SpaceX Falcon 9 v1.1 - Dragon - CRS-5/SpX-5 -Jan. 10, 2015 - DISCUSSION

[IslandPlaya]

Quote from: IslandPlaya on Today at 22:29:07
Ah, yes...
Very funny.
Should that not be in the party thread?

@MeekGee
Not much value will be lost, agreed, but between now and Tuesday, what else are we to do but party?

[laika_fr]

Ok, regarding the pad :


(forecast)

- cloud cover improving
- light wind (<10km/h)

[ChrisWilson68]

The diagram above shows a re-entry burn, but no boost-back burn.

For forward recovery, this makes sense.

However, I think this test includes some boost-back.  I say that, because on the nav charts, the intended landing zone is #3 out of 4.  It makes sense to me that the furthest zone (#4) is a contingency zone in case the boost-back burn does not happen, and zone #2 is in case the re-entry burn does not happen, and zone #3 is where the X is.

Why are they doing boost-back when they can just park the barge further out?  Maybe it's part of practicing RTLS, so going through all the motions.  When it's time to do FH center cores, they might skip the boost back.

I think Zone 4 is where the stage comes down if re-entry burn doesn't happen.

The stage is going pretty fast when the re-entry burn happens.  If it fails, it will keep going quickly and end up overshooting the intended landing zone.

[deruch]

The diagram above shows a re-entry burn, but no boost-back burn.

For forward recovery, this makes sense.

However, I think this test includes some boost-back.  I say that, because on the nav charts, the intended landing zone is #3 out of 4.  It makes sense to me that the furthest zone (#4) is a contingency zone in case the boost-back burn does not happen, and zone #2 is in case the re-entry burn does not happen, and zone #3 is where the X is.

Why are they doing boost-back when they can just park the barge further out?  Maybe it's part of practicing RTLS, so going through all the motions.  When it's time to do FH center cores, they might skip the boost back.

I think Zone 4 is where the stage comes down if re-entry burn doesn't happen.

The stage is going pretty fast when the re-entry burn happens.  If it fails, it will keep going quickly and end up overshooting the intended landing zone.

Compare the zones with the hazard maps from earlier CRS flights.  You can see that they are clearly boosting back towards land.  That boost back happens before re-entry.

[georgegassaway]

The grid fins that Russia uses on the Soyuz, to be deployed in case of an abort, are purely for aerodynamic stability so it does not tumble. After all, the abort rockets are not gimbaled either, so there is no active guidance that does anything once begun (same was true for Mercury and Apollo Launch Escape Systems). 

The grid fins are not steerable. 

On Falcon, they steer.  So they create stability, and as-desired can also help to steer

As to whether they deploy for the re-entry burn, it would really depend on how accurate the previous re-entry burn tests have been, how the vehicle was responding to the attempts to steer it with the thrusters and engine thrust. 

IIRC, the stated landing error was as much has 10 miles from the intended spot, and for an error that big, it would seem to be introduced from pretty high up.  It seems that during re-entry (and after too), the rocket body wants to wobble around a lot aerodynamically, even if it is tail-heavy (or nose-heavy in the relative  sense that it's falling backwards). So while the engines plus thrusters were able to keep it pointed the right way, and TRY to keep on course, obviously not enough to make it accurately follow a precise path.

 Given that the re-entry burn is happening from 43 to 25 miles, and the error from previous test being 10 miles, it would be an awful lot to expect grid fins could steer it well enough to make up for a 10 mile error from only 25 miles up, if they are not deployed until after the retry burn is completed.   This of course does not necessarily mean the grid fins would be deployed before the re-entry burn. In theory they could be deploy at some point during re-entry. Though that could cause issues if all four did not deploy at exactly the same angles at any given instant.   The grid fins have been described as working at hypersonic speeds.  So, they may very well be deployed before re-entry. 

It is certainly possible for grid fins to be fabricated of material that can take the heat of this kind of re-entry (far less intense than an orbital re-entry), and for the hinges and servo-mechnical control mechanisms are built sturdily enough to withstand Max-Q coming down.  Depends partly on  the cost of  making them to be able to hold up for use on re-entry,  and mass.

 Well, its also a question of how badly the Falcon stage needs the aerodynamic steering early. But given the stated 10 mile error it just seems awful hard for aero steering to be able to make up for a 10 mile path error in the last 25 miles.  A cylinder flying thru air at a given angle of attack WILL produce lift, of course, but a "glide" of 1 mile per 2.5 miles traveled seems unlikely.

