefficients. Since the model is aligned with the axial and normal directions, I had to do the trig to resolve the forces into the L and D directions.

(Fun fact: Automotive CD is measured along the vehicle axis (direction of travel), not the relative wind.)

The shape of these curves seems reasonable.

Last edited:
Nice work! Is it possible with the software to get a plot of the surface? When I was in college, I was a CO-OP student at the university wind tunnel. We ran tests for an airplane company that was experiencing stall on their flaps during landing. The techs mixed up diesel and some black powder, painted the model (which was painted white or light gray), and then made a run. Immediately after they took pictures while the flow pattern remained. The results were similar to the picture I attached, except it was not multi-colored I made surface plots of an Alpha running the Solidworks product "Flo Works" many moons ago.

#### Attachments

• DLR-F11-with-flow-visualization-paint-in-the-Low-Speed-Wind-Tunnel-of-Airbus-at-Bremen.jpg
55.3 KB · Views: 0
• StdNoLLPr2.jpg
34.1 KB · Views: 0
Yes, anything is possible in CFD post-processing. Is there anything in particular you want to see on the rocket surface?

Yes, anything is possible in CFD post-processing. Is there anything in particular you want to see on the rocket surface?
I was just curious what the surface pressure plot on the windward and leeward sides looked like. I marked up your diagram from earlier. I've long been curious about Openfoam when a colleague at work recommended it. I'm not a Linux user, and tried to run a bootable Ubuntu session but could not get OF to install properly...I gave up then. My career work has been structural FEA, but I dabbled with CFD at work. Since I retired, I've lost access to the software.

#### Attachments

• Fatboy_CFD angle-02.png
378.3 KB · Views: 0
I was just curious what the surface pressure plot on the windward and leeward sides looked like. I marked up your diagram from earlier. I've long been curious about Openfoam when a colleague at work recommended it. I'm not a Linux user, and tried to run a bootable Ubuntu session but could not get OF to install properly...I gave up then. My career work has been structural FEA, but I dabbled with CFD at work. Since I retired, I've lost access to the software.
Here ya go. Leeward and windward sides at 12 degrees angle of attack

If you are a noob and want to play with OpenFOAM, you really want a friendly UI front end. Most are licensed, but a few are freeware. I am using FreeCAD with CfdOF Workbench. There are Windows and Linux installations. Paraview is used for post-processing results.

Here ya go. Leeward and windward sides at 12 degrees angle of attack

Really illuminating stuff, thanks!
Might you have the time to also take an image of the pressure contours on the aft end? Speaking to base drag... Disregard, found it on your other thread

Last edited:
After some searching and a tip from drewnickel, I now have the proper way to compute CP at zero AOA from the CFD models. From the OR thesis and user guide:

For zero AoA, I will be using the slopes as in eq. 3.5, and not the classical center of pressure integral as in post #1. For the Fatboy model, CP = 221 mm (was 222 mm). This is not a big difference, but longer models will be.

For the last week or so, I've been working on a 4-fin model to address that issue. I'm having some issues with getting a test model meshed, so I need to do some further troubleshooting. In an ideal world, I'll run a bunch of AoAs and possibly different lengths of models (current testbed is 5:1 L).
I've often wondered why most real rockets have 4 fins while most hobbyist rockets have 3.

I know the number 1 answer is weight and drag. But then, I've also thought that 3 fins always induce rotation too.

I've often wondered why most real rockets have 4 fins while most hobbyist rockets have 3.

I know the number 1 answer is weight and drag. But then, I've also thought that 3 fins always induce rotation too.
I think there's a decent chance that it's a heck of a lot easier to put guidance on a 4-fin rocket. One pair of fins controls the y axis, and the other controls the z.

CFD simulation of a flat disk at Mach 0.3

I wanted to predict the CP of a flat disk, which gives some insight to the stability of spools and saucers. Bruce Levison claimed that published data shows the CP for a flat plate in a perpendicular flow is about 2.2 diameters behind the plate. He also noted that CP values for a flat plate vary widely. No references were given for either statement.

The disk is 66 mm in diameter and 5 mm thick. The CFD set up is the same as my previous FatBoy studies, except that I increased the grid resolution a tad immediately around the disk to hopefully better capture any organized vortex shedding.

First, some flow analysis:

Here is the steady-state RANS drag convergence. The result is CD = 1.18, similar to Hoerner's 1.17:

Cut plane of velocity streamlines. Note that the flow massively separates off the leading edge of the disk, and throws a wide wake, with the disk base not doing much.

Here is the pressure drag on the front and back of the disk, along with some streamlines. The drag due to stagnation on the front of the disk is much greater than the low pressure base drag. "Base drag stability" is a misnomer, imo. A better description of spools and such would be "vortex shedding stability."

I also ran DES transient simulations for more realistic time accuracy. CD is starting to oscillate around 1.18 as well:

The small amplitude oscillations have a period around 0.0004 s. Here are some animations at this capture rate. Looks like some organized low-pressure vortex shedding sets up the fluctuating wake. (Loop the animation and adjust speed as needed for best viewing.)

View attachment anim08.mp4

View attachment anim05.mp4

Finally, CP calculation. I ran the disk pitched at 2 deg and 4 deg angles of attack. Pitching moment about Y and normal force along Z were recorded and plotted. Assuming zero normal force and moment at zero AOA, I fit a 2nd order polynomial through 3 points:

Evaluating the slopes and using the l'Hopital's rule as in post #44 yields the CP location at zero angle of attack:

X = 290 mm or about 4.4 diameters behind the disk.

So, it seems plausible that a disk with a long sting on the back would measure CP to be a few diameters behind the body in a wind tunnel.

