Cold-formed metal products are increasingly serving as high duty machine pa
rts. Designers and users need to know their properties as accurately as pos
sible. One such product property is the new yield strength, which can be ap
proximated by the final flow stress of the workpiece material during formin
g. Vickers hardness measurements provide an easy and inexpensive method of
evaluating the new local yield stress in cold-formed workpieces. The well-k
nown available models given in literature to convert the measured hardness
number into the corresponding yield stress have an error of up to 25%. This
is basically due to the facts that cold formed material experiences large
plastic strains in the main forming stage, the hardening behaviour is aniso
tropic and, moreover, the material properties are inhomogeneous especially
at the workpiece surface. The purpose of this study is to improve the accur
acy of the well-known available correlation models between Vickers hardness
measurements and yield stress. This is achieved by utilizing finite elemen
t simulations of the indentation process. The models currently incorporate
only the isotropic strain-hardening behaviour of the work material. The new
suggested model decreases the theoretical conversion error to less than 10
%.
The improved model has been verified by experiments. The difficulty in veri
fying the models is realizing an experiment with a precisely known high pla
stic strain. In this study, the forward extrusion process was utilized for
this purpose. In the forward extrusion process there is one location in the
workpiece where the plastic equivalent strain and hence the yield stress i
s exactly known: the axis of the extrudate. By this method it is possible t
o obtain strain-hardening states up to an equivalent plastic strain of 1.6
(!). Hence, making hardness measurements at the axis of extruded workpieces
, it was possible to verify the improved relationship up to realistic strai
n values. The results have shown that the new relationship supplies convers
ions with a drastically reduced error as compared to the classical relation
s.