Friday, April 17, 2015

Hovering capability for the reusable Falcon 9.

Copyright 2015 Robert Clark

 We have now two landing attempts by the reusable F9 first stage onto the SpaceX barge. Both were unsuccessful. From the appearance of both failed landings it would appear that the capability to hover could have made both landings successful:

 About this latest test landing, Elon Musk in a Tweet has acknowledged that not being able to hover will result in a high g landing:


   This is confirmed by the rather high rate of descent apparent in the video, even though, as has been reported, the video is slowed down.

 Another disadvantage of not having hovering capability is apparent in the video. In correcting for mistakes in the angle of tilt, the engine having limited throttle capability will tend to over correct. That is, without hover and its low thrust capability, you can't make fine adjustments to the rocket orientation. Then at low altitude with little time to make corrections to the over-corrections, this can lead to failed landing due to the need for repeated adjustments and readjustments.

 In the blog post "Merlin 1A engine for a hovering Falcon 9 v1.1 first stage" I suggested one possible solution to the hovering question would be by using the lower thrust Merlin 1A engine. However, it would have to be made throttleable for this to work. In further updates to that blog post, I suggested either using the preburner exhaust or using variable size nozzles.

 Indeed all the proposals discussed in the "Altitude compensation attachments for standard rocket engines, and applications" post could also be used to make variable nozzle attachments to the engine nozzles to reduce the thrust when needed to allow hovering. For instance the carbon nanotube "rubber" attachment could be made to restrict the exit area to reduce the thrust and the "internal spike" proposal could be made to flare out to direct some proportion of the thrust laterally outwards rather than downwards to reduce the downwards thrust.

 These would reduce the efficiency of the engines, i.e., the Isp would be decreased for the hovering proportion of the flight. However, the altitude compensation attachments actually increase the payload perhaps as much as 25% for multi-stage rockets so overall the result will still be an improvement of the payload capacity.

 These altitude compensation attachment proposals do need more R & D work however, and SpaceX might want a quicker fix that can be attached quickly to the engines, or likely just the central engine for the landing phase.

 A possibility is suggested by this collapsible vegetable steamer:

 You would make an attachment like this that could flare out or be closed up, except it would have no holes on the sides. The open position would be usual formation used during the flight. The closed up position would be used only during hover to restrict the thrust.

 Another simple attachment might be the exhaust steering vanes used on rockets prior to the advent of engine gimbaling for steering:

      They could be used to direct a portion of the thrust laterally to reduce the downward thrust.

    Bob Clark

 UPDATE, April 21, 2015:

 Someone suggested to me another method to restrict the thrust to allow hovering, the variable nozzles put on some fighter jet engines:


  1. Yep, they have a landing design flaw to overcome, as you have pointed out before. Some of us old farts could have foreseen this, but they hire no one over age 45. There are two short-term solutions: at least one deep-throttling but lower thrust engine among the 9, or else add attitude thrusters. Longer-term, I'd recommend reworking the Merlin-D for deeper throttle capability, and adding it as soon as it is ready. -- GW

  2. Sorry, I disagree.

    IMO what's lacking isn't control authority, it has gobs of that or the stage wouldn't have hit the edge of the target circle from100 miles up even with the malfunction below.

    The problem was what chopper pilots call a phase lag, this due to a malfunctioning propellant valve. Musk stated the problem was static friction (stiction) in said valve.

    Because of this stiction the throttle response was insufficient vs the guidance signal. This resulted in an instability which the system increasingly tried to compensate for with an increased gimbal angle, causing the stage to deviate.

    In this case the stage slid out, landeing heavy on only 1-2 legs which failed causing the stage to topple.

  3. Note that having hovering capability was considered as a given for vertical
    landing reusable launchers. See for instance this discussion between noted
    space historian Henry Spencer and Mitchell Burnside Clapp, who worked on
    both the DC-X and X-33 programs:

    Horizontal vs. vertical landing (Henry Spencer; Mitchell Burnside Clapp).

    Looking at the videos of both attempts at the F9 barge landings, note that
    they all looked fine as they initially approached the landing point. It was
    only when they were a few tens of meters above the barge when they were
    attempting to make their final corrections that the landings went awry.

    My opinion, this is an inherent aspect of the "hover slam" approach because
    you have so little margin of error. In the videos as released by SpaceX the
    stages already look like they are approaching too fast. But in point of fact
    they are returning even *faster* then this because the videos are slowed

    It is notable then that all of the several Grasshopper tests that used
    hovering went perfectly even those that had to make horizontal translations
    before landing. Of the two Grasshopper tests that used the non-hovering
    "hover slam" approach though, one was successful, and one failed.
    Then to get a successful barge landing or return-to-launch-site landing, use
    the current approach until just a few tens of meters above the landing spot.
    Then switch to a hovering approach.

    BTW, Jeff Findley on the forum suggested another way to give the Falcon 9 first stage hover capability. Give it the SuperDraco thrusters to be used on the Dragon V2. The Dragon will use 8 of them, but probably 4 would be sufficient to give the F9 stage hover ability when it is nearly empty upon landing.

    Bob Clark

  4. Yes, at about 7 t-f each 4 SD's should hover a 20-25t slightly wet stage but a the cost of reduced payload mass and the messiness of dealing with hypergolics.

    Still, there's nothing inherently wrong with the hover slam. Like any landing system it's at the mercy of engine reliability and engine controls. Even a hover-able helicopter (or rocket) would go bump with a serious lag in the throttle system.