LIVE: SpaceX Falcon 9 (Flight 2) - COTS-1 - Launch Updates - December 8, 2010

[hop]

If they really did get within less than 800m of their target from orbit, that wasn't by chance.

Unwarranted assumption.

but it seems to suggest to me that they probably have active guidance via the small lift/drag you get from having an angle-of-attack with a capsule and/or Dracos firing while reentering.

Of course they flew a guided re-entry. AFAIK every crewed capsule since Voskhod has attempted to do so. Soyuz does, and sometimes it achieves kilometer-ish accuracy like this, and other times it does not. See Aniks wonderful map here http://forum.nasaspaceflight.com/index.php?topic=19141.0

Essentially, SpaceX needs only the last mile.

Not proven.

[martin hegedus]

Reentry control for low L/D vehicles is typically simply based on damping the vehicle body rates (capsules tend to be quite stable about the trim angle-of-attack).

Therefore, the only real question related to aerodynamic model uncertainty is the trim AOA profile that the vehicle followed during entry. That's why Elon specifically mentioned in the press conference that they had predicted a maximum of 20% dispersion and only had a 2% difference wrt to the predicted trim AOA. This is significant enough to suggest that the L/D profile they predicted in their aerodynamic analyses was very very close to the actual L/D profile flown.

As I mentioned earlier, I don't know the ins and outs of their control system.  Of course they can get very complex.  But lets say I simplify it by saying there is an inner loop maintaining the angle of attack and an outer loop maintaining the trajectory.  If the inner loop is maintaining the commanded trim angle of attack, why would a 2% difference in AoA indicate they had predicted the aero correctly?  I would think the deviation of the control command would be a better indication of how well they predicting the aerodynamics.  Or did they fly with the inner loop open?  Or are they commanding attitude rather than flow angles?

As far as my deviation of AoA statements, I was thinking about the outer loop.  The one that commands the required trim angle of attack based on errors in the trajectory.  Of course if this is open loop and they are following a schedule, then my comments don't apply.  I was assuming the outer loop was closed.  But if the outer loop is open, then I don't see how the Dragon will be able to land on a helipad.

Of course I could be totally over or under simplifying this, or just clue_less.

[zaitcev]

Judging from the sea state, there was very little wind. Soyuz experience shows that wind can blow a capsule up to 15 km from the expected landing point. The question is if Dragon can compensate properly for the wind drift by changing the no-chute aimpoint on the surface, based on last-minute weather reports by the recovery team.

[Xentry]

As I mentioned earlier, I don't know the ins and outs of their control system.  Of course they can get very complex.  But lets say I simplify it by saying there is an inner loop maintaining the angle of attack and an outer loop maintaining the trajectory.  If the inner loop is maintaining the commanded trim angle of attack, why would a 2% difference in AoA indicate they had predicted the aero correctly? 

It doesn't make much sense to "command" the trim angle of attack. The stabilising moments are typically too high about the actual trim AOA (too much fuel consumption needed to keep the AOA at the wrong trim point), so what you're supposed to do is just damp oscillations about the actual trim AOA. That's why the 2% difference wrt to the prediction is so significant: because since they are not actively controlling the AOA, it means that their predictions of the trim AOA were really good.

[Xentry]

Or did they fly with the inner loop open?  Or are they commanding attitude rather than flow angles?

As far as my deviation of AoA statements, I was thinking about the outer loop.  The one that commands the required trim angle of attack based on errors in the trajectory.  Of course if this is open loop and they are following a schedule, then my comments don't apply.  I was assuming the outer loop was closed.  But if the outer loop is open, then I don't see how the Dragon will be able to land on a helipad.

Of course I could be totally over or under simplifying this, or just clue_less.

Low L/D vehicle guidance and control is typically structured as an outer loop which generates roll/bank commands in order to track a given drag, sink rate and/or range profile (switching the roll sign everytime the heading error exceeds a certain value to limit the lateral error), and an inner loop which is meant, on one hand, to track the outer loop-commanded roll angle, and on the other to simply cancel any angular rates - which will result on stabilising the vehicle about its actual trim AOA.
That's it. No AOA control is exerted, since it would be too costly in terms of fuel to fight the actual trim AOA.
This is more than enough to keep you within a few kms from a target parachute deployment location. More precision can be obtained using an online trajectory update, a fine heading control scheme, a smart chute deployment logic, a numerical propagation scheme, etc, but the underlying flight control structure won't differ much from this.

