"Flat Plate" flight profile calculation =advanced=

[DynaSoar]

I'll start.

Recently I've heard someone say that most rocketry speed/altitude flight profile calculators were "flat plate" estimates, meaning they pretty much just took frontal area into account (as well as weight, of course).

This seems to work fairly well, or at least fairly consistently. I've gotten results from Rocksim, SpaceCAD and Scott Ghiz's calculator that were all but indentical. But, as Scott Ghiz himself says, there's a lot of other factors involved, they just tend to balance each other out. Fine and well, but I need more.

For instance (and it was part of the discussion which prompted this) length matters. The more surface, the more turbulence, the more drag. How does this affect that number we're told is 0.75, the CD (coefficient of drag)? And I guess that what this comes to: what and the major factors that come into play for CD? What can we measure, or at least estimate, to convince ourselves that 0.75 CD actually applies?

 

[illini]

Drag = 1/2 * air-density * velocity-squared * Cd * area

where the Cd is referenced to the area...usually frontal (cross-sectional) area.

Cd has a number of variables. Shape factors into it. Turbulence factors into it. Turbulence over a flat plate generally transitions at Reynolds numbers around 30,000 to 100,000 (see Schlichting - Boundary Layer Theory, 7th edition, page 453). Separation factors into it. Basically, Cd=0.75 is a very gross estimate. So many other factors. So design dependent. Which factors dominate depend greatly on your design.

Drag of flow over a sphere or cylinder vs. Reynolds number is actually pretty illustrative of the issues. Most aero undergrads sooner or later do the sphere in a windtunnel experiment. Here's a good page from Princeton that I found that gives a good explanation:

https://www.princeton.edu/~asmits/Bicycle_web/blunt.html

 

[Fore Check]

When I flight simulations on RockSim, I select the choice for "Calculate Cd at simulation time." If you do this, run a sim, and then re-run the sim with the constant Cd of .75, you get two quite different results. I have found after doing the sims and actual launches that the calculated Cd is more accurate (or, I should say, the predicted altitude from that sim is more accurate.)

I haven't messed with it enough to know how RockSim calculates that Cd, or what the final Cd result of the calculation is.

 

[illini]

RockSim has a Cd calculator under one of the menus that allows you to see what it thinks the computed Cd of your design should be vs. velocity. I don't know everything they do under the hood with RockSim, but my guess is that they compute a nominal Cd using "rules-of-thumb" based on shape, etc. Then they have some equations to adjust that for "in-flight" conditions. How good the RockSim Cd numbers will be for any design will vary. I recall my senior design project where we had to use Air Force DATCOM to supply estimated values for lift and drag coefficients. Basically, DATCOM was an enormous database of historical data. So the theory was that you could find a design or configuration similar to yours and use the numbers for that. As long as your design didn't deviate too much from something in the database, it could give you reliable numbers. I expect that RockSim is doing something similar. Would be nearly impossible to do more than that.

What could be done to get a more reliable Cd estimate? CFD calculations? To some degree these might help you with pressure drag, but not as much for friction (viscous CFD calculations are much more complex, and once we go viscous then we have to start thinking about turbulence. Turbulence modeling in CFD is pretty dicey). At the end of the day, your best Cd calculator is a wind tunnel or flight test.

 

[Ryan S.]

I am suprised base drag does not come into account as well. I know the drag caused by airflow over the airframe does come into effect, however base drag is certianly a larger factor in most rockets (not so much in minimum diameter)

Also, this "flate plate" simulation you describe also does not seem to take into account nosecone shape, which is also another large factor in predicting altitude

 

[illini]

Base drag is a form of pressure drag. Take a look at this: https://aerodyn.org/Drag/vxdrag.html. And yes, you're right, base drag certainly is a big contributor.

When DynaSoar is referring to a flat plate, what he's saying is that the drag calculation (the formula I gave you) basically reduces the entire rocket to a simple flat plate (the area in the equation). All other contributors to drag - pressure and friction - are accounted for in Cd...this includes the nose cone shape, the base drag, launch lug. Everything gets lumped into that lowly coefficient...the Cd.

 

[GL-P]

I find that the number 0.75 is a little overkill. All the drag programs I've used state lower numbers (Aerolab, AeroDRAG) Can't validate the drag coeffs. right now. Got a rocket that could be a good test bed for rocket simulation software accuracy. It seems to make a huge difference.

 

[illini]

Originally posted by GL-P
I find that the number 0.75 is a little overkill. All the drag programs I've used state lower numbers (Aerolab, AeroDRAG) Can't validate the drag coeffs. right now. Got a rocket that could be a good test bed for rocket simulation software accuracy. It seems to make a huge difference.

Just to throw another little wrench in the works (that's right, we're gonna kick it up a notch...BAM!!!), many of the components of Cd are dependent on velocity. So just when you think you've figured out the "overall" Cd for your design by flight testing and backing it out, another motor will invalidate that answer due to the different flight profile it provides. The real answer is that Cd is continuously changing throughout flight. RockSim tries to model this, but its still an approximation for all the reasons above. What would be interesting (maybe as a NARAM R&D project for somebody?) would be to take 3-4 different designs and collect wind tunnel data on them at various speeds to determine what the "actual" Cd is for each vs. velocity. Then, compare those with RockSim's predictions. My guess? RockSim probably does reasonably well on somewhat conventional designs, but probably misses the mark by more than a little on the less conventional ones.

