The use of supercomputer modelling of high-temperature failure in pipe weldments to optimize weld and heat affected zone materials property selection

Abstract

The creep deformation and damage evolution in a pipe weldment has been mod­elled by using the finite-element continuum damage mechanics (CDM) method. The finite-element CDM computer program Damage xx has been adapted to run with increased speed on a Cray XMP/416 supercomputer. Run times are suffi­ciently short (20 min) to permit many parametric studies to be carried out on vessel lifetimes for different weld and heat affected zone (HAZ) materials. Finite-element mesh sensitivity was studied first in order to select a mesh capable of correctly predicting experimentally observed results using the least possible com­puter time. A study was then made of the effect on the lifetime of a butt welded vessel of each of the commonly measured material parameters for the weld and HAZ materials. Forty different ferritic steel welded vessels were analysed for a constant internal pressure of 45.5 MPa at a temperature of 565 °C; each ves­sel having the same parent pipe material but different weld and HAZ materials. From these studies it has been concluded that good creep rupture ductility and strength of the weld material are necessary to achieve long vessel lifetimes. Vessel lifetimes are not strongly influenced by the HAZ material properties provided that the rupture strength is maintained above that of the parent pipe material, and that the ductility is greater than one half of that for the parent pipe material. Also it is found that the optimum design can be achieved with both the weld and the HAZ materials having properties of high primary creep strain, mean sec­ondary creep strain, mean rupture strain, and long rupture lifetime, relative to the base or parent pipe material. A lifetime improvement has been demonstrated of 30% over that obtained for the initial materials property data. A methodology for weldment design has been established which uses supercomputer-based CDM analysis techniques; it is quick to use, provides accurate results, and is a viable design tool.