It just also occurred to me that if the grid fins are deployed before or at the start of the reentry burn, then the wobbling of the rocket should not be much of a problem anymore,e so that the rocket steering can also be more effective in keeping the stage on path for the landing sot. So then the grid fins will not need to steer as much after engine shutdown.

It will be interesting to see. Really  looking forward to it.

- GeorgeG

[deruch]

The grid fins that Russia uses on the Soyuz, to be deployed in case of an abort, are purely for aerodynamic stability so it does not tumble. After all, the abort rockets are not gimbaled either, so there is no active guidance that does anything once begun (same was true for Mercury and Apollo Launch Escape Systems). 

The grid fins are not steerable. 

On Falcon, they steer.  So they create stability, and as-desired can also help to steer

As to whether they deploy for the re-entry burn, it would really depend on how accurate the previous re-entry burn tests have been, how the vehicle was responding to the attempts to steer it with the thrusters and engine thrust. 

IIRC, the stated landing error was as much has 10 miles from the intended spot, and for an error that big, it would seem to be introduced from pretty high up.  It seems that during re-entry (and after too), the rocket body wants to wobble around a lot aerodynamically, even if it is tail-heavy (or nose-heavy in the relative  sense that it's falling backwards). So while the engines plus thrusters were able to keep it pointed the right way, and TRY to keep on course, obviously not enough to make it accurately follow a precise path.

 Given that the re-entry burn is happening from 43 to 25 miles, and the error from previous test being 10 miles, it would be an awful lot to expect grid fins could steer it well enough to make up for a 10 mile error from only 25 miles up, if they are not deployed until after the retry burn is completed.   This of course does not necessarily mean the grid fins would be deployed before the re-entry burn. In theory they could be deploy at some point during re-entry. Though that could cause issues if all four did not deploy at exactly the same angles at any given instant.   The grid fins have been described as working at hypersonic speeds.  So, they may very well be deployed before re-entry. 

It is certainly possible for grid fins to be fabricated of material that can take the heat of this kind of re-entry (far less intense than an orbital re-entry), and for the hinges and servo-mechnical control mechanisms are built sturdily enough to withstand Max-Q coming down.  Depends partly on  the cost of  making them to be able to hold up for use on re-entry,  and mass.

 Well, its also a question of how badly the Falcon stage needs the aerodynamic steering early. But given the stated 10 mile error it just seems awful hard for aero steering to be able to make up for a 10 mile path error in the last 25 miles.  A cylinder flying thru air at a given angle of attack WILL produce lift, of course, but a "glide" of 1 mile per 2.5 miles traveled seems unlikely.

It just also occurred to me that if the grid fins are deployed before or at the start of the reentry burn, then the wobbling of the rocket should not be much of a problem anymore,e so that the rocket steering can also be more effective in keeping the stage on path for the landing sot. So then the grid fins will not need to steer as much after engine shutdown.

It will be interesting to see. Really  looking forward to it.

- GeorgeG

The previous accuracy was 10 km, not miles.  So about 6.2 miles.

http://www.spacex.com/news/2014/12/16/x-marks-spot-falcon-9-attempts-ocean-platform-landing

[meekGee]

The diagram above shows a re-entry burn, but no boost-back burn.

For forward recovery, this makes sense.

However, I think this test includes some boost-back.  I say that, because on the nav charts, the intended landing zone is #3 out of 4.  It makes sense to me that the furthest zone (#4) is a contingency zone in case the boost-back burn does not happen, and zone #2 is in case the re-entry burn does not happen, and zone #3 is where the X is.

Why are they doing boost-back when they can just park the barge further out?  Maybe it's part of practicing RTLS, so going through all the motions.  When it's time to do FH center cores, they might skip the boost back.

I think Zone 4 is where the stage comes down if re-entry burn doesn't happen.

The stage is going pretty fast when the re-entry burn happens.  If it fails, it will keep going quickly and end up overshooting the intended landing zone.

If there's only one burn (apart from the landing burn) then there should be only 3 zones, no?

[mvpel]

And if SpaceX is really confident in their modeling (and it is true that grid fins are fairly well-understood analyticly) and in the strength of their actuators, then they may well keep them extended through the transonic transition region. It would be pretty gutsy to do so, considering the aerodynamic and mechanical unknowns, the risk of control inversion, etc.  Perhaps the actuators can be locked through transition. SpaceX do take risks, and the extra transonic drag might make it worth it.  Maybe not on the first flight, though?

Take a closer look at how fierce their actuator and mounting hardware is:



Six big old bolts on a beefy mounting bracket for the longitudinal actuator, and two more arms on either side. I'm certainly no expert in the field, but I wouldn't sweat a drop having those things deployed for some extra drag during the trans-sonic choke-out period.

