Nao

Very informative post Smidge , but i'll nitpick at one thing: During straight up burn, you constantly burn "prograde" so you technically don't loose energy going up the way normal launch to orbit does as this is your escape direction anyways. No horizontal velocity needed. While the method is indeed "dumb" if we have very High TWR rocket the difference between direct ascent and straight up is very small. There was a thread like this many months ago. During testing the least dV to SOI escape (at some speed) was something in the middle of direct ascent and the dumb one. Also I think the dumb way doesn't loose any more energy due to gravity than a normal launch. As you say yourself "Work against gravity is path independent.". I'm not well versed in orbital mechanics to show it in math (and there isn't any real life calculations to draw from) but i believe the difference in performance between dumb and direct path is more to because of the Oberth effect due to having much less time to accelerate when going straight up.

Turn on the Jetpack and chase the sucker! Land on top of it and board (If you use mouse for attitude control and only one direction of thrust ex [shift] catching up should be no problem.)

On my old PC (DDR2) i did some Ram testing in regards to KSP and Ram timings. Since low timings give quite big performance increase for Dwarf Fortress, (even to a point where DDR2 is faster than DDR3 due to lower innate timings) I thought it could help with KSP too. After several tests one thing noted was that lowering the timings (and frequency a little to keep it stable) did not give any noticeable performance changes. So in your case 2,4 GHz CL11 seems pretty good. Still i didn't see anybody post actual results of changing timings on modern ram for KSP so you can be the first! Go forth and do science!

With current drag model lifting the payload to orbit on planets with atmosphere with best mass efficiency is around 1.7 TWR (1.8 for best fuel eff.), that means you start the stage at ~1.4 and end it at 2.1. Ascent from atmosphere less bodies more depend on ISP of engines used, and its from only 1.2 for LN-N to 1.5+ standard ~350-370 Isp engines. When in orbit the less engines the better but burns below 0.2 TWR start to suffer from increased dV demands due to Oberth effect and steering losses. I would say 0.1 TWR should be a lower bound for standard burns. For ascent on Kerbin TWR lower than 1.3 is quite wasteful as you burn much more fuel to fight gravity than you would burn fighting aerodynamic drag. For a rocket with unlimited thrust the best ascent TWR is around 2.1 . 2 for balancing the aero drag with gravity drag one to one, and 0.1 for acceleration of the craft itself. But because engines have weight too, its more efficient to use less powerful engines at 100% constantly than throttling down much heavier engine that can give above 2 TWR at launch. Oh and a fun fact for atmosphere-less planets: If you want to land and ascent on only one stage with LV-N engines, best starting TWR is actually below 1 (0,8-0,9 depending on planet/moon). edit: agree with above ninja

Thanks Amram for the links! I worked with CFD for a while so i do know how wings work, but i've never been able to explain it well in simple terms. And since most of the "magic" is happening at the top of the wing i've always tried to put emphasis on pressure instead of air deflection as the latter lead to questions like "then why doors aren't good wings". Never thought about using the notion of Conada effect thou , that will help getting people to understand laminar flow. @5thHorseman ROTFL good one !

The Zisteonian Empire is overtaking Kurt Republic the space race quite badly, even thou it started from 0 only recently. Hes getting better quite slowly but that's part of his charm .

Also you can barely ascent form Munar surface to above 5x5km orbit on that. Just like it was in previous versions.

They showed source code because the studio was disbanding if i remember correctly. And it was long after the game was released. KSP have a long term development approach so its kind of hard to compare them. That said and considering many other things, I think there would be more good than harm if the source code was leaked right now probably. Also while FS2 Legacy is great, does any original creators actually benefit from gog sales?

Welp, he plays Starcraft mostly, where being in orbit, means being higher (also some maps are in space with standard surface physics ) but that doesn't mean the game is bad or casual (quite the opposite). Personally I blame the lack of space knowledge on movie industry as it still sets the bar for other media like games. Watching Elysium (2013) gave me cancer (mostly because it had so much contrast with super simple physics and the advanced realistic concept of "open" colony).

The two main problems of multiple wings are simple: 1) As the wings work on pressure (top of the wing having less pressure than bottom) another wing right above the first one will cancel the effect out (thats why on modern 2 wing aircraft like acrobatic Pitts top wing is placed more forward) and 2) Up to 40% of wing drag comes from induced drag that forms at the wingtips, more wingtips more drag, thats also why gliders have very long wings and many modern planes have winglets.

