Experimental and numerical analysis of creep cracking under secondary and combined loading

Abstract

The UK’s advanced gas-cooled reactor power plants contain welded components which were not stress relieved following fabrication. The presence of weld induced residual stresses, in combination with the plant operating under creep conditions and the material’s low creep ductility has caused cracks to form during service due to a process known as reheat cracking. To investigate this cracking process and to develop assessment procedures to evaluate the structural integrity of such components in operation, fracture specimens are required to simulate this loading condition and subsequently perform crack growth studies in laboratory controlled conditions. Two new fracture mechanics specimen designs were proposed in this study: an electron beam (EB) welded compact tension, C(T), specimen and a wedge-loaded C(T) specimen. EB welding had previously been used as a fabrication process to manufacture C(T) specimens reconstituted from ex-service components. However in this study, this weld process was used specifically to introduce residual stresses in the specimens. Both specimen designs were fabricated using ex-service Type 316H austenitic stainless steel. Extensive residual stress measurements were made using the neutron diffraction, contour method and slitting techniques on EB welded and wedge-loaded C(T) specimens. This data was used to develop and validate numerical simulations of the fabrication processes enabling residual stress predictions to be made and stress intensity factors, which define the crack driving force, to be determined. Large initial stress intensity factors due to the residual stresses were determined as up to 22.2 MPam^(1/2) and 43.6 MPam^(1/2) for the EB welded and wedge-loaded C(T) specimens respectively. Accurate estimates of the weld residual stresses in the EB welded C(T) specimen required detailed weld simulations to be developed. These were created by following guidelines recently published in the R6 fracture assessment procedure for modelling arc welding processes. The stress predictions made by the weld simulations were in close agreement with the experimental measurements, which showed the advice in the R6 guidelines may be followed to produce accurate numerical simulations of EB welding. Creep crack growth (CCG) tests were conducted using these new specimen designs at 550°C for up to 1,300 h, under secondary and combined loading conditions, where large crack extensions of up to 5.4 mm occurred. Such large crack growth was achieved by pre-conditioning the material by uniform pre-compression prior to specimen fabrication. This process reduced the creep ductility of the material enabling CCG to occur during relaxation of the residual stresses. It was shown that this pre-conditioning process was necessary to perform such experiments, as the creep ductility in ex-service Type 316H stainless steel in the as-received condition can be high. The experimental measurements were used to validate crack growth predictions using the new revision of the R5 assessment procedure. Crack extensions were predicted using C(t) estimated using a reference stress based approach under secondary and combined loading conditions. This assessment was shown to be very conservative where no plastic deformation was assumed, as predicted crack lengths were up to 10 times larger than experimental measurements. By including the effects of crack tip plasticity in the assessment, estimates of the crack extensions were in close agreement with the experimental data. The creep crack growth rates were sensitive to the material’s creep ductility, which differed between the test specimens due to weld induced plasticity and variability of the microstructure in the ex-service material. Use of the upper bound crack growth properties is recommended to ensure conservative assessments are made. Damage models developed using the ductility exhaustion approach were also used to predict CCG. Crack length predictions in the wedge-loaded C(T) specimens were generally in good agreement with the numerical models. However non-conservative crack growth predictions were made in the EB welded C(T) specimens which was due to the presence of plasticity.