Land on Tylo with low Thrust to Weight Ratio

[RocketPilot573]

Hi, I'm trying to land on Tylo with this lander:

It has 4 Atomic engines and 2 aerospikes, with a thrust to weight ratio of around 1.4 according to mechjeb. My plan is to use the Nukes to slow down the horizontal velocity, and then use the aerospikes to land. So far, I've tryed 3 times, each of which ended in crashing into the ground at around 600m/s.

Any ideas on good decent profile?

[DMagic]

I did something like this, I had a TWR of 2 or maybe 3 by the time I got close to the surface, lower before that. It was really hard, but I think you have the right idea. If you have enough fuel just get into a somewhat high orbit, maybe 40 or 50km to kill your horizontal velocity. Once that is almost down to zero then slowly come down to the surface. This is really fuel inefficient, but if you keep your vertical speed low, maybe below 50 m/s above 10km, and gradually decrease it below that you should be fine. Use the nukes as long as you can if you have problems with fuel. Just use the aerospikes once you get closer to the surface.

[bang_coi]

Tylo has the same size as Kerbin. Its surface gravity is about the same as Kerbin (0.8g). You might wanna try a more powerful descent engine

It doesn't have its own atmosphere to slow down your craft, so landing on Tylo is quite tricky...

[stupid_chris]

Hmm. The thing is, having engines that don't thrust is very inneficient. They'Re really just acting as dead weight. So you might want to fire them all at once. Then what is your DV? the minimal DV to land from 50km is at least 3100m/s

[Nao]

When using LV-N the most mass efficient TWR's for atmosphereless bodies is quite low. I did calculations for Mun and it was as low as 1,25. I think it would be similar for Tylo. So you have quite good craft for it.

As you slow down while in Tylo orbit, you need to increase attitude to keep altitude constant.

If we compare engines for this task, it turns out that up to attitude of 60deg LV-N's give more energy (speed) just for horizontal slowdown per unit of fuel than aerospikes while working fully horizontal ( 800s ISP * cos60 = 400s , and aerospikes have 390s ).

That's why i would advise slowing down only on LV-N, while maintaining constant altitude, up to ~65deg attitude and then engaging aerospikes.

Also its best to start from very low orbit (~11km? was it) - just high enough you won't suddenly hit a mountain.

[tavert]

MechJeb's probably telling you the TWR assuming all the engines are active, it doesn't always properly figure things out when you disable some engines. The landing delta-V on Tylo depends pretty strongly on your TWR. With a good TWR and a low starting orbit you can land on Tylo for less than 2500 m/s delta-V.

The issue with trying to kill horizontal speed first and then do a vertical landing is as soon as you go any slower than orbital speed, you will start losing altitude. You can try to keep vertical speed from going too high by pointing above the horizon, but that can only counteract the sine of your angle above horizontal. And pointing anywhere but retrograde incurs a steering loss. You're probably better off using all the available engines for Tylo landing - even though the aerospikes aren't as fuel efficient as the LV-N's, the higher TWR will reduce the gravity and steering losses incurred in landing.

I agree with Nao that a low altitude is better in principle (Mr Oberth sends his regards), but you'll need some buffer for Tylo.

Edited June 6, 2013 by tavert

[stupid_chris]

If we compare engines for this task, it turns out that up to attitude of 60deg LV-N's give more energy (speed) just for horizontal slowdown per unit of fuel than aerospikes while working fully horizontal ( 800s ISP * cos60 = 400s , and aerospikes have 390s ).

The ISP aren't units of energy or speed, but only of fuel efficiency. The thrust is the unit that calculate your braking speed.

[Nao]

The ISP aren't units of energy or speed, but only of fuel efficiency. The thrust is the unit that calculate your braking speed.

Well i tried to be short. And ISP is one element that is easy to see right of the bat in KSP. It would be more appropriate to use exhaust velocity and all that but the effect would be the same.

What I'm trying to say is that both ISP and attitude angle can be measures of efficiency if we count work done over just one direction (horizontal here).

