Hittade en bra förklaring på en facebooksida om Spacex Hovering:
Citat:
Been meaning to write this essay for a while now: Why it's irrelevant if the F9 can hover as it lands.
Lots keep getting said about the F9's inability to hover, and that it would have a higher chance of success on the ASDS if only it could slow down, do a final translate for perfect centering, and descend gently onto the deck. Here is a list of what I believe to be good technical reasons why the "hoverslam" or "suicide burn" is actually the preferred tactic.
1) The faster the descent through the last 1000 feet, the less time the stage is exposed to potentially changing wind speed and direction close to the water compared to higher up. Unless perfectly modeled or factored into the algorithms, the landing solution is set into motion a lot higher up, at terminal velocity. All low level winds could do is add error to the solution.
2) Any time the stage is falling at terminal velocity, perhaps 250-300 MPH, is time it's not using propellant. For the same reason (gravity losses) that rockets take off with as high a thrust to weight ratio as possible, the optimum landing profile is to burn as late and as hard as possible. When a stage is in free fall at terminal velocity, it is NOT experiencing zero G. Its speed is constant, so it feels 1 G. Wind resistance keeps it from accelerating any more downwards, so burning propellant to slow it down too soon, means gravity starts fighting that slower speed to bring it back up to terminal velocity.
3) Intuitively, you should see that it's wasteful to light up the single landing engine at 10,000 feet, let's say, and (if you could) burn it at 5% thrust all the way down. Or 100,000 feet and burn it at 2% all the way down. They have plenty of data by now about how much the stage should decelerate given a commanded thrust level, and that algorithm can be constantly adjusted in real time because they have fantastic accelerometers on the stage. These are so sensitive that when the COPV Helium tank let loose on the failed CRS7 launch, the telemetry indicated the tiny slowdown in the rocket as the Helium tank accelerated FORWARD. If the commanded thrust results in a deceleration different than the target, the control loop is plenty fast enough to compensate.
4) You want the stage to come to zero vertical velocity (or less than some defined value related to the strength of the legs) when it reaches zero altitude over the deck. This is IDENTICAL to the challenge faced thousands of times a day by private, commercial, and military pilots when they land. Nobody except helicopters, Harriers, and private pilot noobs try to come to zero vertical velocity when they're not at zero altitude. I'm a private pilot, and sometimes I messed up my own landing solution, and my descent was aimed at a point short of the runway....so I had to add power, flatten out, and make sure I was over pavement before I descended the last 50 feet to the runway. This is generally regarded as suboptimal - a screwup.
Military pilots especially on carriers, don't even bother to flare and slow the decent rate much. The aim point is critical on a short carrier "runway". The landing gear is beefed up. You enter a descent that intersects the runway plane and smack into it to catch the cable with your tailhook.
Landing smoothly for passenger comfort is an art that even autoland systems aren't perfect at. But autoland systems do exist, don't depend on hovering, and with proper throttle/energy management, it's not inconceivable that rocket guidance systems can be programmed to find a landing solution as well.
5) The ASDS deck is not made of ablative heat shield material or ceramic flame trench cement. It's steel, and gets sprayed down with water before landing, but you want to minimize the potential for Mach 10 rocket exhaust damage to both the deck and the equipment on the ends of the barge. Another reason to perform the landing with all deliberate speed and not hover.
6) I did some math on the ballasted weight of the Marmac series of barges, and it's millions of pounds. And I've seen discussion of the typical sea states and swell lengths out where it will normally be picking up a stage. If the distance between swells is shorter than the barge, it's not going to be rocking up and down as badly as people have speculated. It's not going to be jumping up and down 6-10 feet or tilting that much. It's not steaming into the wind like an aircraft carrier to add to the motion. Since the landing is done by GPS, and not by a radar altimeter system, the returning stage really doesn't have any idea of the barge tilt or change in height over time. But you have to figure that the SpaceX designers took that into account when they set the specs for the descent rate at intended touchdown, and if the deck is within some anticipated height above and below that altitude, then the legs won't break.
Besides, they've had PLENTY of time to gather data during sea trials and especially towing the ASDS out and back multiple times. ALL of that time could have been spent characterizing the barge performance - how much DOES it tilt and wiggle by wave and wind action? I have every confidence some engineer in Hawthorne knows the statistics on ASDS deck movement for all possible sea states, and that data has been used to inform the leg designers how strong they need to be, and the landing algorithm designers how much error budget they have to work with, and the grid fin and engine gimbaling actuator people and the thrust valve designers how much bandwidth they have to work with to make the landing solution possible. It's all been designed and profiled - and they wouldn't even be trying this tactic if they hadn't done a lot of modeling before they built the first Falcon 9.
Thrust management is no easier to hover a stage than to match a specific level that causes accurate deceleration. It might even be harder because hovering a stage that's constantly getting lighter involves a steady ramp DOWNWARD of thrust, instead of one set value. It was always my goal for a smooth landing in a light aircraft, to set the throttle ONCE for a targeted descent rate from traffic pattern altitude, and then not touch it again until I was in the flare about to touch down. It's inelegant to need to fiddle with the throttle all the time unless you have gusty and unpredictable winds.
If all the hardware works to the intended response rate, and the response of the stage to those control inputs is as modeled and adjusted by the 5-6 ocean landing tests they ran before now, then hovering is only needed by Kerbal noobs who don't have insanely fast computer autopilots. When every gram counts on a launch vehicle, everything gets optimized. SpaceX isn't full of dummies - if all other reasons are not compelling to you, give them the benefit of the doubt and admit they might know more about their hardware, and their solution is optimum.
And after they manage to land one, and tow it back to Jacksonville, we'll (hopefully) discover what their optimum solution is to the "how do they fasten it to the deck during transit" challenge.
Sorry for the length of this but I thought it was important, and I have a bit of experience with flying and control algorithms, as do some other members.
och Blue Origin har gjort något stort och SpaceX kommer göra det förhoppningsvis under december.