Efficient landing procedure?

[qoonpooka]

I saw it mentioned in another thread that landing is super inefficient and can consume upwards of 2-3x the dV given on a dV map.

That seems like it makes dV maps basically useless. How do we trim as much of that manuvering loss as possible while still having a reasonable margin for safety and surprise terrain issues?

Edited November 7, 2015 by qoonpooka

[Oafman]

The most efficient landing is a 'suicide burn'. One single burn, started as late as possible, to bring your vertical velocity to zero at the exact point that you reach the ground.

Of course, if you're doing a manual landing, starting this burn as late as possible is very risky (hence the name), because there's the chance you start it too late, and still have remaining vertical velocity by the time you reach the ground, leading to often catastrophic lithobraking (a crash).

Imagine you start your descent burn 30 seconds before the latest possible point. If you keep burning, you'll kill your vertical velocity to zero 30 seconds before you needed to. So you'll be briefly hovering at some altitude above the planet, before gravity starts to accelerate you downwards again. Then you'll need to burn more to counter this effect of gravity, and that extra burning is where the inefficiency comes in.

Realistically, you'll burn for a while, realise you're slowing down your vertical velocity too quickly, and ease off the throttle to allow gravity to increase your vertical velocity. That's fine, everyone does it, even Mechjeb to an extent. But minimising the need to do that is key to maximising your landing efficiency.

[Starhawk]

Well, it depends how you land. I find landing generally inefficient compared to ascent, but not twice as much. Or, Kraken forbid, three times as much.

I often find that if I have a 20% safety margin compared to the maps I can land quite comfortably. Unless something goes wrong.

I'm not usually one to gainsay Slashy, but twice as much dv as the map says seems like a lot. OTOH, if you use the 'burn retrograde until stopped and then fall down 20 km to the surface' method then I suppose you could spend twice as much as listed.

Happy (and efficient) landings!

Edited November 5, 2015 by Starhawk
clarity

[Retread]

Great topic! I haven't seen much discussion of this before, and I've often wondered how others optimize landings.

Theoretically, I believe the most efficient landing is essentially a gravity turn in reverse. All thrust takes place exactly retrograde, with the retrograde vector smoothly descending from horizontal to vertical as you burn, timing the burn so that your velocity relative to the surface reaches zero just as you touch down. In practice, the burn would need to be timed and throttled almost perfectly, or else you would reach zero velocity somewhere above the surface, requiring more dV to land, or below the surface, with obvious severe negative consequences.

My landings are typically broken down into a deorbit phase and a descent phase.

In the deorbit phase, my goal is to reduce my Pe a point low enough above the surface that I can begin the descent phase. I burn retrograde briefly at the beginning to lower my Pe, and then again shortly before the intended landing point to reduce my trajectory closer to vertical, beginning the descent. Because both burns are retrograde, I don't think there is too much inefficiency in the deorbit phase.

During the descent phase, burns have two purposes - the first is to reduce my speed enough to land safely, the other to fine tune my landing location. Any dV spent fine tuning the location is going to be over and above that necessary to land, and this can be reduced by making the initial deorbit burn more accurate. Most dV is wasted because I try to give myself enough time to respond as the landing develops, limiting my vertical speed to something I'm comfortable with. Ideally, you would do a suicide burn as above.

Since almost all of the dV losses happen during the descent phase, I try to minimize how long this phase is by bringing the Pe as close to the surface as possible during the initial deorbit burn. When landing on the Minmus flats, for example, I burn retrograde on the opposite side of the moon from my intended landing so that my Pe decreases from 20km to 3.5km, just enough to skim the mountains near the landing spot.

[Alshain]

GoSlash27's "Reverse Gravity Turn" is the most efficient non-suicidal method. I do it a bit differently but the concept is the same.

http://forum.kerbalspaceprogram.com/threads/104638-Reverse-gravity-turn-landing-technique-for-airless-bodies

[Clipperride]

IF you don't mind using quicksaves you can experiment with "suicide burns". The way I do it is I) make a named save while still in a nice safe orbit, so you can revert to that save if it turns out you don't have enough fuel for the ascent later on. II) make a second named save after you have done the deorbiting burn so you can start experimenting. III) decide on an altitude to start your burn (this gets easier with practice) as close to that altitude as possible, by applying full throttle.

