Corrosion Rate Calculation: Long-Term vs Short-Term Rates in API 510/570

Short answer: In API 510 and API 570, the corrosion rate for thinning is the wall loss between two thickness readings at the same condition monitoring location (CML) divided by the years between them. The long-term rate uses the initial reading and the latest reading; the short-term rate uses the previous and latest readings. The codes do not say "always use the higher one": the inspector, with a corrosion specialist, compares both and selects the rate that best reflects current conditions, then uses it for remaining life and the next inspection date.

That last sentence is where most programmes go wrong, so this guide spends most of its time on rate selection: when the long-term rate is the honest number, when the short-term rate is warning you about something real, when the short-term rate is just measurement noise, and what the code expects for new equipment that has no history yet. It also covers the arithmetic and units (inches per year, mils per year, millimetres per year), how weight-loss coupons and probes fit alongside CML thickness data, and what any corrosion rate spreadsheet or software has to get right. For how the rate then drives remaining life and inspection intervals, see our companion article on remaining life calculations.

Code references in this guide come from API 510 (11th edition, 2022) and the current API 570 body of knowledge published by API. API codes are copyrighted; the requirements are described here in our own words, and you should confirm details against your licensed copy of the current edition.

The basic formula for corrosion rate

For thinning damage, API 510 defines corrosion rate as the difference between two thickness readings divided by the time between them. Written out in plain terms:

• Long-term (LT) rate = (initial thickness minus actual thickness) divided by the years between those two readings.
• Short-term (ST) rate = (previous thickness minus actual thickness) divided by the years between those two readings.

The definitions matter as much as the formula. The actual thickness is the reading at the CML from the most recent inspection. The previous thickness is the reading at the same location from the prior inspection. The initial thickness is either the first thickness measurement at that CML or the thickness at the start of a new corrosion rate environment. That last clause is easy to overlook and very useful: if the service changed, a process upset introduced a new mechanism, or the equipment was repaired, the owner can reset the baseline rather than averaging two different environments together.

The code also allows corrosion rates to be determined from thickness data collected at more than two different times. In practice that means a CML with five or six readings can be trended, not just differenced, which is far more robust than any single pair of readings.

API 570 treats piping the same way in principle. The current API 570 exam body of knowledge places corrosion rate calculation at clause 7.1.2, remaining life at 7.2, and inspection intervals at 6.3.3 with its interval table, so a piping inspector works with the same long-term and short-term concepts at the CML level. API 653 uses its own corrosion-rate-based approach for tank shells and bottoms; see our API 653 guide.

Units: inches per year, mils per year and mm per year

A frequent source of confusion is simply units. Thickness readings in the US are usually recorded in inches, so the raw rate comes out in inches per year (in/yr). Corrosion engineers usually talk in mils per year (mpy), where one mil is one thousandth of an inch. Outside the US, and in many corrosion-testing standards, the unit is millimetres per year (mm/y).

Keep the unit explicit in every record. A remaining-life calculation that mixes an in/yr rate with millimetre thicknesses will be wrong by a factor of 25, and the error is easy to miss in a large spreadsheet.

A worked example: one CML, two rates

Consider a carbon steel nozzle CML on a vessel. The numbers below are illustrative, chosen to show the arithmetic rather than to represent any real equipment.

• Initial reading (first measurement, 15 years ago): 0.500 in
• Previous reading (5 years ago): 0.452 in
• Actual reading (this inspection): 0.436 in

Long-term rate: (0.500 minus 0.436) / 15 = 0.064 / 15 = 0.00427 in/yr, about 4.3 mpy or 0.108 mm/y.

Short-term rate: (0.452 minus 0.436) / 5 = 0.016 / 5 = 0.0032 in/yr, about 3.2 mpy or 0.081 mm/y.

Here the long-term rate is higher. That pattern often means corrosion was faster earlier in life (perhaps before a process change, a chemical injection programme or a metallurgy upgrade upstream) and has since slowed. Whether the inspector should adopt the lower short-term rate depends on whether the reason for the slowdown is known, documented and still in place.