[marsavian]

He's talking about keeping the (SDLV, presumably) SLS, but using some of the money for a Falcon-series HLV.

No, he's talking about the upper and lower stages of the HLV, not the SLS ...

http://web02.aviationweek.com/aw/mstory.do?id=news/awst/2010/11/29/AW_11_29_2010_p28-271784.xml&channel=space&headline=NASA%20Studies%20Scaled-Up%20Falcon,%20Merlin

Based on a roughly evenly split $10 billion budget for heavy lift, with half for the boost stage and half for the upper stage, “we’re confident we could get a fully operational vehicle to the pad for $2.5 billion

SpaceX believes this arrangement could allow the use of an unchanged Falcon 9 upper stage. “That way you get a three-stage super-heavy-lift vehicle, and all you’ve done is scale up the Merlin and Falcon 9 first stage. You essentially get a second stage for free,” says Musk.

One sentence says $5B for upper stage, the other says upper stage is free because re-use the F9 one. Implies $10B doesn't refer to the 'X'.

So, does the "roughly even split" refer to SLS? IE: "assuming SLS core would be $5B, we could do ours for $2.5B, and re-use the Falcon upper stage for free"?

But F9 u/s would also limit this to LEO only, even with the Raptor upper stage.

cheers, Martin

Yep looks like I misunderstood what he was saying, so it's his version of Phase 2 using minimum changes from where he is now, clever. Take your point about the Merlin upper stage but there is a way though you could make quick effective use of this cheap kerosene LEO HLV and that would be to stick an Ares I type J-2X upper stage on top of the Falcon 9 upper stage and use that as the EDS with its Orion/mission module payload on top of it. So your EDS would not be your upper stage but be sized solely for its mission. I don't know how both 'sides' would feel about such a hybrid but seems to me the cheapest effective way to an exploration HLV if he can deliver his LEO HLV for under $5bn with a working large Merlin 2.

[martin hegedus]

Low L/D vehicle guidance and control is typically structured as an outer loop which generates roll/bank commands in order to track a given drag, sink rate and/or range profile (switching the roll sign everytime the heading error exceeds a certain value to limit the lateral error), and an inner loop which is meant, on one hand, to track the outer loop-commanded roll angle, and on the other to simply cancel any angular rates - which will result on stabilising the vehicle about its actual trim AOA.
That's it. No AOA control is exerted, since it would be too costly in terms of fuel to fight the actual trim AOA.
This is more than enough to keep you within a few kms from a target parachute deployment location. More precision can be obtained using an online trajectory update, a fine heading control scheme, a smart chute deployment logic, a numerical propagation scheme, etc, but the underlying flight control structure won't differ much from this.

Thanks.

This off topic discussion (Sorry moderators, delete if necessary) started from an earlier post that stated "What was truly incredible was Elon Musk saying that a future Dragon is being planned to do a full powered landing and land on a helipad similar to landing a helicopter."  Granted the more fuel you have the more you can do, but using a control system such as you mentioned (and I believe you) it seems very hard (impossible?) to target a helipad well enough to reliably be where you want to be for the end game...  Just scratching my head.  Of course this future Dragon may have a completely different control system.

[Comga]

Of course they flew a guided re-entry. AFAIK every crewed capsule since Voskhod has attempted to do so. Soyuz does, and sometimes it achieves kilometer-ish accuracy like this, and other times it does not. See Aniks wonderful map here http://forum.nasaspaceflight.com/index.php?topic=19141.0

Wow! That's a great map, and I have sent thanks to Anik for it.

Looking at the map, I can't find any pairs of targets and actuals within a kilometer of each other.  (Anik obviously has the data, and could give us maximum, minimum, and average offsets.)

It seems that Soyuz has an average landing error of several kilometers, after dozens of landings. Dragon came down less than a kilometer from its aim point on its first flight.  While they must have gotten lucky, the idea that it was simply luck is not credible. 

For now, give SpaceX some credit, people.  We will know the dispersion in a year or two when there are enough data for statistics.  Then  you can look back and see who was a foolish amazing people and who was overly skeptical.  For now we have one data point, although some doubt that, and it's a remarkable one.

PS Discussions of future Dragon mods, and offers by Musk to build an HLV for a fixed price, while mentioned in the post-COTS-1 press conference, are really off topic here.  They are worthy of their own threads, so can we keep this one to the COTS-1 launch, flight, and landing, like where to find video of the parachutes opening? :-)

[NotGncDude]

Of course they flew a guided re-entry. AFAIK every crewed capsule since Voskhod has attempted to do so. Soyuz does, and sometimes it achieves kilometer-ish accuracy like this, and other times it does not. See Aniks wonderful map here http://forum.nasaspaceflight.com/index.php?topic=19141.0

I think 5 kilometer-ish is a better description based on this plot, which is what I'd expect.

[hop]

I think 5 kilometer-ish is a better description based on this plot, which is what I'd expect.

You're right. I thought I saw some km-ish ones in there, but looking closer I don't 3ish seems to be about the minimum (TMA-7 for example).
Still, I don't think this negates the basic point: The difference between "good" Soyuz landings and average is quite large, even if you throw out the really off nominal ones. One data point from Dragon doesn't tell us a whole lot about the average case.

Dragon came down less than a kilometer from its aim point on its first flight.  While they must have gotten lucky, the idea that it was simply luck is not credible. 

Certainly it wasn't *all* luck. Clearly they did a very good job. I'm not saying that SpaceX can't achieve this kind of accuracy reliably. All I'm saying is that this one landing doesn't justify the assumption that they can. Maybe they can, or maybe conditions were particularly favorable on that day.

[Comga]

Of course they flew a guided re-entry. AFAIK every crewed capsule since Voskhod has attempted to do so. Soyuz does, and sometimes it achieves kilometer-ish accuracy like this, and other times it does not. See Aniks wonderful map here http://forum.nasaspaceflight.com/index.php?topic=19141.0

I think 5 kilometer-ish is a better description based on this plot, which is what I'd expect.

Using data provided by hop, covering 39 flights since TM-12, excluding the three ballistic reentries, I calculate the average is 13.1 km and the minimum is 2.7 km.

[edkyle99]

U.S. tracking shows eight objects in LEO, not including the two-orbit Dragon spacecraft, and an upper stage in a 278 x 11,083 km x 34.54 deg orbit. 

 - Ed Kyle

[Lars_J]

So 8 cube/nano sats were launched on this flight?

[docmordrid]

Yup, one being the first US Army sat since 1960: SMDC-ONE

Link....

[Lobo]

Hey, quick question.

Elon mentioned wanting to do a propulsive landing eventually.  At other times he mentioned having the LAS as a pusher system that would then be used for landing it it was a successful launch.
In the press coference, I got the impression he was talking about a 100% propulsive landing that could land on an area as small as a helicopter pad, and basically not needing parachutes except as backup?

Can someone maybe elaborate on this?  It seems like you'd need a lot more propellant than the LAS system would have to decellerate from terminal velocity of 755 fps to about 24fps, which is what the Soyuz parachutes do for it, before the landing rockets fire for a soft landing.  A 100% propulsive landing would also need maneuvering to hit a target rather than just landing in a wide open dry lakebed at Edwards or something.  Just seems like that'd take a lot of propellant for a capsule and crew weighing several mt, without a parachute at all.

I always figured when Elon was talking about dual purposing a pusher LAS system for a propulsive landing, that they'd use parachutes to decellerate from terminal velosity, and just use the LAS system for touchdown like Soyuz does. 

But I don't really know how much propellant is needed for a 100% propulsive landing, or if that is about as much needed for an LAS system?  Seems like it'd be more, but maybe not.
So thus the question.
How much propellant/thruster mass we talking about here compared to the mass of the parachutes?  (which sound like they'd be there anyway for backup).
Am I understanding Elon correctly that he's talking about 100% propulsive landing with the chutes only as backup?  And wouldn't that be pretty mass inefficient?

[Robotbeat]

Hey, quick question.

Elon mentioned wanting to do a propulsive landing eventually.  At other times he mentioned having the LAS as a pusher system that would then be used for landing it it was a successful launch.
In the press coference, I got the impression he was talking about a 100% propulsive landing that could land on an area as small as a helicopter pad, and basically not needing parachutes except as backup?