If your burn brings you to a stop short of the surface, reload and start the burn a little lower likewise, if you hit the ground, note the altitude of the local terrain and try again from higher up. If you make multiple landings on the same body, looking back at previous burns will give you a good starting point for your next attempt.

Happy landings.

[Snark]

Back to the original poster's questions:

The dV maps aren't "useless"-- think of them as the price of admission. They tell you the necessary minimum, but "necessary" isnt' the same as "sufficient".

It takes a certain minimum dV to transfer from orbit around one planet to orbit around another planet.

For a planet without atmosphere, it takes a certain minimum dV (i.e. the planet's escape velocity) to go from "enter the SoI" to "be on the surface".

The reason why dV maps can't give you the "extra" part that you're complaining about is because it depends on ship design. So all you can do is use the dV map as a guide for a starting point, and then add on an "appropriate" amount of extra based on your ship design. And design your ships for the most efficient landing possible.

What "efficient" means will be situation-dependent:

• For thick-atmosphere worlds (Kerbin, Eve, Laythe): aerobrake, then use parachutes to land (or make an airplane-style landing, for spaceplanes). Requires zero dV (though depending on your arrival speed at the planet, may require you to spend significant mass on a heat shield).
  
• For thin-atmosphere worlds (Duna): aerobrake, then use some combination of drogue chutes and standard chutes to slow to a few dozen meters per second. Then use a quick blip of rocket thrust to slow to a survivable speed just before impact. (Or make an airplane-style landing, if you've got enough wing surface.) Requires minimal dV.
  
• For vacuum worlds (most places in KSP): set a periapsis as low as possible without scraping your toes, then right at periapsis do a deorbit burn, then suicide-burn to the surface. This uses scads of dV, but there's no way around it. Ship design is a compromise: high TWR makes for more efficient suicide burns (and more efficient takeoff afterwards, if it's a return mission), but also means you're lugging around heavier engines. So it's a compromise between engine mass, Isp, and TWR.
  

[ElWanderer]

I saw it mentioned in another thread that landing is super inefficient and can consume upwards of 2-3x the dV given on a dV map.

I think 2-3x times is a Major Exaggeration (*salutes*) but landing on an atmosphere-less body is the one place where you definitely want a comfortable fuel margin. When I visited the moons of Jool, my landings were all just inside the figures quoted by the delta-v map, except for Tylo where some of my early attempts failed due to running out of fuel, even though I had 3800m/s to play with (the others failed due to hitting the ground at 500+m/s). Tylo is hard, especially if you don't have an efficient landing technique. I lacked that at the time and had no idea when to burn, but came back with a constant altitude landing* having read about that online and nailed a landing first time for about 2700m/s. I think you can theoretically do it with under 2300m/s but I was happy enough with that.

.

* set a periapsis as low as you dare (helps to know roughly how high the terrain is in the area you're aiming for), then burn retrograde at periapsis. As you slow down, angle the craft up enough that your vertical velocity stays close to zero (i.e. so your altitude stays constant) - this lets you bleed off a very high orbital speed safely, but can be inefficient if you end up hovering 5km up. It's better to allow yourself to descend at a few metres per second during the burn if you have the altitude to play with (keep an eye on the radar altimeter) and be prepared to switch to a controlled descent or drop-and-suicide-burn once your horizontal velocity is low.

[davidpsummers]

I've been tracking the dV needed to land on Minmus. (Its where I get my fuel.) If find I pretty much need 320-340 m/s (or 80% above ideal.) The best I've ever done was 288 m/s. (From a 10-20 km orbit)

One poster said he gets by with 20% above ideal. I'm both impressed and little skeptical.

I do a "no quite" suicide burn where I kill orbital velocity, drop down, and begin burning at about 1.5-2x the suicide burn alt. that Kerban Engineer gives. (Or 20-30 seconds before "impact".) After that I might drop again, but I pretty much do it as normal (keep a descent velocity that gets smaller as I near the surface).

