Simulation differences

[astrowolf67]

I've been sitting here running some sims on my PML Endeavor, in both Wrasp, and the demo version of RocSim. Using the same weight in both programs, and launching with a J350, RocSim shows an altitude of around 3800 feet, while Wrasp shows around 3100 feet. Why such a difference, and which one is more accurate??

 

[hokkyokusei]

Are you using the same cd in both programs?

 

[astrowolf67]

No, I wasn't. But, I changed it in Wrasp, to match RocSim's, and the Wrasp altitude dropped to 2750 feet. That makes it over 1000 of feet difference. That's quite a bit!!

 

[Johnnie]

David,

Todd Jurhs of HARA, who certified L3 this past weekend sim'ed his rocket over and over again with the purchased version of rocsim. It simmed every time to 7,200 ft. His acyual flight was to 5,295 ft.

I believe that Rocsim err's on the side of caution. I have not used wrasp in a while, I may have to dust my copy off.

3,000 ft sounds pretty good for the Endeavor on a J350.

 

[teflonrocketry1]

What equations were you using in the simulation, RockSim or Barrowman and what method were yo using Explicit Euler (RASP Style) or 4th Order Runge-Kuta? If you take the time to make accurate measurements and you use the appropriate launch rod length and angle. You can actualy calcualte the flight altitude to within a few percent. Your altimeter should also be calibrated to and appropriately vented to the atmosphere. For accuracy I prefer RockSim eqations using the Explicit Euler method and a simulation resolution of 1000 samples per second. I have bracketed consecutive flights around a RockSim calculated altitude value with an Adept A1 altimeter in a vented payload bay. I believe that most people won't spend the time I do making accurate measurements for a good simulation. How many times have you weighed a rocket motor, or measured the actual angle of a launch rod, or the length of a lauch rod from the lowest point on the launch lugs to the rod tip? If you want a simulation close to reality you have to spend the time doing things like this; it all adds up and makes a big difference.

Bruce S. Levison, NAR #69055

 

[cls]

let me second what Bruce says - accurately weigh your rocket and all its components, set the altitude and atmospheric conditions, etc.... a percent here a couple percent there ... it all adds up.

 

[Micromeister]

OK I'll start the furor again.
Roc-Sim simulations.. NO matter which way they are run or equations uses are alway WAY off. In tests by as much as 10 percent. Accuracy in input is important but roc-sim programing makes assumptions that cause the errors.

Bruce I have to say accuracy is important but come ON it's a simulation program.. I haven't had to measure the Actual rod length between the last lug and tip, or have an exact LOwt to a tenth of a grams, or measure the rod angle to the degree to get Wrasp to give simulation data that can by varified by tracking and altimiters within the couple percentage points you've reported must be done to get Roc-sim to do the same.
Good weights, Yes, approximate rod length and angle, sure. accuracy of sim within 3-5% should be required of the manufacturer, Not requiring the user to field test every detail. Good science? maybe not but Tim's gotta remember this is a hobby! Seldom do I carry a protractor with me to the range.

Stick with Wrasp or deduct 10percent from your Roc-sim guestimates to be in real world 'realistic" simulations.
This has been one of my main problems with Roc-sim since it's introduction.... the simulations are petty much useless. Tim knows it. I guess his programmers can't or won't fix it.

I'm NOT saying don't use Roc-sim.. Just don't believe everything it gives you. Verify, Check re-check.... its only a tool! not THE answer.

 

[Stymye]

This has been one of my main problems with Roc-sim since it's introduction.... the simulations are petty much useless. Tim knows it. I guess his programmers can't or won't fix it

Gee Micro,,I wouldn't go that far ,,, I've found the results to be pretty close with the proper input...once the model is constructed you can go back into rocsim if necessary and make adjustments to fit the actuall weights and cg/cd.. ect.. to match the model more closely...I don't rely totally on rocsim..but I've found it to be consistently within the 10% range for me
Obviously a simulation is only as good as the input

 

[illini]

I said it before in a previous thread a few months ago, and I'll say it again here:

What questions do you want the simulation to answer? The validity of the simulation is directly tied to the questions. Just because the simulation *can* produce a number doesn't mean that number is worth anything.

