Capstone: documenting a pressure vessel repair

In June, V-412 finally gets its deferred internal inspection during the Northgate turnaround. On the second day, the owner's inspector runs magnetic particle along the nozzle welds and stops at N3, a 200 mm (8 in) nozzle on the lower shell. A thin red line runs along the toe of the weld for about 90 mm (3 1/2 in). Ruth Ambrose writes the finding, the turnaround planner calls Ironbridge, and by the afternoon Kiera is standing beside the nozzle with Harjit, Grant Mahler and the crew that will do the repair. "This is the whole course on one nozzle," Harjit says. "Code, plan, materials, welding, examination, test and report. You're going to write it up as we go." This capstone follows the repair from finding to return to service, with the documents produced at every step.

The finding

The finding report records the vessel, its registration and serial numbers, the location of the indication, its length, the method used to find it and the inspector's requirement: remove the crack, repair to the original design, and examine. Ruth adds the engineering context. V-412 is a Section VIII, Division 1 drum, 16 mm (5/8 in) SA-516 Grade 70, design pressure 1,700 kPa (250 psi), corrosion allowance 3 mm (1/8 in), built by Ironbridge in 2027. N3 connects to a line that was rerouted in the last turnaround, and the operators have reported vibration on it since.

The first decision is whether this is a repair or an alteration. Removing a crack and restoring the original weld is a repair, as long as the nozzle, its reinforcement and the shell stay as designed. If the excavation went so deep that the reinforcement calculation was affected, or if Ruth wanted to change the nozzle, it could become an alteration. For now the plan is written as a repair, with a note that the engineer will be consulted if the remaining wall falls below what the design needs.

The repair plan

Kiera drafts the plan from Ironbridge's repair procedure template, and Harjit reviews it before it goes to Grant. Every heading in the plan points back to a lesson in this course.

Grant reads the plan and accepts it with one change: because the drum is in place with internals installed and Northgate cannot fill it with water during the turnaround without a long delay, he will decide on the test after he has seen the excavation, choosing between a hydrostatic test and additional examination. He signs the acceptance, and the work can start.

Removing the crack

Before any grinding, the crew confirms the vessel is isolated, blinded, purged and cleared for hot work and entry under Northgate's permits, and the welding area is protected. The crack is ground out with a carbide burr and a small grinder, working along its length in shallow passes and keeping the groove smooth and boat-shaped so that it can be welded without trapping slag. After each pass the inspector's technician runs magnetic particle on the groove. At 4 mm (0.16 in) there is still a short indication at one end; at 5.5 mm (0.22 in) the groove is clear.

Kiera records the finished excavation: its length, width and depth, a sketch with dimensions from the nozzle centreline, the MT results at each stage, and the ultrasonic thickness readings around it. The lowest thickness reading near N3 is 15.4 mm (0.605 in), a little under the nominal 16 mm (5/8 in) plate, which is expected after two years of service.

# V-412 nozzle N3 repair: can the excavation be weld repaired, and what goes on the report? (metric)
P, R, S, E, CA = 1700, 900, 138000, 1.0, 3
t_req = P * R / (S * E - 0.6 * P)            # required thickness without corrosion allowance
nominal, ut_min = 16, 15.4                # nominal shell and lowest thickness measured near N3
depth = 5.5                               # deepest point of the excavation after the crack was removed
remaining = ut_min - depth
print(f"required {t_req:.1f} mm, measured {ut_min} mm, excavation {depth} mm, remaining {remaining:.1f} mm")
print("remaining wall is below the required thickness: weld repair needed" if remaining < t_req
      else "remaining wall meets the required thickness: blend and examine")
length, width = 90, 20                  # excavation size
vol = 0.5 * length * width * depth            # rough boat-shaped groove volume
print(f"weld metal about {vol * 0.00785:.0f} g (groove {length} x {width} x {depth} mm)")
report = {"repair organization and permit number": "Ironbridge Industrial, ABSA permit on file", "owner": "Northgate Upgrader", "vessel registration": "V-412, CRN and ABSA number", "description of work": "crack at N3 toe removed, weld repaired, MT and UT", "materials and filler": "E4918 (E7018), heat recorded", "welders and WPS": "J. Okeke, WPS 12", "pressure test or NDE": "MT and UT, inspector accepted", "safety codes officer acceptance": ""}
blank = [k for k, v in report.items() if not v]
print("report fields still blank:", ", ".join(blank) if blank else "none")

