Talk-Polywell.org

Here is some good video of the landing attempt

https://www.youtube.com/watch?v=rAzwuEmZcmE

Looks like it wasn't able to get position, velocity, rotation, and rotation rate to all hit nominal for touchdown at the same time. The vehicle was rotating sideways across vertical as it touched down. Some kind of lateral thruster set at the top would help a lot there, but at the expense of more weight.

It looked to me like similar dynamics to this event, also in the news yesterday:

http://www.usatoday.com/story/news/nati ... /25810827/

But with more flames. The bigger they are, the harder they fall. The tail end was swaying rather badly, probably under-anticipating moment of inertia of the craft. I suspect the controls group I worked with at Athena could help him get the system's control laws tuned up. And at this point, I'd say that's the issue, and it is correctable.

hanelyp wrote:Some kind of lateral thruster set at the top would help a lot there, but at the expense of more weight.

Look again, it's there! Poor little thing tried as hard as it could, bless its heart.

Maui wrote:

hanelyp wrote:Some kind of lateral thruster set at the top would help a lot there, but at the expense of more weight.

Look again, it's there! Poor little thing tried as hard as it could, bless its heart.

The key, I think, is making sure the control system can predict the situation well enough that it NEVER has to try as hard as it can. It must stay ahead of the situation well enough that small corrections keep it coming down correctly. If the excursions get to gimbal lock, trying as hard as it can equates to crashing.

The problem is not that the sway was more that the control system could deal with, it is that the sway developed at all. When they get this to work right, it will look easy.

Thinking about it some more, the following sequence seems to fit:
- a gust of wind blows the rocket off the desired path.
- to "correct", the rocket tilts into the wind. Appropriate if the top thrusters aren't there and the rocket has time to make the correction. But this pushes the rocket further out, and out of plumb.
- Now both well off center and tilted, the rocket makes a desperate attempt to center itself on the barge, right itself, and bring rotation to a stop when it reaches vertical, a combination that conflicts given the program above.

Sounds like good practice would be to build another Grasshopper take out on the ocean and bounce it back and forth between two barges a few times.

They could test it here in Wyoming too, we got wind. The tests in Texas didn't have these sorts of crosswinds. Landing Falcons will get there eventually, but might be faster and easier to use a Grasshopper.

Some analysis:

http://www.wired.com/2015/04/analysis-f ... =social_fb

He notes that the rocket was coming down about 86 mph. Based on my experience landing this sort of craft (which consists mostly of a lunar lander game on the HP-67 programmable calculator), I offer a little insight.

After crashing a bunch of landers either by hitting too hard or running out of fuel, my roommate found the solution, which worked every time and took all the fun out of it. You let the lander free-fall until the last move before impact, and then burned most of the fuel in the last second or so. Perfect landing every time.

The fuel margin on these landings is pretty tight. If you want the last thousand feet or so of the landing to be slow, that requires burning a lot of fuel over that last distance. Instead, they've undoubtedly decided to come down relatively fast, though considerably less than terminal velocity (my guess is that terminal would be around 200 mph at sea level density). The rocket is very light at landing so they probably can scrub off the last of the downward velocity over the last second or so, and it looks to me as if that is exactly what happened, although the tail of the rocket is obscured by smoke, flame, steam, etc. at touchdown so it is hard to be sure.

Tom Ligon wrote:You let the lander free-fall until the last move before impact, and then burned most of the fuel in the last second or so. Perfect landing every time.

Were they hitting a small target area? Seems to me adding the requirement to hit a barge or any small area means you'll need to do some active maneuvering early--where "active" means burning as opposed to merely using the fins they've added. I don't really know.

Skippy? Skippy would know.

GIThruster wrote:Some analysis:

http://www.wired.com/2015/04/analysis-f ... =social_fb

The video at the bottom of that shows the gimbel of the engine by flame direction a lot clearer that what I've seen before. Some serious lag, sticking, or insufficient derivative term trying to correct orientation. The engine isn't moving to damp rotation until after passing vertical. I recall mention of stiction in one of the tweets. Hoping they recovered the engine assembly in good enough shape for a direct postmortem.

hanelyp wrote:The video at the bottom of that shows the gimbel of the engine by flame direction a lot clearer that what I've seen before. Some serious lag, sticking, or insufficient derivative term trying to correct orientation. The engine isn't moving to damp rotation until after passing vertical.

Here's a video from the barge itself that illustrates your point very well:
https://vid.me/i6o5

EDIT: And here's a 600fps version of the chase plane video:
https://www.youtube.com/watch?v=3oJF8mmorbQ

... if nothing else, SpaceX's landing attempts sure do provide for some great entertainment.

There is one other factor that I do not believe they are accounting for, and that is the transition turbulence from water to a solid platform. This was a huge factor that the pilots had to deal with when doing the crossover maneuver for the landing on LHA. I suspect they have not enough experience with that part of the transition.
AV8B harrier is what I am talking about.

Given the apparent steering problem, I doubt a level and stable landing pad would have saved the rocket.

I am wondering about the "servo bandwidth" for the system that moves the rocket motors. With UAVs, the controls engineers were always greedy for more servo bandwidth, i.e. the ability of the servos to track a sine wave command for small deflections. They considered 5 Hz to be pretty slow. Granted, we usually were working on fairly small UAVs with a much lower moment of inertia about any given control axis than this rocket, but the principle will apply. If the steering apparatus is inherently slow compared to dynamics of the vehicle itself, then the phase lag problem is inevitable, and flight will be either sloppy or unstable. For the control system to function correctly, they must be able to swivel the motors a lot faster than the rocket tends to sway. The servos may be adequate at launch when the rocket is carrying a full load of fuel, but when it is nearly empty they may not be fast enough.

If they have to make the steering faster, that could require a fairly significant redesign of the controls rather than a simple tuneup.