TY - JOUR
T1 - Influence of specimen preparation, microstructure anisotropy, and residual stresses on stress-strain curves of rolled Al2024 T351 as derived from spherical indentation tests
AU - Heerens, Juergen
AU - Mubarok, F.
AU - Huber, N.
PY - 2009/3
Y1 - 2009/3
N2 - In the present work, a previously developed neural network approach for analyzing spherical indentation experiments is applied to prestressed specimens to determine the effect of residual stresses on the identified stress-strain curves. Within this scope, a comparison to other measurement errors has been made, which are caused by surface preparation and anisotropy of the material. To validate the experimental and analysis approach, the effect of compressive and tensile prestresses was also simulated using a three-dimensional finite element model. The material investigated is a rolled 2024 T351, which is widely used for manufacturing airplanes. It is shown that the existing neural network approach is able to determine the stress-strain behavior in agreement with that obtained from tensile tests. The method is robust against most error sources, such as surface roughness, coarse grain structure, and anisotropy, if a sufficient number of experiments are available. The most important influencing factor can be the residual stress causing errors up to 20% in the identified stress-strain curves.
AB - In the present work, a previously developed neural network approach for analyzing spherical indentation experiments is applied to prestressed specimens to determine the effect of residual stresses on the identified stress-strain curves. Within this scope, a comparison to other measurement errors has been made, which are caused by surface preparation and anisotropy of the material. To validate the experimental and analysis approach, the effect of compressive and tensile prestresses was also simulated using a three-dimensional finite element model. The material investigated is a rolled 2024 T351, which is widely used for manufacturing airplanes. It is shown that the existing neural network approach is able to determine the stress-strain behavior in agreement with that obtained from tensile tests. The method is robust against most error sources, such as surface roughness, coarse grain structure, and anisotropy, if a sufficient number of experiments are available. The most important influencing factor can be the residual stress causing errors up to 20% in the identified stress-strain curves.
UR - http://www.scopus.com/inward/record.url?scp=63149131387&partnerID=8YFLogxK
U2 - 10.1557/jmr.2009.0116
DO - 10.1557/jmr.2009.0116
M3 - Article
AN - SCOPUS:63149131387
SN - 0884-2914
VL - 24
SP - 907
EP - 917
JO - Journal of Materials Research
JF - Journal of Materials Research
IS - 3
ER -