Economical descent profile for Mun landing?

[jadoo1989]

So, I have no problems getting to the Mun from a Kerbin parking orbit. I can land pretty easily. The problem is, I use a LOT of fuel on the descent path because I end up freaking out and trying to cancel out some of the orbital velocity way sooner than I probably should, which basically leads to me fighting physics when I could be using the high velocity to my advantage to speed the landing process up. Theoretically, I know this. When I see the Mun below me and my distance toward it decreasing, I start trying to fight it anyway. This leads to me having very little fuel to establish a Munar orbit that's efficient, and leads to me having to find creative ways to make it back to Kerbin with little fuel after I reach a stable low Munar orbit.

Basically, at what point do you start bleeding off some of the orbital velocity during your descents?

[Frederf]

The optimum descent profile is the time-reversed optimum ascent profile with a negative propellant flow rate. It'll look mostly like a good takeoff in reverse.

However that's hard to do without teh maths so I found a method that's nearly as good ad a lot easier. The idea is to place your periapsis about 3-5km above your intended landing site. Upon approaching your landing site (how far determined by TWR) you should be near your 3-5 km close scrape height. Burn retrograde with enough radial positive component to cancel any falling but not enough to gain height. As you decelerate pitch up more and more to direct more of your thrust to maintaining altitude. When you're over the landing site and your horizontal speed is manageable then let down at 10-30 m/s vertically until close and land per normal. I budget about 60 seconds hover time (Mun's gravity times 60 = hover dV) for the landing.

[Wesreidau]

I land by burning nearly all of my orbital V off after a circular 10km orbit.

Edit: Derp. I should probably have mentioned I get to that orbit in a service module from which my lander decouples after the pilot EVA's over to it. I should probably also have mentioned I come in at about 45 degrees but level off at about 1km, from which I descend vertically. 3 AM posting fail.

Come down nearly vertical and use RCS to keep the prograde pipper centered on the ball. Don't tilt the rocket to correct horizontal drift if you can help it at all, and avoid the mistake of false halts as you descend. You may also want to drop your landing gear and any empty tanks upon liftoff from the Mun (not gonna need legs in space) and rendezvous with a craft orbiting the Mun that performs your orbital maneuvers and transfer burns so the lander can be reserved for landing and takeoff.

Edited July 29, 2013 by Wesreidau
3AM Derp Correction

[Stochasty]

I land by burning nearly all of my orbital V off after a circular 10km orbit. Come down nearly vertical and use RCS to keep the prograde pipper centered on the ball. Don't tilt the rocket to correct horizontal drift if you can help it at all, and avoid the mistake of false halts as you descend. You may also want to drop your landing gear and any empty tanks upon liftoff from the Mun (not gonna need legs in space) and rendezvous with a craft orbiting the Mun that performs your orbital maneuvers and transfer burns so the lander can be reserved for landing and takeoff.

This is nearly the least efficient descent profile that you can manage. About the only way to make it less efficient is to decide to hover at 1m altitude above the Mun rather than actually landing. It is also the hard way to land, because your velocity is always vertical, so if you wait too late for your burn you're out of luck.

The reverse gravity turn method is easier, more efficient, and much more forgiving to mistakes (if you wait too late, you just wind up burning a bit more fuel rather than testing the viscosity of your spacecraft).

[njd80]

This is a good example of what Frederf described. I have tried to re-create this, but can only watch in awe and hope to be this good one day.

[Smoke]

The (theoretical) optimum would be similar to a Hohman transfer from your parking orbit to an orbit just above the surface:

At the apoapsis of your parking orbit you burn retrograde till your periapsis almost touches the ground. At periapsis you then burn retrograde again (parallel to the ground) until l you lost your orbital speed. Since you're burning at the lowest possible altitude you get the best use of the Oberth effect.

How low you can put your periapse (i.e. at what height you start the break) will mainly depend on your TWR. Just autosave and try some values. For your average mun lander 500m should be a good guess.

This also much easier than the "vertical drop" method, since errors tend to make you over- or undershoot your target position. Using a vertical drop you're more likely to be hovering in the air wasting fuel or vertically overshoot the lithosphere.

[Somerled]

The idea is to place your periapsis about 3-5km above your intended landing site.

Beware that there are features on the Mun's surface above 3km, and the 0.21 Munar surface may have features above 4km [citation needed]. Frederf's plan is what you want to aim for, but be prepared to add some altitude during your final burns if you see you're flying too fast straight into a mountain.

[gilflo]

I design my ship with a supplmentary stage for retro burning on Mun landing. I eject the stage between 5000m and 2000m before lading so that I burn a very little with the ship itself and i get enough for take off, docking on an orbital station for refuel or even to eject from mun to kerbin and land to Kerbin.

The sup stage is too big for Mun and Minmus and I usually burnt half tank before jettisson, I could use a smaller one, but my design is supposed to be used on farther planets and this stage could be use also for establish orbit and then retro burn before landing

Uploaded with ImageShack.com

[Immashift]

Okay so I'll admit I personally use the inefficient kill orbital velocity all at once and plummet straight down, then cancel vertical velocity when I think I have little time left.

