Degree Name

Doctor of Philosophy


School of Mechanical, Materials and Mechatronic Engineering - Faculty of Engineering


This investigation is concerned with the characterisation of actual welded samples of P91 - a 9%Cr heat resisting steel that is widely used for pressure vessels in the power generation industry. Although the creep resistance of the normalised and tempered base steel is excellent, weld fabrication compromises creep performance because of degraded properties in the heat affected zone (HAZ). In order to elucidate the structure and properties of the HAZ, dilatometer heat treatments and simulations of HAZ sub-zones were carried out using both a dilatometer and a Gleeble thermo-mechanical simulator. The simulated samples were used to study the microstructure of the HAZ sub-zones and to subject them to creep testing. In actual welding, post weld heat treatment (PWHT) is used to produce a uniform tempered martensite structure across the weldment, making it difficult to distinguish the HAZ sub-zones and the boundary with the unaffected base plate. Nevertheless, hardness profiles across the HAZ and measurements within the sub-zones confirmed that softening occurs below the base plate hardness after PWHT in the intercritical (IC) and grain refined (GR) sub-zones. The creep fracture times of cross-weld creep samples were also lower than parent metal because of type IV fracture in these HAZ sub-zones. Dilatometric investigations shed new light on the sensitivity of the properties of the martensite to the thermal cycle associated with austenitisation and subsequent cooling. The AC1 and AC3 temperatures were increased with increasing heating rate and MS was lower for a lower heating rate. MS was found to vary from 420°C to 370°C and the hardness of the martensite from 365 to 480 HV, depending on the thermal cycle. This varialibilty is due to the extent of carbide solution. There was a marked increase in hardness with increasing peak temperature of the thermal cycle, but subsequent simulated PWHT substantially decreased the hardness and the hardness range of the simulated sub-zones. An excellent correlation was found between the structures and properties of the HAZ of the actual welds and the simulated sub-zones produced by both dilatometric and Gleeble techniques. It was established that the heat input (1.6 or 2.6 kJ/mm) had only a minor effect on the microstructure and hardness of Gleeble simulated sub-zone samples. TEM results confirmed the presence of coarse Cr-rich, M23C6 and fine V-and Nb-rich, MC in all simulated sub-zones, before and after PWHT, except for the as simulated GCHAZ in which carbide solution occurred. Accelerated creep testing showed rapid creep failure of both AR and simulated IC and GR sub-zone samples for testing at higher temperatures in the range of 630°C to 670°C and at a higher stress, 100 MPa compared to 80 MPa. Failure was associated with high creep ductility and the phenomenon of rehardening in the region adjacent to the neck due to rapid work hardening prior to fracture. This type of failure has been labelled Mode 1 and is characterised structurally by grains and creep cavities that are strongly elongated parallel to the tensile axis. Another characteristic type of creep failure, Mode 2, exhibits a low creep ductility and transversely aligned creep cavities. This mode was found in notched AR samples and notched simulated GCHAZ samples tested at 630°C and a stress at 80 MPa. However, for both Mode 1 and 2 failures, non-metallic inclusions were found to play a significant role in the nucleation of creep cracking and cavition. The research work identified that the GRHAZ is the most creep susceptible HAZ sub-zone because the thermal cycle results in carbide coarsening, reduced precipitation and solid solution strengthening and a high γ prior grain boundary surface area/unit volume. The creep resistance was most marked for the simulated GCHAZ samples and this property relates strongly to the hardness of the P91 steel prior to creep testing. The important role of non-metallic inclusions in the nucleation of creep cracking and cavitation indicates that control of the type, size distribution and content of inclusions should have a significant effect on the creep life of P91 steel.