Now change one number. Suppose the previous reading had been 0.460 in. The long-term rate is unchanged at about 4.3 mpy, but the short-term rate becomes (0.460 minus 0.436) / 5 = 0.0048 in/yr, about 4.8 mpy. The short-term rate is now higher, which may mean something has accelerated recently. Before accepting it, the inspector will want to know whether the readings are trustworthy and whether something in the process changed.

If the required thickness at this location were 0.300 in, the remaining life at the long-term rate would be (0.436 minus 0.300) / 0.00427, roughly 32 years; at the higher short-term rate it would be about 28 years. The interval rules in each code then cap how that remaining life translates to a next inspection date; that step is covered in the remaining life article.

Choosing the governing rate: what the code actually asks

API 510 is explicit that the inspector, in consultation with a corrosion specialist, shall select the rate that best reflects current conditions. It is a judgement requirement, not a rule to take the maximum. The current edition lists factors to consider when deciding which rate to use for remaining life and the next due date. In summary, they are:

1. Whether the damage mechanism is general or localised.
2. Whether the area sees impingement, erosive flow or erosion-corrosion.
3. When the corrosion probably started, if not from first operation, so wall loss is measured over the right time span.
4. Whether a process change may have triggered it, such as water wetting, chlorides entering the process or lower pH.
5. Whether scale is protecting the surface, or protection has been lost (for example, higher velocity stripping scale).
6. Whether stagnant areas could see accelerated attack, such as where iron sulfide accumulates.
7. Whether the unit has stayed within its integrity operating windows (IOWs).
8. Whether a high short-term rate came from a one-off event, and whether the cause has been corrected.

Read together, these factors describe a short investigation rather than a lookup. A higher short-term rate with a plausible process cause (new crude slate, lost inhibitor injection, a velocity increase after debottlenecking) should normally govern until proven otherwise. A higher short-term rate with no plausible cause, at a CML with poor reading history, should prompt re-measurement before it drives a costly early outage. A lower short-term rate should only govern when the reason for the slowdown is understood and controlled, for example an IOW that has been held consistently since a known change.

Some owners write a default into their procedures, such as "use the higher of LT and ST unless the inspector documents a justification." That is a conservative and defensible local practice; just be clear that it is the owner's rule layered on top of the code, not the code itself, and that it still requires someone to look at the data.

Why short-term rates are noisy

The short-term rate divides a small thickness difference by a short time. Any error in either reading is magnified. Suppose (illustratively) each UT reading could be off by 0.005 in because of probe placement, couplant, surface condition or temperature. Over a five-year span, two readings that are each 0.005 in out in opposite directions change the short-term rate by 0.002 in/yr, which is 2 mpy, comparable to the corrosion rate itself in many services. Over a fifteen-year long-term span, the same error moves the rate much less.

API 510 lists sources of ultrasonic measurement inaccuracy that inspectors and examiners should work to eliminate, including improper calibration, coatings or scale, rough surfaces, probe rocking on curved surfaces, laminations, temperature effects at elevated temperature, and doubling on thin material. The practical controls are well known:

• Repeatable locations. Mark CMLs physically or with precise isometric references so the next reading is in the same spot. A reading taken 50 mm away on a pitted nozzle is a different measurement, not a trend.
• Consistent technique. Same probe type, same calibration approach and, where used, the same echo-to-echo or pulse-echo mode through coatings. Record the technique with the reading.
• Temperature correction. Readings on hot equipment need appropriate procedures or correction; API 510 points to metal temperatures typically above 150°F (65°C) as where accuracy is affected, with hot calibration blocks or correction factors as the usual remedies.
• Scanning where corrosion is localised. API 510 notes that ultrasonic scanning or profile radiography is preferred where corrosion is localised or thickness approaches the required minimum. A grid or area scan at the CML, rather than a single spot, gives a minimum that is far more stable over time; see corrosion mapping.
• Validation before use. The current API 510 states that thickness and rate data used in remaining life should be validated, because bad data leads either to unexpected failure or to premature retirement.