- - - Updated - - -

"I'm not usually one to gainsay Slashy, but twice as much dv as the map says seems like a lot. OTOH, if you use the 'burn retrograde until stopped and then fall down 20 km to the surface' method then I suppose you could spend twice as much as listed."

I do this, but mostly because I find it the only way to hit a landing spots with any precision.

[qoonpooka]

So I've heard two (three if you count atmospheric methods but let's assume no atmo for now) methods here so far that are interesting:

Method A: Chase the retrograde marker all the way down (suicide burn), under thrust the whole way, looking to attain 0m/sec just as your gear hits the dirt.

Method B: De-orbit from gonzo-low Periapsis first, then over-burn on your way to the ground.

I've been using Method C: When you're near where you want to land, burn laterally to cancel out your horizontal velocity and then regulate your descent based on available TWR. This sounds rather like it splits the difference between these two, is it more or less efficient than either, do we think?

[GoSlash27]

Well, it depends how you land. I find landing generally inefficient compared to ascent, but not twice as much. Or, Kraken forbid, three times as much.

I often find that if I have a 20% safety margin compared to the maps I can land quite comfortably. Unless something goes wrong.

I'm not usually one to gainsay Slashy, but twice as much dv as the map says seems like a lot. OTOH, if you use the 'burn retrograde until stopped and then fall down 20 km to the surface' method then I suppose you could spend twice as much as listed.

Happy (and efficient) landings!

Starhawk,

No harm, no foul

A reverse hohmann transfer can come very close to the minimum calculated DV, but we usually have a very specific touchdown point in mind. This precision costs a good amount of DV to achieve. Plus translating around in the landing zone, gravity losses from low t/w, pilot error, and safety margin... a good approach can get well- over the minimum calculated DV. A bad approach can exceed twice the calculated value.

For safety's sake, I like to err to the side of caution for airless body landings.

I'm not saying that all approaches should expect twice the listed DV expenditure, only that it *can* happen and it's best to plan for it so you don't end up short.

For example, see davidpsummers' example.

If he designs his lander from the outset with 2x the DV budget for landing, he's safe even if he's never shot the approach before.

Best,

-Slashy

[Snark]

The sources of inefficiency:

• Cosine losses
  
• Gravity losses
  
• Poor use of Oberth effect
  

So to get the best efficiency, you want to optimize all three:

• Burn only surface-retrograde, so cosine losses are zero.
  
• Do a suicide burn, which minimizes gravity loss and maximizes Oberth effect.
  
• Approach the surface almost horizontally, which gives good Oberth effect.
  

In practical terms: From a very low orbit, you only need a trivially small retro-burn to drop your trajectory to intersect the surface. Then just do a suicide burn to the surface. That's as good as you'll get.

[GoSlash27]

Snark,

This is absolutely true... *if* you're landing on a very smooth body and if you aren't particular about where exactly you want to touch down.

If you have to avoid terrain and/ or want to land precisely, you will have to intentionally take those losses. DV expenditure goes up as a result.

Best,

-Slashy

[Snark]

This is absolutely true... *if* you're landing on a very smooth body and if you aren't particular about where exactly you want to touch down.

If you have to avoid terrain and/ or want to land precisely, you will have to intentionally take those losses. DV expenditure goes up as a result.

Point taken.

However, I find that terrain isn't all that big of a deal, usually. Unless I'm trying to land somewhere topographically extreme (like that canyon in Dres, or on the eastern slope of a steep mountainside), the descent path generally has enough verticality to it to avoid terrain. I use this technique for every biome on the Mun all the time, and I wouldn't describe the Mun as especially flat or smooth.

And it works reasonably well for precise landings, too. When setting up my suicide burn, I aim for an impact point slightly to the east of my target (to allow for planetary rotation, plus the fact that my path is going to steepen as I thrust), and I can generally get pretty close. A purely straightforward maximal efficiency suicide burn will get me within a kilometer or so.

If I want to fine-tune it so that I come down a dozen meters away from my base instead of a kilometer, then it's true that I have to spend extra fuel by playing with thrust on the way down... but it's only a little, and any landing technique that wants a precise landing is going to have essentially the same issue. I was mainly trying to comment on the "gross motor skill" aspect of the "big burns" for landing, rather than the "fine motor skill" of pinpoint landings.

