I'm building a MD project and I've elected to use hefty fillets rather than tip-to-tip. For the mass, I created a mass object the length of the fin root, and I have before and after measurements to tell me how much they weigh. Probably a silly question, but I'm curious how to model the drag of my fillets?
OR - how to simulate the drag of fillets?
- Thread starter matthew
- Start date
Help Support The Rocketry Forum:
TheTellurian
In space no one can hear you fart
- Joined
- Mar 3, 2009
- Messages
- 339
- Reaction score
- 14
AFAIK sharp 90* corners create vortex that increase drag. Filleting delays or removes the vortex reducing drag. You should not need to worry about it.
Richard
Richard
Titan II
Well-Known Member
- Joined
- Aug 23, 2013
- Messages
- 2,020
- Reaction score
- 1,157
Does not answer your specific question, but you may find this interesting.
https://www.rocketryforum.com/showthread.php?120320-Drag-effects-of-fin-fillets/page2
https://www.rocketryforum.com/showthread.php?120320-Drag-effects-of-fin-fillets/page2
markkoelsch
Well-Known Member
- Joined
- Mar 18, 2009
- Messages
- 4,362
- Reaction score
- 160
Simply, it would at best be tough to simulate. If you were to look at Barrowman's work, which OR and Rocksim are both based on, the configuration that fit are pretty basic. I read the Barrowman stuff a decade ago and did a couple rockets by hand, and I do not believe there is a method to simulate a fin fillet within.
You likely would have to use a CFD package.
You likely would have to use a CFD package.
Using the recommended 4-8%, gives a roughly .5" to 1" fillet in a 12" root chord, which fits with my build. I'll consider it done and dusted.
I had forgotten about that thread, or that I had read it when it was active. It's probably what planted the seed in my head to consider it in my case. Apologies for the redundant thread, and thanks, gentlemen, for the sanity check.
Xrain
Well-Known Member
- Joined
- Nov 26, 2013
- Messages
- 282
- Reaction score
- 30
Most of the quick drag calculations that are used are pretty basic, but usually will give you a pretty good idea of how much drag your rocket will have. If you are getting to the point where you are getting detailed enough to worry about drag effects of small things like fin fillets, CFD is pretty much the way to go other than empirical data.
Thanks again. On this project I'm not trying to squeeze out the most performance by simming and adjusting then building. The rocket is mostly built. Our club's waiver requires that flights above 17k' AGL need technical review (current OR estimate is ~24k'). I just wanted to have a fairly accurate sim to submit to the review panel and was trying to cover all bases. My fillets are toward the upper end of the given rule of thumb, so I was curious if OR may be over estimating my altitude and velocity.
markkoelsch
Well-Known Member
- Joined
- Mar 18, 2009
- Messages
- 4,362
- Reaction score
- 160
Overestimation of altitude is not a problem. The best thing to do is to sim it as accurately as you can including wind, temperature, and launch angle. Fly it, and get the altitude from your avionics. Adjust the Cd of the rocket in the sim to meet the altitude of the rocket hence giving you a fairly accurate Cd. All future flights and sims will benefit.
- Joined
- Sep 5, 2009
- Messages
- 3,532
- Reaction score
- 1,655
The motor data file is the most unreliable thing in the simulation. Your big fillet drag vs. small fillet drag is a mere blip in the noise of motor variation. Don't sweat it.
A single-value, back-tracked Cd is a fudge factor for one flight only, and likely not repeatable for a rocket going 24k ft. At those conditions, the rocket is probably using a motor much different than your test flight, going above Mach for a longer time, possibly flying at a larger angle of attack over time, and the atmospheric winds are nothing like ground level.
I would put +/-10% bands on your simulation as a best case range of prediction as compared to real life, assuming you input everything correctly. Thus, it is safe to say that your model will exceed the 17k ft by a good margin. You need to re-design if you want lower altitude.
A single-value, back-tracked Cd is a fudge factor for one flight only, and likely not repeatable for a rocket going 24k ft. At those conditions, the rocket is probably using a motor much different than your test flight, going above Mach for a longer time, possibly flying at a larger angle of attack over time, and the atmospheric winds are nothing like ground level.
I would put +/-10% bands on your simulation as a best case range of prediction as compared to real life, assuming you input everything correctly. Thus, it is safe to say that your model will exceed the 17k ft by a good margin. You need to re-design if you want lower altitude.
Last edited:
markkoelsch
Well-Known Member
- Joined
- Mar 18, 2009
- Messages
- 4,362
- Reaction score
- 160
It maybe a fudge factor, but it will likely be closer to a real Cd than the estimate the program comes up with. Real world extrapolated into a sim.
- Joined
- Sep 5, 2009
- Messages
- 3,532
- Reaction score
- 1,655
The back-tracked Cd is a fudge factor for everything wrong in the simulation - weather, mass, motor - not just drag coefficient. I think the Cd simulators can do better than this on basic rocket shapes. The only really good Cd from flight data comes from accelerometer measurements to extract Cd vs. Mach Number. I have seen on this forum some good Cd curve comparisons of RASAero to Featherweight altimeter data, for example.
markkoelsch
Well-Known Member
- Joined
- Mar 18, 2009
- Messages
- 4,362
- Reaction score
- 160
Perhaps. As I alluded to previously, you need the sim to not only be accurate for the rocket itself, but the motor, the weather, and the launch. If these are correct, then the Cd adjustment is the only way to really go. There is no way to adjust for motor variance ( trust me as I create a lot of motor files for all) with the current simulators unless one is modified to randomly vary the motors parameters within a certain range. I think Rocksim Pro might do that, but then you are introducing another variable.