Negative rates (a later reading thicker than an earlier one) are the clearest sign of measurement scatter. They should not be averaged away silently; they should trigger a check of the location, the technique and the record. Our article on building a CML register that survives ten years covers the data-management side, and ASME Section V Article 5 thickness measurement covers technique.

Multiple readings, statistics and circuit rates

Once a CML has several readings, a least-squares trend line through all of them is usually a better estimate than either the LT or the ST pair alone, because it uses all the data and is less sensitive to one bad reading. The code language allowing corrosion rates from data collected at more than two times supports this.

API 510 also permits statistical analysis in corrosion rate and remaining life calculations for vessel sections, for example to support substituting on-stream inspection for an internal inspection or to set the internal interval. It cautions that the statistical treatment must reflect the actual condition of the vessel section, especially where corrosion is localised, that statistics may not be applicable where there is random but significant localised corrosion, and (in the current edition) that the analysis method must be documented.

For piping, owners commonly compute rates per CML and then summarise per circuit (average, a high percentile, or the worst CML), because a circuit is meant to group piping in similar service. The danger is that a circuit average hides one CML in a dead leg or downstream of an injection point that is corroding much faster. The governing rate for each CML's remaining life should still be that CML's rate, with circuit statistics used to spot outliers and to decide whether the circuit boundaries are right. See piping circuit and CML inspection.

New equipment and changes of service

A new vessel, or one whose service is changing, has no history from which to calculate a rate. API 510 requires the owner to establish a probable corrosion rate by one of several routes. In the current edition, those are: data from vessels in the same or similar service, appropriately placed ultrasonic sensors on the equipment, an estimate by a corrosion specialist, or published data for the same or similar service. Remaining life and the first inspection interval are then estimated from that rate.

Where none of those routes can be applied with confidence, the current edition requires the inspection plan to include an on-stream wall-loss check by direct measurement after six months of service, so that an unexpectedly high rate does not go unnoticed. It notes that, because of measurement error over such a short interval, this may not give a true corrosion rate, but it provides data to direct the plan until a credible rate is established. Older editions used a different early-life timing (the 2006 edition referred to approximately 1,000 hours of service), which is a good example of why you should not work from an outdated public copy. Subsequent determinations continue at appropriate intervals, and if the assumed rate turns out to be wrong, the actual rate replaces it in the remaining life calculation.

The "start of a new corrosion rate environment" definition of initial thickness ties in here: when service changes, the thickness at the change becomes the new baseline for long-term rate purposes.

Coupons, probes and other ways to measure corrosion rate

CML thickness readings measure what has happened to the pressure boundary. Corrosion monitoring devices measure what the process is doing now. Both produce a "corrosion rate", and they are complementary rather than interchangeable.

For weight-loss coupons, the widely used laboratory formula from ASTM G1 is corrosion rate = (K × W) / (A × T × D), where W is mass loss in grams, A is exposed area in square centimetres, T is exposure time in hours, D is density in grams per cubic centimetre and K is a units constant (8.76 × 10^4 for mm/y; 3.45 × 10^6 for mpy). Check the current ASTM G1 for the full method. Coupon and probe rates help the corrosion specialist interpret CML trends; the code remaining life is still based on the thickness of the component.

Pipeline operators work in a different framework: transmission and distribution pipelines regulated by PHMSA under 49 CFR 192 and 195 typically derive growth rates by comparing successive in-line inspection runs and direct assessment data, under their integrity management programmes. The API 510/570 LT/ST definitions apply to plant equipment and piping, not to regulated pipelines.

What any corrosion rate calculation tool must get right

Many people search for "corrosion rate calculation software" or a "corrosion rate calculator". A simple calculator does the arithmetic above; an inspection data management system does it for thousands of CMLs. Whatever tool is used, the code requirements set the checklist:

• Store every reading with date, CML identity, technique and inspector, so LT, ST and trends can be recalculated and audited.
• Support resetting the initial thickness at a new corrosion rate environment, with the reason recorded.
• Show LT, ST and trend rates side by side, so the inspector can make and document the selection the code requires, rather than silently picking one.
• Flag negative rates, implausible jumps and readings below the previous minimum for verification.
• Keep units explicit and consistent from reading to remaining life.
• Record who selected the governing rate and why, including any corrosion specialist input.

