SpaceX Falcon 9 : CRS-16 (Dragon SpX-16) : December 5, 2018 - DISCUSSION

[meekGee]

So this leaves the engine. 

Wrong, there is the RCS.  Which does roll control on single engine burns above the atmosphere and during slow speeds in the atmosphere.

I'll look into the RCS later, but are we at least done saying the leg extension did it?

-----
ABCD: Always Be Counting Down

[Jim]

So this leaves the engine. 

Wrong, there is the RCS.  Which does roll control on single engine burns above the atmosphere and during slow speeds in the atmosphere.

I'll look into the RCS later, but are we at least done saying the leg extension did it?

No,  Read the original post.

https://forum.nasaspaceflight.com/index.php?topic=45881.msg1884475#msg1884475

[OxCartMark]

So this leaves the engine. 

Wrong, there is the RCS.  Which does roll control on single engine burns above the atmosphere and during slow speeds in the atmosphere.

I'll look into the RCS later, but are we at least done saying the leg extension did it?

No,  Read the original post.

https://forum.nasaspaceflight.com/index.php?topic=45881.msg1884475#msg1884475

No.

[Cinder]

What I would like to know is how precisely did the stage hit the water target. It seems pretty clear that if it had landed on a hard surface it would have landed upright, so could it have landed on the ASDS?

The stage looked like it was busier staying vertical than aiming for an area the size of an ASDS, so it'd've'd to be something larger like some vacant lot of land.

[marsbase]

What I would like to know is how precisely did the stage hit the water target. It seems pretty clear that if it had landed on a hard surface it would have landed upright, so could it have landed on the ASDS?

The stage looked like it was busier staying vertical than aiming for an area the size of an ASDS, so it'd've'd to be something larger like some vacant lot of land.

The center core of the FH Demo did real damage to OCISLY so maybe it's not a good idea to aim a malfunctioning rocket at an expensive vessel.  I suspect this booster could have landed at LZ1 if given the chance, but of course that is not possible.

[Joffan]

What I would like to know is how precisely did the stage hit the water target. It seems pretty clear that if it had landed on a hard surface it would have landed upright, so could it have landed on the ASDS?

The stage looked like it was busier staying vertical than aiming for an area the size of an ASDS, so it'd've'd to be something larger like some vacant lot of land.

The center core of the FH Demo did real damage to OCISLY so maybe it's not a good idea to aim a malfunctioning rocket at an expensive vessel.  I suspect this booster could have landed at LZ1 if given the chance, but of course that is not possible.

I strongly doubt this booster could have realigned itself to the pad. It just managed to beat the spin right at the end but it couldn't have translated the target point during the spin with any great control.

[CorvusCorax]

the engine was able to recover a wildly spinning ship .

Not that incorrect statement again

What I can’t conceive of is how there was any kind of control authority with the landing burn. The stage is rotating - the center engine is gimbaling to balance the booster during descent, but the rotation the changes the relative angle of thrust so the control algorithm has to gimbal in a new angle, but the booster rotates, so it has to gimbal again to correct, but the booster rotates... you get the idea. This looks like it would quickly lead to a runaway instability situation as the mechanical actuators fall behind the control inputs. How they kept that thing under control is pretty impressive.

The answer to that is "feed forward control". In a nutshell, the computer knows how quick the TVC gimbal can possibly respond. As such they don't "chase" the control loop by moving the TVC where it would have to be now to fix the current situation (atitude, rotation speed) but to the orientation it needs to be in a short time in the future if the system keeps transitioning based on physics (inertia, current control output)

with some control algorithms (model predictive control for example) the computer simulates the behavior of the entire system a few seconds into the future and tries to find the sequence of control outputs that leads to the best possible output.

if its really good, it even "knows" (based on sensors, hydraulic pressure, or inherent system analysis based one expected and actual outcome) that the grid fins aren't doing their job and altered the model based on observed behavior as opposed to a hardcoded model.

you can do that kind of thing arbitrarily complex, but at some point it becomes too computationally expensive, so you take shortcuts with lookup tables based on values you expect not to change (like system geometry and resulting aerodynamic forces) so obviously there's a point when even that sophisticated system can't cope.

but that SpaceX is doing more than just a simple PID controller, I think this landing demonstrated quite impressively :-)

[marsbase]

What I would like to know is how precisely did the stage hit the water target. It seems pretty clear that if it had landed on a hard surface it would have landed upright, so could it have landed on the ASDS?

The stage looked like it was busier staying vertical than aiming for an area the size of an ASDS, so it'd've'd to be something larger like some vacant lot of land.

The center core of the FH Demo did real damage to OCISLY so maybe it's not a good idea to aim a malfunctioning rocket at an expensive vessel.  I suspect this booster could have landed at LZ1 if given the chance, but of course that is not possible.

I strongly doubt this booster could have realigned itself to the pad. It just managed to beat the spin right at the end but it couldn't have translated the target point during the spin with any great control.

Yes, I think you're right. The grid fins play a major role in translation as I understand it.  I just meant that if the guidance had targeted LZ1 from the beginning the booster, which was almost vertical and not rotating much at touchdown ,might have remained standing on the pad. 

[meekGee]

What I would like to know is how precisely did the stage hit the water target. It seems pretty clear that if it had landed on a hard surface it would have landed upright, so could it have landed on the ASDS?

The stage looked like it was busier staying vertical than aiming for an area the size of an ASDS, so it'd've'd to be something larger like some vacant lot of land.

The center core of the FH Demo did real damage to OCISLY so maybe it's not a good idea to aim a malfunctioning rocket at an expensive vessel.  I suspect this booster could have landed at LZ1 if given the chance, but of course that is not possible.

I strongly doubt this booster could have realigned itself to the pad. It just managed to beat the spin right at the end but it couldn't have translated the target point during the spin with any great control.

Yes, I think you're right. The grid fins play a major role in translation as I understand it.  I just meant that if the guidance had targeted LZ1 from the beginning the booster, which was almost vertical and not rotating much at touchdown ,might have remained standing on the pad.

Remember grasshopper? 

I'm pretty sure the translation from ocean to pad occurs (gently) over the course of the landing burn, principally by the engine.

The fins control the booster mainly between the burns.


-----
ABCD: Always Be Counting Down

[georgegassaway]

Remember grasshopper? 

I'm pretty sure the translation from ocean to pad occurs (gently) over the course of the landing burn, principally by the engine.

The fins control the booster mainly between the burns.

I remember Grasshopper. The translation it did was maybe 200-300 feet. Not several MILES.

The F9 re-entry is very steep, nearly vertical, ballistically falling towards a spot a few miles out from shore.

After reentry burn, the grid fins aerodynamically steer it in a steep diagonal dogleg path towards the landing zone (the translational distance from ballistic path in the ocean towards to the LZ).

By the time that the landing burn begins , the translation begins to slow and the path begins to curve from a steep diagonal dogleg towards vertical. In other words, there is LESS translation once the landing burn begins, not more. And the grid fins are still helping with the steering during that time. Indeed the first part of the landing burn the grid fins should still be dominant in steering.  When it gets slower, then the thrust vectoring has more of an effect in steering than the grid fins do, more and more effective as it gets slower and slower.

Tim Dodd (Everyday Astronaut) posted a very good video which explains the landing process, with the grid fins steering the dogleg translation, among other things. Maybe it has already been posted in this thread, but apparently some have not seen it.



With the grid fin steering out, the best that the Falcon could do, was to land slowly somewhere, totally incapable of steering to a specific spot (especially with the rolling). And that somewhere was in the ocean as it was  too far out to make it to land (thank goodness). It was surprising it ended up as close as a half mile away from shore, that is far closer than was supposed to possible. I won’t be surprised if CCAFS investigates how it could get that close. Most likely that it did already start the aerodynamic translation for a short time before things went to hell, then once the grid fins went out the transition ended to become mostly ballistic again (albeit rolling, but averaging out any pitch/yaw effects from the stuck fins). But by then it had moved the ballistic path closer to the  coastline.

[Jim]

Remember grasshopper? 

I'm pretty sure the translation from ocean to pad occurs (gently) over the course of the landing burn, principally by the engine.

The fins control the booster mainly between the burns.

I remember Grasshopper. The translation it did was maybe 200-300 feet. Not several MILES.

The F9 re-entry is very steep, nearly vertical, ballistically falling towards a spot a few miles out from shore.

After reentry burn, the grid fins aerodynamically steer it in a steep diagonal dogleg path towards the landing zone (the translational distance from ballistic path in the ocean towards to the LZ).

By the time that the landing burn begins , the translation begins to slow and the path begins to curve from a steep diagonal dogleg towards vertical. In other words, there is LESS translation once the landing burn begins, not more. And the grid fins are still helping with the steering during that time. Indeed the first part of the landing burn the grid fins should still be dominant in steering.  When it gets slower, then the thrust vectoring has more of an effect in steering than the grid fins do, more and more effective as it gets slower and slower.

Tim Dodd (Everyday Astronaut) posted a very good video which explains the landing process, with the grid fins steering the dogleg translation, among other things. Maybe it has already been posted in this thread, but apparently some have not seen it.



With the grid fin steering out, the best that the Falcon could do, was to land slowly somewhere, totally incapable of steering to a specific spot (especially with the rolling). And that somewhere was in the ocean as it was  too far out to make it to land (thank goodness). It was surprising it ended up as close as a half mile away from shore, that is far closer than was supposed to possible. I won’t be surprised if CCAFS investigates how it could get that close. Most likely that it did already start the aerodynamic translation for a short time before things went to hell, then once the grid fins went out the transition ended to become mostly ballistic again (albeit rolling, but averaging out any pitch/yaw effects from the stuck fins). But by then it had moved the ballistic path closer to the  coastline.

The video states no roll control from single engine and leg deployment reducing roll.

[meekGee]

Remember grasshopper? 

I'm pretty sure the translation from ocean to pad occurs (gently) over the course of the landing burn, principally by the engine.

The fins control the booster mainly between the burns.

I remember Grasshopper. The translation it did was maybe 200-300 feet. Not several MILES.

The F9 re-entry is very steep, nearly vertical, ballistically falling towards a spot a few miles out from shore.

After reentry burn, the grid fins aerodynamically steer it in a steep diagonal dogleg path towards the landing zone (the translational distance from ballistic path in the ocean towards to the LZ).

By the time that the landing burn begins , the translation begins to slow and the path begins to curve from a steep diagonal dogleg towards vertical. In other words, there is LESS translation once the landing burn begins, not more. And the grid fins are still helping with the steering during that time. Indeed the first part of the landing burn the grid fins should still be dominant in steering.  When it gets slower, then the thrust vectoring has more of an effect in steering than the grid fins do, more and more effective as it gets slower and slower.

Tim Dodd (Everyday Astronaut) posted a very good video which explains the landing process, with the grid fins steering the dogleg translation, among other things. Maybe it has already been posted in this thread, but apparently some have not seen it.



With the grid fin steering out, the best that the Falcon could do, was to land slowly somewhere, totally incapable of steering to a specific spot (especially with the rolling). And that somewhere was in the ocean as it was  too far out to make it to land (thank goodness). It was surprising it ended up as close as a half mile away from shore, that is far closer than was supposed to possible. I won’t be surprised if CCAFS investigates how it could get that close. Most likely that it did already start the aerodynamic translation for a short time before things went to hell, then once the grid fins went out the transition ended to become mostly ballistic again (albeit rolling, but averaging out any pitch/yaw effects from the stuck fins). But by then it had moved the ballistic path closer to the  coastline.

The video states no roll control from single engine and leg deployment reducing roll.

Whose video?  EDA's?  He's just a webcaster, and he uses Kerbal, and his simulated F9 does not have the real F9's control software, so of course he can't do fancy stuff like coupled rotational axes.

The man in charge, meanwhile, said it was the engine.

And nobody argued the legs reduce roll.  I even calculated it above.  Somewhere between 1.5, and 2.  Clearly a lot less then we saw in practice.

The fact is, roll decreased before the legs deployed.

How long was the RCS firing, comulatively?

-----
ABCD: Always Be Counting Down

[Lars-J]

The man in charge, meanwhile, said it was the engine.

If you are going to quote the man in charge, don't mis-quote. engineS. (so ... including RCS)

https://twitter.com/elonmusk/status/1070399755526656000

[meekGee]

The man in charge, meanwhile, said it was the engine.

If you are going to quote the man in charge, don't mis-quote. engineS. (so ... including RCS)

https://twitter.com/elonmusk/status/1070399755526656000

OK fair enough, so RCS is fair play based on Musk's statement.  Also, unlike the leg deployment, RCS can indeed bring the roll to zero.

Let's figure out how much they did tho.  Based on the videos, how long IYO were the RCS firing?

[Kabloona]

The man in charge, meanwhile, said it was the engine.

If you are going to quote the man in charge, don't mis-quote. engineS. (so ... including RCS)

https://twitter.com/elonmusk/status/1070399755526656000

OK fair enough, so RCS is fair play based on Musk's statement.  Also, unlike the leg deployment, RCS can indeed bring the roll to zero.

Let's figure out how much they did tho.  Based on the videos, how long IYO were the RCS firing?

I get about 12 seconds of roll-correction thrusting visible from the one thruster in the camera's field of view, starting from just prior to landing burn ignition through touchdown.

The thruster pulses appear to be around 0.75 seconds long each, regularly spaced in time, though some are partially obscured by saturation from the main engine plume, and then there's a multi-second firing just before touchdown.

But, the thruster pulses are ineffective during most of the landing burn because overwhelmed by grid fin forces. It's only at the end when the stage has been slowed enough that the grid fins lose force and the thrusters can dominate. So where do you draw the line? It's not clear.

Having said that, it looks to me that only the last 4 seconds or so of RCS thrusting had an effect on roll, in concert with leg deployment.

[meekGee]

The man in charge, meanwhile, said it was the engine.

If you are going to quote the man in charge, don't mis-quote. engineS. (so ... including RCS)

https://twitter.com/elonmusk/status/1070399755526656000

OK fair enough, so RCS is fair play based on Musk's statement.  Also, unlike the leg deployment, RCS can indeed bring the roll to zero.

Let's figure out how much they did tho.  Based on the videos, how long IYO were the RCS firing?

I get about 12 seconds of roll-correction thrusting visible from the one thruster in the camera's field of view, starting from just prior to landing burn ignition through touchdown.

The thruster pulses appear to be around 0.75 seconds long each, regularly spaced in time, though some are partially obscured by saturation from the main engine plume, and then there's a multi-second firing just before touchdown.

But, the thruster pulses are ineffective during most of the landing burn because overwhelmed by grid fin forces. It's only at the end when the stage has been slowed enough that the grid fins lose force and the thrusters can dominate. So where do you draw the line? It's not clear.

Having said that, it looks to me that only the last 4 seconds or so of RCS thrusting had an effect on roll, in concert with leg deployment.

Yes, agreed on the observations.

I can't find a good estimate on the force of the thrusters.  It's v*dm/dt, and v is probably 500-600, but I can't find dm/dt.   Given how quickly the thrusters flip a still-loaded stage around, they're not chump change, and 12 seconds is a lot of time.   From this, I draw the same conclusion you did - that during most of the time, the fins overwhelm the thrusters.

But the fins lose effectiveness towards the end (probably very non-linearly), and then both things happen - the legs deploy and cut the rotation speed as I calculated above, and the thrusters finally get their way.

As for the main engine - as explained by several people above - when there are external forces on the booster (such as gravity, aerodynamic loads, etc) a center engine can absolutely affect axial spin.  I think during that time, the fins are powerful enough that only something like the main engine can really fight them.

As an aside - I'm surprised by how much spare capacity the RCS thrusters had. 

[Rocket Science]

The fins have lost most of their control authority toward the final landing burn due to the deceased airflow though them as the velocity bled off as is drag on the entire vehicle. The gimbled throttle changes also imparted another torque though the CoG before touchdown righting the vehicle along with my previously mentioned torque (gyroscopic) during leg extension and the conservation of angular momentum. The roll RCS also had a great effect in reduction of rotation as well. Fuel load being burned off also has it's own contribution. Thus compound effects at play which is to be expected in complex motion...

[jpo234]

Obviously, the reaction time for the engine gimbal is quicker than the rate of change imparted by that spin.
Which, yes, means that is one ridiculously nimble engine!
I expect they did this to make the propulsive landing possible in the first place, you do not want a large actuator lag for gimbal control on top of all the other control delays in the last few milliseconds of a landing.

Does anyone here have actual info on the max gimbal speed/rate of the center engine? For example, from first command to position achieved, how long for the engine to gimbal 4 degrees?
For that matter, just what is the maximum deflection achievable on that engine?

Oldie but goldie:

[Jakusb]

Obviously, the reaction time for the engine gimbal is quicker than the rate of change imparted by that spin.
Which, yes, means that is one ridiculously nimble engine!
I expect they did this to make the propulsive landing possible in the first place, you do not want a large actuator lag for gimbal control on top of all the other control delays in the last few milliseconds of a landing.

Does anyone here have actual info on the max gimbal speed/rate of the center engine? For example, from first command to position achieved, how long for the engine to gimbal 4 degrees?
For that matter, just what is the maximum deflection achievable on that engine?

Oldie but goldie:

I understand how just 1 engine is not able to counter a spin as it can only target one direction at a time, but having it gimble in certain way (counter clockwise?) could it not help counter some spin? And what is the effect of the jet hitting the water? Would it not feedback in some way that might also kill any movement relative to the water?
Or does the jet only give upward energy and no counter spin energy as feedback?

Please Note: I only have a basic education in Theoretical Physics, so forgive me if above is a complete garbage reasoning..

[friendly3]

I understand how just 1 engine is not able to counter a spin as it can only target one direction at a time, but having it gimble in certain way (counter clockwise?) could it not help counter some spin? And what is the effect of the jet hitting the water? Would it not feedback in some way that might also kill any movement relative to the water?
Or does the jet only give upward energy and no counter spin energy as feedback?

How could a center engine gimbal counterclockwise?