Now the approach in the video a few posts up is extremely impressive, but let's say you want to land somewhere specific. Like if perhaps you're using Kethane and need to come down over a deposit. How would some of you more skilled / crafty people land somewhere very specific?

I've watched MJ do it, but I prefer manually doing things. Like the OP though, I tend to see the ground coming up mighty fast and jam on the throttle...

I shall resolve to practice the elegant landing in that video at least today sometime practice practice practice is how I learned to dock... That and watching ORDA do it. Eventually stopped using it and even MJ I mostly only use for info screens now, but landing is something I'm still relatively bad at. Half the time I end up rolled the wrong way. And during the final few meters I pitch the wrong way and smear over the landscape.

[Somerled]

... let's say you want to land somewhere specific.

The shallow descent method works pretty well for hitting a target in no atmosphere. Adjust your orbit until it intersects the surface at your target. A shallower descent is more efficient, but runs the risk of hitting higher elevation areas. You burn retrograde as you get close, pitching up and down slightly to keep your intersect at your target location. As you near the target, you can switch to vertical descent, which is practically necessary in rough terrain. How close you are when you switch to vertical is up to you and the surface.

The only concern is when you have long burn times to reduce velocity. The same shallow descent method works, but needs more care. Scott Manley showed a cool way to do it. Set periapsis [edit: see note below] directly above the target by only a few km. Set a maneuver node that kills all horizontal velocity so you drop down directly on the target. Take the reported burn time, and when you're that far away (in time) from the maneuver node, that's roughly the closest you can get before you need to start burning at full throttle. That keeps you from overshooting when you follow the shallow descent method.

[Note: doesn't have to be periapsis, but the less vertical speed you have in the orbit above your target, the more accurate the estimate of your burn time will be. At periapsis, the vertical speed is zero.]

Edited July 29, 2013 by Somerled

[Khrissetti]

I'm a fan of the 'drop like a stone then suicide-burn at the last minute' technique.

[Hiddos]

I'm a fan of the 'drop like a stone then suicide-burn at the last minute' technique.

I can sympathise a lot with this fellow forum member and flight commander

[Geschosskopf]

Beware that there are features on the Mun's surface above 3km, and the 0.21 Munar surface may have features above 4km [citation needed]. Frederf's plan is what you want to aim for, but be prepared to add some altitude during your final burns if you see you're flying too fast straight into a mountain.

I've been to Mun about 10 times in 0.21 and there is a LOT of stuff above 4km now. Vast areas of the ground are at least 3km. I'm pretty sure the big bulge in the huge crater just north of the equator is above 5km with a 4km rim all the way around.

Anyway, I do the "very low orbit stop and drop" method myself. However, my hard deck for the low orbit is 6km and I feel more comfortable with 7km.

In 0.21, I strongly advise doing the "very low orbit stop and drop" method, not because it's efficient but because it's safer. There is no such thing as level ground on Mun anymore, and the slope changes in both direction and severity within just a couple hundred meters. This is because of all the little craters, the ones you can't see from more than about 1km above them, which are literally everywhere. These small craters are usually "fresher" so have higher rims and deeper basins for their size than bigger craters, which means steeper slopes, and they're too small to have any real flat area at the bottom.

Because of this, it's really a good idea to pick your landing spot as best you can from very low orbit and then come down carefully, ready to move sideways at any moment to avoid landing in a small crater you couldn't see before. It also helps to land near the terminator so the low-angle sunlight helps you see the craters better, and don't use powerful downwards floodlights if you can help it for the same reason.

[MBobrik]

The optimum descent profile is the time-reversed optimum ascent profile with a negative propellant flow rate. It'll look mostly like a good takeoff in reverse.

Wrong. because gravity drag loses are biggest near launch/landing position and decrease with speed. But on the way up your mass starts high and decreases and during landing your mass starts high and decreases. So you will spend more time in high gravity drag during ascent than during descent. This can be very significant for small craft with low TWR.

Edited July 29, 2013 by MBobrik

[theflyingfish]

One trick that I have found much use for is putting a maneuver node on the point where your projected orbit intersects the ground, and using it to zero out your dV. The game will figure out when you need to burn and will put you just above the lunar surface if you plan it correctly. It takes the guesswork out of a vertical descent trajectory.

[Mr Shifty]

The (theoretical) optimum would be similar to a Hohman transfer from your parking orbit to an orbit just above the surface:

The reason this isn't true is because you're not going orbital speed when you land; you're going much slower.

[MBobrik]

The reason this isn't true is because you're not going orbital speed when you land; you're going much slower.

the full maneuver actually goes like Hohmann to the surface and then kill your horizontal velocity while compensating for the lost centrifugal force.

It is actually slightly suboptimal because you could start killing your horizontal velocity right off the bat and at the same time descend by first undercompensating the lost centrifugal force and then overcompensating to slow down near the surface.

It has also one big con - with low TWR and high gravity like tylo you can end up grazing the surface for dozens and dozens kilometers till you bleed all horizontal speed.