The problem is that while one wing has a set of properties, if we add several wings together just adding their values won't create realistic properties for the construction. Real life example of that are multi-wing aircraft, just as we saw in aircraft development in first world war, they were adding more wings on top of each other to increase lift, and there were severe diminishing returns from each one added. In the end we ended up with one wing as the most best option. This means we need new base properties model for each plane configuration. Modern simulators have limited module count that allow every part to be computed for limited set of flight states it can get itself into. In KSP we can have 200 wing plane hit the launch tower and disassemble almost randomly, only the simplest flight models can compute flight properties on the fly for each separated element. ...argh didn't see Eric answering that one >_< , aw whatever posting my answer as well away ~weee

Having meaningful CFD computation requires tons of power. Maybe supercomputer could run simple simulations in real time. And simplified models like the ones used in flight sims use a lot of precomputed parameters. In KPS the problem would occur when the ship itself changes, for example during staging, as it would freeze the game for a moment to create new model for each part even if this was done on the other seven cores. I don't have much knowledge on modern flightsim physics models so there might be tricks that could be used to bypass these problems, but i'm sure it wouldn't be easy to implement in KSP, and even harder to put in a thread separated from main physics thread.

And yet here are a lot of people who can only get to Mun with Mechjeb, and are having great fun. Hard is good, but the Formula for good game is "easy to learn hard to master" So it's either some compromise to make things manageable for everyone or having an option to select simulation accuracy. (but that begs for more changes in other parts of the game as well, so not going to happen soon) There are two notable problems with accurate airflow simulation. 1) Real time computing of advanced flight model it would require pre-computing the flight characteristics of a plane. Since KSP planes can have many parts, it would be almost impossible to do, as there is to much configurations for one plane. And computing things as the situation changes is out of the questions on PC's for probably maaany more years. 2) Most KSP part's and general rocket shape resulting from them is quite suboptimal (simplified shapes, lack of many transition parts etc). That could cause lots of problems, forcing implementation of internal "fly be wire" system to have somewhat predictable and consistent controls, and that has its own can of problems. For example currently wings are just boards with sharp edges. Without preset properties for each part they wouldn't work as good wings at all, and then if they had one, interaction with other parts like more wings around or other parts placed on wing would be almost impossible to predict and include in the parametrers. This is why only simplified modes will work, and Ferram has made excelent work creating one. But I think before it can be implemented it would need some tweaking to balance characteristics for both untrained pilots, and game veterans. Actually i think a Ferram model + some sliders for adding simplifying counter-forces forces that would help getting predictable and forgiving flight characteristics would be the way to go (default: on, with some information popup, like the one for physics acceleration).

Unfortunately things have changed and releasing the source code isn't practiced anymore. I'm sure thou that the recent success of many sandbox games, means we will get some competition to KSP, (like OrbiterSpaceProgram or MinecraftSpaceProgram ) and maybe even KSP2 after some more years. With both the budgets going up (including kickstarters, and early access sales) and more advanced game engine toolkits being developed and sold at lower prices, the need for some "original code to build upon" is diminishing anyways.

This thread has made me realize KSP forums are really quite serious, normally i would made "ROTFL" upon seeing such request but I'm giving "Good luck with your project!" instead <3

Ohhh man ... finding Day[9] on a kerbal forum, what a crazy Day <3<3<3 his original work on Starcraft, thanks for sharing! (I guess i need to pop up in SC2 forums on TL.net from time to time to find more of such gems hehe.)

I wonder if it's a bug as for me they always switched at the same time, even thou one of them was starved for air and lacking in thrust. With no other parts KSP devs showed interest in creating performance tiers, why would they make one now? Lets also not forget about sandbox mode. I'm not arguing current RAPIER is perfect, but it's not as bad as you make it look like. I bet performance loss over choosing it instead of turbojet+spike is like 10%.

