1985 Volume 3 Issue 1 Pages 196-203
A measuring method for triaxial residual stress distributions that are axisymmetrical but not uniform both in radial and in thickness (namely axial) directions, was proposed and the accuracy of the measurements was confirmed. This method needs thin arch-specimens, of their length about three times as the thickness of a joint to be concerned, and a thin r-specimen, with their long axes along circumferential and radial directions, respectively, to be cut loose from the joint. The residual stresses of the joint are estimated both from the strain changes induced in the long axis of each thin plate when it is cut out, that is the strain change in the eircumferential direction for the arch-specimen Δεθ and the radial strain change for the r-specimen Δεγ and from the stresses left in those thin plates, in which the stresses are in the plane stress state.
The first approximation for the residual stresses is obtained by assuming that the Δεθ and Δεγ are linear across the thickness of the plate and that the relieved stress in the thickness direction in the r-specimen when it is cut out, that is σ'z, is neglected. Then, the distribution both in the radial and thickness directions of this first approximated circumferential residual stress σθ, in other words the stress distribution acting on the section of the r-specimen before its cutting, can make it possible for the distribution in the thickness direction of Δεγ and σ'z to be calculated by finite element method and results in the more accurate estimation of the residual stresses.
This second approximation is almost attainable limit by this proposed method and depends strongly upon the discrepancy of the Δεθ from a straight line joining the measurements both at top and bottom surfaces of the arch-specimen. This nature of nonlinearity concerning Δεθ becomes stronger as the weld diameter of the joint decreases. The accuracy for the residual stresses, however, maybe maintained good enough even in the weld diameter being 300 mm.