 

[teflonrocketry1]

AeroCFD software https://www.aerorocket.com/aerocfd.html is based on wind tunnel testing and has been validated. If you want to plug in an accurate number for Cd into RockSim use the AeroCFD values!

John Cipolla the owner of AeroRocket has done a study on spool rockets that suggests what the actual drag coefficient for a flat plate might be. The base drag is one factor that contributes to the overall altitude attained for a given flight. Base vortex on the other hands affects stability (the center of pressure) in a dynamic manner, that's why spool rockets are stable. The base vortex puts the CP behind the aft plate of the spool.

Most of the simulation programs assume that the body tube of a rocket has no effect on its stability, this is true only at low angles of attack for an ideal rocket with a length to diameter ratio (L/D) of about 10:1. If the L/D is much less than or much greater than 10:1 these assumptions don't hold true.

The body tubes affect on flight are related to its lift (which is analogous to boayancy), its drag from the frontal surface area and its base vortex. With wider shorter body tubes, base drag and base vortex are more important. With long narrow body tubes, body lift becomes a major factor as soon as the angle of attack deviates from zero.

Bruce S. Levison, NAR #69055

 

[DynaSoar]

Originally posted by teflonrocketry1
AeroCFD software https://www.aerorocket.com/aerocfd.html is based on wind tunnel testing and has been validated. If you want to plug in an accurate number for Cd into RockSim use the AeroCFD values!

Or for SpaceCAD or Scott Ghiz's calculator, or or...

THAT'S what I'm talking about. It's got most of the factors one should ever need to consider. It's easy enough for anyone to use. And figuring out what all those factors are and how they act and interact is an education is itself.

 

[illini]

Originally posted by teflonrocketry1
AeroCFD software https://www.aerorocket.com/aerocfd.html is based on wind tunnel testing and has been validated. If you want to plug in an accurate number for Cd into RockSim use the AeroCFD values!

AeroCFD is a good step forward, but - like anything - its still limited by its assumptions. It uses a panel method with potential flow equations, which are significantly reduced from the full Navier-Stokes equations. Not bad where it applies, but need to understand the limitations. Won't do friction drag...no boundary layers, no turbulence (large shed vortices as a component of pressure drag being an exception). Again, a good step forward, but must recognize the limitations. I'll have to look into it more to understand the extent of their validation. Is there a link to a validation report somewhere? Still think it would be interesting to do independent wind tunnel testing with a variety of designs to compare wind tunnel, RockSim, AeroCFD, and whatever else is out there.

 

[EMRR]

This reminded me of a program from way back. And I found the demo. Attached.

(SEE NEXT POST)

Nick

 

[EMRR]

Here's the screen shot.

 

[rocket72175]

Originally posted by illini
RockSim has a Cd calculator under one of the menus that allows you to see what it thinks the computed Cd of your design should be vs. velocity. I don't know everything they do under the hood with RockSim, but my guess is that they compute a nominal Cd using "rules-of-thumb" based on shape, etc. Then they have some equations to adjust that for "in-flight" conditions. How good the RockSim Cd numbers will be for any design will vary.

To the best of my knowledge, RockSim's Cd estimator is based on Estes TR-11 and Hoerner. These are cited as references in the user manual.

AeroDrag from AeroRocket uses the same data with extensions, like higher Mach number. Some details are on the web site.

Ken

 

[rocket72175]

Originally posted by illini
AeroCFD is a good step forward, but - like anything - its still limited by its assumptions. It uses a panel method with potential flow equations, which are significantly reduced from the full Navier-Stokes equations. Not bad where it applies, but need to understand the limitations. Won't do friction drag...no boundary layers, no turbulence (large shed vortices as a component of pressure drag being an exception).

All these assumptions make the software viable on a home computer and available to the masses. Not bad for goofing around with 3FNC designs.

I do automotive CFD for a living, and these assumptions don't fly, of course! Automobile aerodynamics (bluff body) remains a difficult simulation, more difficult than a lot of aerospace (streamlined) problems. On average, we correlate to 5% in Cd when comparing to wind tunnel measurements (which have their own problems, too!)

Ken

 

[illini]

Cool! I used to do CFD of internal flowfields of rocket motors for a living - mostly developing proprietary CFD codes, but occasionally playing with commercial packages as well. Good to have another CFD guy in here! I agree AeroCFD looks pretty good for a "home CFD package" (never thought I'd utter those words ). I issue the caution regarding the assumptions mostly because there is a tendency - not just among the "masses", but also among the professionals - to believe that anything that carries the CFD label must be right. Even more generally, there is a tendency to attribute more accuracy to any kind of simulation than is warranted. I say that as someone who has spent much of the last 20 years making my living developing simulations (amongst other things). If *I* don't believe in their infallibility, why should anyone else?

But you are correct. Packages like AeroCFD are pretty cool for home use.

