Don't Just Burn Prograde! (Or: How I Learned To Stop Worrying And Burn Radial Too)

[Superfluous J]

After having this discussion with @herbal space program regarding the maneuver nodes and whether or not they accurately predict where you will end up (I still think they do), I decided to test something that I'd kind of taken for granted since starting KSP2:

On long ejection burns, is it actually most efficient to burn prograde the whole way?

Why I started to doubt this is because of how KSP1 handles maneuver nodes. Because the node is instantaneous but your ship is not, you end up burning a portion of your burn "radial in" for a while at the start of your long burn, then as you pass Periapsis you start to burn a portion "radial out." In KSP2, though, the entire burn is Prograde, and what is Prograde changes over the course of the burn.

So I set up a simple test. I got a craft with a poor engine and a lot of fuel into a 100km orbit and set up an ejection burn to 50Mm (halfway between Dres and Jool). I only picked this number because it was reasonably large and reasonably easy to remember, and also caused my ship to need to do about a 7 minute burn. Here is the projected orbit, and the projected ejection, of that burn:

Note it takes 2050m/s and a large portion of the burn is going "up" away from Kerbin, due to prograde changing over time.

Note also that I set the time of burn start to be the exact time in which the Ap over Kerbol was the highest. This is THE MOST EFFICIENT all-prograde burn with this rocket to get to 50Mm (barring gravity assists, I suppose )

                                                                                                                             

I then went crazy, and did something I figured would not work at all: I set up another node with EQUAL AMOUNTS of Prograde and Radial In. I mean, if you're going to do it go crazy, right?

Note I also changed the time of the burn to the ideal time, so the ONLY change is the direction of the burn. Again, this is the MOST EFFICIENT way to get to 50Mm around Kerbol while burning radial and prograde equally, with this rocket.

Note: THIS WAY IS BETTER! It only takes 1818m/s (a savings of 12% or so) and the burn is similarly shorter. Just to make sure, I did this burn and the ship ended up in the projected orbit.

                                                                                                                             

I fiddled with the ratios, and really couldn't get it much better. There's surely an ideal ratio of radial to prograde (which likely changes per ship, ejection amount, and all other variables) but I think a first-off estimate of "equal radial and prograde" may always be better than "just go prograde" for any long burn.

Here's one that's about 3/2 prograde/radial. note it's actually worse than the 1/1 ratio, but not by an amount that is worth worrying about:

What does this mean to you? I don't know. But it means to me that if I'm burning more than a minute or two, I think I'm going to give it a shove radial to see if that's better.

(Note: This should also work for INJECTION burns as well, only they should be retrograde and radial OUT)

[herbal space program]

I appreciate what you're doing here, but for the record, if you look down-thread in the linked discussion, you'll see that I already retracted my initial statement about how the KSP2 maneuver node behaves, although it still doesn't do what it's supposed to do perfectly IMO. I also stated in that thread  (as you should know) that  for an optimal ejection burn using this new system,  especially if it is a longer one, you will have to add a radial inward component to your maneuver node, as you have shown here. I still however stand by my contention that at least for advanced levels of inter-body navigation, especially if multiple periapsis kicks are required, this system is not as easy to work with as the KSP1 system for generating the most efficient possible burns. That's because if you really know how to use the KSP1 node system, you can figure out how to closely approach the optimal burn profile considerably more easily, because 1) you don't have to fiddle about a bunch to optimize how far inward of prograde you need to aim at the outset to put your burn vector on the optimal secant to approximate an instantaneous impulse at the best spot, and 2) If you need to do multiple periapsis kicks, the KSP2 node is really unhelpful in terms of where to place those, whereas the KSP1 node pretty much tells you exactly where that should be right off the bat. As to why I made my original, incorrect statement of what the KSP2 node does, I think it's because it gets confused and doesn't really give you the right trajectory in some situations,  like space planes that have multiple parallel engines that can be toggled on and off for very different TWR (my most common vessel), and burns where two stages with widely disparate TWRs are used sequentially. At any rate, in order to dispel any notion that I just live to disparage KSP2, I will see what I can do to adapt my  navigational approach to the new system and report on what I find here.  I will also investigate what situations might confuse the trajectory prediction algorithm. And AFAICT that maneuver node controller you use above is not a part of stock, so I will not make use of that.

...And the last thing I'll say, before coming back with some documentation of these efforts, is that the Devs could instantly make my problem as a player with ingrained KSP1 habits go away by allowing us the option to arbitrarily set the TWR when placing nodes, rather than only using whatever value it decides. Then I could set it to infinity to determine the ideal instantaneous impulse and multi-kick burn center, ala KSP1, and also to its actual value according to my specific configuration to determine my trajectory after any particular burn.

Edited March 1, 2024 by herbal space program

[herbal space program]

...So what I've come up with to help me plan efficient multi-kick burns is the following: 

1) Put  a bare bones probe, with very high (>=3) initial TWR and enough dV to reach any intercept in the Kerbolar system, on a  circular 100km parking orbit.

2) Use this probe, of which my version can do a 2500 m/s burn in 44 seconds, as a tester to plot the whole orbital transfer as if it were a KSP1 node, since at that high TWR it's essentially equivalent to one. 

3) Place your long-range  mission with its pokey transfer stage on the same parking orbit as the tester craft.

4) Start planning the real maneuver by placing the node for your first PE kick at the spot where the midpoint  of the  high TWR test burn would have been.

5) Pull on the prograde handle at that spot until you have a short (<15s) burn set up,  then drag your node backwards along the orbit to place the midpoint of that burn where its origin had been before.

6) Execute that short burn. Your resulting periapsis will now be at a good approximation of where your ideal periapsis for a near-instantaneous burn would have been.

7) For subsequent, longer PE kicks, start your node half of the desired burn length before your PE, then drag on the prograde handle until your burn is bisected by your current PE.

8) From that position, you will see that your projected new PE is closer to your node than your current PE is.  Drag on the radial inward handle of your node until your new PE is again on top of the old one.

9) Execute the burn from there, which should leave your PE in the same spot it was in before, and then just go back to step 7 for all subsequent burns.

This series of steps worked very well for me going to Jool  with a transfer stage that had a TWR of 0.3, resulting in a total dV expenditure of only 50-70 m/s more than the nominal 1,989 m/s dV required for the ideal burn as plotted here:  https://alexmoon.github.io/ksp/#/Kerbin/100/Jool/100/true/optimalPlaneChange/false/1/1.  Not bad! I can definitely live with that.

Edited March 8, 2024 by herbal space program

[Superfluous J]

My current plan is:

1. Make a node for your ship using the maneuver tool (aka the Maneuver Node Controller mod) 100% prograde.
2. If the burn is less than 2-3 minutes, do that burn and be happy.
3. If the burn is 8+ minutes, consider doing a Pe kick or redesigning the craft because no thank you.
4. Between 3-8 minutes, skew your burn radial inward, removing some prograde to compensate. At 8 minutes, it's just about 50/50 prograde/radial. at 2 minutes, it's basically all prograde. Each minute should be 20% more radial, I guess. Haven't gotten that far into it and doubt it's needed.
5. You'll need to adjust the time to burn and the total amount of the burn iteratively to find the most efficient burn, as well as the prograde/radial ratio.
6. Other planets and orbits have different times. I am 99% sure it has more to do with the percentage of an orbit you are burning than "raw time" or anything like that.

Oh and another thing I've learned that's not strictly related to this but will affect test results: If you stage mid-way through a burn the game recalculates where to aim, which makes the path you travel after that point INCORRECT. So if you start the burn aimed at the maneuver node, you should take that aiming off (hit T twice to toggle the computer) before you stage. I consider this a bug and would report it but I hate the bug forum and don't want to put the time and effort into reporting a bug, only to find out it was reported 8 months ago and I just couldn't find it with the garbage search tools. I miss ksp1's bug reporting page. But that's a rant for another thread.

Edited March 8, 2024 by Superfluous J

[herbal space program]

41 minutes ago, Superfluous J said:

If you stage mid-way through a burn the game recalculates where to aim, which makes the path you travel after that point INCORRECT.

Aha! I knew there was something about staging in the middle of a burn that threw it off.  Well done!

[BechMeister]

I often fly vehicles that are "underpowered" or have long burn times - I have in my mission blogs been asking question about the efficiency and fuel usage when you burn and I could see that by the end of the burn I was burning "up" a lot (as you mentioned). I did not know any other way... But I guess this also explains why I tend to need about 100-200Δv more than what the Δv maps say - In the beginning I was just assuming that my Δv readings were wrong... and was considering doing the rocket equation to be sure..

But just to be sure I understod it right (and pardon me for using rookie terms):

1. You start the burn before the optimal position on PE. place PE on the halfway point of your burn?

2. You then burn prograde and down until you reach the mid point. After which the maneuver node will start burn you Up.

But because you start by burning down, less Δv is spend burning up, and your average burn time is more prograde than if you just tell the maneuver node to burn prograde. Correct?  

Edited March 25, 2024 by BechMeister

[Superfluous J]

54 minutes ago, BechMeister said:

1. You start the burn before the optimal position on PE. place PE on the halfway point of your burn?

Pretty much, except this one is true of all burns. You ALWAYS start "before" the "ideal" point - even in KSP1. Just in KSP2 you place the maneuver node at the START of the burn, instead of (as in KSP1) somewhere in the MIDDLE of the burn.

55 minutes ago, BechMeister said:

2. You then burn prograde and down until you reach the mid point. After which the maneuver node will start burn you Up.

I'm not 100% sure where the point where you're slightly burning down levels off and you start slightly burning up, but yes that will happen somewhere in the middle of the burn.

Basically this method give you what we lost from KSP1's nodes without losing what we gained from KSP2's nodes.

KSP1's nodes aim you (about) the right direction, but you have to guess when to burn (usually 50/50 is fine but not always) and your final path won't match the planned line exactly (but it will be close enough usually).

KSP2's nodes tell you exactly when to burn and what will happen when you do, but you have to figure out the correct direction yourself (which is what this thread is about).

 

[BechMeister]

1 hour ago, Superfluous J said:

Pretty much, except this one is true of all burns. You ALWAYS start "before" the "ideal" point - even in KSP1. Just in KSP2 you place the maneuver node at the START of the burn, instead of (as in KSP1) somewhere in the MIDDLE of the burn.

I must admit that when I plan my maneuvers.. I often just pull the prograte node until I have a path that intercept the object I desire to transfer to.. Then I move the node around on my orbit until I have a interception. And then I use the Normal/Anti normal and or Radial in/out to adjust where i land on the other bodies SOI... So I am very dependent on the nodes.. Its only if I have to circularise or correct inclinations that I eye ball it.

But I guess I can almost do the same on this approach. Just make sure that the initial prograde arrow pull is also pulled down a bit - As seen on this screenshot:

On 2/29/2024 at 9:48 AM, Superfluous J said:

If I should put myself into the spotlight... I can tell you that the maneuver I did which I questioned was the "right" way to do it.. But I guess the only way I knew how to do it... Looked like this: The quote is from my "blog" about building two space stations (one around the Mun and one around Kerbin)

On 12/6/2023 at 5:59 PM, BechMeister said:

The refueling went smooth this time - and the H.T.D. (Heavy Tanker Drone) returned to K.G.01 (Kerbin Gate 01 - my space station) with 94 Δv left - a pretty tight margin, Just how I like it.

I was originally a bit afraid that the LV-1000 engines were "too" fuel efficient, in the sense that the fact it took them 5 min to do the transfer burn to The Mun - ment that they were burning the craft 30° off the prograde.

Here is a reference photo of the 30° difference between being on course and prograde.

Maybe it's just because my understanding is flawed? But I imagine that it is not very efficient to burn 30° off prograde to stay on course - a lot of that fuel is spend staying on track rather than pushing the Apoapsis right? 

I hope its okay I include an example from my own mission report... I just remember specifically asking this out into the void, not getting a correction, and assuming that was how it was done x)

[Fizzlebop Smith]

On 3/7/2024 at 6:05 PM, Superfluous J said:

Another thing I've learned that's not strictly related to this but will affect test results: If you stage mid-way through a burn the game recalculates where to aim, which makes the path you travel after that point INCORRECT. So if you start the burn aimed at the maneuver node, you should take that aiming off (hit T twice to toggle the computer) before you stage. I consider this a bug and would report it but I hate the bug forum and don't want to put the time and effort into reporting a bug, only to find out it was reported 8 months ago and I just couldn't find it with the garbage search tools. I miss ksp1's bug reporting page. But that's a rant for another thread.

Thank you for this. I have had a very difficult time trying to determine how to counteract this.
I had no idea that turning SAS on would call for an immediate calculation adjustment. Makes sense with he CPU attempting to simulate pilot skill.

So many long,  multi stage maneuvers have resulted in horrible inaccurate end destinations. I never could determine a definitive way to combat this. 

I usually stop burning for a short second, use MNC to adjust a variable of my end point by 1/ms and make sure 'node' is still selected.

Like @BechMeister  I "feel" like something is off whenever i make a long burn. I have learned to correct things by watching my DV calcs and playing around with the nodes. 

Whenever i would bring questions regarding node/ burn executions & use my experience with KSP1 to frame the question, I would be told its 'better' without being told why.

There is a tendency for some of those without years of love for KSP1 to view that portion of the community as antiquated or 'out of touch'. My concerns and questiosn would be dismissed out of hand without any of these wonderful visual aides & articulate responses. I like reading these tests that you all conduct.. and especially enjoy the discourse that follows.
 

[herbal space program]

On 3/25/2024 at 1:40 AM, BechMeister said:

I often fly vehicles that are "underpowered" or have long burn times - I have in my mission blogs been asking question about the efficiency and fuel usage when you burn and I could see that by the end of the burn I was burning "up" a lot (as you mentioned). I did not know any other way... But I guess this also explains why I tend to need about 100-200Δv more than what the Δv maps say - In the beginning I was just assuming that my Δv readings were wrong... and was considering doing the rocket equation to be sure..

But just to be sure I understod it right (and pardon me for using rookie terms):

1. You start the burn before the optimal position on PE. place PE on the halfway point of your burn?

2. You then burn prograde and down until you reach the mid point. After which the maneuver node will start burn you Up.

But because you start by burning down, less Δv is spend burning up, and your average burn time is more prograde than if you just tell the maneuver node to burn prograde. Correct?  

Yes, that's about right AIUI.