That is an aerospike engine burning horizontally will add X amount of Dv out of one liter of fuel.

At the same time LV-N burning at 61deg angle will also add X amount od Dv out of one liter of fuel to the horizontal direction.

So until that 61deg attitude the steering losses used to counter gravity (or should i say gravity losses) are fully countered by higher fuel efficiency of nuclear engines.

edit: also the specific impulse is kind of a measurement of speed, as its exhaust velocity divided by earth gravity which is just an constant.

Edited June 6, 2013 by Nao

[stupid_chris]

Well i tried to be short. And ISP is one element that is easy to see right of the bat in KSP. It would be more appropriate to use exhaust velocity and all that but the effect would be the same.

What I'm trying to say is that both ISP and attitude angle can be measures of efficiency if we count work done over just one direction (horizontal here).

That is an aerospike engine burning horizontally will add X amount of Dv out of one liter of fuel.

At the same time LV-N burning at 61deg angle will also add X amount od Dv out of one liter of fuel to the horizontal direction.

So until that 61deg attitude the steering losses used to counter gravity (or should i say gravity losses) are fully countered by higher fuel efficiency of nuclear engines.

edit: also the specific impulse is kind of a measurement of speed, as its exhaust velocity divided by earth gravity which is just an constant.

True, but if your TWR is too low, this means that even burning straight down, you wouldn't be able to slow yourself down. Having a low TWR is fatal on the final descent. The utopic most efficient landing would be one that after deorbiting, the craft would be able to cancel all it's velocity at once a few meters from the ground. And this is barely taking account of the ISP. Else landing with ions should be the best bet. But their thrust is just not enough to slow you down. Same here with the LV-Ns, their thrust is often too little to land alone. Yes, they will give a lot more DV per liter of fuel, but if it takes them 15 minutes to produce the same amount of thrust as an aerospike, you're already smashed in the ground

[Markus Reese]

The trick to Tylo is the ability to decelerate while maintaining your altitude. Essentially, you want to get to a lower altitude first. After that, just begin to cut speed and start pitching up to prevent loss of altitude. As you get close to low transversal velocity, you can cut engine thrust and lower down.

Essentially, the slower you go, the more gravity is in effect. Use the nukes to cut speed until you are using your nukes mostly to maintain altitude then engage aerospikes. No matter what, you will need enough PTW to counter gravity of course.

Start by trying on the moons this practice. It is how I land VTOL style on the moons.

[Nao]

True, but if your TWR is too low, this means that even burning straight down, you wouldn't be able to slow yourself down. Having a low TWR is fatal on the final descent. The utopic most efficient landing would be one that after deorbiting, the craft would be able to cancel all it's velocity at once a few meters from the ground. And this is barely taking account of the ISP. Else landing with ions should be the best bet. But their thrust is just not enough to slow you down. Same here with the LV-Ns, their thrust is often too little to land alone. Yes, they will give a lot more DV per liter of fuel, but if it takes them 15 minutes to produce the same amount of thrust as an aerospike, you're already smashed in the ground

Yes the LV-N TWR (probably around 0,6 in this case) is too low for whole descent but the idea is to use them alone just for the first part. And then his 1,4 TWR is probably close to the most mass efficient one, it's just hard to fly due to constant need of adjusting attitude to maintain constant altitude.

Also about thrust, i think it's a common misconception that big TWR is good for landings/ascents. I did calculations some time ago and the most *mass efficient* TWR for Mun landing was only 1,25 for LV-N engines. For which more than half of the flight was done above 30deg attitude.

This mostly because engines themselves in KSP have quite low TWR (compared to their Earth equivalents). Having lower craft TWR means having less engine mass. That is countered by lower fuel efficiency of descent profile, but until TWR's of ~1,4 the additional fuel burned will still weight less than difference in engine mass compared to higher TWR's.

(I'm not sure about Tylo optimal TWR thou, i would assume it would be similar to Mun's but i didn't tested it so i might be wrong saying ~1,4 is good.)

