Torsion test simulation

Hi all,

I am simulating torsion tests (after being successful with tension and compression tests). For tension and compression, I got a good comparsion with experimental and behaviour in tension is same as in compression (of course leaving the necking part).

For torsion tests,  I have taken equivalent tensile stress = shear stress*sqrt(3) and equivalent tensile strain as = shear strain / sqrt (3).

This is based on Von Mises criteria.

But I have a problem, doing this I find that there is a variation of about 6-8 % from the uniaxial tension test and compression test . IS there some approxiamtion in Von Mises criteria of equivalent tensile stress and strain?

I tried to google and more and this what I found:

http://books.google.co.in/books?id=FZtsnxC8zFEC&pg=PA189&lpg=PA189&dq=torsion+test+equivalent+tensile+strain&source=bl&ots=W9l9Yvj3Rq&sig=r4Isl6IiEVqNn3KlgC5IGcAYsfs&hl=en&sa=X&ei=UBqWT9KdKYSGrAeG563QDQ&ved=0CFoQ6AEwBzgK#v=onepage&q=torsion%20test%20equivalent%20tensile%20strain&f=false

From this ; am I correct below (you can read the abstractabove as well):

 “Torsion equivalent stresses obtained using Von Misesrelationship (as I have done) are 6.5% lower than those obtained using BishopHill formulation which is more accurate”

So, can we say that torsion equivalent stresses will bearound 6.5% less than that obtained using uniaxial tension ?

Warm regards,

Kajal