TG5: Edge Welded Beam (Ferritic Steel)

 

Task Group 5 has been studying ferritic steel (grade SA508) beams with dimensions 180x50x10 mm3 with an autogenous weld along one edge. This geometry and welding method avoids the addition of any (foreign) material to the system and also the introduction of more than one thermal cycle allowing analysts to concentrate on modelling phase transformations in the base material. Early FE simulation work has shown that phase transformations are sensitive to the welding parameters, so one half of the test specimens have been welded at a 'slow speed' and the remainder at a 'fast speed'.

 

TG5_spec.JPG
TG5_model.JPG

Fig. 1. TG5 – edge welded specimen shown with run-on and run-off tabs

Fig. 2. TG5 – macrograph of weld cross section vs. simulated phase distribution after welding

The residual stress measurement round robin in Task Group 5 comprises three sets of neutron diffraction measurements and one set of synchrotron X-ray diffraction results for each of the two welding speeds.  The measurement results are in good agreement and they clearly show that the two welding speeds are resulting in differing stress distributions with the “fast” weld exhibiting an about 100-150 MPa higher peak stress than the “slow” weld.

 

Three different numerical predictions of the residual stress distributions have been submitted for each specimen. The results are in reasonable, albeit not perfect, agreement with the measured stresses whereby the predictions for the “fast” weld agree better with the corresponding measurement data than those for the “slow” weld.  All results clearly show that solid state phase transformation has a major impact on stresses in the SA508 beam edge-welded specimen.

 

A number of parametric numerical studies have revealed that these welding simulations in ferritic steel are quite sensitive to the material hardening model considered, to the phase transformation temperatures and to the grain growth behaviour. On the other hand, the predicted residual stresses are relatively insensitive to the choice of phase-dependant material properties and annealing temperatures.

TG5_RS_longitudinal_meas_fast.jpg
TG5_RS_longitudinal_meas_slow.jpg

Fig. 3. TG5 slow weld – measured longitudinal residual stresses and their Bayesian mean

Fig. 4. TG5 fast weld – measured longitudinal residual stresses and their Bayesian mean

Network on Neutron Techniques Standarization for Structural Integrity