Intro_Course_6_3 - nasa/gunns GitHub Wiki

Intro Course 6.3: Tuning

In many applications, particularly the fluid aspect, GUNNS is not predictive of real-world behavior without being tuned. Tuning means adjusting the configuration data of your links until they give the desired results.

  • Most fluid conductor links are configured not with real-world hardware parameters, like valve coefficients & duct geometry, but the more generic conductance parameter instead.
  • Even in cases where we have links that are configured with real-world hardware parameters, there are cases in projects where the hardware specifications aren’t known or can’t be given, but you have the expected outputs of the system instead.

Parameters that usually need to be tuned are:

  • conductance of fluid conductors, to create the right flow rates and pressure drops in a fluid circuit
  • reactor & absorber efficiencies
  • heat transfer coefficients of heat exchangers, coldplates, tank walls, etc.
  • expansion and compression scale factor, in very specific cases where compressible fluid cooling & heating are desired.
  • thermal conductance of conductive and radiative flow paths

We won’t go into how to tune all of these parameters, but the basic process is the same:

  • Make an initial guess in your drawing
  • Run the network, change values of tuning parameters, observe the network response, and iterate to converge to the desired results
  • Keep in mind that the network is a system, and changing one parameter often has effects everywhere else in the system. So tuning usually involves multiple parameters at the same time to keep everything in sync.
  • Often you can apply circuit theory to reduce the number of simultaneous variables you are tuning:
    • Reduce flow branches of multiple conductors in series or parallel to equivalent single conductors representing that flow branch
    • With the reduced number of variables in the equivalent single-conductor branches, tune them to get the overall circuit flow properties for the system as a whole
    • Expand each branch to its original series & parallel components, one at a time, such that the tuned branch flow rates are unaffected.

Previous Page / Next Page

⚠️ **GitHub.com Fallback** ⚠️