Does anyone have any information on how KSP models structural stability?

[Rusty6899]

I quite often find that with larger rockets, my crafts are fairly structurally unstable.

Sometimes I get strange failures where a fuel tank will detach from both radial decouplers and Stability Enhancers, while still on the launch pad.

I'll get the odd mid-flight failure where I assume there is a structural failure leading to a large explosion.

I quite often find that for more slender rockets, my SAS tends to encourage a resonance in the rocket, leading to it snapping. I guess that's more of a obvious design flaw, though.

Basically, the things I am interested in are roughly how the failure of connections between parts are calculated (no need for precise equations, I was more wondering which parameters it considered; velocity/acceleration/mass etc/properties of the part). Also I was wondering how connections between struts were modelled. Is it simply modelled as one part connected to another, or does it take into account the exact position and angle of the strut? Also, can too many struts cause a more precarious rocket by restricting motion at all?

[Specialist290]

Believe it or not, someone did an investigation into how KSP's rigid-body physics works just yesterday. Harvester himself added a few remarks to that thread that may prove enlightening as well.