Problems related to stress corrosion cracking reduced by controlled shot peening

Neutron diffraction has been used by the nuclear industry for measuring residual stresses in structures for which integrity safety cases must be made. However measurement of stresses in materials containing large grains, or in anisotropic weld metal, or approaching air-metal interfaces, or metal-metal interfaces using neutron diffraction is particularly challenging. These types of measurements can give rise to errors (often termed “pseudo-strains”) that can be significantly larger than the residual stress actually present. Such errors arise when the gauge volume is partially filled (e.g. air to metal interface for measuring near-surface stresses), or when the gauge volume composition is inhomogeneous (metal to metal interface), or the gauge volume material is anisotropic (welds with bundles of elongated grains). To mitigate such errors several approaches have been proposed using numerical and analytical methods to calculate the magnitude of pseudo-strain. Whether these methods can be applied depends on the instrument used and the type of sample being measured. In this context a different approach based on the Monte-Carlo method, as embedded in the neutron ray tracing software package McStas is proposed.

The aim is to validate by neutron measurements the ability of McStas to simulate pseudo strains associated with a gauge volume that is partially filled, or a gauge volume that samples inhomogeneous and anisotropic material. The instrument modelled for this research project is ENGINX a time-of-flight instrument located at ISIS (RAL). To understand and isolate the effects of traversing an interface made up of different materials a stepwise approach was taken. The initial phase involved building a new model of ENGINX in McStas and to validate its correctness by analysing the characteristics of the beam and gauge volume by simulating a steel pin scan. The subsequent phase focussed on virtual experiments to study the pseudo strain arising when measuring strain in an air-to-material interface, a material-to-material interface and when large grains or pores are present as in the vicinity of welds. The results presented are not intended to be used as a quantitative prediction of pseudo strain but to demonstrate how McStas can be used to model virtual experiments to study the pseudo strains occurring. Several virtual sample models have been built to demonstrate how this could be useful/interesting for beamline scientists and users of neutron diffraction. One such virtual sample model is used to demonstrate that the mitigation technique of rotating the sample 180° works when the detector is in transmission but not in reflection, and that it is strongly dependent on the attenuation of the investigated material. Another virtual sample model is used to demonstrate that when the ratio of a cavity (hole, pore, etc) to the gauge volume is more than 2%, then significant pseudo strains can arise. Moreover, this work delivers a new model for ENGINX.

John Brokx

Papers

Modelling Peak Shifts not Resulting from Lattice Strain in Anisotropic and Inhomogeneous Materials using Monte-Carlo Simulation of Neutron Diffraction