December 12, 2002
Editor's Notes: In-service weld defects found in Australian refineries by an inspection team required assessment to determine the best course of action - repair, replace, or do nothing. This is the author's first-person account of the team's findings and solutions.
The post-World War II period to the mid-1960s was the beginning of a building surge in major refining, petrochemical, and similar plants in Australia. While the welding quality in pressure equipment was excellent in most cases, at least at Bulwer Island, the field construction teams were under pressure to complete fabrication on time, and nondestructive testing (NDT) was more cumbersome than it is today. As a result, welding quality sometimes was compromised. In addition, it was possible welds would deteriorate over time, particularly when the material used (carbon steel unless otherwise indicated) was not best suited for the service.
When an in-service weld is found to be defective because of faulty construction or service deterioration, our assessment will result in one of three possible alternatives: Do nothing, repair, or replace.
The reactor was scheduled to be upgraded to increase its service temperature, a process that involved internally insulating the shell to lower the steel temperature to about 200 degrees C.
We took plate samples from the shell to perform a series of tests, including creep tests, to assess the significance of the defects.
The tests showed that the material still had adequate properties and was still fit for service with no specified time limit once the modifications were completed and the temperature reduced. Therefore, we took no action to repair the affected welds.
In contrast, inspection of piping, particularly by radiography, often revealed defects in welds that had been in service for decades. Usually such welds are incompletely penetrated. Assessing each weld would be a major undertaking.
When the piping is not subject to pressure cycling or severe vibration, is not highly stressed from external loads, shows no indication of corrosion in the affected area, and does not have lethal contents, it does not require repair.
On the other hand on most of the heat exchangers in an HF alkylation unit, the butt welds on almost every neck flange were underpenetrated, forming a crevice. Because the unit used HF as a catalyst, it was subject to severe corrosion. We repaired these welds.
We conducted an inspection of a deaerator in response to the well-known industry scare in the 1980s. As had been recommended the internal weld reinforcing was ground smooth but not quite flush with the walls to perform the inspection. The welds were examined by the wet fluorescent magnetic particle technique (WFMT), with the yoke running off a 12-volt DC transformer for safety reasons. No cracking was found.
A few years later we repeated the inspection -- this time running the yoke off 240-V AC. This inspection revealed a classic case of deaerator cracking, with thousands of cracks transverse to the welds.
We tried several other inspection techniques, including a repeat attempt with WFMT using a DC yoke. Only WFMT with 240-V AC showed the cracking. We considered it unlikely that the cracking had developed in the interval between the inspections and were almost certain that the first inspection had missed the cracking.
Subsequently we found that the vessel originally had been fabricated by rolling the plates and butting the square-cut ends together, then welding over the join. No attempt had been made to penetrate the weld. Grinding off the inner weld reinforcing during the first inspection had inadvertently removed nearly half the weld. The potential implications hardly need stressing. In this instance the welds were ground out and replaced with full penetration welds.
Note: The lessons learned from this experience are incorporated in the Welding Technology Institute of Australia (WTIA) Guidance Note on deaerator cracking.
The deaerator mentioned previously was manufactured by a small company; however, purpose-built vessels constructed by large manufacturers are not immune to error either. Breechlockheat exchangers are constructed for very high-pressure service. In one particular exchanger, the tube sheet and channel were forged in one large piece made of 5Cr 1/2Mo material. The alloy tubes were seal-welded to the 200-mm-thick tube sheet. We suspected the exchanger of leaking. Inspection of the seal welds revealed cracks and a high degree of hardness.
During manufacture the forging had been stress-relieved before the tubes were installed. The seal welds, therefore, were locally stress-relieved by placing heating pads against the tube sheet and heating the face of the tube sheet up to the required temperature.
The heat treatment time and temperature were lower than we would have specified for this material, and we strongly suspected that the thermocouples were placed on the tube sheet under the heating pad where they may not have properly indicated the temperature. In any event, the process had not adequately softened the welds, and it was not surprising that they had cracked.
We ruled out weld repair for a number of reasons, and a new exchanger bundle was purchased at a cost of about $500,000.
What we had identified as the major problem -- WIG from high-temperature service -- proved not to require repair (admittedly after an extensive investigation), whereas the simple error in the stress-relieving technique used resulted in a very costly replacement.