Can someone maybe elaborate on this?  It seems like you'd need a lot more propellant than the LAS system would have to decellerate from terminal velocity of 755 fps to about 24fps, which is what the Soyuz parachutes do for it, before the landing rockets fire for a soft landing.  A 100% propulsive landing would also need maneuvering to hit a target rather than just landing in a wide open dry lakebed at Edwards or something.  Just seems like that'd take a lot of propellant for a capsule and crew weighing several mt, without a parachute at all.

I always figured when Elon was talking about dual purposing a pusher LAS system for a propulsive landing, that they'd use parachutes to decellerate from terminal velosity, and just use the LAS system for touchdown like Soyuz does. 

But I don't really know how much propellant is needed for a 100% propulsive landing, or if that is about as much needed for an LAS system?  Seems like it'd be more, but maybe not.
So thus the question.
How much propellant/thruster mass we talking about here compared to the mass of the parachutes?  (which sound like they'd be there anyway for backup).
Am I understanding Elon correctly that he's talking about 100% propulsive landing with the chutes only as backup?  And wouldn't that be pretty mass inefficient?

I get less than 500 feet per second terminal velocity, assuming (quite conservatively) a drag coefficient of only .6 and an altitude of 10,000 feet.

[Xentry]

Hey, quick question.

Elon mentioned wanting to do a propulsive landing eventually.  At other times he mentioned having the LAS as a pusher system that would then be used for landing it it was a successful launch.
In the press coference, I got the impression he was talking about a 100% propulsive landing that could land on an area as small as a helicopter pad, and basically not needing parachutes except as backup?

Can someone maybe elaborate on this?  It seems like you'd need a lot more propellant than the LAS system would have to decellerate from terminal velocity of 755 fps to about 24fps, which is what the Soyuz parachutes do for it, before the landing rockets fire for a soft landing.  A 100% propulsive landing would also need maneuvering to hit a target rather than just landing in a wide open dry lakebed at Edwards or something.  Just seems like that'd take a lot of propellant for a capsule and crew weighing several mt, without a parachute at all.

I always figured when Elon was talking about dual purposing a pusher LAS system for a propulsive landing, that they'd use parachutes to decellerate from terminal velosity, and just use the LAS system for touchdown like Soyuz does. 

But I don't really know how much propellant is needed for a 100% propulsive landing, or if that is about as much needed for an LAS system?  Seems like it'd be more, but maybe not.
So thus the question.
How much propellant/thruster mass we talking about here compared to the mass of the parachutes?  (which sound like they'd be there anyway for backup).
Am I understanding Elon correctly that he's talking about 100% propulsive landing with the chutes only as backup?  And wouldn't that be pretty mass inefficient?

I am not sure about the specific configuration of the Dragon capsule, but at low supersonic/subsonic speeds capsules tend to be unstable, so at least a drogue chute is probably necessary. And the terminal speed under a drogue chute is certainly smaller than 755fps, so delta-v requirements in the end could be much smaller...
However, precision landing requirements will eat into these savings, depending on the accumulated errors until throttle up.
As for the trade-off btw parachute and fuel+thrusters mass fractions, I honestly have no idea.

[Robotbeat]

There are other options besides a drogue chute... another kind of (small) deployable aerosurface could be employed for stabilizing the capsule. Jon Goff, at least, doesn't like drogue chutes. As I said before, the terminal velocity is going to be considerably less than 755 fps either way (it will be somewhere between 100 and 150m/s... use metric!!!).

[Xentry]

Granted the more fuel you have the more you can do, but using a control system such as you mentioned (and I believe you) it seems very hard (impossible?) to target a helipad well enough to reliably be where you want to be for the end game...  Just scratching my head.  Of course this future Dragon may have a completely different control system.

See  http://trs-new.jpl.nasa.gov/dspace/bitstream/2014/40343/1/06-3509.pdf for a good example of the state of the art in entry flight control.
This approach, combined with the modified Apollo entry guidance they are using, should be enough for a sub-km precision at parachute deployment (as can be seen in the entry guidance research work done by both UC Irvine and JSC). The problem in this particular case is just that MSL is limited on a number of aspects which affect the overall landing precision.

[docmordrid]

A couple months ago I used a terminal velocity calculator, all the published Dragon info I could dig up and some minor guesswork and came up with 323.386 kph/200.943 mph.