Thermal fatigue benchmark final - research report

DNV (Det Norske Veritas) has analysed a 3D mock-up, loaded with variable temperature. The load is applied to the internal of a pipe, and deviates from the axisymmetrical case. The calculations were performed in blind in an international benchmark project. DNV's contribution was funded by SKI. The calculations show the importance of taking the non-axisymmetry into account. An axisymmetrical analysis would underestimate the stresses in the pipe. The temperature field in the mock-up was measured at several locations in the pre-test condition. It turned out to be difficult to capture the measured field by applying only convection, adjusting heat transfer coefficients. The adjustment of the heat transfer coefficient proved to be a major problem. No standard estimation of these parameters were capable of satisfyingly capture the temperature fields. This highlights the complexity of this kind of problems. It was reported by CEA that modelling of radiation was required for accurately resolving the stresses. The time to crack initiation was computed, as well as crack propagation rates. The computed crack initiation time is significantly longer than the crack propagation time. All results by DNV in terms of maximum stress range, computed design life and crack propagation time are comparable to those obtained by other contributors to the benchmark project. The DNV computed maximum stress range is Δσ = 715 MPa (von Mises). The contribution by other members range from 507 to 805 MPa. The DNV computed fatigue life (from two mean curves, ASME and CEA) range from 100.000 to 1.000.000 depending on different assumptions