Creep analysis of a main steam pipe system

Abstract

ABSTRACT The present work is performed on the main steam pipe system in Heleneholmsverket, a CHP in Sweden and consists of the following parts (i) numerical analysis of the in-service creep behaviour of the pipe system, (ii) creep testing of new and service exposed materials from welded components, (iii) characterisation of the creep damage distribution in creep tested welds of the actual piping. The entire system has been modelled for creep evaluation to make it possible to compare the simulated creep stress and strain distributions in selected welds with observed amounts of creep cavitation, which can be correlated to the accumulated creep strain. Creep data for the analyses were produced by creep testing of service exposed base and weld metals from a pipe weld and a T-piece branch weld from the system. In addition, the creep tested welds were studied metallographically to map the creep damage and make it possible to compare the damage development with the resulting creep stress and strain distributions in the weld. In the previous project also a T-piece branch weld was investigated in a similar way and those results were used for verification of the re-analyses in present project with the updated system model. The following results were achieved: The model of the entire steam pipe system was created in Abaqus and the strain distributions were verified in comparison to a corresponding elastic Caepipe model. The Norton creep law was used for the simulations. In addition, also primary creep was analysed. The effects of primary creep on the long-term creep behaviour was significant and the results shows the importance of including primary creep into the model. There is no effect of starts and stops on the stress and strain distributions in the system during creep. The system analysis results showed enhanced strains up to 2.1% at one bend and 0.5–1.0% in some parts of the system. Although replica testing had not been conducted directly at the bend the high strains indirectly agreed with the observations of small creep cracks had been observed in replica testing of in a weld at one of the ends of the actual bend. Furthermore, several components in the system have been exchanged due to creep crack formation. Moderate levels creep damage was observed in the pipe weld. The analysis of this pipe weld gave somewhat lower creep strains than expected. The stress and strain distributions matched with the maximum principal stress criterion but not with the von Mises stress that Abaqus uses for creep analyses by default. The analysis of the branch weld matched well with observed creep damage distributions whereas the maximum strain level of 0.4% appears to be rather low in comparison to the quite extensive creep damage. However, local constraint and multiaxiality in welds lead to significantly lower creep ductility compared to uniaxial creep and contribute to a reasonable agreement between the strain and the damage levels. The creep tests of service exposed material resulted in relatively high Norton creep law exponents and no shift to lower values at the lowest tested stresses. It is hardly possible to perform tests at even lower stresses and therefore the simulations at service conditions resulted in unreasonably low creep strains.