One last comment: I'm not about to change the way I fly just so that my flights can come close to Rocsim's predictions. Its the other way around. If I care about such things, I want the simulation to be able to capture the way I fly. Don't forget, the *flying* is the real world, not the sim.

Not gonna beat a horse that was killed dead in a previous thread. I'll leave it at that.

 

[Micromeister]

Originally posted by stymye
Gee Micro,,I wouldn't go that far ,,, I've found the results to be pretty close with the proper input...once the model is constructed you can go back into rocsim if necessary and make adjustments to fit the actuall weights and cg/cd.. ect.. to match the model more closely...I don't rely totally on rocsim..but I've found it to be consistently within the 10% range for me
Obviously a simulation is only as good as the input

There we go taking things out of context! your experience MAY Very, MY peeve is WE shouldn't have to pay HUGE bucks for software that DOES NOT do a decent job of what it was written for. Wrasp is FREE and does a much better job. NO! I'm sorry.. I just can't condone the fleecing of the Mod-Roc community without letting the perpetrators know it!
That R&D work well i'm sure at some point be completed,
showing which Sim's program(s) and which altimeters perform best, reporting Near real world Tracked altutides. To date the evidence on a level playing field show Roc-sim off by AS MUCH AS 10 percent on repeated flights. They we not all 10% some more some less but pretty consistant. its not my project so I can't really say much more about it. Except to say the finding Have confirmed what I've been saying about this sim's program since it's introduction. I use Roc-Sim for some design, some stabiltiy and Some Sim work. I simply Adjust and Verify the results with other data! It just makes me sick seeing folks quoting "Roc-sim" like the Bible.. It's not, It's flawed, It's got problems like every other program.. overlooking them or attributing it to "pilot error" just Ticks me off.

Good point Illini: This horse has been dead for awhile! nuf said

 

[astrowolf67]

The main reasons I'm running the sims, is, I plan to do my L2 either late this year, or early next year. I'm caught between dual deployment, or motor deployment at apogee. Since I don't yet have an altimeter, I try to keep my highest flights at 2500 feet maximum. I also know that the altitude will drop if I go with dual deploy, with the added weight. I've always used wrasp, cause it's quick and easy, but on my L1 flight, wrasp seems to have totally missed the altitude, and the demo of RocSim was more like what it appeared. But, with the Endeavor, I'm getting reverse results. Who knows, I'll just fly it, and keep my fingers crossed.

 

[teflonrocketry1]

Micromister,

I respectfully disagree with you on these issues about RockSim software. I am not a programmer, just a major (ab)user of the software. Paul Fossey, the actual RockSim programmer, and Tim Van Milligan at Apogee have ben kind enough to listen to me and publish some articles I wrote about extending the use of the RockSim software to more than just basic rocket designs.

I am not sure how to convince anyone that RockSim is accurate and even more so than wRASP. Perhaps some real data will help; this data is from flights made at a launch that was conducted on Sept. 29, 2002 at the Summit County Fairgrounds in Talmage, Ohio.

Rocket: Estes Hijax
This model was measured with a Vernier caliper, the values used in the simulation were an average of three measurements.

Weight fully preped with payload and motor 102.4g

3 Motors C6-5, same lot, all weights matched to within 0.050g of stated weight in the motor file. All flight weights to within same tolerance and CG adjusted to 12 and 25/32" from nose tip.

Altimeter Adept A1, three 1/8" vent holes in payload section
Temp 75F measured at site
Humidity 80% measured at site
Lat 41.12356 degrees
Elevation 1150feet ASL
Wind variable 5-10mph
Launch angle measured 0 degrees
Launch Rod length used 3 feet

RockSim version 5
highest predicted altitude (5mph wind) 722.45' AGL
lowest predicted altitude (10 mph wind 701.13' AGL

wRASP version 2.6
781.4' AGL

Actual Altimeter Readings

Flight 1, 725'AGL
Flight 2, 695'AGL
Flight 3, 715'AGL

Bruce S. Levison, NAR #69055

 

[bobkrech]

It's like comparing apples and oranges. They are 2 very different programs but both are pretty good at what they are designed to do.

wRASP is a simple altitude simulation program that only requires a rocket weight, diameter, estimated CD, and .eng files to generate an altitude calculation.

