View Single Post
      08-24-2007, 02:04 PM   #39
swamp2
Lieutenant General
 
swamp2's Avatar
 
Drives: E92 M3
Join Date: Sep 2006
Location: San Diego, CA USA

Posts: 10,201
iTrader: (1)

Improved M-DCT simulations

Thanks to lucid for triggering my thinking on a small error with the M-DCT sims.

Why were the time periods of deceleration too long with the tweaked M-DCT runs?? It was obvious. Not sure why I missed this - the time step is one of the most important things in a time domain simulation. The time step must be small enough to capture the actual physics (time scales) of the event! The std. time step in the software is .05 seconds. You typically get changes in in results only in the second decimal point or after by bumping the time step down drastically (say by say a full order of magnitude). Using a smaller time step for a typical AT or MT just gets you no better results and takes a lot longer - totally not necessary. But now with a .01 s time step I have some much better sims for the M-DCT! I did try much smaller time steps as well but again typically only changes in the second decimal place for most relevant results. I'll have to update the main post on this topic as well!

You'd think a guy in this business would get convergence...Oh well really happy to have some better figures. Not only do we see better performance but the results make more sense and there is sound physical reasoning as to why these sims are better. Thanks again to lucid for the "trigger" to get me on the right track here.
Attached Images
 
swamp2 is offline   United_States
0
Reply With Quote