Mistake in boundary function (boyde2009) #1
Description
The factor (1-nu*epsilon) must be removed from the boundary function, because it assumes that the semi-minor b=c and semi-major a axes are defined as:
a = a' * (1+ epsilon)
b = c = a' * (1- nu*epsilon)
with the initial (before stretching) radius a'. However, in the method boyde.core.stress they are actually defined as:
a = a' * (1+ epsilon) / (1- nu*epsilon)
b = c = a'
- This makes sense, because it removes the necessity of the Poisson's ratio for computing the stress (and thus validates the high-level wrapper methods).
- This leads to slight changes in the stress profiles which leads to about 5% overestimation of the GGF.
- This does not affect the high-level wrapper functions or the LUT, because they set the Poisson's ratio to zero and only use it for computing the GGF.