So, you could improve the above by flying multiple flights and comparing them to accurate sims. Far from perfect I will admit, but better than not doing it.
- Joined
- Sep 5, 2009
- Messages
- 3,532
- Reaction score
- 1,655
Right. One's ability to do this is just as faulty as Cd prediction. I can't tell if a motor will burn at 1, 2, or 3 standard deviations of its thrust curve, or simulate the wind profile at 15k ft.
I know. I am the one complaining about proliferation of dubious motor files in other threads!
The splashdown simulation is the only thing with some statistical significance that I have seen.
I would like to see this data of backed-out constant Cd. I never had a lot of success with the approach.
Last edited:
CarVac
Well-Known Member
- Joined
- Feb 12, 2012
- Messages
- 5,704
- Reaction score
- 37
Constant Cd only works if you are subsonic. When in the supersonic regime, a constant Cd will underestimate altitude for slow burning motors and overestimate altitude for fast burning motors. You cannot adjust a fixed Cd for all flights when your rocket goes supersonic. It's just not possible.
What I would do is tweak parameters of the rocket (fin thickness particularly) in order to match altitude. This way OpenRocket still does its variable drag estimation, while coming closer to the actual performance of the rocket.
markkoelsch
Well-Known Member
- Joined
- Mar 18, 2009
- Messages
- 4,362
- Reaction score
- 160
Right. One's ability to do this is just as faulty as Cd prediction. I can't tell if a motor will burn at 1, 2, or 3 standard deviations of its thrust curve, or simulate the wind profile at 15k ft.
I know. I am the one complaining about proliferation of dubious motor files in other threads!
The splashdown simulation is the only thing with some statistical significance that I have seen.
I would like to see this data of backed-out constant Cd. I never had a lot of success with the approach.
Maybe you should get off your butt and contribute motor files if you think you can do better. Anybody can contribute motor files on Thrustcurve. I can tell you that I have developed relationships with CAR, TMT, and NAR. I use the cert data to make my files-what do you use in the files you share with others in the hobby?
It is a very manual process, and yes, the rare mistake is made. Once notified of an error, it has never taken me more than a couple days to verify the error and to fix it.
Frankly, it is really easy to complain while doing absolutely zero to help. So, either step up and help or pipe down.
Also, it would be nice to know who this actually is? Buckeye does not really help in the regard, so if complaining a name would be nice.
I am also not saying a constant Cd is perfect. I think adjusting to match flight data over multiple flights can be useful especially if you have made sure the sim is as accurate as you can. To do it well takes data, several flights, and is in essence a bracketing technique. You will get close. If you think you have a better method using commonly available tools i am listening.
Xrain
Well-Known Member
- Joined
- Nov 26, 2013
- Messages
- 282
- Reaction score
- 30
A standard technique with CFD and other simulation programs is to take exiting empirical data, then rework your simulation until you can match the results like Mark suggested. You can do this a bit with OR, if you have any friends who used similar fillets, and similar speeds, and recorded good flight data. You can make his rocket in OR until the simulation matches the real data, then make the modifications to the model convert it to your rocket. This should at least get you well within the ball park, and if you record data on your flight you can compare it with your simulated results to get yourself a better idea on what adjustments in OR will have what effect.
This kind of self checking is what is done in industry as well, since depending on what mistakes you make in your modeling, any number of errors can occur.
This kind of self checking is what is done in industry as well, since depending on what mistakes you make in your modeling, any number of errors can occur.
markkoelsch
Well-Known Member
- Joined
- Mar 18, 2009
- Messages
- 4,362
- Reaction score
- 160
Exactly what I was trying to get at.
Both Rocksim and OR are based on Barrowman. Read Barrowman once- there are some basic assumptions about velocity being subsonic to start with so many flights are pushing those equations beyond their intent a bit or a lot.
Drag calculation hence Cd is at best an approximation. CFD is best way to simulate it, but none of the aforementioned simulators use it. To use the method suggested above to better approximate your real Cd is a way to use real world data to improve your simulation. Yes, they are all approximations.
The absolute best method to determine a real Cd curve would be with a full scale model in a supersonic wind tunnel. Considering that is not accessible to 99.99% of hobbyist I will stick to my method, which has proven to work adequately.
- Joined
- Sep 5, 2009
- Messages
- 3,532
- Reaction score
- 1,655
Excellent idea.
- Joined
- Sep 5, 2009
- Messages
- 3,532
- Reaction score
- 1,655
I am poking fun at myself (the complainer), not you (the doer). Maybe the smiley emoticon wasn't enough to convey that message. Sorry. I have given you kudos for your contributions, like rocketryfiles, in various forum discussions. Thanks again.
I like CarVac's idea of tweaking a drag component, not the entire value. I suggested the acclerometer data, as others have done. I tried the constant Cd, and I could not get consistent results. That is why I am curious to see some successful data.
Last edited:
markkoelsch
Well-Known Member
- Joined
- Mar 18, 2009
- Messages
- 4,362
- Reaction score
- 160
Sorry, I am not really looking for kudos, but will take them. I cannot see some stuff at times like emoticons as I read mostly on my phone.
All these methods can work I think. The thing is that the tools we as hobbyists have are not professional tools- at least in most cases. I think the trick is to constrain as many variables as you can before making whatever modification you will make.
Also, my method does work for the rockets I have tried it on. I would have to delve into a lot of notes, simulations, and such to dig it out and do not have the time now. I will put the caveat though- if you are basing an Cd approximated on flight data you do need to have data from similar performing flights.
Similar threads
