Author Topic: Falcon 9 1st stage landing accuracy and progress towards ITS landing cradle?  (Read 6901 times)

Offline hkultala

  • Full Member
  • ****
  • Posts: 640
  • Liked: 173
  • Likes Given: 105
Is there any compilation of F9 landing precision statistics?

The latest was 0.7 meters off-center from the platform.

This is quite good but might not be enought for the cradle landing.

What about earlier ones?

Has the accuracy been about the same with some random variation, or been improving?

(and has it always been off-center to same direction or to random direction?)


I wonder if thy are constantly tweaking their landing control algorithms(so that they can in the future get it to the precision required by the cradle), or have they decided that the current algorithm is accurate enough for F9 and don't want to risk introducing some bug by changing it?
« Last Edit: 08/25/2017 08:29 AM by hkultala »

Offline meberbs

  • Full Member
  • ****
  • Posts: 1324
  • Liked: 1219
  • Likes Given: 312
The latest was 0.7 meters off-center from the platform.

This is quite good but might not be enought for the cradle landing.
I assume this post was prompted by Musk's tweet of landing accuracy.

He also had a follow up where someone asked what accuracy is needed for ITS landing mount.
Quote from: Elon Musk
Probably 2m or so
The rest of your questions bring up some good points though, for a landing cradle, you would want to make sure you can be within the accuracy even in a worst case situation.

The "probably" in Musk's response is almost certainly due to it still being way to early in the design for the accuracy needed to be fixed to an exact number.

Offline IRobot

  • Full Member
  • ****
  • Posts: 1170
  • Portugal & Germany
  • Liked: 207
  • Likes Given: 196
The cradle itseft might move to match the stage. I've seen a 400mT crane in an well intervention ship (oil & gas) moving sideways and up/down by 2.5 meters to each side, to account for sea motion.
« Last Edit: 08/25/2017 08:33 AM by IRobot »

Offline cppetrie

  • Full Member
  • ***
  • Posts: 392
  • Liked: 214
  • Likes Given: 3
Is there any compilation of F9 landing precision statistics?

The latest was 0.7 meters off-center from the platform.

This is quite good but might not be enought for the cradle landing.

What about earlier ones?

Has the accuracy been about the same with some random variation, or been improving?

(and has it always been off-center to same direction or to random direction?)


I wonder if thy are constantly tweaking their landing control algorithms(so that they can in the future get it to the precision required by the cradle), or have they decided that the current algorithm is accurate enough for F9 and don't want to risk introducing some bug by changing it?
There was a plot recently from the landing bingo game that showed the winning location for each landing. (BTW did I miss landing bingo for these last two launches?) it showed there was a consistent drift direction that we suspected was wind related. It is probably possible to calculate the distance from center for each one and plot over time to determine how landing accuracy has changed over time. Can anyone find that plot to post into this thread? I'm on my phone only and searching abilities are limited.

Continuing the trend of landing on the aft half.


Edit: found the post
« Last Edit: 08/25/2017 02:34 PM by cppetrie »

Online wannamoonbase

  • Elite Veteran
  • Senior Member
  • *****
  • Posts: 2709
  • Denver, CO
    • U.S. Metric Association
  • Liked: 534
  • Likes Given: 922
0.7 meters on an ASDS is seriously impressive.

They are getting better and more consistent.  Very cool to be watching this evolution.

This is an exciting time to be a spacenut.
Excited to be finally into the first Falcon Heavy flow, we are getting so close!

Online Lar

  • Fan boy at large
  • Global Moderator
  • Senior Member
  • *****
  • Posts: 8601
  • Saw Gemini live on TV
  • A large LEGO storage facility ... in Michigan
  • Liked: 5362
  • Likes Given: 3545
Landing bingo may be getting kinda pointless...

I don't know that anyone did this analysis for land landings.
"I think it would be great to be born on Earth and to die on Mars. Just hopefully not at the point of impact." -Elon Musk
"We're a little bit like the dog who caught the bus" - Musk after CRS-8 S1 successfully landed on ASDS OCISLY

Offline Jim

  • Night Gator
  • Senior Member
  • *****
  • Posts: 31277
  • Cape Canaveral Spaceport
  • Liked: 9561
  • Likes Given: 299

There was a plot recently from the landing bingo game that showed the winning location for each landing. (BTW did I miss landing bingo for these last two launches?) it showed there was a consistent drift direction that we suspected was wind related. It is probably possible to calculate the distance from center for each one and plot over time to determine how landing accuracy has changed over time. Can anyone find that plot to post into this thread? I'm on my phone only and searching abilities are limited.


