API 653 tank inspection: floor, shell, and settlement

An atmospheric storage tank fails in three places: through the floor, through the shell, or by settling until the shell or the floating roof no longer works. API 653 addresses all three, and it expects the inspection record to hold three different kinds of evidence. Floor plates are judged by a projection of remaining thickness. Shell courses are judged against the API 650 design basis. Settlement is judged by a geometric survey. A tank file that holds a floor scan report, a shell UT sheet and a surveyor's PDF has all three pieces and none of the connections. This paper sets out what each evaluation needs, how the interval follows from it, and what the record must carry to make the interval defensible.

What inspections does API 653 require and how often?

API 653 covers tanks built to API 650 and its predecessor API 12C, and it uses API 575 for inspection guidance, API 651 for cathodic protection and API 652 for linings. It defines four inspection activities with different frequencies.

• Routine in-service inspection. A monthly external walkaround by owner personnel: leaks, shell distortion, settlement signs, corrosion, foundation condition, coating, insulation and appurtenances.
• External inspection by an authorized inspector. At least every five years, or RCA/4N years where RCA is the difference between the measured shell thickness and the minimum required thickness in mils and N is the shell corrosion rate in mils per year, whichever is less.
• Ultrasonic thickness of the shell. When the corrosion rate is not known, at intervals not exceeding five years. When it is known, RCA/2N years or fifteen years, whichever is less. This is often done during the external inspection.
• Internal inspection. The interval is governed by the bottom, so that the minimum remaining thickness at the next inspection meets the code's table. Without RBI the interval cannot exceed twenty years, and where the corrosion rate is unknown and cannot be established from similar service it cannot exceed ten. With a documented RBI assessment the interval may extend to thirty years.

Each interval is derived, and the record must hold the derivation: the shell readings and rate behind RCA/4N, the bottom rates behind the internal interval, or the RBI assessment and its approvals behind an extension.

How is the tank bottom evaluated?

The bottom is where most tanks leak, and API 653 evaluates it with a projection rather than a snapshot. The internal inspection collects thickness data, usually by magnetic flux leakage floor scanning with ultrasonic prove-up of indications, from both the product side and the soil side. The record needs both, because they corrode by different mechanisms and at different rates.

The governing quantity is the minimum remaining thickness at the next inspection:

MRT = (lesser of RTbc or RTip) minus Or multiplied by (StPr plus UPr)

where RTbc is the minimum remaining thickness from bottom-side corrosion after repairs, RTip is the minimum remaining thickness from internal (product-side) corrosion after repairs, Or is the interval in years to the next internal inspection, StPr is the maximum rate of corrosion on the top side not repaired, and UPr is the maximum rate of corrosion on the underside. MRT must not fall below the code table: 0.10 inch where there is no means to detect and contain a bottom leak, or 0.05 inch where a release prevention barrier and leak detection are in place. Where a reinforced lining meets API 652, the lining's own criteria apply.

Two things follow for the record. First, repairs during the outage change RTbc and RTip, so the bottom evaluation must be recomputed after repairs, not before. Second, the underside rate depends on cathodic protection, soil and the presence of a barrier, so the evaluation should carry the CP survey status and the bottom design as inputs. The critical zone, the bottom within a few inches of the shell, carries its own minimum thickness rule because the shell-to-bottom weld sees the highest stress and the worst edge settlement effects.

How is shell thickness evaluated?

API 653 sends the shell back to its construction standard. For tanks under 200 feet in diameter the one-foot method applies:

tmin = 2.6 (H minus 1) D G / (S E)

where H is the height in feet from the bottom of the course under consideration to the maximum liquid level, D is the tank diameter in feet, G is the highest specific gravity of the contents, S is the maximum allowable stress for the course, and E is the original joint efficiency. Larger tanks use the variable design point method from API 650. Because H is measured from the bottom of each course, every course has its own tmin, and the record must hold one per course, not one per tank.

For a locally thinned area the code allows averaging. The thicknesses along a vertical line through the thinnest point are averaged over a critical length L = 3.7 times the square root of (D t2), with D in inches, t2 the least thickness in the area in inches, and L limited to forty inches. The average must meet tmin and the least thickness t2 must not fall below sixty percent of tmin. Widely scattered pits may be disregarded when no pit leaves less than half of tmin and the pit dimensions along any vertical line total no more than two inches in an eight-inch length.

Older tanks raise a second shell question: brittle fracture. API 653 provides a decision path for tanks that may operate below the design metal temperature, based on material, thickness, service history and whether a hydrostatic test has demonstrated fitness. The record must show that the assessment was made, because it is not repeated at every inspection.

How does API 653 treat settlement?

Settlement does not appear on a thickness sheet, which is why it is the most commonly neglected part of a tank record. Annex B of API 653 describes the survey and the evaluation.

