Comment by danielvf | Hacker News Reader

Comment by danielvf | Hacker News Reader

Parent Full thread

I noticed a major difference in the landing sequences of the New Shepard and Falcon 9 first stage.

The Falcon 9 first stage, if I recall correctly, is incapable of either hovering or slowly decending. The engine has simply too much power. A Falcon 9 first stage, stopped mid air has two choices, keep the motor on and go up, or turn the motor off and fall - and I don't think the motor has too many extra restarts available.

Because of this inability to hover the Falcon 9 first stage, SpaceX is attempting to have the rocket's vertical velocity reach zero at the exact moment the rocket reaches the pad. This is why when you watch the grasshopper or other SpaceX landing videos, you always wonder for a split second if the rocket has just smashed into the ground. In order for this to work, all nine axis (three each of position, rotation, and velocity) must be brought to zero at exactly the same fixed time. This is insane level control theory here.

Now this type of landing is theoretically possible - and I think it has been tested on the grasshopper at lower speeds, but it scares the willies out of me. There's almost no room for error nor for the chaos of the universe.

New Shepard on the other hand, comes to a hover about 100? feet above the pad, moves horizontally to be above the pad, stops, then lowers itself down. This is tremendously simpler since the rocket only really cares about one or two set of axis at a time, it does not have to be nearly as precise, and you have time to fix anything that's not lined up.

I'm still curious to see if SpaceX can pull off their landing style, or will instead change so that their first stages will be able to hover.

That's not entirely true. Grasshopper is entirely capable of hovering, there are plenty of videos of it doing precisely that. The Falcon 9 landings that have been attempted are a different story because they actually weigh much less at the time of landing because they have burned off most of their fuel. This makes the thrust to weight ratio much higher, making hovering impossible.

The important difference between this New Shepherd flight and those Falcon 9 tests is that the Falcon 9s launched payload to orbit. They burned all their fuel, leaving them with the incredibly difficult slam-landing manoeuver to manage at the end. They have gone close, very close, but have not yet got things entirely right.

So, although this was an impressive technological demonstration, it's a long way from demonstrating a relaunch capability. I look forward to seeing new developments.

Which part is not true?

"Because of this inability to hover the Falcon 9 first stage..."

The notion that Blue Origin has accomplished something SpaceX hasn't quite worked out is quite a laugh.

SpaceX has said repeatedly that the current Falcon 9 first stage cannot hover when landing after returning from a launch.

The SpaceX Grasshopper can hover, but that's a different rocket and its not following a trajectoy returning from space.

Hovering is also an insane balancing act for a rocket. At speed a rocket can use aerodynamic forces to keep everything lined up. In a hover everything literally turns on the engine gimbals. So there is something to be said for coming to an abrupt halt and landing immediately, before things start getting tipsy.

The hardware and engineering to make a hovering a rocket is amazing. But the control theory behind the hover is just not that tricky - It's very similar to hovering a quad copter.

If you had a JavaScript rocket simulator, with clean position, rotation, and velocity data, you could probably code a hover in five or six lines. Now it's not quite that easy in practice, but the theory is still simple.

The Apollo lunar modules were capable of hovering, using only a portion of a 0.001Ghz 16 bit computer.

Hovering in 1.622 m/s² gravity in a vacuum is significantly simpler than hovering in 9.807 m/s² gravity plus wind.

That said, it's clearly possible.

The rocket is mostly empty and decelerating at that stage; there should be little tipping going on, given that the center of mass is very close to the engine.

But that big empty fuel tank above the engine is like a sail for wind. Not to mention the possibility of fuel sloshing around inside nearly-empty tanks.

When it comes in for a landing, SpaceX seems to be using the engine gimballing a lot. The disadvantage of not being able to hover is that you don't get a second chance if things aren't looking good.

https://www.youtube.com/watch?v=BhMSzC1crr0

Hovering wastes energy though. The most efficient (theoretical) landing would be an instantaneous impulse that brings your velocity to whatever your landing gear can support. The Falcon economics might not support anything less efficient (than the hover-slam technique).

spacex is a company that has an insane level of dedication to product and extremely lofty aspirations somewhere near the very edge of what is currently capable by humans and allowable by physics.

From purely a process standpoint, I think that trying to tightly control 3 axis to this degree of precision is a daunting task. The need to do it this way could be entirely built into the architecture of the system, but from a "first make it work, then make it work better" stand point, the ability to hover or throttle trajectory would be an easier step one.

Outside of this single piece (reusability) and all of the vectors surrounding it, the rocket still has other challenges (e.g no launch since last failure is concerning) which need to be overcome.

SpaceX is scaling too fast and is doing an amazing job at complex coordination, however they don't seem to have the talent bandwith nor the time to tackle every problem 100% and the way they have structured reusability makes it really difficult to be successful even at ~98%.

> The need to do it this way could be entirely built into the architecture of the system, but from a "first make it work, then make it work better" stand point, the ability to hover or throttle trajectory would be an easier step one.

Of course, we can apply the same "make it work, then make it work better" stand point to the Falcon, where "work" is delivering payloads to orbit and "work better" is reusability :)

I think SpaceX can make it work. Their previous attempts have failed for pretty mundane reasons (insufficient hydraulic fluid, sticky valves) and yet have come quite close just the same. I wouldn't be surprised if the next try works, and even if it doesn't, then a few more tries should have it in the bag.

The genius of SpaceX's approach is that their launcher is perfectly viable as an expendable rocket. They are basically being paid by their customers to test the landing technique. That means they don't need a technique which can be pretty certain to work the first time, but instead they can go through a lot of build/test/crash/refine cycles without bleeding away investors' money.

If it ends up that the Falcon 9 has a 10% crash rate when landing even after tons of refinement and practice, well, that's still great! No problem, cost of access to space has still been brought down tremendously.

> SpaceX is attempting to have the rocket's vertical velocity reach zero at the exact moment the rocket reaches the pad

It's definitely hard. It's also the fuel-optimal solution (referred to as a "suicide burn", http://space.stackexchange.com/questions/10307/what-is-a-sui...).

So while it's a very difficult control solution to accomplish, being able to do it reliably translates to a significant savings in fuel cost and therefore one step closer to sustainable reusable rocketry.

Reply on news.ycombinator.com