Actually the difference in efficiency between asparagus staging and lack of it is only, ~10% for Kerbin ascents. The real benefit of asparagus lies in the actual design process. It's easy to just slap "booster" stages around until we get enough deltaV for orbit. The non asparagus method requires more calculation (every stage needs to be set with correlation to each other) and more "conical" rocket design that creates problems with wobbliness etc. Agreed, but that's in return needs fairings and other aerodynamic parts to come to stock game first. Personally i think it would be best to add proper lift drag polars like CombatWombat suggests (an lift-speed polar actually, similar to the thrust-speed polars of jet engines) to wing surfaces instead of current constant coefficients and stick to stock drag model until the game can support new aerodynamic model with dedicated (hopefully procedural) aerodynamic parts. (sorry if that was already mentioned, ill try reading the whole thread later)

The system for calculating lift is quite simple right now. The lift formula is similar to the real life one with two main exceptions: 1) velocity is not squared in the lift formula and 2) lift coefficients are increased to compensate. As the velocity in drag equation is squared like in IRL, wings generate much more lift than drag at small speeds, and vice versa. This simulates lowering of lift ratio for supersonic speeds well but it make the lift disappear quite quickly right after. One thing easy to implement would be to use is an "lift coefficient - surface speed" curve, instead of constant a value, just like the jet engines thrust-speed curve. That would both eliminate the silly low speed "infinite glide" of wings and allow for realistic supersonic gliding without excessive wing part count. This wouldn't increase CPU load and would work well with the current dumb-as-rock drag model. Thanks for pointing out that there is some more in-depth discussion in the above mentioned thread. Ill try to read that tomorrow.

I think the first part of jet engine balance should start at new aerodynamics model. Currently the lift to drag ratio of wings i falling quite fast as speed increases. It means that at high altitudes the "lift" comes almost entirely from angled engine thrust. So current planes need a lot of engine power, almost following rocket ascent flight profile. If we could have high altitude/speed aerodynamics working better, lowering jet engine thrust wouldn't be a problem. That said, I really like the current lift/drag formula as it is clever, elegant, simple and more importantly works well in creating realistically looking flight profiles and craft designs. Right now modifying the speed-thrust curve could be the best way to balance turbojets. Probably having max thrust at ~700m/s (down from 1000m/s) and half thrust at ~1500m/s (down from 2000m/s) would be enough to bring it more in line with rapier while retaining its unique super high thrust and maximum speed characteristics. As for RAPIER, in the end i think it all comes down balance of jet-rocket power. If the new RAPIER had 200 jet and 150 rocket thrust, with the same other stats, it would have much warmer welcome.

Yup have this problem also, both in space and on launchpad/other planets. Around half of the time, even with the simplest crafts, Kerbals just disconnect from the craft as soon as they appear when going EVA. (playing stock game)

Except when it has 10+ 175 /9,81/1,75 = 10,2 I agree that it has less power for more mass than the Turbojet, but frankly its the Turbojet that's overpowered as it was the "one engine for all" thing and now as we add more parts there is no room for variety without power creep. It's rocket TWR is 11 to aerospikes 11,9 big difference here . Also while its jet twr at launch is only 5, but we should not forget that its actually two engines not one, numbers wise its twr is fine, its the balance between jet and rocket power that is "lacking".

The breathable air in resource tab is only the air that is not used, when it drops to around 0 we get flameouts. Now in 0.23 there is some new logic concerning resources distribution. It looks like as the engines are reduced in thrust and liquid fuel flow accordingly to how much breathable air they can get. One engine gets it all to function properly, and the second one is choking but still producing thrust. This *could* happen with turbojets in 0.22 patch but it required closed radial air intakes on the ship. (glitch) There are also some more bizarre things happening as we reduce the thrust a little, both engines could start choking at the same time, or rapidly switching the choking from one engine to other. With these glitches it's possible to ride the RAPIRERS as well as other jet engines well beyond their normal capacity. As much as this is bad design wise, i have ton of fun getting these stubborn things to do my bidding and surpass what was thought to be impossible, feels like really pushing the boundaries of a new not yet understood technology <3.

Oh just as a heads up. The values for tweakables are saved for each craft individually in craft file. It's quite easy to edit them to whatever you desire. It's not really a solution but works as a temporary fix

Speaking of witch, I think the new update brings some changes to the way air intakes work. Using only 4 ram intakes on a test plane (at least 1t of dead weight) i got to 2326m/s. Also Radial intakes seem to be quite good now. I think it would be good time to post a new thread for the 0.23, it would clean things too Oh and as long as the craft doesn't have any oxidizer on board i see no problems using RAPIERS, they aren't as good as Turbojets, as their max speed is only 2200m/s.