Monte Carlo simulation.

Steve - that link is where you go if you click on the "records" tab at the top of the TRA site....It's been "updated."

OverTheTop said:

Been there, done that, but not that many times.

For reasonable stats you need a minimum of about 30 data points. Confidence with a sample of 5 is a bit scratchy.

Click to expand...

It is a trivial matter to vary the integer variable "noi" in the appropriate script subroutine. 5 iterations is just the test case. These images represent a run
of 31 iterations of the same NOAA data as the previous test rocket.

I work with some people that are quite pedantic about stats, so I just thought I'd mention it.

Sometimes in conversations I trot out the comment that "half the people are below average" just to bait them and get "half the people are below median" back

ecarson said:

The main warnings I got this time, were that the differential equation solvers were not reaching total convergence after 10,000 iterations each test.

Click to expand...

I don't know how Cambridge is choosing to solve the equations, but this sounds odd. The rocket trajectory ODEs are very straightforward to solve numerically. Even with time marching explicit schemes (Runge Kutta), and big time steps, the solutions are very stable. I have never seen RS, OR, or RA complain of convergence.

Buckeye said:

I don't know how Cambridge is choosing to solve the equations, but this sounds odd. The rocket trajectory ODEs are very straightforward to solve numerically. Even with time marching explicit schemes (Runge Kutta), and big time steps, the solutions are very stable. I have never seen RS, OR, or RA complain of convergence.

Click to expand...

I did not get these warnings on the original Monte Carlo altitude example file settings. When I put in the actual NOAA data at the various altitudes, there was differences in bearing, velocity, and temperature not evident in the example file. The Octave odepkg diff.eq. solver is probably the culprit here. I suspect the transitions between the various altitude levels was the factor in these warning messages.

In using SU2 CFD on my sim aircraft, I often got failure to converge problems at high angles of attack using supercritical wing sections. As long as I kept angle-of-attack within reason (-2 to +10 degrees) they would converge after 200 to 300 iterations. So, I am left to hypothesize that the sharp transitions between the reported altitude levels may be at fault. Right now, I am looking at the Cambridge scripts that call the Octave odepkg in an effort to see what is causing this. The resultant output though does not appear to be in gross error.

If I could get the original Cambridge Simulator FlightPlotter to work properly, I would not have to resort to Octave at all, as that is a standalone program. It worked much faster than going through the Octave script calls for sure. I have not tried the Linux version of that though, and that is another avenue I'm going to explore in the coming days.

In the meantime, I'm working on about 2 dozen other projects both related and unrelated to the rocket. Not enough time to do it all, but that is a blessing in retirement as I had zero time while doing the "indentured servant" for industry gig the last 30 years.

OverTheTop said:

I work with some people that are quite pedantic about stats, so I just thought I'd mention it.

Sometimes in conversations I trot out the comment that "half the people are below average" just to bait them and get "half the people are below median" back

Click to expand...

Think of how stupid the average person is, and realize half of them are stupider than that. --George Carlin

OverTheTop said:

I work with some people that are quite pedantic about stats, so I just thought I'd mention it.

Sometimes in conversations I trot out the comment that "half the people are below average" just to bait them and get "half the people are below median" back

Click to expand...

It's only half below median if there's an even or infinite population size and none are exactly the median.

ecarson said:

I did not get these warnings on the original Monte Carlo altitude example file settings. When I put in the actual NOAA data at the various altitudes, there was differences in bearing, velocity, and temperature not evident in the example file. The Octave odepkg diff.eq. solver is probably the culprit here. I suspect the transitions between the various altitude levels was the factor in these warning messages.

Click to expand...

Ah, yeah, the NOAA data is probably tripping it up.

ecarson said:

In using SU2 CFD on my sim aircraft, I often got failure to converge problems at high angles of attack using supercritical wing sections. As long as I kept angle-of-attack within reason (-2 to +10 degrees) they would converge after 200 to 300 iterations.

Click to expand...

For sure. The PDEs of CFD solvers can have big time convergence issues.