Data for landings on the barge shouldn't be included since there are errors in the location of the barge.  Only data on ground based landings should be looked at.

Offline Comga

  • Senior Member
  • *****
  • Posts: 3910
  • Liked: 1232
  • Likes Given: 1047

There was a plot recently from the landing bingo game that showed the winning location for each landing. (BTW did I miss landing bingo for these last two launches?) it showed there was a consistent drift direction that we suspected was wind related. It is probably possible to calculate the distance from center for each one and plot over time to determine how landing accuracy has changed over time. Can anyone find that plot to post into this thread? I'm on my phone only and searching abilities are limited.


Data for landings on the barge shouldn't be included since there are errors in the location of the barge.  Only data on ground based landings should be looked at.

I am not sure I agree, Jim.

Of course these should be considered. They bound the accuracy and we have been told the ASDS finds the target position to a meter or so.
Plus we have this data. We don't get as good data for the pad landings.
Mediocre data with a fixed target or very accurate data with a compounding error.  Both have value especially with such a small number of data points.

edit/Lar: PoliteJim/3000 works for inbound as well as outbound messages.
« Last Edit: 08/25/2017 06:11 PM by Lar »
What kind of wastrels would dump a perfectly good booster in the ocean after just one use?

Offline Lars-J

  • Senior Member
  • *****
  • Posts: 3454
  • California
  • Liked: 2690
  • Likes Given: 1700
The cradle itseft might move to match the stage. I've seen a 400mT crane in an well intervention ship (oil & gas) moving sideways and up/down by 2.5 meters to each side, to account for sea motion.

It doesn't even have to move... If the cradle has some sort of vanes/glove that can shift the landing vehicle to its final resting position. Building something like this that can handle a 2m horizontal error is certainly possible. And more likely than a dynamic moving system, IMO.

Think of the landing cradle as the equivalent of the passive cone in a probe & drogue docking system. (see image)
« Last Edit: 08/25/2017 06:00 PM by Lars-J »

Online matthewkantar

  • Full Member
  • ****
  • Posts: 678
  • Liked: 425
  • Likes Given: 437
The cradle itseft might move to match the stage. I've seen a 400mT crane in an well intervention ship (oil & gas) moving sideways and up/down by 2.5 meters to each side, to account for sea motion.

It doesn't even have to move... If the cradle has some sort of vanes/glove that can shift the landing vehicle to its final resting position. Building something like this that can handle a 2m horizontal error is certainly possible. And more likely than a dynamic moving system, IMO.

Think of the landing cradle as the equivalent of the passive cone in a probe & drogue docking system. (see image)

Seems to me that tis would require some large, robust (heavy) surfaces on the returning booster. The equipment on the ground can be arbitrarily robust, the modifications to the booster have to be significantly lighter than landing gear for the effort to pay off.

Matthew

Offline Lars-J

  • Senior Member
  • *****
  • Posts: 3454
  • California
  • Liked: 2690
  • Likes Given: 1700
The cradle itseft might move to match the stage. I've seen a 400mT crane in an well intervention ship (oil & gas) moving sideways and up/down by 2.5 meters to each side, to account for sea motion.

It doesn't even have to move... If the cradle has some sort of vanes/glove that can shift the landing vehicle to its final resting position. Building something like this that can handle a 2m horizontal error is certainly possible. And more likely than a dynamic moving system, IMO.

Think of the landing cradle as the equivalent of the passive cone in a probe & drogue docking system. (see image)

Seems to me that tis would require some large, robust (heavy) surfaces on the returning booster. The equipment on the ground can be arbitrarily robust, the modifications to the booster have to be significantly lighter than landing gear for the effort to pay off.

Matthew

Not necessarily. The catching mechanism could have dampening and very low friction (even as low tech as heavy duty rollers on the surface)

And the base is already *incredibly* robust on rockets... The octaweb equivalent will need to support the holding the entire fully loaded rocket before takeoff (launch hold downs attach to it), and with the the purpose of the three fins in the ITS concept is to make sure that only they need to be the strongest, they would guide the booster into the proper position. So for the base to handle the an empty booster being pushed slightly sideways is nothing compared to the loads before and during liftoff.

Offline intrepidpursuit

  • Full Member
  • ****
  • Posts: 460
  • Orlando, FL
  • Liked: 277
  • Likes Given: 193
The landing fins on the BFR are exactly the hard point vanes you are talking about. We already know they are there and can reverse engineer what the other side would look like.

Elon said they need about 2m of accuracy. I'm sure he's put some thought into it.

