I am full of grief and coffee

[Whackjob]

Ok. I'm a big man. And part of being a big man is admitting when you need help.

Pictured above is my prototype heavy lifter platform. The thrusters are up top hanging from the carosel, and the fuel stacks hang below, dropped off in sections as they empty. The cargo it'll haul will hang protected (?) in the center. Flight and cohesiveness isn't a problem. I can get that sucker off the ground and screaming through the air with no trouble. The problem is stability. I suspect that the piloting systems by default calculate adjusting the gimbal mainsails by assuming they're at the bottom of the craft. Might be, maybe not. Even with that off, I quickly go off centre and cannot recover. What starts out looking promising quickly devolves as the crew of the spaceport scatter.

The Ferris Wheel of Doom cometh.

What can I do to fix? Suggestions?

Sad bump. Anyone got suggestions?

Edited July 22, 2013 by Aphox
Merged multiposts

[Spatzimaus]

Ok. I'm a big man.

With a big appetite?

One thing about gimbaling is that it's not very effective if your engines are level with your center of mass, because it becomes impossible for the difference in vectors from gimbaling to create a torque around that point. (It'll do a translational motion instead, which doesn't do you much good as SAS doesn't even try to do that.) Without a way of adding a torque, you can't keep yourself aligned. In reality we'd simply throttle down the engines on one side of the ship and leave the others going at full throttle, but KSP doesn't allow for that sort of thing.

An engine at the back end won't have that problem, and neither will one far in front of the CoM; "puller" designs have been done by quite a few people without problems. A similar effect happens with flaps on a spaceplane; if they're even with the center of mass then you can't use them to pitch the plane, but canards at the front are just as effective as flaps at the back if they're far enough from the CoM. So, your rocket shifts a little during thrust, and it can't gimbal to fix it (possibly even making it worse, if it tries to gimbal anyway), and RCS can't even come close to providing enough force to make up for that. In-atmosphere, flaps can help make up for that, and outside an atmosphere your SAS can do the job. Putting even one good engine down at the bottom would help a lot, as would moving them further upward.

Which brings us to this episode of Crazy Spatz's Knee-Jerk Design Workshoppe:

1> You don't have any flaps or SAS, from what I can see. The SAS, especially, should go on every design; put an SAS on top of each of your eight mini-stacks, and an ASAS controller somewhere in he middle. This'll make it MUCH easier to stay stable while thrusting, without using any RCS. Even if you do have an ASAS tucked away in there somewhere, a large ship needs a lot of SAS torque, so add more anyway.

2> I see 8 batteries, but I don't see anything that draws power (although presumably you've got an unmanned controller hidden somewhere on there). You also don't seem to have any way of generating electricity outside of the engines themselves. One RTG goes a long way, although I also like throwing on a few of the non-deploying solar panels as well.

3> That is a metric buttload of struts. While I'm glad you've taken the MOAR STRUTS mantra to heart, they're not perfectly rigid. Things CAN still flex a little bit, so it's not impossible to still get destructive resonances. And struts, unfortunately, make your part count go up quickly. As we all know, part counts are the real limiting factors in KSP...

4> It's hard to tell if the gray cylinders you've attached the engines to are structural fuselages (which are just empty dead weight) or the type 1 jet fuel tanks which don't carry oxidizer. Either way, you should replace them with standard rocket fuel tanks if possible.

5> The thing about RCS jets is, they're sort of the reverse of the gimbal thing I said above; for RCS, you want the quads to be exactly alongside the CoM if possible. If all of your RCS jets are forward of the center of mass, then two things will happen: you'll burn through fuel faster, since it'll be trying to fire more than one jet for each rotation, and you won't be able to do any translational motion (which really only matters if you're trying to dock with something in space, but is just good design in general).