required 11.2 mm, measured 15.4 mm, excavation 5.5 mm, remaining 9.9 mm
remaining wall is below the required thickness: weld repair needed
weld metal about 39 g (groove 90 x 20 x 5.5 mm)
report fields still blank: safety codes officer acceptance

Grant comes at the hold point, looks at the clear groove and the MT record, checks the thickness figures and agrees that a weld repair is needed and that the original design is not affected. He also decides on the test: because the repair is a small, shallow weld at a nozzle toe, he will accept magnetic particle and ultrasonic examination after welding in place of a hydrostatic test, and records that decision on the plan.

Welding

J. Okeke makes the repair. Before he strikes an arc, Harjit checks three things in front of the class. The WPS: WPS 12 covers SMAW, P-No. 1 to P-No. 1, F-No. 4, the thickness and the preheat. The welder: J. Okeke's SMAW qualification is current, with his last code weld in May. The filler: the E7018 electrodes come from the rod oven, and their lot number goes on the weld record. The area is preheated, checked with a temperature indicator, and the groove is welded in stringer beads, each cleaned and inspected, with the final layer slightly above the surface and then ground flush to the shell and blended into the nozzle weld so that there is no sharp toe left to start a new crack.

Kiera fills in the weld record as the work goes: the date, the welder, the WPS, the electrode size and lot, the preheat reading, the interpass temperatures, the number of passes and the time. It takes less than an hour of welding, and more than an hour of paperwork, which Harjit says is about the right ratio for a code repair.

Examination and acceptance

After the weld has cooled for the time the shop's procedure requires, the technician examines the repair by magnetic particle on the surface and ultrasonics through the weld. Both are clear. The NDE reports are signed by the technician, with his certification level, and attached to the repair record. Grant reviews the reports, looks at the finished repair, checks the weld record against the WPS and the welder's qualification, and accepts the work.

Ruth adds one more action that is not part of the code repair but is the reason it should not happen again. The crack ran along the toe of a nozzle weld on a line that vibrates, which points to fatigue. She has Northgate's piping group add a support and a vibration check on the N3 line before start-up, and she sets a shorter interval for the next MT of N3 in V-412's inspection plan.

The report

The repair is reported on ABSA's repair and alteration report. Kiera fills in each field from the records: Ironbridge and its permit number, Northgate as owner, V-412 with its CRN, ABSA registration and serial numbers, a description of the work, the filler metal and its lot, the welder and WPS, the examinations and their results, the inspector's decision on the test and his acceptance. The program shows the one field she cannot fill: the safety codes officer's acceptance, which Grant signs once the package is complete. The report goes to Northgate for V-412's records and to ABSA.

On the last night of the code block, Harjit hands back the blacked-out V-412 calculation from lesson 4 and the N3 repair package side by side. "Two years ago this drum was a stack of plate in the yard," he says. "Somebody had to register it, buy the right steel, keep it traceable, weld it with qualified people, examine it, test it, stamp it and write it all down. Now it's cracked, and the same chain starts again. That's the job: not just making the weld, but making sure the next person can trust it."

The time line of the repair

Kiera's write-up ends with a short time line, because the turnaround schedule depends on how long code repairs take, and the planner wants to know where the time went.

Less than two days went into the repair itself, and most of that was waiting for permits, for the steel to cool and for examination. The planner's lesson, which Ruth passes on, is that a code repair is fast when the repair organization has its procedures ready, its welders current and its paperwork templates prepared, and slow when it has to work any of that out on the day.