Case Study: Full-containment LNG storage tank, 240 ft diameter, 9% Ni inner shell with carbon steel annular plate — LNG storage and export

Asset & context

LNG storage and export — US Gulf Coast (Louisiana). Full-containment LNG storage tank, 240 ft diameter, 9% Ni inner shell with carbon steel annular plate. Crew of 8 for 41 days.

The challenge

A 240-ft full-containment LNG tank with a 9% Ni inner shell and 6 mm carbon steel annular plate ring (the perimeter plate at the shell-to-floor connection) was opened for its 10-year out-of-service API 653 internal inspection.[1] The tank had been in service for 22 years with one prior internal inspection at year 10. The inspection plan required: (a) 100% MFL of the inner floor and annular plate, (b) UT thickness verification of every MFL indication ≥ 30% wall loss, (c) VT and MT of the shell-to-bottom inside fillet weld, and (d) bottom-plate-to-grade VT of the outer surface where access via crawl-space hatches was possible. The campaign ran on a 45-day shutdown window — every day of overrun cost the terminal $720,000 in deferred LNG export ship loadings against contracted offtake commitments.

Approach

Method selection: robotic MFL for floor screening, PAUT and UT for verification

API 653 §4.4 and API RP 575 §10.5 specify MFL as the standard screening method for atmospheric storage tank floors.[1][2] MFL on a 6 mm carbon steel floor plate gives roughly 90% POD on a 20% wall loss feature with the modern generation of robotic crawler tools (Silverwing RMS-PA, Aqualis FloorMap, Eddyfi VersaScan). The operator selected a Silverwing RMS-PA running 250 mm × 250 mm magnet head coverage and integrated UT confirmation, with onboard data logging tied to a tank-floor grid coordinate system.

The floor coverage plan gridded the 12,200 m² floor into 245 lanes of 250 mm width. Robotic MFL throughput at the tool's rated 0.4 m/sec scan speed gave full-coverage time of 102 hours wall-clock, distributed across three 18-hour shifts. The annular plate ring — the 600 mm perimeter band at the shell-to-floor weld — required manual MFL with a hand-pushed scanner because the robotic tool could not maneuver into the corner geometry. Manual MFL on the annular ring added 16 hours.

Verification of MFL indications ≥ 30% wall loss used PAUT corrosion mapping per ASME V Article 4 Mandatory Appendix IV on a 200 mm × 200 mm window centered on each MFL call.[3] The PAUT scan resolved actual pit depth and shape, plus identified whether the pit was soil-side (under-floor, the typical case for tank-floor pitting driven by chloride-bearing groundwater) or product-side (top-of-floor, much rarer on a clean LNG service). For the 9% Ni inner shell, separate UT thickness scans were run on the lower three courses by climbing scaffold — 9% Ni cannot use MT due to the alloy's low magnetic permeability, so MT was substituted with PT on critical welds per ASME V Article 6.[4]

Procedure: robotic floor scan, manual annular ring, shell course UT

The robotic crawler was deployed through the central manway after the tank was confirmed gas-free (LEL < 10%, O2 > 19.5%, target gases for LNG service < 1 ppm). Surface prep was minimal — the floor was washed with clean water and dried, with any loose debris removed by hand. MFL is tolerant of light surface contamination but heavy mill scale or paint > 2 mm causes signal attenuation. The 22-year-old floor had no remaining paint and a thin uniform mill-scale layer, well inside the MFL tolerance band.

Each lane was scanned with the crawler running parallel to the previous lane at a 50 mm overlap. The MFL signal was streamed to the analysis station in real-time, with auto-flagging at the 20% wall-loss threshold. A Level III MFL analyst reviewed every flagged indication during acquisition, with hold points to spot-check the crawler navigation accuracy by re-locating fixed reference points (column footings, sump grates, manway).

