Avoiding "Ghost Forces" below 750 m/s

[Ten Key]

I am having a problem with what can best be described as "turbulence" any time a craft drops below 750 m/s orbital velocity. It is an abrupt transition. . .751 m/s is fine, 749 m/s is not. This is not an RCS/SAS issue. With the RCS disabled, SAS disengaged and all engines deactivated, I can watch the turbulence start right at the 750 threshold, with the NAVball suddenly pulsing in random directions. It's almost like the ship is colliding with something.

And if the reading I've been doing is correct, it is colliding with something. The ship is colliding with itself.

This seems related to the Krakensbane fix, whereby KSP switches how velocity is handled to avoid physics errors. According to the API entry on the wiki, the transition point is 750 m/s orbital velocity.

After digging around on the internet for a bit, the general consensus seems to be that part clipping is a major contributor. My first Mun probes did not suffer from this, though there was some minor part clipping involved. My later Mun ships were heavily part clipped, and suffered from these "ghost forces" quite a bit. It seemed to get worse as more and more objects were placed in Munar orbit, to the point where the landers, transit ship and supply station were behaving like they were full of angry cats.

I found the following in a wiki page regarding "closed loop" construction work arounds. . .

"Key to the paired-port strategy is correct assembly. If parts overlap, even slightly, then "ghost forces" arise that cause uncontrolled attitude changes, even changes in velocity vector."

Hoping to avoid this, I built a new transit ship for Minmus that was completely devoid of part clipping. I sent it on along an elliptical Kerbin orbit as a test, and sure enough, at exactly 749 m/s the 150 ton ship started tumbling every which way, like it was being poked by an invisible giant every few seconds.

Let's go back to my first Mun probes for a moment. There was some part clipping baked into them, but they did not have this problem. But they also did not have any struts on them. I started adding struts to probes and spacecraft after one of the probes had serious flexing issues during a burn. And after that, everything I've sent out that drops below 750 m/s has wanted to gyrate all over the place. Having taken considerable care to avoid part clipping on this new ship, the struts stand out as the obvious culprit.

This is driving me nuts. A big part of my enjoyment of this game is "OMG SPACE!!", and I don't get that from watching missions play out from the tracking station, or from having to constantly pump time compression to hold a ship steady enough to interact with it. Manual IVA docking in Munar orbit is an absolute slog, to say nothing about having to account for ridiculous reserves of monopropellant.

Obviously, struts cannot be avoided on ships of any real size. I've been reading through some of the threads devoted to Kraken Drive development, and I've been wondering if the design community has found a way to deal with the "ghost force" issue.

Is there a way of placing struts so that they do not trigger collision forces inside a ship? Are there specific parts that tend to cause this? Has anyone else noticed that ghost forces tend to get more pronounced as the number of vessels in orbit goes up?

Edited May 5, 2015 by Ten Key

[panzer1b]

This must have something to do with kraken-drives, as those essentially explode when you drop from above 750 to below 750m/s. So there is a transition point at 750m/s for something (k-drives go from anti-grav below it, to 20+ Gs of linear acceleration above it.

Anyways, ive never had major issues with phantom forces except for some parts such as landing gear, legs, claws. If its animated and movable, it can cause some issues when clipped or partially clipped into other parts.

Then again, with conventional parts, i do not believe you can even have issues unless something detaches and remains inside your craft. Even stacks of 10+ intakes all inside each other dont case any issues, until ofc one of them breaks off and decided to explode the rest of the ship.

[Ten Key]

Did a bit more digging, and found this. . .

I've seen two causes for phantom forces causing this kind of rotation.

The first is as mentioned, part clipping.

The second is more annoying since it's harder to avoid. Sometimes, struts just do that. The first time I encountered this and it wasn't clipping was when I was doing a mothership to Jool type mission, and the phantom rotation started up in LKO. I eventually found that by changing the position of the struts that the phantom rotation would go away. So I did a shakedown cruise to the Mun and Minmus, and the phantom rotation returned in Minmus orbit. I deleted all the struts and put them back in approximately the same place and went on the shakedown cruise again, and had no phantom force problems.

The second time was on a grand tour mission, where there was no phantom force problems in LKO, the Mun, or Minmus (yes, because of precisely this issue, those are now required stops on any mothership-style mission that I do). However, it started up in Moho orbit, the first stop of the grand tour. This time, I did more thorough testing. I created a new drive section, hyperedited it into Moho orbit, and it started spinning. I stripped off all of the struts in the drive section, hyperedited that into Moho orbit, and there was no spinning, but the craft wasn't stable enough to launch.

I then returned to the original design and stripped off the bottom-most 8 struts (out of 16 struts total, if I remember correctly), hyperedited that into Moho orbit, and there was no spinning. Several attempts at replacing the eight struts all resulted in spinning. So I went back to the reduced strut drive section, found it stable enough to launch, flew it out to Moho where it picked up the mission sections and completed the mission. The drive section with the problem is still in Moho orbit in that save file. In copies of that save file, I've both hyperedited the drive section back to LKO (and low Minmus orbit), and just flown it back, and the phantom forces only appear in Moho orbit. Maybe someday I'll take that drive section on it's own grand tour to see if there are any other planets/moons that cause that particular one to have phantom forces.

What I'm specifically looking for is a sort of "best practices" for strut placement to mitigate issues like this. If it really just comes down to repeated trial and error, I may go mad.

[Ten Key]

So, after repeated trial and error (), it looks like removing all of the struts from the spacecraft has mostly solved the problem. It still wants to bobble a little bit, but it's far more controllable and the ship no longer bleeds monopropellant.