Corrosion rate: short term, long term, and which governs

Two inspectors look at the same CML history and produce two different next inspection dates. Neither made an arithmetic error. One used the long-term rate, one used the short-term rate, and the two differ by a factor of three because the last reading was taken fourteen months after the one before it and the wall lost 0.3 millimeters in between. The corrosion rate is the single number that drives remaining life and therefore the inspection interval, and the standards define two ways to compute it. This paper sets out both, explains why they disagree, and describes how the governing rate should be chosen and recorded so the decision is visible to the next person who reads the record.

How do API 510 and API 570 define corrosion rate?

Both standards describe two rates at each corrosion monitoring location (CML), and they use the same arithmetic.

Long-term (LT) corrosion rate:

Rate = (t initial minus t actual) divided by the time in years between the initial and actual readings.

Here t initial is the first thickness measured at the location (the baseline, or nominal thickness less mill tolerance if no baseline reading exists) and t actual is the latest measured thickness.

Short-term (ST) corrosion rate:

Rate = (t previous minus t actual) divided by the time in years between the previous and actual readings.

Here t previous is the reading immediately before the latest one.

The long-term rate averages the whole history and is stable. The short-term rate looks only at the most recent interval and reacts to change. The standards then instruct the inspector, in consultation with a corrosion specialist where the choice is not obvious, to select the rate that best reflects the current process. The conservative default in most programs is the higher of the two, which produces the shorter remaining life. That default is right most of the time, and wrong often enough that the choice needs to be a recorded decision rather than a formula.

Both standards also allow the rate to be established statistically over a set of CMLs in the same circuit or on the same equipment, provided the corrosion is general rather than localized, and both address new equipment with no history: estimate from similar service or published data, or measure after approximately 1,000 hours to establish a probable rate.

Why do the short-term and long-term rates disagree?

They disagree because they measure different things. Three causes account for most of the gap.

• The process changed. A feed change, a temperature increase, a new injection point or a wash water upset raises the corrosion rate. The long-term rate dilutes the change across years of benign history; the short-term rate shows it. In this case the short-term rate is the truth and the long-term rate is a comfortable lie.
• The process improved. Inhibitor injection started, a corrosive feed was removed, or a lining was applied. Now the long-term rate overstates current loss and the short-term rate is closer to reality. Using the long-term rate here is conservative, and wastes inspection effort on equipment that is no longer corroding.
• Measurement noise. Neither reading is exact. A short interval with a small thickness change produces a short-term rate dominated by gauge and operator error, in either direction. This is the case that produces the wild short-term rate, and it is the most common.

The inspector's job is to decide which cause explains the gap. The record's job is to make that decision reviewable. A CML history that shows the readings, dates, gauge, technician and surface condition lets a reviewer recognize the third cause without a site visit. A history that shows only the rate does not.

When is the short-term rate really just gauge noise?

Ultrasonic thickness gauges on prepared surfaces are typically good to a few hundredths of a millimeter. In the field, with coating, scale, temperature, couplant variation, probe angle and a different technician than last time, repeatability of 0.1 to 0.25 millimeters is realistic. That is the noise floor.

Consider a CML measured at 12.50 mm, then 12.35 mm one year later. The apparent short-term rate is 0.15 mm per year. If the repeatability is 0.2 mm, the true change is anywhere from wall growth to 0.35 mm per year. The number is not wrong; it is unresolved. The same 0.15 mm change over five years is 0.03 mm per year with error bounds of plus or minus 0.04, which is informative.

Signs that a short-term rate is noise rather than corrosion:

• The interval between readings is short (under two years) and the change is within one or two gauge resolutions.
• Neighboring CMLs on the same component show no similar change.
• The rate is negative, meaning the wall grew, which it did not.
• The two readings were taken by different technicians, with different gauges, or on different surface conditions.
• The location is on a coated or scaled surface where couplant and probe seating vary.

A negative rate should be treated as zero for calculation and as a flag for investigation: either the earlier reading was low, the current reading is high, or the probe was not on the same spot. The last is common enough that CML marking and drawing location matter as much as gauge calibration.

How should the governing rate be chosen and recorded?

Choosing the governing rate is where the calculation becomes engineering. A defensible procedure:

1. Compute both rates at every CML, every time a new reading is entered. Do not compute only one.
2. Compute the maximum of the two as the default governing rate. This is the conservative position the standards point toward when there is no reason to prefer one.
3. Screen for noise. Where the short-term rate exceeds the long-term rate and the interval is short, the change is within measurement resolution, or the rate is negative, mark the short-term rate as unresolved and hold the long-term rate as governing, with the reason recorded.
4. Screen for process change. Where the short-term rate exceeds the long-term rate across several CMLs on the same circuit, or a known process change occurred, adopt the short-term rate and record the change as the reason.
5. Record the decision on the CML: the governing rate, which of the two it is, who chose it, and why. A rate without a reason is a rate that will be argued about.
6. Re-evaluate at every reading. The governing rate is not fixed. Each new measurement moves the short-term rate and, more slowly, the long-term one.

Where a corrosion specialist overrides the default, the override is itself a record: the specialist, the date, the basis (a process change, a corrosion study, a comparable circuit) and the date the override should be reviewed.

What does the corrosion rate feed downstream?

The reason the rate matters is what it drives. Remaining life is the measured thickness minus the required minimum thickness, divided by the governing corrosion rate. API 510 and API 570 then cap the internal or thickness inspection interval at half the remaining life (subject to the standards' maximum intervals), so an error in the rate propagates directly into how long the equipment runs between inspections.

The chain looks like this at a single CML:

• Latest thickness, from the gauge, with date and technician.
• tMin, from the design code for the equipment type (ASME B31.3 for piping, ASME Section VIII Division 1 for vessels, API 653 for tank shells), with the formula and inputs visible, and compared against any structural minimum.
• Governing corrosion rate, chosen as above.
• Remaining life, as (t actual minus tMin) divided by rate.
• Next inspection interval, at no more than half the remaining life, capped by the standard.
• Next inspection date, from the last inspection date plus the interval.

Each step should be visible on the record, because each is a place where two inspectors can reasonably disagree and where an auditor will ask why. A system that shows a next inspection date without the rate and formula behind it has hidden the decision that mattered most.

How intelliSPEC handles it

Each CML record in intelliSPEC carries 64 fields, including nominal thickness, tMin and the method used to derive it, design pressure, material, corrosion allowance, latest thickness and three corrosion rates: short-term, long-term and maximum. The maximum is the governing rate by default, and remaining life is computed as (t minus tMin) divided by that rate. intelliINTEGRITY routes tMin by equipment type to ASME B31.3, ASME Section VIII Division 1 or API 653 and returns the design code and the formula alongside the number, then caps the API 510 or API 570 interval at half the remaining life. Every result is provisional until an engineer verifies it, and the product never auto-derates equipment.

Readings arrive from gauges such as the Evident 38DL Plus, the Elcometer MTG and the DeFelsko PosiTector UTG over Bluetooth or keyboard wedge, so the value on the record is the value the gauge displayed, with the device identity attached. intelliNDT carries the UT procedure and the technician's qualification with the reading, which is what lets a reviewer judge whether a suspicious short-term rate is a process change or a probe seating problem. Field capture works fully offline with no time limit on the local store, and where two technicians touch the same CML offline, the conflict is recorded at field level with the local value, the remote value, who resolved it and when.

Completed records pin the field configuration they were executed with, so the rate decision recorded today renders unchanged at the next inspection. Definitions for the terms used here are in the glossary.