 

[rocket72175]

Funny, it is just the opposite in my business! All simulations are met with skepticism, while any kind of "measurement" or "test" is treated as gospel.

Most engineering development is a "model" of some sort, since we can't field test everything. Wind tunnels and CFD are both simulations. Both are tools in the design process.

Ken

 

[illini]

You know...been so long since I actually did CFD that I almost forgot, but you're right! Back then the sims were viewed with skepticism while the test data was gospel (and you're right about that being an approximation as well). I remember solving one problem with CFD where I had the answer down cold. No reasonable person could reject the results, and all the tests we did failed to shed any light on the problem. CFD nailed it. Testing did not. My results were uniformly rejected. Of course, my results also put the company in a bad light, so that might have had something to do with it...

In the business I'm in now, the sims are often accepted a little too readily. I think the reason is that the types of environments I model now are difficult to collect significant test data on without starting a war. So the sims are often the *only* way to get to the answer. And, the customers of the sim results are often not sufficiently technically savvy to understand either the capabilities or limitations of simulations. They tend to buy off on the results too quickly, especially if they support their position.

 

[teflonrocketry1]

If you brose through the AeroCFD web site you will find that John Cipolla has taken the time to validate his programs against data from real rockets and present the issues/compairisons on his web site. https://www.aerorocket.com/index.html You can obtain even more sophisticated CFD programs like AeroEuler and VisualCFD that are based on different underlying assumptions. Alternatively, you can even hire John to run wind tunnel tests on your designs to help refine simulations.

I find the accuracy of my model rocket simulations is related to my ability to make measurements of the all variables. Since motors themselves are manufactured to a +/- 20 tolerance, you might think that a given simulation should reflect this error. While this is true in most cases, careful selection of motors form the same lot, can yield simulation data that match altitudes and range data very closely and well within the 20% range. In addition, I have yet to find anyone else that has an angle finder or some other mechanism built into their launch pad to determine launch guide angle. It seems to me most rocketeers ignore this important variable for a given flight and then complain that their simulations are way off!

Bruce S. Levison, NAR #69055

 

[DynaSoar]

Originally posted by rocket72175
Funny, it is just the opposite in my business! All simulations are met with skepticism, while any kind of "measurement" or "test" is treated as gospel.

Most engineering development is a "model" of some sort, since we can't field test everything. Wind tunnels and CFD are both simulations. Both are tools in the design process.

Ken

Same in my field. If you couldn't stick a wire in it and measure the signal, you were just guessing. Then along came complex simulation software and neural network models. Now, a neuroscientist that won't at least consider modelling is old fashioned and headed for retirement.

 

[rocket72175]

Originally posted by illini
Just to throw another little wrench in the works (that's right, we're gonna kick it up a notch...BAM!!!), many of the components of Cd are dependent on velocity. So just when you think you've figured out the "overall" Cd for your design by flight testing and backing it out, another motor will invalidate that answer due to the different flight profile it provides. The real answer is that Cd is continuously changing throughout flight.

Illini, I put your theory to the test (albeit a limited test). Using a 3FNC design in RockSim, I used a few sufficiently different motors. I ran 1D sims with Cd vs. velocity (estimated by RSim) and fixed Cd=0.75. Let's assume the flight test is identical to the Cd vs. velocity simulation. I then backtracked an "overall" fixed Cd to match the flight observation. Here are the altitude (ft) and backtrack results:

Motor, alt@Fixed Cd, alt@Calc Cd , backtracked Cd
------------------------------------------------------------------
G69, 902, 963, 0.539
H48, 2105, 2406, 0.513
I69, 3234, 3745, 0.539
I360, 4667, 5613, 0.548
J400, 4829, 5826, 0.546

While the varying thrust makes an impact, the backtracked Cds are not grossly different, with the max range of 35 counts (7%). This is analogous to many cases where using an average thrust over burn time gives the same altitude simulation as following the thrust profile exactly.

I think most hobbyists would be satisfied with Cd backtracking using different motors. Adding wind to the simulations probably presents greater issues than the thrust profiles. What do you think?

You are correct. This stuff would make for interesting R&D! Too bad RockSim doesn't allow for user-defined Cd curves.

Ken

 

[illini]

Cool experiment! You're right, 7% isn't much at all and well within the tolerance of other factors...some of which teflon pointed out (like motor variances). Also interesting that the lowest Cd was with the motor with the lowest average thrust. It really would be a neat R&D project. For a range of designs, from 3FNC to the more exotic, determine Cd vs. velocity using wind tunnel testing, RockSim's native Cd calculator, AeroCFD, more advanced CFD methods, "backed out" from flight test, and compare them both in terms of their raw values and in terms of their impact on altitude predictions. Armed with this information, we can begin to understand more fully the impact of approximating Cd one way versus another. It may very well be that the conclusion is that other factors dominate. Not only would it be a good R&D project, it would probably be enough to get someone a Master's thesis (one of you guys still in college take note!).

Teflon posted the link to the AeroCFD validation information and some other CFD codes. I'm going to look those over, then probably start a separate CFD thread to discuss the differences between different CFD techniques (after all, this thread was only about Cd ).