• The survey. Elevations around the shell at a minimum of eight points, with the number of points not less than the diameter in feet divided by ten, and spacing not exceeding about thirty-two feet. Bottom elevations along radial lines are added when interior access is available.
• Uniform settlement. The whole tank drops together. It does not stress the shell but affects piping connections and drainage.
• Planar tilt. The tank tips as a rigid body. It increases hoop stress on one side and, for floating roof tanks, threatens roof seal performance and travel.
• Out-of-plane settlement. The component that remains after uniform settlement and tilt are removed. This is the one that distorts the shell. Annex B fits the measured points to a cosine curve and compares the deviation with an allowable deflection that depends on the arc length between survey points, the yield strength, the modulus and the shell height.
• Edge settlement and bottom settlement near the shell. Evaluated against limits that depend on the radial width and depth of the settled area and whether the bottom welds run parallel or perpendicular to the shell.

The survey is only meaningful in comparison with prior surveys. A tank that has settled uniformly by two inches over thirty years is behaving. A tank whose out-of-plane component grew by half an inch since the last survey is not. The record must hold the points, the datum, the date and the surveyor for each survey so the comparison can be made.

What about roofs, nozzles and appurtenances?

The parts of the tank that are neither floor nor shell still generate findings and still need a home in the record.

• Fixed roofs. Roof plates have a minimum remaining thickness rule, and supports, rafters and columns are inspected for corrosion at the plate interface and at the bottom where they meet the floor.
• Floating roofs. Pontoon integrity, deck thickness, seal condition and gap, roof drains, guide poles, anti-rotation devices, legs and landing conditions.
• Nozzles and shell penetrations. Reinforcement, weld condition, and the effect of settlement on connected piping. Shell penetrations larger than a threshold size count as major alterations when added.
• Foundations. Ringwall cracking, drainage, anchor bolts, and the shell-to-foundation seal that keeps water away from the bottom edge.
• Cathodic protection and linings. CP potentials to API 651 and lining condition to API 652 feed directly into the underside and product-side rates used in the bottom evaluation.

Repairs and alterations to any of these have their own rules in API 653, including when a hydrostatic test is required after major repairs and the conditions under which it may be exempted with an engineering evaluation on file.

What does the tank record need to carry?

Put together, an API 653 tank file has a definite structure, and each part must be traceable to the others.

• Construction basis: API 650 edition, diameter, height, course thicknesses and materials, joint efficiency, design specific gravity, bottom design and leak detection arrangement, roof type.
• Shell: per-course tmin with the method, per-CML readings over time, corrosion rate and RCA, the external and UT intervals derived from them.
• Bottom: floor scan results with prove-up readings, top-side and underside rates, repairs performed, MRT calculation with inputs, and the internal interval derived from it.
• Settlement: every survey with its points, datum and evaluation, and the trend between surveys.
• Brittle fracture assessment and its basis.
• Roof, nozzle, foundation and CP findings with photos bound to location.
• Repairs and alterations with classification, procedures, NDE and hydrostatic test or exemption evaluation.
• Approvals and signatures with the qualification of the signer.

The test is whether a new inspector can pick up the file and rerun the internal interval without a phone call. If the MRT inputs are in the file, they can. If the interval is a date on a schedule, they cannot.

How intelliSPEC handles it

intelliSPEC holds the tank as an asset node with its courses, bottom, roof and appurtenances as components, so shell readings, floor scan prove-up points and roof findings all attach to one record with one history. Ultrasonic readings arrive in intelliNDT from supported gauges without retyping, and each CML carries its nominal thickness, tMin and method, material, allowable stress, joint efficiency and three corrosion rates.

The calculation engine routes tMin by equipment type, and tanks route to API 653, with the design code and formula returned alongside the number. The corrosion summary produces short-term, long-term and maximum rates and computes remaining life as thickness minus tMin over the governing rate. Because the platform is metadata-driven, with forms and workflows configured through the form designer rather than code, a settlement survey, a floor scan prove-up sheet or a floating roof seal gap record is a configured form with structured fields, photos bound at capture, and GPS and drawing annotation where they help. Completed records pin the form configuration they were executed with, so a survey from a prior outage renders as it was signed.

Internal inspections, external walkarounds and roof inspections are captured in intelliINSPECT, fully offline with no time limit on the local store, which matters on a tank farm. Where a tank is a candidate for an RBI-extended interval, the assessment is carried in intelliINTEGRITY and its approvals are recorded with actor, time and a hash-chained audit trail. The terminal and tank farm context for this work is described on the pipelines and terminals page.

API 653 asks three different questions of a tank. The record that answers all three, with the arithmetic showing, is the one that holds up when the interval is challenged.