Planning Suicide Burns with Manoeuvre Nodes - Why does it not work?

[Andersenman]

Hello,

I am trying to perform suicide burns in a stock game. I figured, if I place the node right onto the surface and pull retrograde, it would eventually show me the burn time needed to stop whatever speed I would have where my original orbit would have intersected with the surface; but instead of performing the burns half-half around the node, lithospherical reasons would dictate to perform the entire burn before the node.

This is (generally) what I'm trying to do.

However, this doesn't work and I don't understand why. I always stop much earlier, canceling ALL orbital speed far above the ground. The discrepancy is larger the higher my vertical speed at the node, and possibly the more my current speed differs from the speed at the node; ie. it appears to be worst when I'm coming straight down. Also, it is always much larger than the orbital speed on the surface due to rotation.

Please help me find the mistake in my reasoning.

Many thanks!

A.

[Bobnova]

The node calculator burn time doesn't seem to take into account the increasing TWR as you burn fuel. If you actually time a burn that it says will be 10 minutes it'll take less time. The larger the percentage weight loss due to burnt fuel, the larger the time difference.

Beyond that, I don't know.

[stoani96]

That is because your average speed is half the initial speed

(initial speed - endspeed)/2; endspeed is zero in your case.

an simple example: your initial speed is 100m/s, the node tells you, that you would need 10s to zero your velocity(a=10ms^-2).

When the node tells you, you will be at it in 10 seconds, the way you will cross equals speed*time=100m/s * 10s = 1000m = 1km

if you start you burn here(100%), your speed will be at zero after 10s and the way you crossed equals average_speed*time (above the average speed equals the initial speed)

the average speed is half the initial speed(see above), so s=500m and you are 500m above the ground or whatever!

i hope its understandable...

EDIT: what Bobnova says does also effect this, of course

EDIT2:

a=thrust/mass - gravity of body

t=velocity/a <--------speed difference(in our case the initial velocity)

s=(a*t*t)/2

=> TimeBeforeNode=s/initial_speed

i think that should work....

No time to proof it, if anyone does, please tell me

EDIT3:

corrected the formulas

Edited January 5, 2014 by stoani96

[MBobrik]

maneuver nodes don't have that precision to begin with. a few meters off in either direction and you end up lithobraking.

[kzauner]

Hello,

I am trying to perform suicide burns in a stock game. I figured, if I place the node right onto the surface and pull retrograde, it would eventually show me the burn time needed to stop whatever speed I would have where my original orbit would have intersected with the surface; but instead of performing the burns half-half around the node, lithospherical reasons would dictate to perform the entire burn before the node.

A.

You are deceleration, thus having a longer and longer flight-time until you reach node (on the ground). The resulting function isn't linear. Mathwise OTOH try aproximating by 2/3 of the suggested burn-time (it should be around 62%, IMO). But I'm not taking the increasing TWR into account, nor your trajectory over ground (and gravitational acceleration).

I'm too lazy to calculate this throughoutly.

[Andersenman]

The node calculator burn time doesn't seem to take into account the increasing TWR as you burn fuel.

But if it didn't, why does the half-half burn work for other orbital changes? If the burn calculator couldn't work with variable TWR, the actual burn time would ALWAYS be less than what was calculated. But that's not the case: When the node says, "Burn X seconds", the orbital change is done after X seconds regardless how much TWR changed during burn - if at all, since it works with Infinite Fuel, too.

because of a=v/t and v=s/t:

a=s/t^2

a=thrust/mass - gravity of body

t=velocity/a <--------average speed!

s=a*t*t

=> TimeBeforeNode=s/initial_speed

But don't those apply to movement in straight lines in a force-free space only?

maneuver nodes don't have that precision to begin with. a few meters off in either direction and you end up lithobraking.

Starting to brake early should solve that. I don't mind suiciding to a few metres above ground. Plus, since I have always been coming in short so far this hasn't become a problem yet.

You are deceleration, thus having a longer and longer flight-time until you reach node (on the ground). The resulting function isn't linear. Mathwise OTOH try aproximating by 2/3 of the suggested burn-time (it should be around 62%, IMO).

I know, but again, shouldn't the node system already be factoring this in when giving me a simple burn time for my current craft and the point on the current orbit?

[stoani96]

But don't those apply to movement in straight lines in a force-free space only?

youre right, the formula was incorrect!!

But i already corrected it, and it should work just fine...

but add 2seconds or so to get the thrust to 100% and other incorrectnesses.