5a> And that brings us to the diagonal issue. You've got eight sets of two RCS quads apiece; assume for the moment that you took my advice and they're EXACTLY alongside the center of mass. Let's number them 1 through 8, with 1 being the "north" end. Now imagine you want to pitch your ship northward during flight (i.e., you hit the up or down key, whichever one's that direction). It'll fire the jets on 3 upward and 7 downward (or vice versa), obviously, since those ones have jets pointing in those directions. It won't fire 1 and 5, since they're perpendicular to that axis and so can't create a torque around it. But it'll also fire 2, 4, 6, and 8, even though since they're 70% closer to that axis they won't be as efficient. It's even worse for translational RCS motion, since jets 2 and 4 would be partially canceling out each others' effects, and 6 and 8 doing the same. Likewise, pushing left or right will fire 1 and 5, and not 3 and 7. But, again, 2, 4, 6, and 8 will fire. In other words, you'll burn through RCS fuel faster than you need to, no matter which way you're trying to rotate, because the four diagonal sets will always fire but have less effect for the same amount of fuel.

6> There is NO rule #6.

Edited June 22, 2013 by Spatzimaus

[Whackjob]

OH, ok. I see. I need those RCS units as close as possible to the centers of the stack. I'll try putting them on the inside of the fuel carosel and see if there's much of a change.

[Spatzimaus]

OH, ok. I see. I need those RCS units as close as possible to the centers of the stack. I'll try putting them on the inside of the fuel carosel and see if there's much of a change.

I think you misunderstood what I meant by "alongside". You want them as far from the center of mass as possible, to maximize the total torque, but you want them to be level with it along the primary axis. That is, in your case up/down is primary, so the RCS jets should be at the same height off the VAB floor as the center of mass is. This won't be a perfect thing, as your center of mass will shift in-flight as your fuel depletes, though. For something intended to do precisions movements in space (i.e. docking), you want one set of RCS "above" the center of mass and one set below. For what you're doing it's a bit less critical.

[Whackjob]

Gotcha. Ok, last test was a failure. Even with two sets of wings with flaps on each tank of the bottom two sections. Something's putting a crazy torque on that lifter, and I can't figure out what it is for the life of me. I built it with the 8x modifier, so it ought to be perfectly balanced. I've examined it very closely and I'm not seeing any differences to one side.

I don't know. I might scrap it and rebuild it... maybe put the engines further off to the side. Tougher build, but I think I can manage it.

#EDIT: Maybe go with a flatter horizontal wheel? Big pancake?

WOOOOO!

Edited July 22, 2013 by Aphox
Merged doubleposts

[Spatzimaus]

I built it with the 8x modifier, so it ought to be perfectly balanced.

There's one notable case where this isn't true: if you disabled the clipping check, to where one part is trying to pass through the space occupied by another. That causes massive asymmetrical torques, which might explain your problems, although without trying out the craft file for myself it's hard to be sure. This sort of thing is easy to test if you can make it to orbit: turn off SAS and RCS, turn on time acceleration for am moment then set it back to x1. If the ship immediately starts torqueing again, it's a clipping issue. This is a lot harder to test on the ground, though.

Beyond that, the gimbal issue I mentioned can lead to this, albeit not in what I'd describe as a crazy amount of torque. Even if your design is perfectly symmetrical, a minor perturbation can lead to that sort of effect if there's no restoring force that can pull it back, and the game's physics model has just enough inaccuracy in it that small perturbations DO happen.

[Whackjob]

Ah! That would explain a lot, actually. Thank you. No, I haven't disabled clipping, but... given the number of struts I use? A lot of times I try to make them hook into the same point. I wonder if that does it.

[Spatzimaus]

It's easily possible that the strut thing is the problem. To be clear, though, you CAN get the clipping issue even if you don't explicitly disable the clipping check. You see, the game only does the full clipping check when you first place an item, but if you then move it to a new location, it does NOT do the full check at the new location. This loophole was used by people who wanted to stack multiple jet intakes, even without the F12 option.

The point is, if the physics engine thinks two objects are trying to occupy the same location, it'll try to separate them by pulling them apart. If for whatever reason they can't be separated, and one's bigger than the other, you get a net torque that can't go away. Now, there are exceptions to this; you can usually place a small object (like an RTG or an avionics package) INSIDE a fuel tank and it won't cause problems, and you won't even need to override the clipping to get it to work. But girders and struts can cause these sorts of problems depending on your exact design; it's just yet another reasons why you're usually better off using a smaller number of large parts than to try building something out of lots of small bits.

