KSP Stress Model & Rockmax Brand Adapters, 1 & 2

[Wcmille]

(Aerodynamics aside) do adapters, like the two Rockomax Brand Adapters affect the structural integrity of a craft, or are they merely aesthetic? In other words, if an adapter is placed between a 2.5m tank and a 1.25m tank, is that design stronger in some way than a 1.25m tank connected directly to a 2.5m tank?

Is there a (hopefully launch pad) experiment that can be devised to prove either case?

What is the KSP stress model? How does it determine when two parts have undergone enough shear, torsional, or tensile stress for their connection to fail?

 

Edited January 8, 2016 by Wcmille

[MaxL_1023]

I am interested in this as well. I know they help aerodynamically, but I am not sure about structurally. I think they might, as usually structural-type parts (like adapters, struts and empty fuselages) are stronger than fuel tanks. 

Mathematically, you should get an advantage by allowing most of the structural load to be transferred to the 2.5m side of the adapter (assuming the load is coming from the 1.25m side). Basically, your load appears on a 2.5m x 2.5m connection with a fractional load (related to force moment arms and other lovely stuff) on the 1.25m x 2.5m side.

I recommend Kerbal Joint Reinforcement though - most KSP rockets are wet noodles compared to what you would expect and need for larger launchers.

[Snark]

Aerodynamics are the reason for them.  Prior to 1.0, when aero wasn't a thing, there was no point in using them.

It's always less strong to add more pieces in a stack, because that means more joint locations where bending can happen.  Individual parts in KSP are rigid, but they can bend at the joints where they connect.

The stiffness of a joint is a function of the node size-- i.e. two 2.5m parts joined together will be stronger than two 1.25m parts joined together.  Docking-port-to-docking-port joints are considerably less stiff than "solid" joints between connected parts.

Some folks like to use Kerbal Joint Reinforcement, but I prefer to steer clear of it myself-- I like the added challenge of the bendy model in KSP, the fact that it requires me to design around structural limitations.  I launch big ships all the time and have no problem with it-- they just require a bit of design thought.

<gratuitous_advice>

If you're relatively new to KSP, my advice would be:  try playing at "full difficulty" for one career game, so you can learn the basics of KSP unaided.  In other words, start out without mods that make things easier or less challenging initially (such as Kerbal Engineer, MechJeb, or Kerbal Joint Reinforcement).  Once you've got a career game under your belt and you have a better idea of how you feel about the way the game is balanced, then you can make your decisions about just how much assistance you'd like, if any.

(The reason I suggest this is that if you start out with the extra assistance, it can be difficult to wean yourself off it later.  I don't use MechJeb or Kerbal Engineer myself, but from reading a lot of forum posts from folks who do, my impression is that the happiest and most successful users of such mods are the ones who don't actually need it, and are just using it to skip tedium, rather than to do things that they couldn't do without it.)

</gratuitous_advice>

[steve_v]

36 minutes ago, Snark said:

The stiffness of a joint is a function of the node size-- i.e. two 2.5m parts joined together will be stronger than two 1.25m parts joined together.

Out of curiosity, how is this applied to a joint between dissimilar node sizes - i.e. a 1.25m part connected directly to a 2.5m part?

On KJR, I got tired of strut-spam and engine shimmy long ago. I hear it's much improved now, but I kept KJR for it's other benefits - such as physics easing for fewer launchpad explosions.  

 

Edited January 8, 2016 by steve_v

[Snark]

5 minutes ago, steve_v said:

Out of curiosity, how is this applied to a joint with between dissimilar node sizes - i.e. a 1.25m part connected directly to a 2.5m part?

I have no idea.    I would guess that it would be a case of "weakest one wins", but I don't know what the actual code does.

[steve_v]

Yeah, that's my guess too. Nevermind, just a passing thought.

[swjr-swis]

Interesting question.

As for experiment: since the lateral forces appear to have a much faster/stronger breaking effect on part connections, I would suggest:

• attach a reference part to two launchpad clamps (only one would still allow the part to swing). Preferably a large rocket fuel tank, to maximize mass and to fuel the engine(s) that will exert lateral forces.
• stack connect the part(s) you wish to test for connection strength.
• surface attach one or more Vectors to the part to test, as close to the conection edge as possible.
• throttle down to zero, stage the vector(s), then slowly throttle up to see when part connection breaks.
• Alternatively: in consecutive tests, preset throttle to increasing values, then stage, to more closely mimic sudden impacts. Observe at what throttle value parts separate.

 

[Snark]

54 minutes ago, swjr-swis said:

Interesting question.

As for experiment: since the lateral forces appear to have a much faster/stronger breaking effect on part connections, I would suggest:

• attach a reference part to two launchpad clamps (only one would still allow the part to swing). Preferably a large rocket fuel tank, to maximize mass and to fuel the engine(s) that will exert lateral forces.
• stack connect the part(s) you wish to test for connection strength.
• surface attach one or more Vectors to the part to test, as close to the conection edge as possible.
• throttle down to zero, stage the vector(s), then slowly throttle up to see when part connection breaks.
• Alternatively: in consecutive tests, preset throttle to increasing values, then stage, to more closely mimic sudden impacts. Observe at what throttle value parts separate.

 

That's an interesting idea for an experiment!

In my experience it's pretty rare for things to actually break-- of more interest is their flexibility.  They can bend a lot before they break loose.  A more interesting test might be to surface-attach a Vector as far from the joint as possible (to give it a decent lever arm), and then throttle it up to observe how much it bends.  Then you can compare bending amounts for a 1.25m join, a 2.5m join, or a 1.25m-to-2.5m join.

[Wcmille]

swjr-swis's experiment seems to verify Snark's comments.
 

My test rig was a full 1600 2.5m Rocket Fuel Tank, firmly attached by 5 structural clamps. I then attached two 1.25 800 Rocket Tanks end-to-end to the fixed 1600 tank, a 1000 1.25m battery and a 1.25 probe core to the end of the tanks for control. I then added 3 vector engines (facing the same direction) to the very end of the 1.25m assembly. The vector engines were set apart so that they just barely touched, at the far free end. Setting the engine strength to 87% each, this caused a failure in the Rockomax adapter, no failure with no adapter, and a close failure with the Rockomax 2 (the lighter, shorter one).

I did this, rather than ramp the vector thrust, because I noticed that changes in how the thrust was applied occasionally created a hard-to-reproduce shock depending how fast I applied it.

I don't think this models real world stress; I believe the adapter version should be stronger, for the reason MaxL_1023 suggests.

 

[swjr-swis]

6 minutes ago, Snark said:

A more interesting test might be to surface-attach a Vector as far from the joint as possible (to give it a decent lever arm), and then throttle it up to observe how much it bends.

So to properly science this: do both, and test breaking torque as well as breaking force.

I think whether one or the other happens more readily is somewhat dependent on the parts (which probably relates to the part breaking force/torque settings). MK3 parts attached to each other seem to have a tendency of simply 'letting go' of each other, rather than wobbling flexibly like most smaller form factors.

[swjr-swis]

11 minutes ago, Wcmille said:

Setting the engine strength to 87% each, this caused a failure in the Rockomax adapter, no failure with no adapter, and a close failure with the Rockomax 2 (the lighter, shorter one).

Indeed, not quite what one would expect from a dedicated structural part.

Now I wish we had a working wind tunnel to also test the aerodynamics properties/differences of those combinations. Considering the less than true-to-real-physics drag model stock KSP uses, I suspect aerodynamics might not be all that much of a benefit of using the (now experimentally proven) weaker connection method.