Effect of spatial variation of stress-free lattice spacings on measured residual stresses M Law* and V Luzin Australian Nuclear Science and Technology Organisation, Kirrawee DC, New South Wales, Australia The manuscript was received on 2 March 2011 and was accepted after revision for publication on 24 May 2011. DOI: 10.1177/0309324711413648 Abstract: When neutron diffraction is used to measure residual stresses, estimates of the unstressed lattice spacing (d 0 ) are often required. These are generally measured on a sample that is considered representative of the component. In the case of welded samples, the possi- bility of variation in weld shape between the d 0 sample and the area where stresses are deter- mined must be taken into account. To assess this, d 0 values are measured on an area considered representative of the sample, and then are re-measured on a sample cut from the actual strain measurement plane. The residual stresses are re-calculated with these new d 0 values and the results compared to those based on the initial d 0 measurement. Methods of estimating these errors are discussed. The errors in residual stress that arise from the spatial variation of d 0 values are greater than the commonly reported peak fitting errors. These errors can be estimated and should be included in the reported residual stress measurements. Cutting a d 0 sample from the measurement plane makes this a destructive technique. Keywords: welding, residual stress, neutron diffraction 1 INTRODUCTION Neutron diffraction is termed a ‘non-destructive’ technique in contrast to destructive techniques, which derive stresses from displacements that occur during cutting, drilling, or slicing. Some diffraction surface techniques use the sin 2 c technique, which assumes the normal stress component is zero and does not require an unstressed lattice parameter (d 0 ). Thicker samples will have a triaxial stress field and require some d 0 value. If the sample contains a weld, there will be a spatial variation of the d 0 values [1], and d 0 samples must either be taken from a ‘duplicate’ weld that is assumed to have identical material and spatial disposition of constituent mate- rials, or the original sample could be sectioned to produce d 0 samples from the actual measured area. Strictly speaking, sectioning the measured sample will make this a destructive technique. Typical methods of deriving d 0 values [2] include: (a) measurement of an area where stresses are expected to be small: this method assumes a single material with no microstructural changes due to localized heating; (b) stress-free powders or filings: these may have significant plastic strains and intergranular stresses that affect their accuracy; (c) stress-free combs or cubes, typically cut by an elec- tro-discharge machining (EDM), which may intro- duce surface residual stresses of up to 0.5 mm depth [ 3, 4], hence samples should be large enough that the gauge volume lies beneath this zone; (d) a thin slice can be cut and the d 0 values derived from this (this technique is described later in the paper). All these methods may also be checked and the val- ues adjusted by use of force or moment balance in the sample. *Corresponding author: ANSTO (Australian Nuclear Science and Technology Organisation), Locked Bag 2001, Kirrawee DC, NSW 2232, Australia. email: mlx@ansto.gov.au 837 J. Strain Analysis Vol. 46