[Whackjob]

Excellent. Thank you greatly for your assistance. I've made great progress tonight.

I have another "wheel" lifter I made again from scratch, this time being extremely careful in placement. The end result is a lifter, then when fired directly up, ran out of fuel with an apogee of 1.2M meters. Steady as heck. Gotta coma till tomorrowsday get, but in the morning, before work, I'm going to destroy my save to clean things up, and build a new one. I've already got some serious plans that'll make my burning wheel up there look like a chump thing. Hehe. Color me cursed, but I do love weird builds!

See ya tomorrow.

[technicalfool]

I'm going to destroy my save to clean things up, and build a new one.

Just in case you don't know, your "save" is a whole directory tree of stuff. You can delete the persistence file to remove ships and debris from orbit, without removing your various ship and spaceplane designs.

At the same time, you can remove .craft files from the ships/vab and ships/sph subdirectories to remove old, useless designs, without hitting the nuke button on your whole game.

Just a heads up!

[Panchopinata]

I began my career as a Kerbalnaut much the same as you did, seeing it pop up in Steam now and again and tossing it aside as something which looked boring or not worth the cost or what have you. Then, one day, I was on YouTube. While looking over my subscriptions, I spied a 0.18 gameplay video. Watching that video made me realize that this thing could have some potential to it, and so I visited the Steam page and grabbed the demo. I was instantly hooked, and after putting hours into the free version, I bought it (around late April or so.)

I haven't looked back once. You're not on the outside, clinging to the windowsill, you're on the inside of an ever expanding rocket blasting off to version 1.0 and beyond. Welcome to the club, we kept the pot on for you.

[Whackjob]

Just thought I'd share my current project:

I am extremely pleased with both how this is turning out, and how it looks. Look at that thing! It is powered by 40, yes forty mainsail engines. You can see the beginnings of the fuel reserves I'm placing in the center carousel, where empty fuel tanks drop happily away. The point of this project is to lift large components into orbit for space ship and station construction in the least efficient and most awesome way possible.

My time draws short, and the office beckons. But I will keep plugging away at it right up until the last minute.

To preempt the question, "Does that thing even FLY?"

Edited July 22, 2013 by Aphox
Merged multiposts

[numerobis]

Nice truss work! My latest rocket failed because the outer motors folded themselves around the core, for lack of sufficient trusses.

[Whackjob]

Much appreciated! The ship shown above isn't perfectly rigid, and before takeoff it tends to flex and wiggle a little bit, but that's fine. Not shown in the inner ring of tanks that I added just before coming to work. I've got the staging and fuel lines all nicely worked out, and I couldn't ask for it to run better. I've so far neglected to put any attitude adjustment nozzles. I didn't need any so far, as I've been testing just by flying straight up. I still want to add yet more fuel tanks, though I worry I might put too much weight centre and cause the stress to cause the rockets to detach. I'm considering adding a final large orange fuel tank on top of the center rocket of each cluster, just so the mass prevents additional flexing.

I did learn something new last night. Namely, I can use WASDQE to change how the parts attach. Which is nice... that's already come in handy more times during that build alone than I can think of.

[korda]

To preempt the question, "Does that thing even FLY?"

Does it fly somewhere outside Kerbin atmosphere then?

My biggest heavy lifter has only 13 Mainsails (and dozens of SRB's) and it uses asparagus staging. I kinda envy you that you can run that monstrosity - my heavy lifters are making my laptop go 2FPS already... If that game could actually use all 4 CPU cores though...

EDIT: also if you still plan landing and going back from Mun remember that you really need only something small (really small) to make to lunar orbit and then back to Kerbin. Big lander will be hard to get to the Mun, better invest in transfer stage. My Mun mission would look like that:

1) Lift rocket that would place me in orbit (or at least high enough to use LV-N to make orbit)

2) Transfer stage with perhaps LV-N as it's very efficient

3) Leave transfer stage to slow down descent, throw it away only as late as you can

4) you have small lander with plenty of fuel left now on Mun, getting back is easy thing now.