Also, I think you may be overestimating the unknowns - they've got a server room stuffed with systems dedicated to computational fluid dynamics and finite element analysis work, and they had . These days you can get over 2,500 3GHz Hyperthread cores in a single 40U rack with HP BladeSystems, not to mention access to tens of thousands of processors via Amazon or Google's clouds if they choose to go that route, so their simulation capacity is undoubtedly formidable. They've probably done hundreds of thousands of Monte Carlo runs leading up to this flight - particularly with the extra couple of weeks they've had - under all sorts of simulated atmospheric conditions. If there's any aerodynamic or mechanical unknowns at this point, someone would have had to have had a failure of imagination.

As an interesting aside, here's a link to patent 5048773 for a curved grid fin.

With the curved grid fins swept back or forward at an angle of about 45° with the longitudinal axis of the missile body, maximum drag is produced for deceleration of the payload.

"Swept forward at an angle of 45 degrees..." reminds me of the F9R-Dev1 fin flight test video at 2:01.

[douglas100]

I thought that the stated error was 10 km not 10 miles. If that's true it makes the grid fins' job a little easier. I think they will be deployed before entry to provide maximum steering. The entry braking burn should reduce the heating and aerodynamic forces on them.

[cscott]

@mvpel I mostly agree with you. But grid fins at 45° are still small compared to the body cross section when entering at nonzero angle of attack.
And the risk of transonic grid fins is (a) physical reality differing from computational realty, and (b) flow asymmetries in the transition region, which can be caused by tiny differences in surface roughness or a rogue insect or a chaotic vortex.  If one fin is low-drag hypersonic and the other side is high-drag transonic things get pretty interesting.
You are right in that they may have modeled the heck out of the problem and come up with some clever sensing and feedback and control algorithms in the avionics to mitigate the problem.  They are certainly doing a lot more principled analysis of transonic flow than Carmack ever did.  But it's the unknown unknowns that always get you.  Maybe their engineers feel cocky and they'll go for it and see what happens, maybe they'll play it safe and keep things locked down through the transonic region.  We'll see!

[ChrisWilson68]

The diagram above shows a re-entry burn, but no boost-back burn.

For forward recovery, this makes sense.

However, I think this test includes some boost-back.  I say that, because on the nav charts, the intended landing zone is #3 out of 4.  It makes sense to me that the furthest zone (#4) is a contingency zone in case the boost-back burn does not happen, and zone #2 is in case the re-entry burn does not happen, and zone #3 is where the X is.

Why are they doing boost-back when they can just park the barge further out?  Maybe it's part of practicing RTLS, so going through all the motions.  When it's time to do FH center cores, they might skip the boost back.

I think Zone 4 is where the stage comes down if re-entry burn doesn't happen.

The stage is going pretty fast when the re-entry burn happens.  If it fails, it will keep going quickly and end up overshooting the intended landing zone.

If there's only one burn (apart from the landing burn) then there should be only 3 zones, no?

Maybe.  Or maybe Zone 2 is for something else.  Maybe it's where the rocket drops if something goes wrong at max-Q.

Zone 2 is small and separated from everything else.  Zone 3 has a tail that stretches to Zone 4.  That seems to imply to me that there's a continuum of possible failure cases that stretch from Zone 4 to the target in Zone 3.  That is consistent with Zone 4 being the failure case for the re-entry burn.  If the re-entry burn fails partway through, the rocket ends up somewhere in the tail between the target point and Zone 4, which is all part of Zone 3.

In your theory, the boostback burn changes the target point from Zone 4 to Zone 2, then the re-entry burn changes the target point from Zone 2 to Zone 3.  So why a tail between Zone 3 and Zone 4?

[meekGee]

The diagram above shows a re-entry burn, but no boost-back burn.

For forward recovery, this makes sense.

However, I think this test includes some boost-back.  I say that, because on the nav charts, the intended landing zone is #3 out of 4.  It makes sense to me that the furthest zone (#4) is a contingency zone in case the boost-back burn does not happen, and zone #2 is in case the re-entry burn does not happen, and zone #3 is where the X is.

Why are they doing boost-back when they can just park the barge further out?  Maybe it's part of practicing RTLS, so going through all the motions.  When it's time to do FH center cores, they might skip the boost back.

I think Zone 4 is where the stage comes down if re-entry burn doesn't happen.

The stage is going pretty fast when the re-entry burn happens.  If it fails, it will keep going quickly and end up overshooting the intended landing zone.

If there's only one burn (apart from the landing burn) then there should be only 3 zones, no?

Maybe.  Or maybe Zone 2 is for something else.  Maybe it's where the rocket drops if something goes wrong at max-Q.

Zone 2 is small and separated from everything else.  Zone 3 has a tail that stretches to Zone 4.  That seems to imply to me that there's a continuum of possible failure cases that stretch from Zone 4 to the target in Zone 3.  That is consistent with Zone 4 being the failure case for the re-entry burn.  If the re-entry burn fails partway through, the rocket ends up somewhere in the tail between the target point and Zone 4, which is all part of Zone 3.

In your theory, the boostback burn changes the target point from Zone 4 to Zone 2, then the re-entry burn changes the target point from Zone 2 to Zone 3.  So why a tail between Zone 3 and Zone 4?

Not sure.

I think Zone 1 captures all of the launch phase.

The boost-back burn happens when the rocket is moving more horizontally than the re-entry burn, so it's failure zone is more drawn out, and thus zone 2 is small.  (since the rocket at that stage is moving at the steepest angle.)

Why zone 3 intrudes into zone 4?  I think the boost-back is very minimal.  More of a systems check (so the number of restarts and thermal cycles is the same) then real boost-back.  Just to increase the fidelity of the test.

We'll see.  If they have the means, and the will, they'll split screen the webcast starting at MECO.   That would be cool.  A 2-ring circus.

[deruch]

The diagram above shows a re-entry burn, but no boost-back burn.

For forward recovery, this makes sense.

However, I think this test includes some boost-back.  I say that, because on the nav charts, the intended landing zone is #3 out of 4.  It makes sense to me that the furthest zone (#4) is a contingency zone in case the boost-back burn does not happen, and zone #2 is in case the re-entry burn does not happen, and zone #3 is where the X is.

Why are they doing boost-back when they can just park the barge further out?  Maybe it's part of practicing RTLS, so going through all the motions.  When it's time to do FH center cores, they might skip the boost back.

I think Zone 4 is where the stage comes down if re-entry burn doesn't happen.

The stage is going pretty fast when the re-entry burn happens.  If it fails, it will keep going quickly and end up overshooting the intended landing zone.

If there's only one burn (apart from the landing burn) then there should be only 3 zones, no?

Maybe.  Or maybe Zone 2 is for something else.  Maybe it's where the rocket drops if something goes wrong at max-Q.

Zone 2 is small and separated from everything else.  Zone 3 has a tail that stretches to Zone 4.  That seems to imply to me that there's a continuum of possible failure cases that stretch from Zone 4 to the target in Zone 3.  That is consistent with Zone 4 being the failure case for the re-entry burn.  If the re-entry burn fails partway through, the rocket ends up somewhere in the tail between the target point and Zone 4, which is all part of Zone 3.

In your theory, the boostback burn changes the target point from Zone 4 to Zone 2, then the re-entry burn changes the target point from Zone 2 to Zone 3.  So why a tail between Zone 3 and Zone 4?

"Zone 2" or Area B on the map is a bit of a red herring.  No one, that I've seen, has yet postulated anything that would adequately explain why it's there.  It's new.  In the sense that there hasn't been an analogous zone in any of the previous CRS launches.  While there have been a previous boost back (sideways really), reentry burns, and landing burns.  None of those necessitated an analogous hazard zone.  Maybe SpaceX is staging some additional asset out there and wants a bit of a "keep away" zone?  The point is that we don't know.  And its small size and location suggests it isn't related to the failure of either the boost back or reentry burns.

 Again.  Look at the difference between CRS-5, CRS-4, and CRS-3 maps:
https://mapsengine.google.com/map/viewer?mid=zp15b_P5ERVk.koWeOnV6-O-o

https://mapsengine.google.com/map/viewer?mid=zp15b_P5ERVk.krdQml9d0HR0

https://mapsengine.google.com/map/viewer?mid=zp15b_P5ERVk.k1KeaKAECPyY  (you can uncheck the other launches to clear up some of the clutter)

You can see that while with CRS-4 they boosted sideways, with CRS-5 they are boosting back about 1/3 of the distance towards the launch site.  On the CRS-5 map, the areas are designated A, B, C, and D so that's what I'll use.  Area D is the hazard zone if the boost back either fails or doesn't happen.  It looks similar to the analogous CRS-3 zone, which was for a non-boosted back, "controlled" landing. 

[mvpel]

And the risk of transonic grid fins is (a) physical reality differing from computational realty, and (b) flow asymmetries in the transition region, which can be caused by tiny differences in surface roughness or a rogue insect or a chaotic vortex.  If one fin is low-drag hypersonic and the other side is high-drag transonic things get pretty interesting.

Indeed they would. 