Offline spacenut

  • Senior Member
  • *****
  • Posts: 2150
  • East Alabama
  • Liked: 310
  • Likes Given: 195
I agree with Jim.  The ship is moving or trying to stay in one position in a moving sea.  It is usually more windy at sea also.  The landing pad(s) are fixed and never move.  It is all up to the rocket to hit the center on land.  I think they need a lot more return to launch site and see how accurate they can get before using a landing cradle. 

Maybe the could land on a movable landing cradle, that can land it away from the launch pad.  The cradle could then carry the rocket over the launch pad for refueling and launching.  Then having a spare movable cradle in case of crash so as not to keep other rockets from launching and landing.  Something like the mobile platform for Saturn V, shuttle, and SLS, but either on tracks for a traditional locomotive to move or wheels. 

Online FutureSpaceTourist

  • Senior Member
  • *****
  • Posts: 4241
  • UK
    • Plan 28
  • Liked: 2838
  • Likes Given: 948
I think they need a lot more return to launch site and see how accurate they can get before using a landing cradle.


Edit to add: itís not clear to me whether youíre referring to F9 RTLS or BFR.

I suspect just this yearís F9 RTLS has given SpaceX a pretty good handle on F9 accuracy, coupled with any planned further improvements for block 5. Given increased frequency of RTLS with block 5, it wonít take long in 2018 to get significant more RTLS experience.

So I think the question here is the extent to which F9 landing accuracy does, or doesnít carry over to BFR. My guess is that SpaceX have a pretty good handle now on what drives/limits F9 accuracy.
« Last Edit: 08/26/2017 02:59 AM by FutureSpaceTourist »

Offline darkenfast

  • Member
  • Full Member
  • ****
  • Posts: 735
  • Liked: 370
  • Likes Given: 718
The wind out at sea is fairly steady (most of the time).  It think some of what we see is the drift that happens right at the end of flight, when the engine starts shutting down.  You can often see a little drift, almost like a skid.  Take a look at where the rocket is when it's 15 feet up and we may get a better idea of what will happen when the vanes enter the guides on the pad.  Pure speculation, etc.

Offline spacenut

  • Senior Member
  • *****
  • Posts: 2150
  • East Alabama
  • Liked: 310
  • Likes Given: 195
I was referring to RTLS landings of Falcon 9 to within the 2m of accuracy that Elon is shooting for.  I didn't think they had 9 return to launch site, most have been on the landing ships.  I know they have had 9 landings this year, but most have been on the landing ships. 

Online LouScheffer

  • Full Member
  • ****
  • Posts: 1554
  • Liked: 1732
  • Likes Given: 197
Is there any compilation of F9 landing precision statistics?

The latest was 0.7 meters off-center from the platform.

This is quite good but might not be enought for the cradle landing.
If they can routinely get 0.7 meters, or even close, it's plenty good enough for landing two boosters on one ship. 

That would result in a big boost in FH payload, at the cost of two ships in one ocean (one to  catch the outside boosters, and one further downrange to catch the central core).

Offline meekGee

  • Senior Member
  • *****
  • Posts: 7679
  • N. California
  • Liked: 3984
  • Likes Given: 822
I agree with Jim.  The ship is moving or trying to stay in one position in a moving sea.  It is usually more windy at sea also.  The landing pad(s) are fixed and never move.  It is all up to the rocket to hit the center on land.  I think they need a lot more return to launch site and see how accurate they can get before using a landing cradle. 

Maybe the could land on a movable landing cradle, that can land it away from the launch pad.  The cradle could then carry the rocket over the launch pad for refueling and launching.  Then having a spare movable cradle in case of crash so as not to keep other rockets from launching and landing.  Something like the mobile platform for Saturn V, shuttle, and SLS, but either on tracks for a traditional locomotive to move or wheels.
Don't forget that unlike the audience, SpaceX knows the internal state of the control systems, so that if the barge was not where it was supposed to be, they can subtract that out.

So they know, and certainly can use the data.

For us, all we know is that barge landing accuracy is degraded by the additional motion of the barge, and by conditions that are typically more adverse, so land landing accuracy will be better.
ABCD - Always Be Counting Down

Online wannamoonbase

  • Elite Veteran
  • Senior Member
  • *****
  • Posts: 2709
  • Denver, CO
    • U.S. Metric Association
  • Liked: 534
  • Likes Given: 922
Is there any compilation of F9 landing precision statistics?

The latest was 0.7 meters off-center from the platform.

This is quite good but might not be enought for the cradle landing.
If they can routinely get 0.7 meters, or even close, it's plenty good enough for landing two boosters on one ship. 

That would result in a big boost in FH payload, at the cost of two ships in one ocean (one to  catch the outside boosters, and one further downrange to catch the central core).

Might want to build the second ship to be larger.

What would happen to boosters that close together?  If one was landed briefly before the other.  What would the affects of the second landing (engine exhaust) be on the first?

It's an interesting idea though.  The performance increase would be pretty impressive but currently unwarranted for anything on the manifest I believe.
Excited to be finally into the first Falcon Heavy flow, we are getting so close!

Online LouScheffer

  • Full Member
  • ****
  • Posts: 1554
  • Liked: 1732
  • Likes Given: 197
If they can routinely get 0.7 meters, or even close, it's plenty good enough for landing two boosters on one ship. 

That would result in a big boost in FH payload, at the cost of two ships in one ocean (one to  catch the outside boosters, and one further downrange to catch the central core).

Might want to build the second ship to be larger.

What would happen to boosters that close together?  If one was landed briefly before the other.  What would the affects of the second landing (engine exhaust) be on the first?

A back of the envelope calculation indicates no problem with the second booster blowing over the first.  So it should work if the accuracy is enough.

Offline meekGee

  • Senior Member
  • *****
  • Posts: 7679
  • N. California
  • Liked: 3984
  • Likes Given: 822
If they can routinely get 0.7 meters, or even close, it's plenty good enough for landing two boosters on one ship. 

That would result in a big boost in FH payload, at the cost of two ships in one ocean (one to  catch the outside boosters, and one further downrange to catch the central core).

Might want to build the second ship to be larger.

What would happen to boosters that close together?  If one was landed briefly before the other.  What would the affects of the second landing (engine exhaust) be on the first?

A back of the envelope calculation indicates no problem with the second booster blowing over the first.  So it should work if the accuracy is enough.
That would be cool, but I don't think I'll work.  The issue is not one booster toppling the other, but just the backwash ruining the landing accuracy.

Also, the upside is limited. Side boosters cut off low and slow, so RTLS is not so expensive. 
ABCD - Always Be Counting Down

Offline envy887

  • Senior Member
  • *****
  • Posts: 2701
  • Liked: 1243
  • Likes Given: 776
Side boosters cut off low and slow, so RTLS is not so expensive.
FH accelerates faster than F9, so the boosters stage while going faster and further downrange for the same fuel burn.

Offline meekGee

  • Senior Member
  • *****
  • Posts: 7679
  • N. California
  • Liked: 3984
  • Likes Given: 822
If that's true, I stand corrected..  but do you have numbers for that? It's payload and orbit dependent, but still.

Initial t/w won't be THAT much higher, since the boosters are much heavier than S2.

Only when the boosters get lighter will the difference begin to grow.  Peak will be just before staging of course, unless thrust is checked.
ABCD - Always Be Counting Down

Online LouScheffer

  • Full Member
  • ****
  • Posts: 1554
  • Liked: 1732
  • Likes Given: 197
If they can routinely get 0.7 meters, or even close, it's plenty good enough for landing two boosters on one ship. 

That would result in a big boost in FH payload, at the cost of two ships in one ocean (one to  catch the outside boosters, and one further downrange to catch the central core).

Might want to build the second ship to be larger.

What would happen to boosters that close together?  If one was landed briefly before the other.  What would the affects of the second landing (engine exhaust) be on the first?

A back of the envelope calculation indicates no problem with the second booster blowing over the first.  So it should work if the accuracy is enough.
That would be cool, but I don't think I'll work.  The issue is not one booster toppling the other, but just the backwash ruining the landing accuracy.
Should not affect landing accuracy of the second booster much.  The force on the first landed booster is only about 500 kg-force.   The back-reaction on the second landing booster will be much less (something on the order of 10 kg-f, assuming the first booster scatters the wind uniformly.)  Anyway, very much less than the forces available from thrust vectoring, so the closed-loop landing guidance will mostly null out even this small effect.

Offline meekGee

  • Senior Member
  • *****
  • Posts: 7679
  • N. California
  • Liked: 3984
  • Likes Given: 822
If they can routinely get 0.7 meters, or even close, it's plenty good enough for landing two boosters on one ship. 

That would result in a big boost in FH payload, at the cost of two ships in one ocean (one to  catch the outside boosters, and one further downrange to catch the central core).

Might want to build the second ship to be larger.

What would happen to boosters that close together?  If one was landed briefly before the other.  What would the affects of the second landing (engine exhaust) be on the first?