CP is abstract enough on streamlined bodies. It becomes even more weird and merely academic on odd shapes like this.

CFD simulation of a flat disk at Mach 0.3

I wanted to predict the CP of a flat disk, which gives some insight to the stability of spools and saucers. Bruce Levison claimed that published data shows the CP for a flat plate in a perpendicular flow is about 2.2 diameters behind the plate. He also noted that CP values for a flat plate vary widely. No references were given for either statement.

The disk is 66 mm in diameter and 5 mm thick. The CFD set up is the same as my previous FatBoy studies, except that I increased the grid resolution a tad immediately around the disk to hopefully better capture any organized vortex shedding.

First, some flow analysis:

Here is the steady-state RANS drag convergence. The result is CD = 1.18, similar to Hoerner's 1.17:

View attachment 658921View attachment 658922

Cut plane of velocity streamlines. Note that the flow massively separates off the leading edge of the disk, and throws a wide wake, with the disk base not doing much.

View attachment 658923

Here is the pressure drag on the front and back of the disk, along with some streamlines. The drag due to stagnation on the front of the disk is much greater than the low pressure base drag. "Base drag stability" is a misnomer, imo. A better description of spools and such would be "vortex shedding stability."

View attachment 658924View attachment 658925

I also ran DES transient simulations for more realistic time accuracy. CD is starting to oscillate around 1.18 as well:

View attachment 658931

The small amplitude oscillations have a period around 0.0004 s. Here are some animations at this capture rate. Looks like some organized low-pressure vortex shedding sets up the fluctuating wake. (Loop the animation and adjust speed as needed for best viewing.)

View attachment 658932

View attachment 658933

Finally, CP calculation. I ran the disk pitched at 2 deg and 4 deg angles of attack. Pitching moment about Y and normal force along Z were recorded and plotted. Assuming zero normal force and moment at zero AOA, I fit a 2nd order polynomial through 3 points:

View attachment 658943View attachment 658944

Evaluating the slopes and using the l'Hopital's rule as in post #44 yields the CP location at zero angle of attack:

X = 290 mm or about 4.4 diameters behind the disk.

So, it seems plausible that a disk with a long sting on the back would measure CP to be a few diameters behind the body in a wind tunnel.

CP is abstract enough on streamlined bodies. It becomes even more weird and merely academic on odd shapes like this.
This is an interesting study. One thing that doesn’t feel “right” is that the oscillation speed seems very high frequency for what I think of for Von Karman vortices.

I don’t think that base stability is a terrible name. The vortices are set up by the large base area.

This is an interesting study. One thing that doesn’t feel “right” is that the oscillation speed seems very high frequency for what I think of for Von Karman vortices.

Maybe. Don't expect a beautiful vortex street like a 2D cylinder at the optimum Reynolds number. This is more complicated.

In this disk case, Re = 4.55E+05

A look at normal force Fy may be a better depiction, since it is indicative of the flow symmetry. It shows that there may be evidence of two Strouhal Numbers. The high frequency corresponds to about St = 1.8. The lower frequency would be about St = 0.13, which is more like the classic value. I am just eyeballing here and not running an FFT.

https://en.wikipedia.org/wiki/Strouhal_number

This paper gives some insights on vortex shedding of a disk and 3D axisymmetric bodies. They used a velocity signal, not forces. ...and an interesting citation similar to my thinking:

https://www.researchgate.net/publication/225773499_On_vortex_shedding_behind_a_circular_disk

"Jefferys (1930) pointed out that the vortex system behind
a three-dimensional body is not simply an extension of the
Karman vortex street of a two-dimensional cylinder."

Last edited:
Are you going to try the same sim with more disk thicknesses? Moving from Disk to Cylinder?

Also, the experimentalist in me cringes at fitting 3 terms to 3 points (I count a quadratic as 3 terms - Ax^2 +Bx + C) It's something I've been trying to get my colleagues to not do because they keep coming to me and saying 'why did my fit explode' - and I have to explain degrees of freedom to them again.

But these aren't measured points - they're values from the sim, and probably count as having zero error. You're also forcing a zero intercept - which both gains you a point, and reduces a term. Sorry for being wordy, I was just thinking it through.

Are you going to try the same sim with more disk thicknesses? Moving from Disk to Cylinder?
Wasn't planning on it.

Also, the experimentalist in me cringes at fitting 3 terms to 3 points (I count a quadratic as 3 terms - Ax^2 +Bx + C) It's something I've been trying to get my colleagues to not do because they keep coming to me and saying 'why did my fit explode' - and I have to explain degrees of freedom to them again.

But these aren't measured points - they're values from the sim, and probably count as having zero error. You're also forcing a zero intercept - which both gains you a point, and reduces a term. Sorry for being wordy, I was just thinking it through.

Yeah, not stochastic data, here. I just need a direct fit polynomial to differentiate and get in the ballpark for slopes. For real rockets, these quantities are assumed linear at small angles of attack, and one evaluation at say, 2 degrees, is typically all that is done (plus assuming 0, 0). Since this is not a rocket, I thought I should do a little better than that. I could add a couple more points between zero and 4 degrees.

The drag due to stagnation on the front of the disk is much greater than the low pressure base drag. "Base drag stability" is a misnomer, imo.
Yup. I've been trying to convince people of this for a while.
A good experiment to run is what happens with a shape that's a flat disk on the front and a cone or streamlined shape on the rear, like the Apollo capsules on reentry. That configuration is also very stable, but it doesn't have nearly as much dead space on the "base".

Replies
343
Views
15K
Replies
131
Views
7K
Replies
20
Views
977
Replies
11
Views
2K
Replies
1
Views
662