BTW, do you use quicksave and quickload?

Edited June 22, 2013 by korda

[Whackjob]

Before I added the center tanks, I was able to get an apogee of over 2M km With the new tanks, that's quite reduced, but that's because I haven't yet done my final tuning. For its final testing, I intend on landing one on the Mun.

[kahlzun]

its actually really interesting watching you go through the stages of KSP development..

If you're trying to land somewhere, Minmus is easier than the Mun (as it has much less gravity) but a little harder to get to.

[Whackjob]

BTW, do you use quicksave and quickload?

Wasn't even aware that was possible! O_o

EDIT: also if you still plan landing and going back from Mun remember that you really need only something small (really small) to make to lunar orbit and then back to Kerbin. Big lander will be hard to get to the Mun, better invest in transfer stage. My Mun mission would look like that:

I've already been to the Mun and back hehe. I actually set a goal for myself. I had to hit that milestone before I'd permit myself to buy the full game.

Edited June 22, 2013 by Whackjob

[korda]

Wasn't even aware that was possible! O_o

F5 to quicksave (doesn't work in some situations, like in atmosphere there will be text informing you that it didn't save though), press and hold F9 to quickload. Useful when you are learning how to land or doing something really hard.

[Whackjob]

Hrm. All of a sudden, inspiration strikes. I've already got the "An Exercise in Absurdity Mk2" planned out in my head. I've thought of a way that should greatly improve the design. If I'm doing all the mental pseudomath right, the Mk2 will be 200% larger, have 96 mainsail engines, be robust enough to go full throttle from the get-go, drop my framerates to one frame per seven seconds, and single-handedly deplete the ozone layer and collapse the food chain.

Oh, man! I can't wait to get home tonight!

Question: Are the physics of this game good enough to replicate gyroscopic forces? If I spin something up, will it be more stable in an atmosphere?

Edited July 22, 2013 by Aphox
Merged multiposts

[Nao]

Short answer: yes

Although parts have rotational drag in atmosphere as well as some limits on rotational speeds in vacuum to prevent physics from exploding. So it might be hard to make something spin super fast.

[Whackjob]

I'm thinking... Triebflugel.

[Nao]

Ive done something similar, it kind of works, one important thing if you want to extract lift from rotating wings is that the force of lift is calculated from and acts at the base of the wing not in the middle.

Also if it ends up being unstable, use struts, as small bending induces wobble that leads to more wobble and destabilization. Oh and if you put control surfaces at the ends remember that it works like in a helicopter - controls moved by 90degrees - that can create a illusion of instability when its just control mismatch between command pod and rotating wings.

[Whackjob]

I just noticed an odd phenomenon. I just caught myself trying to show a co-worker a picture of my absurdity project. And he goes, "Yes, I've seen it, I've seen it." Suddenly I realize that I'm acting like some office worker showing off a new baby picture. I stop in my tracks. I'm one of those people now.

Dumb question: Are there two components that can be pit together along the same axis that can rotate freely in respect to each other? I need an axle.

Follow-up dumb question: If I use an unmanned core, and add a manned can later as empty space, the game puts Kerbonauts in there even though the control core was first. Is there any way of making the game let me launch empty?

Edited July 22, 2013 by Aphox
Merged multiposts

[softweir]

DQ1: Not stock. The Damned Robotics mod has such a part, though the mod doesn't work with KSP 0.20.x (There is a hacked patch, but the "official" developer hasn't released a fixc yet.)

DQ2: There are two ways. 1) EVA your crew while the ship is on the launchpad, and get them to climb down the ship (or jump - they have a cartoonish ability to survive falls). 2) Try the Crew Manifest mod, which allows you to empty your ship, fill all non-command-pod spaces (such as in the Hitchhiker module) and move them from module to module - especially useful when crewing space stations!