[tavert]

I was playing around with Tylo a bunch with 1-man lander can, an aerospike, and various different amounts of fuel to compare delta-V cost to TWR. Here's what I found.

8 tons of fuel (initial TWR=2): 2453 m/s to land from 30km, 2270 m/s to take off back to 15km.

14 tons of fuel (initial TWR=1.25): 2668 m/s to land from 35km, 2303 m/s to take off back to 15km.

Landing site was about 1400m higher altitude for the heavier one. I had another run at 12 tons but doesn't look like I saved pictures.

For the OP's case, I think it depends highly whether his TWR is 1.4 with the aerospikes or without. I can't quite tell what the mass of his ship is by looking at it.

Generally though, I have a hard time seeing how you'd ever want to burn anything other than pure retrograde when you're landing. You'll have to start from a slightly higher altitude, but my gut feeling here tells me the Oberth penalty of a bit higher altitude for the start of the burn should be smaller than the steering costs of holding constant altitude... But I should probably go do the actual math here.

[Nao]

@tavert

I've build a mockup of his ship in VAB and it looks like 1,4 TWR is with all engines "on" over Tylo.

My idea of burning at an high angles came from ascent experiences. For example when you have a craft that has both LV-T30 for ascent/landing and LV-N for interplanetary transfer. During (Kerbin) ascent there was a question of when to shut off LV-T30's and continue only on LV-N's?

I've found out that shutting them early and then burning at high angles with LV-N's used the least amounts of fuel.

This is also applicable to SSTO's which usually lack thrust if their last stage is LV-N.

As for atmosphere less bodies, it's not only Oberth effect that makes low altitude ascents/landings efficient. It's also the the fact that any vertical acceleration gained is (mostly) a wasted fuel. Also timewise if we burn prograde only during ascent, the time to achieve orbital speed increases compared to just horizontal flight, so we get more gravity drag. It's only really visible in low TWR situations thou.

For example a test Mun ascent. TWR only 1,21 Launch mass 60,7t orbit mass 55,3t with payload being 49,4t. Only 735m/s Dv spent even thou at first I had to fly with more than 60deg attitude.

I would be really surprised if you've found a way to ascent into Mun orbit with payload to launch mass ratio greater than 0,81 (49,4t/60,7t for the example).

[Kosmo-not]

Landing with a low TWR isn't too bad. Just use pitch to control rate of descent. Watch this video to see how it's done on Mun.

[tavert]

The question of when to switch to more efficient engines is relevant to the OP, but not for trajectory optimization on a single type of engine.

Nao, I didn't think you were right about "time to achieve [brake from] orbital speed," I was thinking the fastest way to slow down from a given initial orbital speed would be to point directly retrograde. But I just made a quick little Euler-integrating spreadsheet (RK4 is more trouble than it's worth in Excel) to convince myself for sure, and it looks like either you're right or my math is wrong. Or maybe both I suppose. But yeah, it looks like you are better off just taking the steering losses than letting gravity add any vertical speed. I might have to change my landing profiles now...

Edit: And regarding payload fraction for Mun ascent, can't say I've tried that one. I've done tons of Mun craft optimizing for Kerbin mass with all sorts of different combinations of constraints (jets/no jets, pod/seat, etc), but my Mun landers tend to be tiny enough that their design is dictated more by discrete engine/tank choices than anything continuous like optimal TWR or trajectory. Perhaps you should let Scott Manley know so he can change his Mun surface-to-orbit fuel tanker design...

Edited June 6, 2013 by tavert

[RocketPilot573]

Thanks everyone!

Finally I have landed the first Kerbals on Tylo:

As it turned out, pitching up to about 60 degrees (rather than 15 degrees like I originally tried) was the answer. Though the ship only has about 800 liquid fuel, and 1000 oxidizer left. But with only the Atomic Engines the lander has a TWR of 1, so it might reach orbit!

[Markus Reese]

Awesome to hear! Gratz

[Nao]

Thanks everyone!

