Welcome, Guest

Author Topic: Any news on tethers?  (Read 2199 times)

Gordon Freeman

  • *****
  • Posts: 480
Any news on tethers?
« on: November 30, 2014, 11:46:06 PM »
A few months back we heard about strings and even space elevator features. How come we aren't hearing about these anymore?

Greenleaf

  • Thomas Grønneløv
  • Development Team
  • *****
  • Posts: 211
Re: Any news on tethers?
« Reply #1 on: December 01, 2014, 01:06:05 AM »
A few months back we heard about strings and even space elevator features. How come we aren't hearing about these anymore?


You didn't hear, because nothing new has happened, and no one asked. Now you asked, so I can tell...
The first implementation using a 1D finite element method could correctly model different material properties, but there were issues. One was that even strong materials like steel really is irrelevant since it is too weak. The breaking length of steel is about 26km. http://en.wikipedia.org/wiki/Specific_strength
This means, as we all know, that stronger materials are needed, which could easily be defined and used as well.


The thing is just that when doing explicit numerical integration of a segmented tether like this, the more rigid the material, the more sensitive the problem is and the smaller timesteps we need. Otherwise the system "blows up", which essentially looks like a tether starting to wiggle and then become more and more violent and breaking up, since breaking strength was modeled as well. Solving tethers with tiny timesteps means that we need a lot of them to get anywhere and that takes time. In other words, it was too slow for our use, since we have a lot of other things to calculate, and 1x time (which is perfectly doable) is really too slow to be fun.


When I revisit this, and that might not be too far into the future, I will look at writing an implicit solver which should let us run with much greater performance.


Essentially, we could also add in a tether here and now which doesn't have a specific realistic material type and dimensions, and solve this using simple relaxation.