A back of the envelope calculation indicates no problem with the second booster blowing over the first.  So it should work if the accuracy is enough.
That would be cool, but I don't think I'll work.  The issue is not one booster toppling the other, but just the backwash ruining the landing accuracy.
Should not affect landing accuracy of the second booster much.  The force on the first landed booster is only about 500 kg-force.   The back-reaction on the second landing booster will be much less (something on the order of 10 kg-f, assuming the first booster scatters the wind uniformly.)  Anyway, very much less than the forces available from thrust vectoring, so the closed-loop landing guidance will mostly null out even this small effect.
Yeah, but the magnitude of the force is a strong function of the precise timing between boosters.

That said, if they come in 10 seconds apart, then there's no issue and only the blow-over consideration remains.

I therefore withdraw that part of the argument....

Still the payback IMO is low, and still I think it'd be the absolute coolest thing to watch
ABCD - Always Be Counting Down

Online speedevil

  • Full Member
  • ***
  • Posts: 388
  • Fife
  • Liked: 181
  • Likes Given: 165
That said, if they come in 10 seconds apart, then there's no issue and only the blow-over consideration remains.

I therefore withdraw that part of the argument....

Still the payback IMO is low, and still I think it'd be the absolute coolest thing to watch
This was mentioned on the (IIRC) post-launch conference of CRS-12.
Comment was 'the ship looks a lot smaller with a rocket on it' or something.

Offline IRobot

  • Full Member
  • ****
  • Posts: 1170
  • Portugal & Germany
  • Liked: 207
  • Likes Given: 196
The wind out at sea is fairly steady (most of the time).  It think some of what we see is the drift that happens right at the end of flight, when the engine starts shutting down.  You can often see a little drift, almost like a skid.  Take a look at where the rocket is when it's 15 feet up and we may get a better idea of what will happen when the vanes enter the guides on the pad.  Pure speculation, etc.
Yes, the wind at high seas is steady (laminar wind) but very close to the ship (last 10-20 meters) it changes a lot.
That is why high performance sailing boats have two wind vanes: one on top of the mast and one lower.

Offline AncientU

  • Senior Member
  • *****
  • Posts: 4790
  • Liked: 2875
  • Likes Given: 4034
Not sure if this is the correct thread... but this will be interesting!

Quote
Putting together SpaceX rocket landing blooper reel. We messed up a lot before it finally worked, but there's some epic explosion footage …

https://twitter.com/elonmusk/status/903333005527093248

Note response from someone named Chris B...;)
Quote
There's an unreleased one from a successful landing. Apparently danced on one leg before landing it. Hope that's included. #EpicLanding.
« Last Edit: 08/31/2017 07:21 PM by AncientU »
"If we shared everything [we are working on] people would think we are insane!"
-- SpaceX friend of mlindner

Offline AC in NC

  • Full Member
  • *
  • Posts: 172
  • Raleigh NC
  • Liked: 131
  • Likes Given: 146
Quote
DEIMOS IMAGING‏ @deimosimaging 6m6 minutes ago
Replying to @SpaceX

#DEIMOS2 caught Cape Canaveral's Landing Zone1 at around 15:58 UTC, September 7, less than 2hours after #Falcon9's first stage landed there!



https://twitter.com/deimosimaging/status/907255552153194496




Are they intentionally attempting to land with the legs oriented toward the "hash mark" (breaks) in the landing circle perimeter?  I see similar on the drone ship picture in one of the posts above.  And here on the latest landing, it appears the leg orientation was close to perfect.
« Last Edit: 09/11/2017 03:43 PM by AC in NC »

Online zhangmdev

  • Member
  • Posts: 42
  • Liked: 14
  • Likes Given: 0
Supposedly that has something to do with the orientation of its inertial platform.

Online oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 3117
  • Florida
  • Liked: 1513
  • Likes Given: 117
In order for the capture box of 2m on a 9m diameter vehicle the +- rotation angle error must be less than <12.7 degrees.

Are the legs orientation to the marks less than that?

Offline Kabloona

  • Senior Member
  • *****
  • Posts: 4025
  • Velocitas Eradico
  • Fortress of Solitude
  • Liked: 2204
  • Likes Given: 445
In order for the capture box of 2m on a 9m diameter vehicle the +- rotation angle error must be less than <12.7 degrees.

Are the legs orientation to the marks less than that?

When I put a straightedge on the photo and line it up with the upper right and lower left hash marks, the lower left leg lies directly on the straightedge, within the resolution of the photo. I'd call it rotationally aligned to within single-digit degrees of angular error, if not better.
« Last Edit: 09/12/2017 11:29 PM by Kabloona »

Tags: