Author Topic: Yaw steering and RAAN steering  (Read 27076 times)

Offline Kaputnik

  • Extreme Veteran
  • Senior Member
  • *****
  • Posts: 3079
  • Liked: 722
  • Likes Given: 821
Yaw steering and RAAN steering
« on: 01/07/2017 06:22 pm »
As discussed, but off topic for, here:
https://forum.nasaspaceflight.com/index.php?topic=35112.msg1627222#msg1627222

In the context of instantaneous launch windows.

As someone not in the know, may I ask- what is it, how hard is it, and which LVs can do it?

Thanks
« Last Edit: 05/29/2022 09:10 pm by zubenelgenubi »
"I don't care what anything was DESIGNED to do, I care about what it CAN do"- Gene Kranz

Online meberbs

  • Senior Member
  • *****
  • Posts: 3096
  • Liked: 3379
  • Likes Given: 777
Re: Yaw Steering
« Reply #1 on: 01/07/2017 06:26 pm »

A docking spacecraft is in zero g and can take its time with the maneuver. You have all the authority you need, and can abort and try again.

An F9 doing a hoverslam is so completely a different thing.

It is decelerating at multiple g, in wind, with very limited controls, especially towards the end - mostly main engine gimbaling.

Not the same as docking


wrong.  Avionics wise it is the same.  Landing just has more constraints and external influences.  The F9 always knows where it is going to land.   Landing a vehicle is not hard (see lunar and mars landers).   

Landing actually uses less external sensors (altitude radar).  Rendezvous and docking require long range (star trackers, radar, etc) and short range (radar, lidar, TV, etc) sensors.
"more constraints and external influences" makes it harder, by a lot. The F9 knows only the coordinates of where it wants to go for landing, not what it is going to have to do to get there.

Also did you just seriously try to use Mars landers as evidence that landing is not hard?


The secret sauce of the reentry burn and the aerodynamic​ flight segment are unique to SpaceX and already more difficult.


Doesn't require any changes to the avionics to perform those tasks.   Just a little more programming
How does adding additional control hardware and changing the software equal no changes to avionics?

Landing can be done independent of the actual time, unlike yaw steering and rendezvous/docking.  That is a major change to the avionics/ flight software architecture

You know that GPS data comes with accurate measurements of current time for free right? And if GPS is being used at all, they already should have a mechanism for syncing the GPS data time stamp to the current internal clock time. They just have to then feed this already existing data into a yaw steering algorithm. (And this algorithm should be much simpler than a landing algorithm)

Offline Newton_V

  • Full Member
  • ****
  • Posts: 842
  • United States
  • Liked: 822
  • Likes Given: 129
Re: Yaw Steering
« Reply #2 on: 01/07/2017 06:33 pm »
I read some of that discussion.  I think there's some confusion between guidance and navigation software (along with hardware to support it) and control systems (autopilots).  They are 2 separate issues.
The people comparing to landing the booster are pushing the control system aspect and the associated difficulties.  The targeting part is simple.
For yaw (RAAN) steering, it's the robustness of the software and how it converges under highly dispersed conditions (see OA-6).
Atlas has had this capability going back to the Atlas II days.  Delta IV will have it this year.

Offline Robotbeat

  • Senior Member
  • *****
  • Posts: 39271
  • Minnesota
  • Liked: 25241
  • Likes Given: 12115
Re: Yaw Steering
« Reply #3 on: 01/07/2017 06:45 pm »
How is yaw steering hard? You need more on board memory to store a look up table of launch solutions, but other than that, how is it hard? Particularly, how can it possibly be easier than landing?

Landing on a small platform with rockets, in atmosphere, and in Earth gravity is anything but easy.

Time reference
So a decent real time clock synced while on the pad. I'd be surprised if that wasn't already in use for a modern launch vehicle.
« Last Edit: 01/07/2017 08:26 pm by Chris Bergin »
Chris  Whoever loves correction loves knowledge, but he who hates reproof is stupid.

To the maximum extent practicable, the Federal Government shall plan missions to accommodate the space transportation services capabilities of United States commercial providers. US law http://goo.gl/YZYNt0

Offline CorvusCorax

  • Full Member
  • ****
  • Posts: 1860
  • Germany
  • Liked: 4010
  • Likes Given: 2738
Re: Yaw Steering
« Reply #4 on: 01/07/2017 08:27 pm »
Well, I'd say if you claim Yaw steering as a higher level navigational problem, and docking as a higher level control problem, then the hoverslam landing is both.

For yaw steering, if I understand it right you try to match two 4d trajectories. The vehicles current position and velocity versus the velocity and position of a hypothetical "target" vehicle that is on the optimal trajectory and launched at the perfect T-zero. As such you need to cost-optimize the transitioning to have the trajectories converge to 0 - error with minimal extra delta-v expense

A docking approach could be seen as a simplified subset, except your "target 4d trajectory" is completely static. ( unless you want to do something crazy, like dock 2 sattelites with ion engines while both or under acceleration)

Landing on a spot on the surface however is not as simplified. Even if you are in orbit, but on an interplanetary trajectory it becomes much more obvious, that in order to make reentry and landing on a specific point on the surface,due tonthr difference in orbital movement and planet rotation, you have to be on a spoton in a 4d trajectory that leads there on both correct place and time. From the perspective of an orbital vehicle, any point on the surface is under constant acceleration.

But then you have to deal with all the issues of unstable, under-controlled atmospheric flight, reentry, atmospheric uncertanities and wind from the moment of interface on, all of which don't exist on an orbital docking. Even the moon landing didn't have to deal with that.

To make F9s landing extra difficult, for an orbital docking you have an infijite number of solutions. You can hold indefinitely in a multitude of parallel orbits and try again with little extra cost.

For the F9 landing trajectory there is only one correct solution. You cannot recycle or hold anywhere. It cannot hover, and it has insufficient fuel to recycle ( accelerate upwards, turn engine off, then make a 2nd approach ) afaik the engine can't even turn on another time past the landing burn.

Once you end up on a trajectory that no longer converges that's it. Just like with yawcsteering.

But theres another difficulty. The possible solutions to yaw steering that converge can be precalculated in a lookup table.
Landing has to be calculated online.

The final landing approach is both a difficult control problem, and an advanced navigational problem. The calculation of the corrective trajectory if the vehicle is off course is all but trivial.

I'd say docking is the easiest of the problems
Followed by yaw steering
And when you have mastered both, you can join a harder league and try reentry and spot-landing in atmosphere.

Offline Robotbeat

  • Senior Member
  • *****
  • Posts: 39271
  • Minnesota
  • Liked: 25241
  • Likes Given: 12115
Re: Yaw Steering
« Reply #5 on: 01/07/2017 08:45 pm »
Yaw steering requires more analysis to prove safety as well, since you now have a bunch more nominal trajectories to simulate.

I'm not saying yaw steering is trivial, just that it's almost totally a software problem for a modern vehicle. And software can be an enormous pain and very expensive.
Chris  Whoever loves correction loves knowledge, but he who hates reproof is stupid.

To the maximum extent practicable, the Federal Government shall plan missions to accommodate the space transportation services capabilities of United States commercial providers. US law http://goo.gl/YZYNt0

Offline baldusi

  • Senior Member
  • *****
  • Posts: 8356
  • Buenos Aires, Argentina
  • Liked: 2539
  • Likes Given: 8273
Re: Yaw Steering
« Reply #6 on: 01/07/2017 10:39 pm »
I think Newton_V got it right. Yaw steering is a more difficult navigational problem, hoverslam RTLS is a more difficult control problem. Or said in other form, yaw steering is a more difficult "what to do" problem while RTLS is a more difficult "how to do it".
Clearly ULA has decades of working on the use requirement side of their avionics while SpaceX has been concentration on internal requirements.

Offline Kaputnik

  • Extreme Veteran
  • Senior Member
  • *****
  • Posts: 3079
  • Liked: 722
  • Likes Given: 821
Re: Yaw Steering
« Reply #7 on: 01/08/2017 09:57 am »
Is Atlas the only vehicle that can do yaw steering?
"I don't care what anything was DESIGNED to do, I care about what it CAN do"- Gene Kranz

Offline pippin

  • Regular
  • Senior Member
  • *****
  • Posts: 2575
  • Liked: 312
  • Likes Given: 45
Re: Yaw Steering
« Reply #8 on: 01/08/2017 10:55 am »
Shuttle did it, too

Offline Jim

  • Night Gator
  • Senior Member
  • *****
  • Posts: 37442
  • Cape Canaveral Spaceport
  • Liked: 21452
  • Likes Given: 428
Re: Yaw Steering
« Reply #9 on: 01/08/2017 01:17 pm »
Well, I'd say if you claim Yaw steering as a higher level navigational problem, and docking as a higher level control problem, then the hoverslam landing is both.

And that would be wrong.

The landing is fixed and known and never changes.  There is no computing to determine or find it.

And it is just not docking, it is so rendezvous.


Landing on a spot on the surface however is not as simplified. Even if you are in orbit,


No, that is not part of the discussion.  This is specifically about first stage RTLS.  The first stage flies on a trajectory that when the vehicle is operating normally (no engine problems), it will have enough propellant to return to the launch site.  It is purposely staged with that amount.


But then you have to deal with all the issues of unstable, under-controlled atmospheric flight, reentry, atmospheric uncertanities and wind from the moment of interface on,


The landing site is fixed and all that that requires is sufficient  propellant margin to cover dispersions.

and an advanced navigational problem.

Again, wrong.  The landing site is known and fixed.  It stored on board.


I'd say docking is the easiest of the problems
Followed by yaw steering
And when you have mastered both, you can join a harder league and try reentry and spot-landing in atmosphere.


Wrong.

 The docking target locationis unknown until the first sensor lockon.

Yaw steering is not a look up table, it is computed onboard

Landing is no different than achieving orbit.  It is just has tighter constraints.
« Last Edit: 01/08/2017 01:35 pm by Jim »

Offline Jim

  • Night Gator
  • Senior Member
  • *****
  • Posts: 37442
  • Cape Canaveral Spaceport
  • Liked: 21452
  • Likes Given: 428
Re: Yaw Steering
« Reply #10 on: 01/08/2017 01:44 pm »
BTW

we have been using the wrong term

Guidance refers to the determination of the desired path of travel (the "trajectory") from the vehicle's current location to a designated target, as well as desired changes in velocity, rotation and acceleration for following that path.

Navigation refers to the determination, at a given time, of the vehicle's location and velocity (the "state vector") as well as its attitude.

Control refers to the manipulation of the forces, by way of steering controls, thrusters, etc., needed to execute guidance commands whilst maintaining vehicle stability

With IMU's and GPS, navigation is easy. 

Guidance for yaw steering and rendevous and docking is harder, there are more variables.

RTLS has more external influences but that was just beefing up the control portion.

There is no active feed back with propellant utilization for RTLS. 

Offline Silmfeanor

  • Full Member
  • ****
  • Posts: 1254
  • Utrecht, The Netherlands
  • Liked: 403
  • Likes Given: 723
Re: Yaw Steering
« Reply #11 on: 01/08/2017 03:05 pm »
So far this discussion has been rather unfruitfull - instead of asking about Yaw steering, what it requires and which LVs did it, we get 'landing/docking is harder /easier than yaw-steering' combined with various opinions of SpaceX.

Those are other discussions.
I am wondering - just to see if I understand it - that Yaw steering requires just engine gimbals for steering during launch. Most, if not all, launch vehicles have that. The problem lies in the guidance package, perhaps combined with specific sensors.

Some vehicles have proven they can do it - thus, they have both working. Other vehicles have the gimballing, so the only other thing that should make yaw-steering work for them is guidance code, and / or some specific sensors.

Is that correct?

Online LouScheffer

  • Senior Member
  • *****
  • Posts: 3383
  • Liked: 6111
  • Likes Given: 837
Re: Yaw Steering
« Reply #12 on: 01/08/2017 03:26 pm »
But for SpaceX, it's almost surely software.  The first stage is solving convex optimization in real time, a much harder task than yaw steering.
Not true.  Targeting for yaw steering is harder.  Again, the landing pad is a static target and always be in the same place no matter what time it is launched.  Launching into a specific orbital plane at anytime within a launch window is much harder.
The Apollo-Soyuz rendezvous did this (launch anywhere with an 8 minute launch window) in the early 1970s.  This was implemented by the Apollo computer, which had 12,000 instructions per second.  Here's a description by one of the implementors.

Even further back, the Atlas-Centaur could do near-optimal yaw steering.  This was used on the HEAO missions, as detailed in Bilinear tangent yaw guidance:
Quote
This paper presents a parametric yaw steering law which has been used to provide closed-loop yaw guidance for the launch of the HEAO (High Energy Astronomy Observatory) satellite mission using the Atlas/Centaur launch vehicle. This bilinear tangent steering law provides near optimal yaw steering for maneuvers requiring insertion into orbits with a specified inclination and node. [...]  The flight computer implementation of these laws in a rotating coordinate system using real-time integration of the equations of motion is detailed. Explicit solution of the parametric guidance equations requires the inflight solution of (2x2) two-point boundary value problems in the pitch and yaw planes. Excellent results are obtained even for very large (greater than 50 deg) out-of-plane

And even further back, Gemini XII used lots of yaw steering, since they were rendezvousing with a target launched 90 minutes earlier.  From the Gemini 12 mission report
Quote
At spacecraft insertion, the range between Spacecraft 12 and the Gemini XII GATV was approximately 500 nautical miles, and the out-of-plane velocity error resulting from the spacecraft launch-vehicle ascent yaw steering was about 8 ft/sec.
This mission had a 33 second launch window.  So even if you can't update the solution for any time within a launch window, (even though Apollo and HEAO did this), you could make it even easier on the rocket avionics by pre-computing a number of trajectories (say 21 of them, 30 seconds apart, covering +- 10 minutes from nominal), then pick one once you decide where in the window to launch.  This reduces the problem to something that was solved by a 1960's flight computer.

Offline Jim

  • Night Gator
  • Senior Member
  • *****
  • Posts: 37442
  • Cape Canaveral Spaceport
  • Liked: 21452
  • Likes Given: 428
Re: Yaw Steering
« Reply #13 on: 01/08/2017 06:20 pm »

This mission had a 33 second launch window.  So even if you can't update the solution for any time within a launch window, (even though Apollo and HEAO did this), you could make it even easier on the rocket avionics by pre-computing a number of trajectories (say 21 of them, 30 seconds apart, covering +- 10 minutes from nominal), then pick one once you decide where in the window to launch.  This reduces the problem to something that was solved by a 1960's flight computer.

Not feasible unless does before terminal count. There needs to be time to load and verify the program.


Online LouScheffer

  • Senior Member
  • *****
  • Posts: 3383
  • Liked: 6111
  • Likes Given: 837
Re: Yaw Steering
« Reply #14 on: 01/08/2017 07:52 pm »
This mission had a 33 second launch window.  So even if you can't update the solution for any time within a launch window, (even though Apollo and HEAO did this), you could make it even easier on the rocket avionics by pre-computing a number of trajectories (say 21 of them, 30 seconds apart, covering +- 10 minutes from nominal), then pick one once you decide where in the window to launch.  This reduces the problem to something that was solved by a 1960's flight computer.
Not feasible unless does before terminal count. There needs to be time to load and verify the program.
Given that computers in the 1960s could load and verify at least one trajectory, a modern computer can surely load and verify all 21 (in this case) in the hours before launch.  Then it uses the one that corresponds to the selected launch time.

Given that yaw steering was developed and used, on manned mission, in the 1960s, and that the needed methods have been published, I believe the reasons for lacking it are bean counters, not technology.  Unless you have a mission that needs it to reach some otherwise un-available orbit, or some customer demands it, then the decision boils down to how often a few minute window would help, how much a delay until the next window costs, versus the cost to develop and certify.   

Offline Jim

  • Night Gator
  • Senior Member
  • *****
  • Posts: 37442
  • Cape Canaveral Spaceport
  • Liked: 21452
  • Likes Given: 428
Re: Yaw Steering
« Reply #15 on: 01/08/2017 08:59 pm »

Given that computers in the 1960s could load and verify at least one trajectory, a modern computer can surely load and verify all 21 (in this case) in the hours before launch. 

nope, they just load the one

Offline Jim

  • Night Gator
  • Senior Member
  • *****
  • Posts: 37442
  • Cape Canaveral Spaceport
  • Liked: 21452
  • Likes Given: 428
Re: Yaw Steering
« Reply #16 on: 01/08/2017 09:00 pm »

 I believe the reasons for lacking it are bean counters, not technology.  Unless you have a mission that needs it to reach some otherwise un-available orbit, or some customer demands it, then the decision boils down to how often a few minute window would help, how much a delay until the next window costs, versus the cost to develop and certify.   


Nope, it is avionics/software architecture.  Delta could not if they wanted to

Offline hkultala

  • Full Member
  • ****
  • Posts: 1199
  • Liked: 748
  • Likes Given: 945
Re: Yaw Steering
« Reply #17 on: 01/08/2017 09:04 pm »
Given that computers in the 1960s could load and verify at least one trajectory, a modern computer can surely load and verify all 21 (in this case) in the hours before launch.  Then it uses the one that corresponds to the selected launch time.

What makes you think any rocket (except Falcon 9) today is using a modern guidance computer?

Online LouScheffer

  • Senior Member
  • *****
  • Posts: 3383
  • Liked: 6111
  • Likes Given: 837
Re: Yaw Steering
« Reply #18 on: 01/08/2017 10:35 pm »
Given that computers in the 1960s could load and verify at least one trajectory, a modern computer can surely load and verify all 21 (in this case) in the hours before launch.  Then it uses the one that corresponds to the selected launch time.
What makes you think any rocket (except Falcon 9) today is using a modern guidance computer?
Well, "modern" depends on your reference.  The Delta RIFCA uses triple MIL-STD-1750A processors.  Compared to the computers in the 1960s Titan, or those in Apollo (both of which could do yaw steering), that's a modern processor, with enormous improvements in speed and capacity.   But compared to what's in the Falcon, it's a slow and memory-limited antique.

Online meberbs

  • Senior Member
  • *****
  • Posts: 3096
  • Liked: 3379
  • Likes Given: 777
Re: Yaw Steering
« Reply #19 on: 01/08/2017 11:36 pm »
The landing is fixed and known and never changes.  There is no computing to determine or find it.
...

Again, wrong.  The landing site is known and fixed.  It stored on board.

...

 The docking target locationis unknown until the first sensor lockon.

Yaw steering is not a look up table, it is computed onboard
Landing cannot be done from a purely pre-computed path, because it has to account for perturbing forces, some of which happen during "engines off" periods. And just like docking, additional sensors (e.g. altitude sensors) are needed to dynamically compute the final solution, because the exact relative location of the target is not known beforehand. (You seem to think that because the landing location is fixed relative to the surface of Earth it is somehow easier for the rocket to know its relative location to the landing site, but this is not true)

Yaw steering has to deal with the atmosphere on the way up, just as any other launch, so it can only be done from pre-computed tables as much as a direct launch would (I don't know to what extent they do). If regular launches can use tables, then so can yaw steering. Your argument that "nope, they just load the one" is meaningless because just because they don't, doesn't mean they couldn't. You would have to provide a reason they couldn't, such as if Delta doesn't have enough RAM. I doubt that would be the case for a Falcon.

Yaw steering doesn't have to use tables, it can be computed onboard as you said. Either way this still only* is a software functionality, you have yet to provide one bit of hardware that would need to change for a Delta or a Falcon.

*This is not to say that software is trivial. Not wanting to do extra software development/testing is plenty of reason to not implement yaw steering unless it is really needed.

Tags:
 

Advertisement NovaTech
Advertisement Northrop Grumman
Advertisement
Advertisement Margaritaville Beach Resort South Padre Island
Advertisement Brady Kenniston
Advertisement NextSpaceflight
Advertisement Nathan Barker Photography
1