[Empiro]

Fairly precise landings on airless bodies like the Mun should be quite possible with a bit of practice, and even then shouldn't cost more than a bit extra (it's all about knowing how much to overshoot when doing your initial deorbit burn, and keeping that velocity marker a bit higher over the target marker). I think the biggest waste of fuel when I started off is the fuel I spent trying to kill horizontal velocity by eyeballing it. Once I started to use the navball and always burn against the retrograde marker, my landings became much more efficient and easy.

[Warzouz]

The sources of inefficiency:

• Cosine losses
  
• Gravity losses
  
• Poor use of Oberth effect
  

So to get the best efficiency, you want to optimize all three:

• Burn only surface-retrograde, so cosine losses are zero.
  
• Do a suicide burn, which minimizes gravity loss and maximizes Oberth effect.
  
• Approach the surface almost horizontally, which gives good Oberth effect.
  

In practical terms: From a very low orbit, you only need a trivially small retro-burn to drop your trajectory to intersect the surface. Then just do a suicide burn to the surface. That's as good as you'll get.

You're right but your should add that mid/high TWR is mandatory for an efficient landing. If you have a low TWR, suicide burn might not work. You must buy some time by keeping vertical velocity under control. Of course this manoeuvre cost a lot in dV (+25/40% ?). I landed a 0.98 TWR ship on Tylo that way (starting from 30km).

Also, if you deorbit from very low altitude, you're more prone to mountain crash, even with mid/high TWR. Don't forget F5 before attempt...

[Brainlord Mesomorph]

My method for landing has always been “burn it to the ground,†what you guys are calling it a suicide burn.

But I don’t consider it suicidal, my ships have plenty of TWR. So I just lock to retrograde, and make minor throttle adjustments while the way down.

This means that for the most of the decent and I’m running at half throttle or less. And the very end almost none.

My question is, is that maximally efficient? Doing it at half throttle like that?

or would shorter full throttle burns be better?

EDIT:As for landing on target, I just try to keep my target point halfway between my ships current location at my projected impact point. As I continue my retro grade burn, the impact point moves closer to me, and I’m moving closer to it, and, theoretically, everyone meets in the middle.

Edited November 6, 2015 by Brainlord Mesomorph

[Starhawk]

A reverse hohmann transfer can come very close to the minimum calculated DV, but we usually have a very specific touchdown point in mind. This precision costs a good amount of DV to achieve. Plus translating around in the landing zone, gravity losses from low t/w, pilot error, and safety margin... a good approach can get well- over the minimum calculated DV. A bad approach can exceed twice the calculated value.

For safety's sake, I like to err to the side of caution for airless body landings.

I'm not saying that all approaches should expect twice the listed DV expenditure, only that it *can* happen and it's best to plan for it so you don't end up short.

For example, see davidpsummers' example.

If he designs his lander from the outset with 2x the DV budget for landing, he's safe even if he's never shot the approach before.

Very good points, Slashy.

You are, of course, absolutely correct. Especially when considering targeted/precision landings. And it only takes a moment for a routine landing to become problematic and require a whole lot more dv.

And people do use a variety of methods for getting from low orbit down to the surface, including the 'stop and drop' method.

Happy landings!

[jiminator]

it can be somewhat calculated, if you like that sort of thing. scott manley explained how to do it

[Brainlord Mesomorph]

it can be somewhat calculated, if you like that sort of thing. scott manley explained how to do it

http://i.imgur.com/ccTMn4u.gif

That's fascinating.

But why do I need to know length C? or A for that matter.

All it looks like I need is B. - wait do I even need to know that?

All I really need to do is:

1 Plot that test mode.

2 Know how long the burn is.

3 and start the burn that many seconds before the node.

I don't need to know distances at all just speeds and times.

or am I missing the point? (which I usually am)

Edited November 6, 2015 by Brainlord Mesomorph

[Starhawk]

My method for landing has always been “burn it to the ground,†what you guys are calling it a suicide burn.