Force-free.... what do you mean with that? a force ends up in an acceleration (force/mass)

[Kasuha]

Assuming the burn time estimate is as correct as usual, 2/3 of the estimated time ahead is usually good.

However, suicide burns are not very safe way of landing. Particularly because they rely on you starting the burn at the perfect moment. If you're moving at 800 m/s, starting 0.1 s early will dump you 80 m above ground. Starting 0.1 s late will crash you with no chance to survive. And they often involve you landing at a steep angle which is also ineffective unless you are trying to hit one exact point on the surface.

The following video by Kosmo-not describes about the most efficient and safe way of landing. It was recorded on old version of KSP when Mun was significantly flatter than today, but it does not change much on the approach - you just need to start with a bit higher periapsis and keep your vertical speed a bit higher, but still under control while watching for terrain.

Edited January 5, 2014 by Kasuha

[stoani96]

Double post!

Sry

Edited January 5, 2014 by stoani96

[toric5]

are you shire thats an efficient way to land? he spent almost 7 minutes burning nonstop. almost seems like a drop- scuicide burn would be better.

[Kasuha]

are you shire thats an efficient way to land? he spent almost 7 minutes burning nonstop. almost seems like a drop- scuicide burn would be better.

He spent 7 minutes burning nonstop with TWR hardly above 1. If it was a suicide burn, it'd be a 7 minute burn, too - just with the danger that if you started 0.1 s too late, it'll embed you deep into the surface.

[numerobis]

Isn't Kosmo-not's TWR below unity at the start? Regardless, it's the efficient method.

The burn time counter in KSP seems to give me garbage a large fraction of the time. If you need precision, calculate the burn time yourself.

[stoani96]

are you shire thats an efficient way to land? he spent almost 7 minutes burning nonstop. almost seems like a drop- scuicide burn would be better.

of course a "suicide burn" is more efficient!

Just think logically!

The longer you fight gravity, the more fuel you will waste. So if your landing burn is short, the percentage of wasted fuel is less.

And a short landing burn is, of course, a single one with 100% of thrust ---> and i think that is called "suicide burn" (?)

A big disadvantage is, that it is pretty risky.

What i do:

I try to burn one time with ~90%, so i will never crash into the ground(just more thrust) and i also can go back with the thrust so i get down the fastest way

[stoani96]

Isn't Kosmo-not's TWR below unity at the start? Regardless, it's the efficient method.

The burn time counter in KSP seems to give me garbage a large fraction of the time. If you need precision, calculate the burn time yourself.

thats unusual... i never had problems with the burning time...

can you report about some detailed cases?

[Ravenchant]

I'd say this video (by Kosmo-not as well) illustrates it a bit better, and seems to be done stock. Both use the same procedures, but the final approach phase in his older video is really drawn-out and less feasible on the new Mun...unless your landing site is completely flat

A good compromise seems to be to "drift" a few hundred metres in the "air" and slowly reduce your horizontal velocity while keeping altitude more or less constant, then kill your horizontal once at, for example, at 50-100m from ground and 20m/s surface speed. The resulting drop shouldn't lose you too much fuel to fighting gravity, and it's useful when there's a base standing around

[Kasuha]

of course a "suicide burn" is more efficient!

Just think logically!

The longer you fight gravity, the more fuel you will waste. So if your landing burn is short, the percentage of wasted fuel is less.

And a short landing burn is, of course, a single one with 100% of thrust ---> and i think that is called "suicide burn" (?)

I used to think so, too. Then I actually tried that. Yes, it defies logic.

Actually I still think a "perfect" suicide burn is more efficient than this method. But you can't pull even average suicide burn most of the time, and "perfect" means setting up all orbital parameters at optimal values together with starting the burn at exactly the right time. So perfect suicide burn in a game like KSP is more of a pipe dream than real possibility.

Also there are situations where you just can't perform a suicide burn. If your orbit is too low, suicide burn crashes you into the ground before you manage to kill your orbital speed.

And finally - if you look at the video, he's thrusting at 100% all the time except the very end. So he is doing optimal maneuver by your own definition. He has an inefficient rocket, but that is the matter.

Moon landings actually used the same maneuver, too.

[Captain Sierra]

I personally always cancel my horizontal velocity with some vertical room to spare. Then I drop down and make several short burns, never coming to a complete stop (just to guage my acceleration) and eyeball it from there. I almost always settle less than 100m up but well over 50, plenty of room for error. Then I ride the engine down on ~5% thrust to a gentle landing. I find it's easier to cancel one velocity vector component at a time rather than fight both. Also, drop landings are a heck of a lot more precise if you don't have lots of practice with the 'proper' method.

[stoani96]

I used to think so, too. Then I actually tried that. Yes, it defies logic.

Actually I still think a "perfect" suicide burn is more efficient than this method. But you can't pull even average suicide burn most of the time, and "perfect" means setting up all orbital parameters at optimal values together with starting the burn at exactly the right time. So perfect suicide burn in a game like KSP is more of a pipe dream than real possibility.

Also there are situations where you just can't perform a suicide burn. If your orbit is too low, suicide burn crashes you into the ground before you manage to kill your orbital speed.

And finally - if you look at the video, he's thrusting at 100% all the time except the very end. So he is doing optimal maneuver by your own definition. He has an inefficient rocket, but that is the matter.

Moon landings actually used the same maneuver, too.

i did not watch any of the videos

Of course its more precise to kill horizontal speed first! I also try to combine both: killing both vectors at the same time with 90-100%, but the horizontal one a bit more than the vertical one ^^. after killing the horizontal i cut off the thrust for a little moment, get into position and just go on killing my velocity.

(Wow O.o, so much killing in here )

But i think we are coming from the subject here

[Kasuha]

Well... anyone who thinks suicide burn is more effective than the horizontal landing, we can always try the old challenge:

Make a new save and download [this persistent.sfs] file into it. Then open the save, switch to the ship land the way you think is the most efficient and show your remaining liquid fuel.

Old results are probably not relevant anymore as the command pod is now heavier by the monopropellant, but we can always run it again.

Edit: my result - 33.46 LF units left

Edited January 5, 2014 by Kasuha

[Rhomphaia]

To get back to the original discussion.

The maneuver node system is calculated from an instantaneous burn with the time calculated from your TWR. So in your example you set up a maneuver node to kill off all your 868.5m/s orbital velocity in 2m9s. at the point you start burning 35 seconds before the node, you are further from periapsis so your orbital velocity is lower. The steeper your descent the more noticeable this will be.

Basically maneuver nodes are not all that useful for this.

[stoani96]

That is because your average speed is half the initial speed

(initial speed - endspeed)/2; endspeed is zero in your case.

an simple example: your initial speed is 100m/s, the node tells you, that you would need 10s to zero your velocity(a=10ms^-2).

When the node tells you, you will be at it in 10 seconds, the way you will cross equals speed*time=100m/s * 10s = 1000m = 1km

if you start you burn here(100%), your speed will be at zero after 10s and the way you crossed equals average_speed*time (above the average speed equals the initial speed)

the average speed is half the initial speed(see above), so s=500m and you are 500m above the ground or whatever!

i hope its understandable...

EDIT: what Bobnova says does also effect this, of course

EDIT2:

a=thrust/mass - gravity of body

t=velocity/a <--------speed difference(in our case the initial velocity)

s=(a*t*t)/2

=> TimeBeforeNode=s/initial_speed

i think that should work....

No time to proof it, if anyone does, please tell me

EDIT3:

corrected the formulas

Its like with a car:

you are driving, lets say, 20m/s and you know you need, lets say, 5s to break. You will not travel all the 100m you would if you would go on driving with this speed! U know what i mean?

http://en.wikipedia.org/wiki/Acceleration

Of course the node is not 100% correct and the speed is also not the same as at the node itself, but the main porblem is this!

http://lmgtfy.com/?q=braking+distance

[Scarecrow88]

are you shire thats an efficient way to land? he spent almost 7 minutes burning nonstop. almost seems like a drop- scuicide burn would be better.

Ultimately you need to kill all your horizontal velocity before touchdown, whether you do a suicide burn or not. By setting your periapsis just above the surface, using the approach shown in the video, by the time you have killed your horizontal velocity, you are not far off the ground, and don't need to burn a shed load of fuel fighting gravity.

[stoani96]

a true word!

[stoani96]

Well... anyone who thinks suicide burn is more effective than the horizontal landing, we can always try the old challenge:

Make a new save and download [this persistent.sfs] file into it. Then open the save, switch to the ship land the way you think is the most efficient and show your remaining liquid fuel.

Old results are probably not relevant anymore as the command pod is now heavier by the monopropellant, but we can always run it again.

Edit: my result - 33.46 LF units left

http://i.imgur.com/9Tu9gMO.png

Just tried it once (and landed at the slope ) and had 30 units left.

it was not a perfect landing, but (especially for that TWR and Landing orb) a pretty good one.

[Andersenman]

Many thanks for all your input. I will try some more landings, including said old challenge to get some comparable results.

Safe landings!

A.