||Shutong Zhang, Sebastian Romo, Rafael Arthur Giorjao, Jorge Penso, Haixia Guo, Simon Yuen, Lisa Ely, Antonio Ramirez
Coke drums are large pressure vessels used in oil refineries to transform heavy residual oil to light-weight oil molecules and solid cokes through the delayed coking process. Each delayed coking unit consists of at least two coke drums, which work alternatively to allow the continuous injections of residual oil. During the operation cycles, coke drums endure severe thermal and mechanical loadings, which lead to coke drum damage such as bulging and cracking. Low-cycle fatigue has been widely regarded as the major cause that leads to the damage, because the stresses within the coke drum material can exceed the yield stress and induce plastic deformations. Welding repairs using temper bead welding (TBW) techniques are widely adopted to replace the cracked regions and restore the local mechanical properties. The nickel-base alloys including Alloy 625 and Alloy 182 have been used as the filler materials because of their matching thermal expansion coefficient and good chemical compatibility with the low alloy steel base metal (BM) and the 410s stainless steel clad. However, they tend to exhibit significant cyclic hardening beyond the base metal and induce stress concentrations in the joint region due to the strength mismatches. In addition, metallurgical reasons such as welding defects and dilutions can reduce the weld metal resistance against fatigue failures. In this study, two Ni-steel dissimilar joints are evaluated based on the low-cycle fatigue tests and metallurgical characterizations. The goal of this work is to measure the low-cycle fatigue behaviors of the two joints and investigate the crack-susceptible regions through failure analysis.
Two welding mockups were fabricated using the shield metal arc welding (SMAW) process with the TBW techniques. The weld metals are Alloy 625 (ENiCrMo-3) and Alloy 182 (ENiCrFe-3) and the base metal is 1.25Cr-0.5Mo steel with a 410s clad layer. Two types of dog-bone specimens were extracted from the welding mockups transverse to the welding directions. The weld metal (WM) specimens contain a full weld metal region within the gage length, while the transition zone (TZ) specimens include the transition interface at the center of the gage length. WM and TZ specimens of the two different weld metals were evaluated using the Gleeble-based isothermal low-cycle fatigue (ILCF) tests. The tests are performed at a fully-reversed (R=-1) strain-control condition at three different strain amplitudes (1.0%, 1.5% and 2.0%) and the temperature is held isothermally at 250ºC (482ºF). On the other hand, the microstructures of the two welding mockups were characterized using optical microscope (OM) and scanning electron microscope (SEM). The TZ samples were extracted to contain the regions of WM, BM and HAZ and the samples were etched using 2.0% Nital and 10% Chromic Acid electrolytically to reveal ferritic and nickel microstructures respectively. In addition, fracture surfaces of the failed dog-bone specimens were analyzed with an Olympus-DSX510 depth of field microscope for the surface morphologies. Electron backscatter diffraction (EBSD) images of weld metal and transition zone specimens were taken before and after tests to reveal the strain distributions and crystal misorientations of the plastically deformed regions.
Results and Discussion
The cyclic stress responses of the weld metals show that both nickel-base alloys, Alloy 182 and Alloy 625, exhibit cyclic hardening at the initial cycles, followed by a saturation stage before the failures occur. However, Alloy 625 WM showed much higher peak stresses than Alloy 182 WM, overmatching the 1.25Cr-0.5Mo BM; Alloy 182 WM slightly under-matched the BM at the first cycle but quickly over-matched it due to the cyclic hardening. The experimentally obtained cyclic strain curves showed that both nickel WMs have similar fatigue life but inferior to that of the BM. On the other hand, the cyclic hardening observed in the Alloy 182 TZ specimens was in contrast to the flattened cyclic responses of Alloy 625 TZ specimens. In addition, the fatigue life of Alloy 182 TZ was much shorter than that of Alloy 625 TZ. The reason for the results can be rationalized by the unbalanced plasticity within the TZs. For Alloy 625 TZ specimens, most of the plastic deformations occurred in the BM due to the overmatching WM strength. For Alloy 182 TZ specimens, most of the plastic deformation occurred in the WM. This could be supported by the failure analysis, as the macrocrack occurred in the BM region of Alloy 625 TZ while it occurred along the fusion boundary in the Alloy 182 WM. Metallurgical analysis showed that particles such as Nb carbides and Ti oxide inclusions were dispersed along the interdendritic regions and solidification grain boundaries. The inclusion particles were observed at dimples of the fracture surfaces along with the fatigue striations marks. In addition, EBSD strain analysis of the Alloy 182 TZ zone showed the transgranular crack propagations with fine-grain structures on both sides of the crack.
Two Nickel WMs and their TZs were evaluated based on the low-cycle fatigue tests and metallurgical characterizations. Both WMs showed similar cyclic behaviors and fatigue life, but Alloy 625 WMs exhibited much higher strength over Alloy 182 WMs and the 1.25Cr-0.5Mo BM at the same strain amplitudes. Due to the strength mismatches, the low-strength regions became the crack susceptible areas. This led to the differences of failure locations and fatigue behaviors between the TZ zones of the two Nickel welds. Failure analysis was performed by fracture surface characterizations and EBSD strain analysis. The results showed that the cracks propagated transgranularly in the Alloy 182 WM and caused the formation of fine grains along the crack path.