Torsion results don't match hand-calcs

Check this thread: Help Non Linear Aluminium FEM Beginner - #6 by FEAnalyst

Basically, you need true stress vs true plastic strain data points. Here you can find some helpful information, including formulas to calculate true stress/strain from engineering stress/strain and plastic strain from total strain: https://classes.engineering.wustl.edu/2009/spring/mase5513/abaqus/docs/v6.5/books/gsx/default.htm?startat=ch05s02.html

You can use a bilinear model (hardening recommended) like this, for instance:

Re ——- 0
Rm —— A

where Re is yield strength, Rm is tensile strength (UTS) and A is elongation at failure.

Or you can specify more data points to define a while nonlinear stress-strain curve (keeping in mind what I said at the beginning about the required stress and strain measures).

Between them - yes. But outside of the specified data, it assumes plateau (horizontal line). You should avoid this due to potential convergence issues - do the linear extrapolation manually by specifying higher stress-strain values as the last point.