The high-temperature design codes are presently considering the use of
stress reduction factors for designing welded structures submitted to
creep. These reduction factors are derived from creep tests which are
generally made on small specimens and are nor necessarily representat
ive of large-size geometries. These codes are very likely overconserva
tive, consequently uneconomical and need to be improved; an investigat
ion to assess and quantify the supposed size effect is required. This
paper presents an experimental and numerical study on creep behavior a
t 600 degrees C of full-size welded joints taking into account real fu
ll-thickness of weldings. The material investigated is the austenitic
stainless steel 316L(N) with manual metal are welds using the 19 Cr 12
Ni 2 Mo electrode grade. The creep laws used in calculations are thos
e obtained from tests using small specimens, but some coefficients of
their theoretical formulation have been modified to obtain a better co
herence with full-size specimen data. Between small and large full-siz
e specimens, experimental results show no significant difference in ti
me to rupture, and the same location of fracture, at the center of the
weldment, is observed Finite element simulations performed for full-s
ize welded joints provide rupture times that are consistent with measu
red values. The calculated percentage of the damaged volume in the wel
d metal as a function of load levels and of creep-time duration is stu
died; it shows that the creep-rupture times for high stress loading ar
e determined with higher accuracy than for low stress loading.