Spreadsheets can meet those requirements on a small scale; the failure mode at scale is lost history and undocumented selection. The tool does not decide the rate; the inspector does.

Common mistakes

• Applying "always use the higher rate" as if it were the code requirement, without investigating why the rates differ.
• Adopting a lower short-term rate without a documented, controlled reason for the slowdown.
• Computing short-term rates from readings taken at slightly different locations or with different techniques.
• Averaging a whole circuit and missing the one fast CML.
• Forgetting to reset the baseline after a service change, repair or replacement.
• Mixing units, or using the nominal thickness as "initial" when a measured baseline exists.
• Working from an old public edition of the code; early-life determination rules and the list of selection factors have changed between editions.

How Atlantis supports this

A corrosion rate is only as good as the thickness data behind it. Atlantis NDT collects repeatable CML thickness data for owner programmes: UT thickness surveys at marked CMLs, corrosion mapping and phased array area scans where corrosion is localised, and re-measurement of suspect readings, by ASNT-certified technicians under ASNT Level III oversight, with technique and location recorded for each reading. The rate selection, remaining life and interval stay with your API-certified inspector and corrosion specialist. See piping circuit and CML inspection and pressure vessel inspection services. Quotes within 24 hours: request a CML thickness survey quote.

Frequently asked questions

What is the formula for corrosion rate?

For thinning in API 510 and 570 programmes, corrosion rate equals the wall loss between two thickness readings at the same CML divided by the years between them. Long-term uses initial and actual readings; short-term uses previous and actual readings.

What is the difference between long-term and short-term corrosion rate?

The long-term rate spans from the first reading (or the start of the current corrosion environment) to now; the short-term rate spans only the last inspection interval. Comparing them shows whether corrosion has recently sped up or slowed down.

Should I always use the higher corrosion rate?

The code does not say so. API 510 requires the inspector, with a corrosion specialist, to select the rate that best reflects current conditions, considering factors such as process changes, localised damage and episodic events. Some owners adopt "use the higher rate unless justified" as their own procedure.

What are the units of corrosion rate?

Inches per year, mils per year (1 mpy = 0.001 in/yr) or millimetres per year (1 mpy = 0.0254 mm/y). Keep units consistent through to remaining life.

How is corrosion rate set for a new vessel?

From similar-service data, appropriately placed UT sensors, a corrosion specialist's estimate or published data. If none can be applied with confidence, the current API 510 requires an on-stream wall-loss check after six months of service.

Can statistics be used to calculate corrosion rate?

Yes. API 510 allows statistical analysis for vessel sections, with cautions about localised corrosion, and the current edition requires the method to be documented.

What does corrosion rate calculation software need to do?

Keep full reading history with technique and location, calculate LT, ST and trend rates side by side, flag suspect data, support baseline resets, keep units consistent and record the inspector's rate selection.

How is a corrosion rate calculator for coupons different?

Coupon calculations convert mass loss, exposed area, exposure time and density to a rate using a units constant (as in ASTM G1). The result describes the coupon's environment, not the equipment's wall.

How is pipeline corrosion rate determined?

Regulated pipelines usually derive growth rates by comparing successive in-line inspection runs and direct assessments under PHMSA integrity management, rather than API 510/570 CML formulas.

Why is my short-term corrosion rate negative?

The latest reading is thicker than the previous one, which almost always means measurement scatter or a location mismatch. Verify the reading and location before using any rate from that CML.

Need cleaner thickness data before your next rate review? Ask about corrosion mapping at critical CMLs or re-measurement of suspect readings.

Sources: API 510, 11th edition (2022), and API 510 9th edition (2006) for historical comparison; API 570 body of knowledge (February 2026); ASTM G1. Related: corrosion rate (glossary), remaining life.