RockSim is a complete CAD package for rockets that includes an altitude calculation package. It's a whole lot more complex than wRASP, but it has to use the same equations of motion as wRASP to predict altitude.

The problem with all simulations are that they are only as good at the data you supply them. Garbage In will give Garbage Out for any simulation.

The major cause of uncertainty with wRASP is that you have to guesstimate the CD because there is no calculational element for it, and you have to have a measured weight and diameter for input into the model.

RockSim automatically calculates a whole bunch of parameters for the altitude calculation including the model weight and CD, but the calculated values are only as good as the data the user inputs into the CAD model. If you don't have the right parts, and weights and diameters, you will get a less accurate altitude calculation.

I'll estimate that 10-20% altitude errors can arise with either package if you put in the wrong information into the altitude models.

Furthermore, there is a real-life variability between motors. The testing standards allow for a standard deviation of up to +/- 6.67% impulse in any given batch of motors with worst case flyiers of +/-20%. Those percentages directly translate into the same percentage in altitude variation!

The only way you can be sure of your rocket performance is to fly it several times on a given motor and used that data to tweek the simulation. Even the big boys do this.

Finally, I really wish rocket community would stop complaning about the price of products. You have to spend real money to develop a reliable product and to market it. If you think a product's price is too high, don't buy it. If you can develop and market an equivalent product for less, do it. I guarantee that you'll find out quickly it ain't that easy, simple, or cheap.

Bob Krech

The way to make a small fortune in rocketry is to start out with a big fortune.

 

[Zack Lau]

Perhaps these programs need an "intelligent guru" mode.

You get to tell the program it is wrong--that you measured
700 feet instead of the 1000 feet predicted. The program
can then "get smart" and make the necessary corrections,
so it works better the next time.

A truly intelligent program should be able to determine the
stuff that you don't tell it. For instance, ordering lots of
automotive paint products and epoxy indicates that your
rockets will weigh more.

 

[rstaff3]

Yup, and Rsim should automatically download weather history from the launch site and run a predicitive model to determine the wind speed and direction. Of course, it would probably show 0 ft if it knows the launch will be rained out. This would piss off some users and they would want it to automatically determine the date and time of the next launch window and re run its simulation.

 

[teflonrocketry1]

I agree with Bob Krech that putting inaccurate information into any program will only give inaccurate information out. The error of any given measurement is carried throughout the entire simulation. I am impressed that someone else actually sees the difference between the two softwares; which is the coefficient of drag in wRASP is a "guessed" value where as RockSim attempts to calculate this parameter. More accurate coefficients of drag can be calculated using programs such as AreoCFD and these better values can be used in either program, I have yet to get that level of complexity.

The way to get around some of the variability in motor manufacturing might be what I did, where the motors were from the same lot and weighed within +/- 0.05 grams of each other and the actual RockSim motor file weight. I noticed that within a given motor lot the variability in weight can be as much as five grams for a C motor.

The values for the flights I gave for the Estes Hijax are what I typically get; these flights were conducted at a public launch with my NAR section and the data (motor and altimeter readings) were enetered into the flight log and should be verifiable.

Bruce S. Levison, NAR #69055

 

[rocket72175]

Originally posted by Zack Lau
Perhaps these programs need an "intelligent guru" mode.

You get to tell the program it is wrong--that you measured
700 feet instead of the 1000 feet predicted. The program
can then "get smart" and make the necessary corrections,
so it works better the next time.

As a matter of fact, there is such a software tool that does this:

https://www.apogeerockets.com/smartsim.asp

SMARTSim is geared to those of you who like to tinker with RockSim, compare measured and simulated data, and optimize designs.

Ken