Altimeter vs. RockSim

[jcsalem]

I had my first altimeter launch Sunday at CMASS Tewksbury. It was VERY windy on a small field so I flew the HeartRoc on just an F50T-4 with a PerfectFlite MAWD in the payload. It was a great engine for a low flight -- nice and loud plus it no trouble getting the rocket off the pad.

I found that the altimeter measured altitude differed by about 20-30% from the RockSim predicted altitude. I accurately entered the Mass and CG overrides to RockSim. PerfectFlite measured 721ft. vs. RockSim's 850-950ft. (depending on the launch conditions I entered).

Is this typical or is there any way to make RockSim more accurate? Also, RockSim predicted that the F50T-6 would be the best match. I'm glad I stuck with my gut and the 4 second delay -- 6 seconds would've been WAY past apogee.

It was a very windy day, 10-15mph, so I wonder if that had something to do with it.

-- Jim

 

[edwardw]

Did you take into account the drag on your launch lug or rail buttons? I find many times that that's the cause of the lugs. Also, I'd start looking at adjusting the Cd of the rocket until you get something that's close. Though a non or near non windy day would be best for 'calibrating' the sim.

Edward

 

[joepolicy]

Did you enter the launch conditions in RocSim to include wind speed, variability, launch guide length and angle, motor overhang, etc.? Did the weight of your rocket match the weight in RocSim? I have found RocSim to be fairly accurate. A recent launch of my PML Ariel simmed to 902 feet, altimeter reported 884 - that's pretty close. When I sim with high winds, it knocked a good 120 feet off the simulated altitude and at least 2 seconds off the delay time. There are many variables in model rocketry that are covered in RocSim - I trust it for all of my launches.

 

[jcsalem]

I did enter the exact weight and measured CG in RockSim. I also set the launch conditions to make the actual conditions at the field. That still resulted in the 20-30% mismatch.

Haven't tried adding lugs to the sim so perhaps that could be it. It will also be interesting to compare the performance on a day when there is less wind.

BTW, I'm using the latest version of RockSim.

-- Jim

 

[Stymye]

lugs make a considerable difference

also rocket motor thrust can vary +/- 10%

also the Rocsim suggested delay can be missleading.. If you start with too short of a dely in the simulation, rocksim see's this as drag prior to apogee and will recomend a longer delay ...always start with a long delay to get the correct delay displayed
I'm not sure if that was the reason in your case , but It does make a difference.

 

[cls]

probably the biggest factor is the rocksim assumed CD vs the actual. it's a hard problem to solve (requires a team of PhDs).

overall I think you should plan to make 3 or 6 flights to tune your simulation with actual, measured results.


what was the flight profile? in that wind, at apogee did your rocket weathercock and "hover" a bit? that might be worth 50 feet or more.

what is the finish on your rocket? does it match the material finish of parts in the sim? in other words, if your model has a rough paint job and rocksim thinks it is baby's bottom smooth - that will count for 5% or more.


depending on the model of altimeter, it may be off by +/- 100 feet!! just because it reads to the foot doesn't mean it is accurate to the foot. I believe the barometrics in Perfectflites are within 50 feet.

to be really sure about the altitude, get two people with trackers to watch the flight.

there is some motor-to-motor variability, something on the order of 5% I believe. for altitude trials, some people weigh their motors and launch with the heaviest, on the assumption the extra weight is propellant. I am not sure about that but if you want to get really accurate then weigh the motors and account for that in the sim. pick 3 of 10 motors with the closest weights and use those for your trials.

 

[Stymye]

motor thrust can vary +/- 10%

 

[Chilly]

This is an interesting thread to me for a different reason:

Rocksim lost most of my saved files last night! Some of them had quite a bit of tweaking, like my rebuilt VB Javelin and seriously modified Estes Saturn V, both of which will have their first flights this weekend.

Any suggestions?

 

[edwardw]

Ouch! That isn't good. Try looking in the directory in the command prompt mode, or using a free recovery program. I've found those really useful when I have a corrupt memory stick and want the photos.

Edward

 

[Missileman]

Another factor is the altimeter.
What size and how many vent holes?
Where are the vent holes?
You said it was pretty windy, perhaps to some degree the wind blowing into the vent hole kept the internal pressure slightly higher?

 

[edwardw]

I didn't think of the vent holes, but I use three placed 120 degrees apart. Also, is there anything that would cause a disturbance in front of the ports? I know this can make it read differently.

Edward

 

[Missileman]

Mainly the nosecone if your vents are too far forward or any protruberances.
I forget off hand the recommended distance aft for the vents but it is quite a bit.

 

[teflonrocketry1]

You can get RockSim to be quite accurate if you want to spend the time making accurate measurements of all the variables. What version of RockSim did you use? Did you use the Rocksim equations in the simulation? What simulation method did you use, Explicit Euler or 4th Order Runge-Kuta? How many data points per second did you run the simulation at? How did you measure the lauch rod angle? How did you determine the altitude and latitude of the launch site? What did yo use to measure the air pressure, temperture and humidity at the launch site. Was there a lot of variability in the wind that day, how did you account for it? How acurately did you measure your Rocket's dimensions? Did you use a program like Areo CFD to determine the Cd of your model?

Not taking into account the launch guides is a big source of simulation error. To get the motor weight right, find one that has the same weight as the one used to create the motor file RockSim uses in the simulation! If motors have a +/- 20% manufacturing tolerance, why not use a balance to pick the right motor to at least match the weight in the engine profile!

Bruce S. Levison, NAR #69055

 

[teflonrocketry1]

Originally posted by Chilly
This is an interesting thread to me for a different reason:

Rocksim lost most of my saved files last night! Some of them had quite a bit of tweaking, like my rebuilt VB Javelin and seriously modified Estes Saturn V, both of which will have their first flights this weekend.

Any suggestions?

Do a search on your hard drives for files named *.rkt from the "Start" menu.

Bruce S. Levison, NAR #69055

 

[North Star]

Originally posted by jcsalem

I found that the altimeter measured altitude differed by about 20-30% from the RockSim predicted altitude. I accurately entered the Mass and CG overrides to RockSim. PerfectFlite measured 721ft. vs. RockSim's 850-950ft. (depending on the launch conditions I entered).

-- Jim

I have used RockSim since it came on floppies and whilst I love it as a construction and stabilty tool I have always found the altitude sims to be very 'optimistic'

I find that, even with basic inputs, Alticalc and even WRasp give results similar to your altimeter result. I've no idea why and the other posts give lots of useful suggestions. However none would make a difference to the other sim methods as that level of input isn't possible. I just use a good spring balance to get an accurate finished weight. I also use Rocflite and find all the results (except RockSim) very close.

 

[jcsalem]

Thanks for all the great responses!

A few more specifics:
- Like Edwardw I used 3 vent holes placed 120 degrees apart. I also placed them about 8" back from the nose cone which is about 3 times the 2.6" dia.
- I'm using RockSim v7
- I tried changing a number of the different variables (including various wind speeds, launch rod angle, launch site latitude, etc.). All of these simulations came out at least 10% higher than the altimeter measured.

I'll certainly be trying more flights to learn more about comparing the RockSim results with the measured ones.

-- Jim

 

[rocket72175]

Originally posted by North Star


I find that, even with basic inputs, Alticalc and even WRasp give results similar to your altimeter result. I've no idea why and the other posts give lots of useful suggestions. However none would make a difference to the other sim methods as that level of input isn't possible. I just use a good spring balance to get an accurate finished weight. I also use Rocflite and find all the results (except RockSim) very close.

I don't know about Alticalc and Rocflite, but WRasp and RockSim essentially use the same fundamental algorithms. As long as the mass and Cd are fixed to the same constant values, the results are nearly identical for 1D flight. I personally have correlated RockSim with my own code to several significant digits. Here is another comparison:

https://fly.to/mrhq

Other, lesser known, differences between softwares include the atmospheric model and number of digits carried in the thrust profile and calculations. These can amount to a few percent at the end of the simulation. The choice of integration scheme (Euler, RK, Adams, etc) makes little difference as long as the time step is sufficiently small. I always use 0.001 sec, but 0.01 is often good enough.

Ken

 

[teflonrocketry1]

Lets see: Tweksbury, MA, September 19, 2004 (I am guessing about 1:00 p.m.) weather.com gives a temperature around 43 degrees F, humidity around 55%, barometric pressure 30.16 mmHg, Winds 8-14mph (slightly breezy), with some variability, mostly cloudy (low strength thermals I guess). Terra server at https://terraserver.homeadvisor.msn.com gives about 42.600 degrees Latitude and an altitude 50 feet above sea level. I guessed at a 48 inch long launch guide angled 3 degrees into the wind (-3).

I tried my best to duplicate your design in RockSim and my first attempt using the RockSim equations gives an altitude of 723.6 feet AGL on a F50T-4! Note I included the rail buttons on the attached RockSim version 7 file.

I am not trying to flame you here, I just want to show that you need to pay attention to the details. If you post or PM me your RockSim file, I will gladly add or help you fill in these extra details that should show your altimeter and the RockSim simulations can agree very closely.

Bruce S. Levison, NAR #69055

 

[rocket72175]

Well done, Bruce.

 

[Stymye]

teflon,within 2 ' can't beat that!

 

[cls]

as Bruce pointed out, things like barometric pressure altitude and your geometric latitude really make a difference. I always try to get those right.


to follow up on my post above, the manual for my Perfectflite Alt15k claims +/- 4 feet for "resolution". that is probably the "precision" of the baro chip and A/D converter, but it is not the "accuracy".

unsurprisingly, accuracy is not rated in the manual. accuracy would be difficult to measure and is a multidimensional problem: absolute delta, skew, repeatability, changes with temperature, battery, orientation, port holes, speed, boundary layer, mach effects, resonance of the cavity,

I think the barometric altimeters are probably better than +/- 50' but not much better. certainly accellerometer altimeters are not that good. nor could you do better with GPS on Z axis.


bottom line (or apogee line!): make a few theodolites and have the next launch be a "measured altitude" event. that would be fun for everyone! carry a portable wx station out there and use that to help with the sim. oh man, rocketry leads to all kinds of other interesting hobbies.

 

[jcsalem]

Wow! Bruce you continue to amaze! The design is remarkably close to the actual rocket (the weight, length, and fin span were all within 10% of the actual rocket).

I've attached my RockSim file which is very accurate. The sizes are exact and I've used the mass override to enter the exact weight and CG on launch day. Using pretty much the same conditions as Bruce did, I now get an altitude of 846' which is still about 15% away from the altimeter measured.

I've tried changing other parameters (e.g., Euler vs. Runge-Kutta, various wind settings, etc.) as well as adding launch lugs but none of those changed the altitude more than 1 to 3'.

Maybe 15% is as good as I should expect without empirically measuring the Cd?

At some point, I will try to empirically measure the Cd and see if that remains constant for the rocket for a few different motors.

-- Jim

 

[Stymye]

I thought the cd changes during flight , a moving target if you will.
I haven't figured out how to over ride the cd in rocksim if it's even possible. it would be handy in some cases

 

[rocket72175]

Originally posted by stymye
I thought the cd changes during flight , a moving target if you will.
I haven't figured out how to over ride the cd in rocksim if it's even possible. it would be handy in some cases

Yes! This is a very important feature in RockSim. From the "rocket design attributes" tab, uncheck "calculate cd at simulation time" and enter a value(s) for constant Cd. The Cd is now fixed for the entire flight.

This is how you backtrack to "tune" your Cd from altimeter data using SMARTSim. https://www.apogeerockets.com/smartsim.asp

Ken

 

[bobkrech]

to follow up on my post above, the manual for my Perfectflite Alt15k claims +/- 4 feet for "resolution". that is probably the "precision" of the baro chip and A/D converter, but it is not the "accuracy".

Cliff

Resolution Resolution is strictly a matter of the ADC step size and the gain of presure sensor electronics. It has nothing to do with either precision or accuracy.

Precision is the measure of the repeatability to give the same value for a given pressure regardless of whether the value is correct or not.

Accuracy is the measure of how far off the reported value is from the true value.

Both precision and accuracy are easy to measure if you have a calibrated pressure chamber.

The pressure sensor is the weak link in the system. If you go to the manufacturer's data sheet of the pressure sensor, you will find the values they specify for both of these values.

Another source of inccuracy could be the algorithm that does the pressure value calculation from the electrical signal although this can be negligible with a good algorithm.

Bruce

The temperature was about 65F on Saturday at Tewksbury.

Bob Krech