Browse/search for people

Publication - Professor David Knowles

    Comparison of predicted cyclic creep damage from a multi-material weldment FEA model and the traditional r5 volume 2/3 weldment approach

    Citation

    ElAgha, F, Tanner, D & Knowles, D, 2018, ‘Comparison of predicted cyclic creep damage from a multi-material weldment FEA model and the traditional r5 volume 2/3 weldment approach’. in: Proceedings of the ASME 2018 Pressure Vessels and Piping Conference. American Society of Mechanical Engineers (ASME)

    Abstract

    The R5 assessment procedure for Integrity of High Temperature Structures employs a Weld Strain Enhancement Factor (WSEF) (dependent only upon classified weld type) for predicting creep fatigue crack initiation at weldments (V2/3 Appendix 4). This serves to amplify the calculated total strain at the weld toe for full penetration welds to account for geometric concentration and material mismatch between weldment zones. The value of WSEF recommended for fillet welds was derived from a review of a limited number of tests on thin welds which were not wholly representative of a typical fillet weld.

    The objective of this paper is to present a comparison of the predicted cyclic creep defect initiation damage at a fillet weld toe using a multi-material finite element (FE) model of the weldment, against the damage predicted using the traditional R5 V2/3 approach, which uses only the parent material properties to derive the weld toe strain range in combination with the WSEF. In this example, the fillet weld joins a high temperature tube to an anti-vibration strap. There is pressure loading in the tube and displacement loading due to thermal expansion.

    The FE model incorporates material properties associated with both the parent and the weld metal, including elastic modulus, plastic yield properties, creep deformation, and creep ductility (to determine damage via ductility exhaustion). The finite element analysis is run for 30 cycles (pressure and thermal cycling) with an average dwell period of 736 hours, with predicted damages for 100 cycles estimated using extrapolation. Sensitivities considering the stress-strain properties of the weld are included.

    Full details in the University publications repository