PAUT verification was performed on every MFL indication ≥ 30% wall loss — 312 of the 4,180 total MFL indications. PAUT scan time per indication averaged 18 minutes including setup. The PAUT C-scan was compared against the MFL indication shape and depth call, with a Unity Plot generated at the end of the campaign covering all 312 verified indications. Of the 312 PAUT-verified indications, 280 confirmed within MFL tolerance, 26 were over-called by MFL by > 10% wall, and 6 were under-called by > 10% wall — the under-calls were the priority for repair sentencing.

Findings: 18 annular-plate pits drive the repair scope

The deepest indication on the entire floor was an under-floor pit on the annular plate ring with a measured depth of 4.1 mm against a 6.0 mm nominal — 68% wall loss. 18 indications on the annular plate ring measured ≥ 30% wall loss; none of the floor-plate-proper indications (away from the perimeter) exceeded 30%. The damage signature on the annular ring was characteristic of chloride-driven soil-side pitting per API RP 651 §A.2 — discrete deep pits with no surrounding general thinning, concentrated in the 100 mm band closest to the shell-to-floor weld.[5]

Two factors drove the annular-ring pattern: (1) chloride condensate accumulation in the slight depression where the annular ring overlaps the sand-pad foundation, exacerbated by the absence of an under-tank cathodic protection grid (the tank predated mandatory CP for LNG-service tanks under the operator's current asset standards); (2) micro-galvanic corrosion at the 9% Ni shell to carbon-steel annular plate weld, with the dissimilar-metal couple driving local accelerated corrosion on the carbon-steel side.

The shell-to-floor inside fillet weld inspected clean — no surface cracking by PT, no UT-detected sub-surface defects. The 9% Ni lower shell courses inspected at 91-94% remaining wall against nominal — no significant corrosion of the cryogenic shell material. The campaign closed the floor inspection with 18 sentenced repair locations and a confirmed-clean shell.

Disposition: localized inserts, CP retrofit, scope expanded for next interval

Repair scope under API 653 §9.10 specified excavating each of the 18 pits and installing flush-patch inserts per API 653 §9.10.1.4.[1] Each insert was sized as a minimum 300 mm × 300 mm square plate with full-penetration welds inspected by RT and PT. Eight of the 18 pits were close enough to each other to consolidate into three larger insert plates, reducing total weld length and weld-related thermal stress on the annular ring. Welding used SMAW with E7018-equivalent electrodes, preheat 50°C, with the welds qualified per ASME IX and the post-weld inspection per API 653 §11.[1]

The asset team specified a retrofit cathodic protection system under the tank — an under-floor impressed-current grid sized per NACE SP0193 for a 240-ft tank on sand-pad foundation.[6] Installation required removing a 1.5 m perimeter band of foundation soil and trenching for the CP anode strings. CP retrofit added 11 days to the outage but saved an estimated 40-50% reduction in annular-plate corrosion rate over the next 10-year interval based on industry CP-effectiveness data.

The next-interval inspection plan tightened MFL coverage on the annular plate ring to 5-year semi-annual intervals using external rope-access UT through dedicated annular-ring inspection ports (cut into the tank during this outage), with the next full out-of-service inspection scheduled for 10 years out. Total inspection-and-repair cost: $3.8M. Avoided cost: a missed soil-side pit propagating through the annular plate to floor-failure releases LNG into the secondary containment annulus, with associated lost-product, terminal-shutdown, and regulatory-action costs that historically run $40-80M per incident in industry data.

Methods deployed

Defects found

Outcome

18 flush-patch inserts installed and inspected per API 653 §9.10. Under-tank impressed-current cathodic protection retrofit installed per NACE SP0193. Annular-ring inspection ports cut for between-interval external UT. Next out-of-service inspection at 10-year interval, with semi-annual external annular-ring monitoring.

Cost avoidance

$40-80M against the worst-case annular-plate floor-failure scenario per industry incident data. Direct schedule cost avoidance: completed in 41 days against the 45-day budget, retaining 4 days of $720K/day buffer for export ship slot recovery = $2.88M.

Certifications required

Crew qualifications: