The members of the NeT network have undertaken parallel round-robin activities measuring and predicting transient temperatures, fusion boundary development, weld residual stresses and structural distortions in a benchmark specimen manufactured from AISI 316L(N) plate with a central finite length slot filled with three superimposed TIG weld beads. This is a strongly three-dimensional configuration with many of the character. istics of a repair weld. The round robin finite element predictions of residual stresses and distortions are compared with an extensive database of residual stress and distortion measurements made using diverse techniques on multiple TG4 specimens. It is found that high solution accuracies can be achieved.

The members of the NeT network have undertaken parallel round-robin activities measuring and predicting transient temperatures, fusion boundary development, weld residual stresses and structural distortions in a benchmark specimen manufactured from AISI 316L(N) austenitic stainless plate with a central finite length slot filled with three superimposed TIG weld beads. This is a strongly 3-dimensional configuration with many of the characteristics of a repair weld. The round-robin finite element predictions of transient temperatures and the extent and shape of the melted zone are compared with thermocouple measurements made during welding, and with the results of destructive metallography. It is found that the majority of thermal simulations achieve high accuracy at weld mid-length, where quasi-steady state conditions apply, and slightly reduced but still acceptable accuracy at the bead ends, where thermal conditions are much more 3-dimensional. With two exceptions, variability in thermal solutions can be discounted as a cause of variation in subsequent mechanical simulations. The two outlier simulations allow investigation of sensitivity to final pass heat input and bead length.

Accurate prediction and measurement of residual stresses in welds is an important part of assuring their short and long-term structural performance in high value, safety critical engineering components and structures. However, both measurements and predictions of weld residual stresses often exhibit high levels of variability that are not widely appreciated. The mission of the European Network on Neutron Techniques Standardization for Structural Integrity (NeT) is to develop experimental and numerical techniques and standards for the reliable characterisation of residual stresses in structural welds. The NeT Task Group 4 project examined residual stresses in a three-pass slot-welded plate specimen fabricated from AISI 316L(N) austenitic stainless steel plate. Several nominally identical specimens were fabricated under closely controlled conditions, with detailed records kept of the manufacturing history, weld process parameters, transient temperatures during welding, and the resulting geometric distortions. Comprehensive stress-strain material property characterisation was then undertaken, extending to the isothermal cyclic tests necessary to calibrate the mixed isotropic-kinematic material hardening models required for accurate weld residual stress prediction. Parallel residual stress measurement and simulation round robins were performed by a large number of participants from around the world. Residual stresses were measured using neutron and high energy synchrotron diffraction, surface X-ray diffraction, surface and deep hole drilling, the contour method, and ultrasonics. Neutron diffraction measurements were made at eight different instruments. The diffraction measurements database alone is large enough to generate reliable mean profiles, to identify clear outliers, and to establish that there is no statistically significant difference in the residual stress field in the specimens used for the non-destructive measurements. NeT Task Group 4 gives a unique insight into the real-world variability of diffraction-based residual stress measurements, and forms a reliable foundation against which to benchmark other measurement methods. NeT Task Group 4 is also a unique test bed for the development and validation of weld residual stress simulation techniques in austenitic stainless steel. Its combination of extensive materials characterisation, accurately characterized welding temperature transients, and reliable residual stress and distortion measurements is currently unrivalled. About thirty finite element simulations were submitted to the network over the course of the project, giving insights into the required accuracy of welding thermal solutions, the mechanical solution accuracy achievable using optimized material constitutive models, and the level of acceptable error in finite element residual stress simulation results for use in structural integrity assessments of high integrity engineering components.

The NeT international network has undertaken a number of round robin studies examining the prediction and measurement of weld residual stresses. NeT TG4 examined the case of a three-pass slot weld made from AISI 316L steel. The project started in 2007, and is now close to completion. This paper reviews the project as a whole, and presents its achievements.

A single weld bead deposited on a flat plate is a deceptively simple problem that is in practice a challenge for both measurement and prediction of weld residual stresses. Task Group 1 of the NeT collaborative network has examined this problem in an extensive programme of measurement and simulation extending from 2002 to 2008. As a result, the NeT bead on plate forms an ideal benchmark problem for the development of weld residual stress simulation techniques. One of the conclusions of NeT Task Group 1 is that the most accurate predictions of weld residual stresses in austenitic steels are achieved using mixed isotropic-kinematic material constitutive models. However, the use of these models can require both extensive materials data, and compromises in fitting either the monotonic or cyclic responses. This paper reports a detailed matrix of sensitivity studies aimed at optimising the behaviour of mixed hardening models in welding simulation, using the Lemaitre-Chaboche formulation in the ABAQUS finite element code. Predicted stresses and strains in the NeT bead on plate specimen are compared with the extensive database of residual stress measurements. Further studies examine sensitivity to the handling of high temperature inelastic strains, using a novel two-stage annealing functionality implemented within ABAQUS. The results show that, overall, the most accurate predictions are made if the Lemaitre-Chaboche parameters are optimised to fit the monotonic response over the first 2% of plastic strain. However, further improvements in prediction could be achieved if the constitutive model were capable of independently fitting both the monotonic and saturated cyclic response of the material.

A single weld bead deposited on a flat plate is a deceptively simple problem that is in practice a challenge for both measurement and prediction of weld residual stresses. Task Group 1 of the NeT collaborative network has examined this problem in a two-phase programme extending from 2002 to 2008. Ten independent sets of residual stress measurements have been reported using diverse techniques, and over forty finite element simulations have been performed. This paper reviews Phase 2 of the Task Group 1 round robin. Here, the finite element simulations all made use of optimised thermal solutions, in which the global welding parameters, including efficiency, were fixed, and only the detailed heat source geometry was varied. These resulted in accurate far field welding temperature distributions, with significant variability only close to the weld bead itself. The subsequent mechanical analyses made use of kinematic, isotropic, and mixed isotropic-kinematic material constitutive models, and made a variety of assumptions about the introduction of weld filler material to the structure and the handling of high temperature inelastic strains. The large database of measurements allowed the derivation of statistical best estimates using a Bayesian "duff data" approach, and these best estimates were compared with the predictions to establish the most accurate material constitutive models. The most accurate predictions of residual stress were made using non-linear kinematic or mixed isotropic-kinematic constitutive models. The methods used to handle high-temperature inelastic strains influenced the predicted stresses only in regions where very high temperatures were predicted during welding. The results emphasise the importance and value of both well-characterised benchmark problems and international collaboration in the development of technologies to both measure and predict weld residual stresses.