tavert

And here's basically the opposite: http://imgur.com/a/LZBGK#0 Maximalist at 148.64 tons (until someone builds something that needs to burn off a bunch of fuel before it can take off), High Efficiency, Higher than an Eagle, Safety Last. First time I've used a mainsail in a while. Most annoying bit was simultaneously having landing gear low enough to get around the engine, while making sure enough fuel was in front of the center of thrust, otherwise it would be uncontrollable.

"Piloted" in a manner of speaking, Minimalist at 1.64 tons, High Efficiency, Safety Last. http://imgur.com/a/Y4y2m#0

I don't remember if this was exactly the case in MechJeb 1.9.x, but in MJ2 there's a substantial difference between "Land at target" and "Land anywhere." The former maintains a controlled gradually decreasing speed as it goes down, and tends not to be full-throttle so will use a bit more fuel than a suicide burn would. Land anywhere is more aggressive, at least 80-90% throttle for most of its landing burn. I'm guessing it's leaving a bit of a safety factor, but I bet that factor could be made adjustable. Time to head on over to their Github and file another feature request... Not sure about accuracy with Land at target, it will occasionally go nuts and be way off for me. If you find a case where MJ2 goes crazy, it's good to back up the quicksave and submit to via their Github so they can look into what's going wrong.

Finally did it! 2.22 tons to take one Kerbal (on a ladder) to and from the Mun. I kept trying with 2 but found that Kerbals have enough mass that the delta-V calculations were off, and I would just barely run out of fuel on the final powered landing back on Kerbin. http://imgur.com/a/zPS6q#0

What works for me is putting an octo-strut underneath the engine and attaching landing gear to the octo-strut: http://imgur.com/a/SJlfb#1 Ends up looking like a coat rack, but seems to work as long as you land somewhere fairly flat. Feels kinda odd to have your engine firing through a piece of your rocket, but ah well.

Not hard to manage especially with MechJeb 2, but they are way overpowered. Anything judged on mass will be forced to use jets if they're allowed, so the designs are more varied without them. I've been trying for several days now to get a 2.22-ton jet-powered contraption to and from the Mun with 2 Kerbals hanging on to ladders, it's really tedious since you can only use 4x physical warp. Also be very careful at the SOI transition, it's best to let go of the ladders while you change SOI otherwise Krakensbane can tear your ship apart and/or send your Kerbals flying. I probably would have gotten it done by now if I weren't trying to do it using only 60 units of fuel for the entire trip. If you have the patience to go to Minmus too, I applaud you. Though I'm guessing SunJumper won't approve for this challenge. Aerospikes don't have gimbals. And the drag issue tends to make rockets with aerospikes unstable if they're placed behind the CoM (as engines usually are), although with enough winglets that can be fixed. If you're scoring on mass like this challenge, I think aerospikes are legitimate and there's no harm in allowing them, they have their uses but I don't think they are the best choice for this particular task anyway.

I think the mod parts might be confusing MechJeb there. Does it still say 1.57 when you go to the pad? The flight numbers are generally more accurate than the VAB numbers. I see at least 9.1 tons on the parts that I recognize as stock. Edit: nevermind. That's a confusing set of parts, all just probe-sized miniaturizations of the stock versions?

That's the right way to optimize the trajectory and throttle profile for a fixed design. I was thinking of writing something similar myself, I'd be curious to see your code and suggest better optimization solvers. There's a nonlinear interior-point solver I happen to be very familiar with that's extremely good. But if you're trying to optimize the design to achieve a specific task, there are far more variables and constraints and the solutions depend significantly on the choice of objective function. There are some now-gone challenges that demonstrated this, maybe partly salvageable from Google cache - the Smallest Eve Ascent Vehicle was a bit unique due to being Eve instead of Kerbin, but scoring by mass showed some interesting designs. The "best stock liquid fuel rocket efficiency" was scored only on fuel consumption and the designs were quite a bit different than they would have been if scored on mass to achieve the same goal (lifting an orange tank + large RCS). Optimal in that you have no steering losses and horizontal thrust is best for reducing future gravity losses, but waiting incurs gravity losses of its own. Due to the Oberth effect you want as much of your thrust to be spent low in the gravity well and at high speed, but you have to balance that out with atmospheric drag at low altitudes as we all know.