This paper from 2004, starting at page 28, has a very interesting, illustrated discussion of the flow choking in transsonic grid fins which I think many people will find informative:

DeSpirito, J., M. Vaughn, Jr., and W. D. Washington. Numerical Investigation of Canard-Controlled Missile Using Planar and Grid Tail Fins, Part II: Subsonic and Transonic Flow. Army Research Lab, ARL-TR-3162. March 2004. Web. <http://www.arl.army.mil/arlreports/2004/ARL-TR-3162.pdf>.

Check out the mesh diagram on page 5, and the CFD hardware information they provided on page 6. The used an SGI "Origin 3800 with R12000 processors and an IBM SMP P3 with Power3 processors." The 3800's R12000 was 400 megahertz, and posted these HPCC numbers. It shows a 0.4032 Tflops/s theoretical peak for 512 cores. The ARL paper used 64 cores.

The bottom of the Top 500 HPCC list as of November 2014 is a stack of HP rack-mount servers, namely DL360e Gen8 systems with middling Xeon E5-2450 8-core 2.1GHz CPUs and Gigabit Ethernet totaling 18,896 cores. That's about 1,181, or 30 racks worth, of these 1U servers, maybe two aisles of a good-size server room - assuming they're counting physical, rather than virtual cores in this list - and that's 317.5 theoretical peak Tflops/s, or nearly 6,300 times more power than what was used for the ARL paper's SGI Origin CFD calculations ten years ago, with a retail servers-only cost of only about $4 million, ignoring any volume discounts you'd certainly get if you were making your HP account rep's year. She'd probably even throw in iLO Federation licenses for free with an order like that.

Granted, SpaceX's CFD/FEA compute pool wouldn't necessarily make the top 500, but you can get a sense of how much computational power they can easily bring to bear on their simulations these days, and how little computational reality is likely to differ from physical reality.

I've got about 1,100 virtual cores in my pool at work and that's just one rack with mostly four-year-old hardware that's clocked a good deal slower than the desktop machines which get jobs overnight and weekends, and it has 14 bays still empty. They put a replacement for it in the contract proposal this year, so if you hear an odd sound in the distance that's just me cackling with glee as I rip open boxes and throw blocks of styrofoam over my shoulders.  I love it when the scheduler gets tens of thousands of sims and they spin up the fans - they're Chris Hadfield's happy machines.

Maybe on the AMA tomorrow I'll ask Elon about their mod/sim infrastructure.

[StephenB]

NY Times article: SpaceX’s Next Frontier: Landing a Rocket on Earth.

[darkenfast]

In reply #451 above, is that a bit of leg showing in front of the lockers at the bottom of the photo? If so, it's a nice view of part of the inside (when stowed), of the leg.

[MTom]

In reply #451 above, is that a bit of leg showing in front of the lockers at the bottom of the photo? If so, it's a nice view of part of the inside (when stowed), of the leg.

Nice find! The attachment points of the leg to the core are to see there.

[docmordrid]

Documenting with a crop

[Prober]

And the risk of transonic grid fins is (a) physical reality differing from computational realty, and (b) flow asymmetries in the transition region, which can be caused by tiny differences in surface roughness or a rogue insect or a chaotic vortex.  If one fin is low-drag hypersonic and the other side is high-drag transonic things get pretty interesting.

I've got about 1,100 virtual cores in my pool at work and that's just one rack with mostly four-year-old hardware that's clocked a good deal slower than the desktop machines which get jobs overnight and weekends, and it has 14 bays still empty. They put a replacement for it in the contract proposal this year, so if you hear an odd sound in the distance that's just me cackling with glee as I rip open boxes and throw blocks of styrofoam over my shoulders.  I love it when the scheduler gets tens of thousands of sims and they spin up the fans - they're Chris Hadfield's happy machines.

Maybe on the AMA tomorrow I'll ask Elon about their mod/sim infrastructure.

I know this feeling 

you should see what hacking discarded older graphics cards, and some rework with a soldering gun can do.  Priceless 

[ScottMC]

Plotted the CRS-5 broadcast warning areas (colours at the bottom of the graph) with the position of CRS-4 at T+60, 2:10, 2:30 and 3:00 on the launch YouTube video.  Used ruler and map scale to figure out distances of warning areas so it's only a rough approximation.

Image is attached (hopefully )

CRS-4 Data
0:60, altitude 5.8 km, speed 250 m/sec, downrange 0.8 km
2:10, altitude 38 km, speed 1,160 m/sec, downrange 19 km
2:30, altitude 61 km, speed 1,900 m/sec, downrange 40 km
3:00, altitude 93 km, speed 2,120 m/sec, downrange 79 km