Finally I have landed the first Kerbals on Tylo: [img snip]

As it turned out, pitching up to about 60 degrees (rather than 15 degrees like I originally tried) was the answer. Though the ship only has about 800 liquid fuel, and 1000 oxidizer left. But with only the Atomic Engines the lander has a TWR of 1, so it might reach orbit!

That's great!

As for ascent i would advise using aerospike for the first part (at least until you can maintain constant altitude at 60deg attitude on just LV-N's) Maybe even burning most of the RCS to lower the weight and increase Dv A little, as 800l doesn't sound to much.

And from experience i would advise against flying as low as possible (even thou it would be the most efficient), as after you build up some speed these mountains can sneak up on you really fast especially on low TWR.

The question of when to switch to more efficient engines is relevant to the OP, but not for trajectory optimization on a single type of engine.

Nao, I didn't think you were right about "time to achieve [brake from] orbital speed," I was thinking the fastest way to slow down from a given initial orbital speed would be to point directly retrograde. But I just made a quick little Euler-integrating spreadsheet (RK4 is more trouble than it's worth in Excel) to convince myself for sure, and it looks like either you're right or my math is wrong. Or maybe both I suppose. But yeah, it looks like you are better off just taking the steering losses than letting gravity add any vertical speed. I might have to change my landing profiles now...

Edit: And regarding payload fraction for Mun ascent, can't say I've tried that one. I've done tons of Mun craft optimizing for Kerbin mass with all sorts of different combinations of constraints (jets/no jets, pod/seat, etc), but my Mun landers tend to be tiny enough that their design is dictated more by discrete engine/tank choices than anything continuous like optimal TWR or trajectory. Perhaps you should let Scott Manley know so he can change his Mun surface-to-orbit fuel tanker design...

Yeah, the Mun lander has unreasonable mass because the damn Mun has so low gravity.

I'm not that good with math, but i feel like the TWR numbers should be the same regardless of planet's mass (at lest if we assume similar densities and no large rotational speeds). So maybe OP actually had a really efficient ship for the mission.

Just for reference, for Mun I've only done test's for constant altitude ascent but for other popular lander engines it looks like this:

Anyways I'm glad you managed to find something new about ascents through my not so good english .

[tavert]

Nah your English is great no worries. I think I saw that chart last time you posted it, very interesting. Did you get that data from experimental tests and recording results, or by integration?

I'm tempted to make a Mathematica notebook which I could post on Wolfram Demonstrations or somewhere to actually solve the constrained optimal control problem for airless-body single-stage landing or ascent. We'll see how I get on with that.

[Nao]

The chart was done by doing an iterated simulation but i did several in game test's and the result's were close to those predicted. The test from screenshots had launcher to liftoff mass of 18,7% with fuel left, and from computation i got ~17,5% for that TWR. So not ideal but enough to draw conclusions i think. (The integrating method was very ...kerbal as in "i know only basic math and im doing rocket science" but after adding some averaging per step the result looked nice even when tested with very small number of steps.)

It would be great if you manage to solve the ascent/descent problem into usable data.

[tavert]

It would be great if you manage to solve the ascent/descent problem into usable data.

Well I got something written up in Mathematica that almost worked, but turns out Mathematica's optimization solvers are not as good as I was hoping they would be. There goes my easy way of getting a halfway decent GUI.

Next I tried symbolically solving the differential equation for constant-altitude landing to see if there's a closed-form solution. It boiled down to: HSpeed'(t) = -sqrt(Thrust^2 - Mass(t)^2 * (Mu - (Altitude + BodyRadius) * HSpeed(t)^2)^2 / (Altitude + BodyRadius)^4) / Mass(t), where Mass(t) = InitialMass - Thrust * t / (g0 * Isp). Not something Mathematica was able to solve in closed form today.

For the numerical optimizer, I know what solver I'd want to use (which is conveniently open source, though a bit annoying to compile on Windows), but not sure about the best interface. Maybe a Python script or something... I might post and move discussion over to the Tools and Applications section if I get anywhere.