Wrong. One ignores engine mass completely, and can't properly count solids with the same single number. When you optimize for total mass including engines, you'll find the best choice of engines/staging/TWR varies depending on what the challenge is trying to accomplish. If you strictly optimize for fuel, then you just end up maximizing ISP using aerospikes and LV-N's. Minimum mass gives more interesting combinations of solids and varied liquid engines, and TWR is far from constant in those designs. ... and we're way off topic. Whoops.

Pad mass is common over in challenges (otherwise how do you count solids on an even footing?). Cost, eventually. And with the discrete nature of components here, you have to set a specific payload to have a well-defined optimal design. Combinations of engines, fuel, and staging that work best to get a 10 ton payload into orbit won't necessarily be the same as for a 20 ton payload. And you have to consider whether your payload has any engines of its own, if you can use those during the ascent, etc.

The end result of the two burns you've pictured would be different. As Jason said, burning downward is a loss, but I think it's essentially the same as any other steering loss, 1 minus the cosine of the steering angle. And a burn anywhere other than 80km can't put you into an 80km circular orbit, period. But assuming the periapsis in the "old technique" before the final burn is negative or very low, your gravity turn should have flattened out at a lower altitude. Then you would have burnt more during the gravity turn than you did with the "old technique," but the savings in the final burn will more than make up the difference. To think of it in a different way, you don't need to bring your apoapsis all the way up to its final height until after you've gotten to orbital speed.

Optimal for what objective function? You can't possibly be trying to say that every single goal has the same solution here. Sure most parts in KSP have the same drag/mass ratio, but rockets differ hugely in payload, staging, engine types (solids or nuclear engines will throw off your numbers more than a little), etc. Define a task before declaring a universal answer.

Who, me? I was trying to avoid making it any heavier than it absolutely needed to be. Even your "micro" is what, 175 tons? I'd be curious to see its total delta-V though (or more importantly, what it has left once it reaches orbit). My feeling was that engine mass costs delta-V, although the improved TWR might make up for it a bit on the Mun landing and takeoff...

Those are some way too specific numbers K^2. All designs are different. Minimum-mass designs can easily end up with significantly lower TWR of 1.5 or less. A rocket SSTO on the other hand can have its TWR change dramatically through the flight.

Depends on your TWR - if you have a high TWR (above 3 or 4), keeping up with terminal velocity will mean a lot of drag in that 10-20 km stretch, but you don't really need to punch it that much. I believe the Goddard problem only strictly applies if you're going for max altitude, rather than orbital insertion. Waiting until 25 km to even start the gravity turn is probably too late for most rockets though. Above 45 km there's not much drag left since the atmosphere's so thin, so Oberth beats drag - I usually aim for a horizontal flight path around 50-55 km, and my circularization burn ends up much smaller.

The max surface speed at which the turbojet generates any thrust is 2400 m/s. If you somehow had infinite intakes with zero drag so you could get up to this speed purely on jets at the very top of the atmosphere (69078 m), your orbital speed would then be 2594.6 m/s (flying east). Using planet.determineOrbit from https://github.com/numerobis/KSP-scripts, that would give an orbit of 577738 by 69078 m. You're not escaping Kerbin or hitting the Mun's SOI with that. Would be interesting to run the numbers for Laythe though, see if getting back into a Jool orbit only on jets is possible... More realistically, for best results use 1 jet and a total mass of no more than 10 tons. You can use more jets for heavier rockets, but flameouts will cause you to lose control - unless you use MechJeb 2's anti-flameout feature. Add a bunch of ram air intakes (use cubic octagonal struts to mount them), aim for level flight around 35-40 km altitude. Keep a close eye on the engine, kill the throttle immediately if it flames out, that'll bring it back online and you can feather the throttle back up to just less than what you previously were using. With this technique you should be able to get "orbits" of 100 x 40 km, then you only need a tiny rocket burn at apoapsis to bring Pe > 70 km.

Here's my lowest-mass rocket, 1.64 tons, that gets 2 Kerbals (on ladders) to orbit and safely returned to KSC with a powered landing.

