Toroidal Strutting: Preventing Structural Failures by Converting Buckling to Rocking

[Tonnetz]

Discovered this around the 8th tier of my BTSM campaign while trying to fight launchpad wobble. I couldn't find it mentioned already, so I'd thought I'd post.

Rag-doll physics is not terribly kind to large rockets. Some are subject to seemingly random structural failures, especially when the physics kicks in, when engines light, and when engines cut out. Others pick up resonance-like shaking, eventually tearing themselves apart. This latter effect is exacerbated, not countered, by SAS.

To the questions I could find on the subject, answers (other than well-it-works-for-me answers) tended to fall into three categories:

1. Add more struts and more launch clamps, often to excess.
   
2. Use a particular strutting pattern, usually cross bracing.
   
3. Use smaller parts.
   

These are all good suggestions, but none of them is consistently helpful; sometimes you can try all three and still bounce apart before liftoff. So I want to add to category 2 something I've found to be more reliable: Toroidal Struttingâ„¢.

The general idea is that you want your structural connections to lie along the surface of one or more imaginary "donuts". For a typical rocket design, the donut's inner wall will comprise tanks, engines, and the inter-stage stack decouplers, while the lower surface will be the wider early stages.

As for the outer wall, I tend to build rigging towers atop my first-stage tanks, joining them with struts along the toroidal perpendicular. My torus's upper surface is formed by rigging struts from the towers to later stages. However, experimentation suggests that it's the donut shape, not towers and rigging specifically, that matters.

As best I can tell, the reason is that open ends in the structure graph don't damp relative orientation differences ("buckling"). Toroidal strutting, on the other hand, keeps these perturbations propagating around cycles, damping them at each junction. Lateral motion can still accumulate ("rocking"), and will sometimes tear off a launch clamp, but it won't snap the rocket itself.

(Obviously, toroidal strutting alone won't hold a rocket together; you'll still need some bracing. And I've found that some designs require more than one torus.)

An example, my Quebec 0, with traditional strutting (835 parts):

6 seconds after launch (full throttle, RCS enabled, SAS enabled, liftoff):

The same rocket, with toroidal strutting (817 parts):

102 after launch (same procedure):

(Okay, you can't see much, but notice that the log shows no damage until the first stage separates and gets scorched by the second.)

My KSP version: 23.0 (Yeah, I know; I'll update sometime.)

My mods: BTSM (and its dependencies) and a few cosmetic mods

[Pecan]

Nice first post - welcome to the forums :-)

I think you're talking about the way spokes are arranged on a classic lightweight wire bicycle wheel, and for much the same reasons:

http://en.wikipedia.org/wiki/Spoke#mediaviewer/File:Bicycle_wheel.jpg