To evaluate the dip tube failure, a 360° section about 2 feet above the top hat was removed, sectioned, and examined. Selected dip tube panels were also removed from just below the top hat and near the bottom, which were examined for physical features that might explain the attack on the dip tube. Wall thicknesses were measured using an ultrasonic thickness gage, and metallurgical cross sections were taken to examine the microstructure for signs of attack, cracking, and overheating. The following features were evaluated on panels removed from above and below the top hat:
The elongated hole and circumferential fracture above the top hat on the east side,
The top hat-to-dip tube weld,
Cracks found above the top hat on the west side,
Buckling and distortion at various locations above and below the top hat.
The initial failure of the dip tube was found to be an elongated hole on the east side that coincided with a circumferential fracture, which was about 0.5–2.0 in. above the top hat, shown in Fig. 6. A circumferential fracture extended about 3 feet around the dip tube wall on the east side. The circumferential fracture was irregular, jagged, and thick lipped, and it was located from 0.5 to 2 in. above the top hat, as shown in Fig. 9. This area did not have the excessive buckling that other locations showed.
The circumferential fracture just above the top hat was an indication that the axial stresses were high enough to cause failure of the dip tube. The fracture extended from the longitudinal seam weld on the dip tube toward the east for about 3 feet. The irregular shape indicated that cracks had initiated at multiple locations at different times, as the horizontal cracks were found to be both extended and intersected.
The fracture was typically thick lipped and thinned in only a few locations. The fracture edge thickness ranged from 0.22 to 0.28 in. This thick edge indicated that the dip tube fractured in a brittle manner, rather than being thinned by an erosion or corrosion mechanism. Although the metal near the fracture was not excessively thinned, the fracture edge was oxidized and altered by the hot syngas passing over it during the incident. The fracture surface was examined, but no obvious fracture features were discernible because of the surface oxidation. The brittle-like fracture appearance, thick-lipped features, cracks, and the lack of deformation indicated that this location above the top hat lacked ductility.
Various locations along the fracture were examined in detail. Location 1 was just above the top hat-to-dip tube weld, where it coincided with the longitudinal dip tube seam weld. The location of first failure is believed to have occurred in the region that showed the greatest signs of discoloration, which was below the cone plate and baffle that were oxidized and burned away. Between locations 1 and 2, the top hat weld was missing. This is the same location observed in the field where the weld had separated from the dip tube. As its surface had been ground down during the top hat removal after the incident, this region could not be examined.
Location 3 was below the circumferential fracture and about 1 in. above the top hat. Some cracks were found at this location. Additional cracks were observed between locations 3 and 4, as shown in Fig. 10. The cracks were cross sectioned to determine the crack morphology. Figures 11 and 12 show secondary cracks that were about 1 in. above the top hat and perpendicular to the circumferential fracture between locations 3 and 4. These cracks initiated from both the outside and inside surfaces, and extended along the grain boundaries, where many grain boundary precipitates were observed. Although these precipitates were not micro-probe analyzed, the grain boundaries likely were filled with carbides, possibly gamma prime and sigma phases, all indicative of elevated temperature exposure, shown in detail in Fig. 13. Similar microcracks were also found on the dip tube from gasifier no. 2 that had not progressed to failure, shown in Fig. 14.
These cracks were likely caused by “stress relaxation cracking,” which occurs in alloys such as Incoloy 825 operating between 550 and 750 °C [1, 2]. This mechanism is also referred to as “stress-induced cracking,” “reheat cracking,” or “stress-assisted grain boundary oxidation” (SAGBO). The fracture is often brittle in appearance and occurs in cold worked regions, frequently in the proximity of welds. The cracks are located along grain boundaries where at elevated temperatures fine precipitates can form, causing the grain boundary to lose ductility and crack when strained. Often a nickel-rich filament is found in the grain boundary surrounded by a chromium-enriched oxide layer. One crack location, shown in Fig. 15, had a chromium-enriched zone that surrounded a filament. Typically stress relaxation cracking occurs at hardnesses above HV 200. The original hardness of the Incoloy 825 dip tube ranged from HV 215 to 227. These observed features supported stress relaxation cracking as the principal failure mechanism for the Incoloy 825 Alloy dip tube.
The dip tube wall temperatures could have been above 550 °C due to hot syngas on the inside and the lack of cooling on the outside. This could have caused the formation of carbides and precipitates in grain boundaries that would give the material less ductility when stressed and strained.
After the stress relaxation cracks formed, the elongated hole is believed to have developed, and once the through wall elongated hole was present, hot gases could escape into the region outside the dip tube. A metallurgical section through the edge revealed that the Incoloy 825 material had melted. The melting point of Incoloy is about 1,400 °C.
The panel section on the west side of the dip tube had two distinct cracks, one vertical and one horizontal, that were located above the top hat, approximately 180° away from the large hole on the east side. The vertical crack was about 4 in. above the top hat. The horizontal crack was angled at about 20° from the horizontal plane, about 3.5 in. above the top hat, as shown in Fig. 7. The wall thickness was reduced to a thickness of about 0.22 in. from an original thickness of 0.26 in. The horizontal crack was 1.8-in. long and initiated from the inside surface, shown in Figs. 16 and 17. Although the scale inside the crack was not analyzed, it was likely oxides and sulfides. The crack pattern along the grain boundaries was similar to that observed at the circumferential fracture on the east side, indicating that the stress relaxation cracks occurred before a hole was formed.