Transmission guys love to complain about friction coefficient, and with good reason. If you get the wrong kind of friction, those gear changes that you don't think that much about - they can turn into obnoxious events in your car. Traditionally, we complain if the static friction coefficient is a lot higher than the dynamic coefficient. It even gets its own name: "roostertail," because of the shape of the coefficient graph produced by an SAE-2 test.
So, friction is good but "roostertail" friction is bad. Here's a plot of simulated vehicle acceleration I created, and it's showing something that looks a lot like roostertail. It's the little bump at the end of the shift, inside the red oval:
That looks like a roostertail friction effect. The shift is doing its thing and then right at the end there's a bump. Roostertail, my old frictional frenemy.
The only problem with this diagnosis is that my simulation doesn't include roostertail friction. I know this for sure because I typed every line of it. So where's the bump coming from?
Turns out there's plenty of ways to get a bump. One of them is to get your shift and your converter K factor to fight it out.. Here's engine and turbine RPM, doing their gear shift thing. Looks pretty good from here...
...but converter torque is NOT cooperating. In fact it looks like my lookup tables flatlined, and if I had a more complete data set I betcha the bump would be worse.
Also occurs to me those runs are probably a breakpoint in my simulation development, since I'm porting my models into compiled C lately. I'll still be using Gnuplot for visualization... but wow! looking back it's amazing how much work you can get done with Notepad++ & Octave. On to Eclipse & MinGW now...
- nzvyyx
