I don't think you should always assume Fcy is equal to Fty. This would be called an "even" material, which very few are.
Technically failure criteria such as Tresca an Von Mises make the assumption that the material is even because the theory essentially states that failure is based on the deviatoric stress tensor (specifically the second invariant of it), equated to an arbitrary allowable, which is set as the pure shear yield strength (they posited that the shear distortion strain energy was critical for failure). The pure shear yield strength is estimated at s_y_tension / sqrt(3). So this first of all, need to be reasonable, and second... the material needs to be even, otherwise multiple K factors would be needed.
This is the point of von Mises - to be able to use uniaxial tension stress-strain curves by effectively distilling a multiaxial stress state to a scalar value.
However, my point is... even if you don't see any real scenario in which the compressive failure would actually be in a block compression regime (it is possible if the L'/rho is extremely small, but, eh), the compressive yield strength can impact the failure envelope for tension or shear loading, if the part experiences a multiaxial stress state.
This can actually be a problem in fatigue analysis as well... the analog is what's called the Masing hypothesis. For strain-life analysis we need to rely on the cyclically stable stress-strain curve to make hysteresis loops. But usually compressive stress-strain curves are harder to come by. So sometimes we use the Masing hypothesis to effectively "double" or mirror the stress strain curve.
But this can be a problem for materials which kinematically harden (the Bauschinger effect), where we'd need some non-Masing model.
Basically in real structure, we have multiaxial stresses which can effect the material response both statically and cyclically. Most materials, including metals are not truly isotropic or even.
Now, we have to remember that these are all semi-empirical models and failure points on a von Mises envelope might see some scatter for even and uneven materials, but its "close enough". But to my knowledge most of the testing done to verify these failure criteria usually focus on the tensile quadrants of the envelope, so...
There are other modified criteria which open it up a bit by correlating to J3 instead of J2. If you're interested, look into Drucker-Prager or Mohr-Coulomb theories.
Having said ALL that:
Joel_Lapointe said:
I have material sheets in a sandwich pressed by screw.
Seems to me your scenario would actually fall more into the realm of Hertzian contact mechanics which is actually dominated by subsurface shear rather than surface or block compression. I would look into "Advanced Mechanics of Materials" by Arthur Boresi which has one of the best treatments of contact which I have read.
Keep em' Flying
//Fight Corrosion!