Edited July 29, 2013 by MBobrik

[Wesreidau]

This is nearly the least efficient descent profile that you can manage. About the only way to make it less efficient is to decide to hover at 1m altitude above the Mun rather than actually landing. It is also the hard way to land, because your velocity is always vertical, so if you wait too late for your burn you're out of luck.

The reverse gravity turn method is easier, more efficient, and much more forgiving to mistakes (if you wait too late, you just wind up burning a bit more fuel rather than testing the viscosity of your spacecraft).

I didn't describe my mission profile very well, but mostly described my final approach which starts at about 1km. Before then its a short burn to move the descent arc into the target area. I had to make a precision landing on the dark side of the Mun to rescue Harbus Kerman, who was stranded after jettisoning his... well, a lot of things went wrong. But my space program has suffered no casualties, so your speculations as to the viscosity of my spacecraft will go unanswered.

The gist of my post is to drop empty tanks and landing gear and have your return rocket orbiting the Mun for you. A lot of you single-stage lander types probably operate differently.

[Somerled]

Wrong. because gravity drag loses are biggest near launch/landing position and decrease with speed. But on the way up your mass starts high and decreases and during landing your mass starts high and decreases. So you will spend more time in high gravity drag during ascent than during descent. This can be very significant for small craft with low TWR.

You're talking about fuel consumption giving higher TWR over time, yes? It's true that you'll have different thrust profiles, but the optimum trajectories will still more or less match up, you'll just be traversing them at different rates. Optimum takeoff in zero atmosphere is a nearly horizontal burn, and it's the same trajectory for landing. That way you don't fight gravity most of the way up or down, just your own horizontal inertia.

[MBobrik]

but the optimum trajectories will still more or less match up

Yes, they will look similar, but the dv needed might be very different.

[Somerled]

Delta-v shouldn't be very different. The key is you pull against gravity as little as possible by minimizing vertical thrust in favor of horizontal thrust. It's why delta-v for Hohmann transfers is independent of ship mass or TWR [edit: see note]. However, at the beginning of an ascent or end of descent, you need to exchange vertical and horizontal speeds, at which your TWR will be an issue and you will get some delta-v loss one way or another. The optimum trajectory saves that for the last possible moment. It's up to the pilot and the surface when that moment occurs.

[Note: For very low ISP craft, this is most definitely an issue, but it's still the optimal trajectory]

Edited July 29, 2013 by Somerled

[tavert]

Delta-V for optimal landing versus optimal takeoff, assuming the same initial TWR, can have fairly different costs when TWR is low. See http://forum.kerbalspaceprogram.com/showthread.php/39812-Landing-and-Takeoff-Delta-V-vs-TWR-and-specific-impulse

The reason is that you're moving slowest when you're heaviest for takeoff, so you need a more vertical pitch angle to counteract gravity, whereas for landing you're moving slowest when you're lightest. For absolute efficiency you only want enough altitude to clear the terrain, and no more. If your landing site is substantially lower than the surrounding terrain, it's best to allow your altitude to change when moving slowest - the gravity losses are lower that way.

Edited July 29, 2013 by tavert

[MBobrik]

See http://forum.kerbalspaceprogram.com/showthread.php/39812-Landing-and-Takeoff-Delta-V-vs-TWR-and-specific-impulse

Nice analysis.

Question. Did you solve the differential equation analytically or numerically. Because I am currently writing an craft optimizer that can compute the optimal combination of engines and fuel tanks for a given payload and celestial body. It works, I've got some interesting results right now, but it is kinda slow because of having to solve the diff eq numerically.

[tavert]

Numerically. For that particular diffeq, I found it fastest to use Mathematica. I more recently wrote some Matlab code to do Pareto optimization of tank and engine choices based on payload and minimum TWR, didn't publicize it though since I have no idea how to design an appropriate GUI for it.

I'd be curious to look at your approach and what you've got so far, see how it compares to ideas I've had for trajectory optimization given a fixed craft (which would mostly be for the tougher problem of atmospheric ascent)...

[MBobrik]

Numerically. For that particular diffeq, I found it fastest to use Mathematica. I more recently wrote some Matlab code to do Pareto optimization of tank and engine choices based on payload and minimum TWR, didn't publicize it though since I have no idea how to design an appropriate GUI for it.

I will just make some nice graphs for all moons and airless worlds, like payload mass -> engine type, count, fuel tank cout, dv, etc...

Then peraps later, I will make something like VAB advisor - something you slap on the payload and it will say what number of what engines and tanks will be best for this payload and destination.

I'd be curious to look at your approach and what you've got so far, see how it compares to ideas I've had for trajectory optimization given a fixed craft (which would mostly be for the tougher problem of atmospheric ascent)...

Right now, I just simulate the "hohmann to the surface from a parking orbit, and then slow down keeping horizontal" approach. not trying to optimize the full 2D trajectory.

If you are interested too, we may exchange our programs.

I have also partially done something like trajectory optimizer for atmospheric planets using genetic algorithms, but it is very far from working yet.