I think the Dres transfer number on this is overly optimistic. The inclination and eccentricity of Dres make things a little trickier, but I've seen closer to 1500-1600 m/s on the ejection burn rather than the 1300 m/s that this map seems to suggest. I wonder whether the original author got these from simple calculations or in-game experiments...

The MJ2 maneuver planner is pretty awesome for getting transfer windows right (only 85 m/s from 10 km Minmus orbit to transfer to Mun!). And it has a real Lambert solver for course corrections. Even with Minmus at nearly 90 degrees from its AN/DN, it got me there with only 40 m/s correction burn. And I might be imagining this, but it feels like the "Land anywhere" autopilot is much more aggressive in its suicide burn than it used to be.

"Deemed impossible" with flawed math. Here's my post from that challenge, resurrected thanks to Google cache: ---- Skepticism was warranted, but this is actually possible (barely) with a stock design: http://imgur.com/a/6bUK3 This was a damn tricky design to actually pull off. For rockets with only a single type of engine, the max possible single-stage delta-V is 6611 m/s (vacuum, for an LV-T30 with zero payload and enough fuel+tanks to give TWR=1; the best atmospheric number is 6405 m/s for aerospikes). But if you mix engines, then you don't need TWR > 1 for all engine types, as long as the nuclear engine has enough thrust to land on the Mun with all the dead weight of empty tanks and inactive engines. What I wound up with was 4 solid boosters, 4 LV-T30's, and one LV-N. The taller fuel stacks drain first so 2 of the T30's burn out at high altitude, giving better weighted-average ISP, and lower TWR which makes the final circularization burn shorter (the last 2 T30's burn out partway through circularization). The only thing I forgot was a decoupler below the pod, as shown the parachute broke off :-( I also broke the rules (so I won't bother counting a score) by having Mechjeb fly my ascents, do circularization burns, help on the landing (to hold vspeed at the end of the suicide burn that had to start at 24km since the TWR was so lousy), etc. I'm an engineer, not a pilot! I was mostly out to determine whether this is possible, and repeatability on the ascents is vital for fine-tuning borderline designs like this one. Conclusion: You can actually get to the Mun and back without using airbreathing engines (or ions) and without any staging, but it's a ridiculous way to do so. ---- Back in the present, I have a slightly redesigned version that uses massless aircraft gear instead of legs so it can carry enough chutes to survive landing back on Kerbin, but I don't particularly feel like flying it manually.

Wasn't sure if it would be enough, but turns out just adding 2 more tons of fuel to an existing design of mine did the trick: http://imgur.com/a/SJlfb#0 19.22 tons on the pad. Went to Minmus first, then Mun, then just barely had enough fuel left to return (went around for a couple extra aerobraking passes).

Have you played with MJ2 yet SunJumper? It has a wonderfully useful "prevent jet flameout" feature. But I think jets are overpowered, the lightest way to get anything to orbit at the moment is: add enough jets to get the payload off the ground, then add tons of intakes so you can reach orbital speed at around 40 km altitude. Ruling them out for any mass-based challenge seems reasonable to me. It's likely the jets will get nerfed at some point (will probably have more to do with nerfing the intakes, so they don't just stack, than the jets themselves), whereas rocket designs hopefully won't need too much changing when the drag model gets redone.

Here's something a little more practical then, using an LV-N: http://imgur.com/a/TaGvs#0 11177 m/s in orbit, did Gemini, Proper Job, Burn Prograde!, Moar Power, One Small Step, One Medium Hop, and One Giant Leap. It got on a return trajectory too, but only gets credit for Return Ticket if you don't have to land on Kerbin for that one... I think I went for too much fuel, and skipped the chutes, in pursuit of that last tiny bit of delta-V. 350 m/s atmospheric isn't enough to do a powered landing with a TWR barely above 1 in Kerbin gravity. A lot of the flight path was pretty wasteful, especially when MechJeb went nuts a couple times on the Duna transfer and correction burns before I caught it.

K, thanks for the clarification. Also guessing you're requiring a manned capsule, Kerbals on ladders would be far from fool-proof. I'll build something shortly.

It takes less fuel if you land on Minmus first, then the Mun... you want to bring less mass down into the bigger gravity well. Aerospikes are banned but not LV-N's? Otherwise stock+MechJeb?