But I don’t consider it suicidal, my ships have plenty of TWR. So I just lock to retrograde, and make minor throttle adjustments while the way down.

My understanding of the term 'suicide burn' is as follows. When already on a trajectory which intersects the surface, a suicide burn is a burn performed at full throttle at the last possible instant to prevent crashing.

This means that for the most of the decent and I’m running at half throttle or less. And the very end almost none.

My question is, is that maximally efficient? Doing it at half throttle like that?

or would shorter full throttle burns be better?

The method you describe is quite inefficient. A reverse Hohmann is most efficient for the entire landing procedure as discussed in posts above. This entails lowering your periapsis as close to the surface as possible above your intended landing zone. Then burning retrograde at full throttle when close to periapsis.

Using less than full throttle for any burn with a vertical component is less efficient. Of course, one needs to use partial throttle as one nears the surface. But the less of that you use, the more efficient your landing.

Happy landings!

[jiminator]

That's fascinating.

But why do I need to know length C? or A for that matter.

All it looks like I need is B. - wait do I even need to know that?

All I really need to do is:

1 Plot that test mode.

2 Know how long the burn is.

3 and start the burn that many seconds before the node.

I don't need to know distances at all just speeds and times.

or am I missing the point? (which I usually am)

The burn time to stop orbit tells you nothing about how long you are going to be dropping. The distance (A) helps for that. If you look at the triangle you would assume based on the test node & speed that you start your burn 19418m from the target. The calculation however shows 20727, so the delay will probably result in overshooting the target by 1300m

[Just Jim]

That's fascinating.

But why do I need to know length C? or A for that matter.

All it looks like I need is B. - wait do I even need to know that?

All I really need to do is:

1 Plot that test mode.

2 Know how long the burn is.

3 and start the burn that many seconds before the node.

I don't need to know distances at all just speeds and times.

or am I missing the point? (which I usually am)

If you're just using B, it's only the distance to a point high over the object, not the object itself, and your burn time will be too short.

[Brainlord Mesomorph]

My understanding of the term 'suicide burn' is as follows. When already on a trajectory which intersects the surface, a suicide burn is a burn performed at full throttle at the last possible instant to prevent crashing.

The method you describe is quite inefficient. A reverse Hohmann is most efficient for the entire landing procedure as discussed in posts above. This entails lowering your periapsis as close to the surface as possible above your intended landing zone. Then burning retrograde at full throttle when close to periapsis.

Using less than full throttle for any burn with a vertical component is less efficient. Of course, one needs to use partial throttle as one nears the surface. But the less of that you use, the more efficient your landing.

Happy landings!

Perhaps I wasn't clear. AFTER I've placed my PE as low as possible over the landing site. And I AM burning retrograde, and I do start at full throttle. But I quickly have to back off the throttle or I'll end up hovering (or rising) and that MUST be inefficient.

Are you saying I should build a ship with a lower TWR and run all the way down at full throttle? Why would that be better?

or are you saying I should free-fall more and do shorter burns? and again, Why would that be better?

I'm trying to understand why.

[Starhawk]

I'm definitely not saying you should have a lower TWR.

I guess I wasn't clear. The reverse Hohmann I was trying to describe is made up of two burns. The initial burn lowers your peri almost to the surface. The second burn is performed when very close to periapsis and to the surface. As much as possible of the second burn should be done at full throttle. It should also be performed as late as safely possible for highest efficiency. Obviously, when you get close to the surface you need partial throttle for the last several seconds of the descent.

Free-fall more and do shorter burns is exactly what I'm saying. The more time you spend burning with any vertical component, the more fuel is being wasted.

I'm oversimplifying this, but: All the fuel burned horizontally is efficient. All of the fuel burned vertically is inefficient. You want to minimize the latter as much as possible.

One way to think of it is this. When you burn horizontally, you change your velocity relative to the surface. However, you can stay in one place burning vertically and burn all your fuel just hovering and not get anywhere. Time spent burning horizontally fights only inertia. Time spent burning vertically fights gravity. And gravity just eats delta-v.

Hopefully some of that makes sense.

Happy landings!