Pulsed eddy current · Fundamentals
Portrait of Inspector Pepe Grietas, LLOG's non-destructive testing assistant
Ask Inspector Pepe Grietas This entire guide is part of his knowledge base. Ask him to explain a section, redo the calculations with your inspection data, or help you interpret a field reading.
Ask Pepe

Before you start · the five symbols used throughout the guide

LO
LiftoffDistance from the probe to the steel: jacket plus insulation.
WT
Wall thicknessWhat you want to measure.
τ
Time constantHow fast the signal dies out. Thickness is derived from it.
FP
FootprintWidth of what the probe sees sharply. It sets the resolution.
AA
Averaging areaEverything that influences the reading. Wider than the footprint.
01 — Context

The corrosion you cannot see

PEC exists so you do not have to remove the insulation to know the condition of the wall.

To measure an insulated pipe, you normally have to strip the insulation. That is expensive and slow. Sometimes it is also dangerous. PEC exists so you do not have to.

With insulation removal scaffoldingremovecleanUT readingreinstall high With PEC access scan low Line stays in service. Open only where the scan says.
Fig. 1 · The difference is not the measurement: it is everything around it that you no longer have to do. PEC does not replace ultrasonics; it reduces the area you have to apply them to.

The cost is not just the scaffolding and the new insulation. It is the permit, the line shutdown and the asbestos handling when there is any. There is an added risk too: if the insulation is reinstalled poorly, you create the very condition you came to inspect.

That is why risk-based inspection programs look for techniques that open as little as possible, cover the most area and work with the line running.

PEC fills that role. It does not replace ultrasonics: it comes before it. It covers area quickly and without surface preparation, and tells you where it is worth opening.

02 — Definition

What PEC is

It measures average and relative thickness, through anything that is not conductive.

The probe rests on the insulation and tells you how much steel wall is left underneath. Nothing needs to be cut or cleaned.

Formally: it is an electromagnetic technique that detects wall loss in ferromagnetic materials, measuring through coatings that do not conduct current.

Inspector measuring with PEC equipment on insulated pipe with an aluminum jacket
The probe rests on the jacket. Nothing is cut, nothing is cleaned, and the line keeps running.Image: Eddyfi Technologies
Where the name comes from

Eddy is not anyone's surname. It means a swirl: the whirlpool that forms in a river behind a rock. The currents are called that because they circulate in loops inside the metal, just like a whirlpool in water.

The comparison goes further than it seems. A whirlpool takes time to form and keeps spinning for a while after whatever created it has passed. Eddy currents do exactly the same, and that persistence time is precisely what PEC measures.

The distance between the probe and the steel is called liftoff (LO). PEC tolerates large liftoff because what it measures does not depend on that distance. Section 03 explains why.

  • Insulation with aluminum, stainless steel or galvanized jacketing
  • Concrete, fireproofing, composite wraps, paint
  • External corrosion product: scale, blistering, flaking
  • Marine growth on submerged structures and in the splash zone
PROBE LIFTOFF LO WALL WT metal jacket insulation carbon steel external corrosion internal corrosion τ does not depend on LO — only on what happens inside the wall
Fig. 2 · Liftoff is everything that separates the probe from the steel: jacket plus insulation. The technique sees through it because the time constant is generated inside the wall, not on the way to it. PEC also cannot tell whether the metal is missing on the inside or the outside.
What PEC delivers

An average wall thickness over the probe area, relative to a reference zone. It is not a point reading or an absolute value like that of an ultrasonic thickness gauge. It is a screening tool.

03 — Physics

The equation that governs everything

A single expression, τ = μσd²/π², explains everything the technique can and cannot do.

This is the most important section of the guide. One formula explains almost everything else: why the measurement is relative, why it only works on carbon steel, and where false positives come from.

When the pulse is switched off, the currents left inside the steel gradually die out. How fast they die out is described by a time constant:

τ = μ · σ · d² / π² μ magnetic permeability · σ electrical conductivity · d wall thickness
What τ is, without the formula

If you have ever charged a capacitor or measured a coil, you already know τ: it is the time the signal takes to fall to 37 % of its initial value. It is the same here. τ is how long the whirlpool takes to die out.

Thick wall, the whirlpool lasts longer. Thin wall, it dies out fast. The whole method is timing that.

That formula has three consequences. All three recur throughout the guide:

  • τ grows with the square of the thickness. If the wall drops by half, the signal dies out four times faster. That is why thickness is recovered with a square root: √(τ/τref).
  • τ contains no liftoff. Moving the probe away lowers the signal amplitude but does not change the slope of the decay. That is the physical reason PEC can see through insulation.
  • τ depends on μ and σ as much as on thickness. A change in permeability produces the same response as wall loss. With a single data point the two cannot be separated.
Classroom demonstration

Drop a magnet down a copper tube and, next to it, drop another down a PVC tube. The one in the PVC falls straight through. The one in the copper descends slowly, as if the tube were filled with a viscous fluid.

Nobody touched the magnet. What slows it down are the eddy currents it induces in the copper as it passes. It is the same phenomenon as in this guide, visible to the naked eye with hardware-store materials.

Decay curve

Interactive
PredictBefore touching anything, predict: if you move the probe 200 mm further from the steel, does the curve drop, tilt, or both? Write down your answer, then move the liftoff control.
τ nominal
τ corroded
Relative thickness
Amplitude

Move Liftoff all the way up: the whole curve drops, but the two slopes stay parallel and the relative thickness does not move a single point. Now move Permeability: without changing the thickness, τ changes. That is where a false positive comes from.

Technical detail: where τ comes from and what is an approximation
Source of τ: Chinese Physics B 24(3) 030301 (2015), diffusion time constant in a ferromagnetic plate from ξ₁ ≈ π/2. The PEC literature confirms the form: for a uniform planar field and constant thickness, the exponential decay constant is proportional to 1/(μσd²).
Comparison with published values: a manufacturer's patent gives, for carbon steel, τ ≈ 10 ms at 10 mm wall and τ ≈ 3 ms at 5 mm. With μr=150 and σ=6 MS/m this model gives 11.5 ms and 2.9 ms. Note that those two published values are NOT consistent with each other under pure d²: their ratio is 3.3, while the exact square would be 4.0.
Why that difference matters little and teaches a lot: μ and σ are not constants of steel; they vary with grade, temperature and applied field. During the pulse PEC magnetizes strongly and the effective permeability drops. That uncertainty in μ and σ is precisely why the method is RELATIVE and requires calibration against a reference zone: the absolute scale cannot be predicted, only compared.
Validity limit: proportionality to d² holds for thin walls. With real probes of finite size the decay tends toward an asymptotic value in thick walls, because the field no longer penetrates uniformly. The published values above already show that effect.
Declared approximations: mode series with 1/n weights truncated at 15 and normalized to 1 at t=0 — what is anchored to the source is τ of the dominant mode, which is the late portion the instrument actually measures; the start of the curve does not attempt to reproduce the early phase, which in reality follows a power law. Amplitude drop with liftoff uses a dipole form A₀/(1+LO/r₀)³, which does not affect the slope.
If you double the liftoff and change nothing else, what happens to the relative thickness the instrument reports?

Nothing. τ contains no liftoff, so the decay slope does not move and the relative thickness comes out the same. What does drop is the amplitude, and with it the signal-to-noise ratio: the reading is the same but noisier.

04 — Acquisition

The cycle, in three steps

Switching off the pulse induces the currents; the decay that follows is the measurement.

Every reading is three steps. They repeat dozens of times per second.

PhaseWhat happens
1 · The pulseThe transmitter coil produces a magnetic field held long enough to saturate the entire wall thickness. That time is the pulse width; if it is too short, the wall is not fully magnetized and the reading is wrong.
2 · Switch-offThe field is cut off abruptly. That abrupt change is what induces eddy currents in the component.
3 · ReceptionThe decay of those currents is measured as they diffuse through the thickness. Faster decay = thinner wall, exactly as τ ∝ d² predicts.
TRANSMITTER pulse width relaxation acquisition cycle RECEIVER active measurement time 1 the pulse magnetizes the full thickness 2 3 abrupt switch-off the decay is measured
Fig. 3 · The switch-off is what induces the currents; the pulse only prepares the ground. The slope of the red section after switch-off is the measurement: steeper means a thinner wall.

The time between switch-off and the end of the measurement is called the relaxation time. Thicker walls need longer windows. An example: at 10 mm, τ is about 12 ms; at 40 mm, it rises to about 200 ms. The thickness was quadrupled and τ was multiplied by sixteen.

Cross section showing the probe field crossing the insulation and concentrating in a corroded zone, with decay curve and thickness reading
The full cycle in one image: the field crosses the insulation, concentrates where metal is missing, and the curve dies out faster. Result: 43.9 % wall.Image: Eddyfi Technologies
05 — Resolution

The footprint

The footprint grows with liftoff, and with it the smallest defect that can be resolved.

The probe does not measure a point: it measures a circle. That circle is called the footprint, and its size determines the smallest defect that can be resolved.

Formally, the footprint is the full width at half maximum of the response, measured over a localized defect.

The footprint is not fixed: it grows with liftoff. And the minimum detectable defect is defined not by diameter but by volume. That is why every millimeter of insulation costs much more resolution than it seems.

FP = 0.65 · LO + FP₀ FP₀ probe footprint at zero liftoff · LO liftoff
scan −6 dB(50 %) FP grid resolution = FP LO WT FP = 0.65 · LO + FP₀ → more insulation, wider footprint, less resolution
Fig. 4 · The footprint is the full width at half maximum of the response. Spacing the grid at exactly one FP guarantees 50 % overlap between neighboring points: that is what keeps a defect from falling between two readings.

Footprint and averaging area calculator

Interactive
PredictPredict: to double the footprint of a 22 mm probe, how much insulation does it take? Estimate first, then check with the control.
Footprint FP
Averaging area
Suggested grid
Minimum defect

With a 22 mm probe over 65 mm of insulation the footprint is already 64 mm: almost three times the nominal size. The probe does not change; the geometry does.

A probe with FP₀ = 30 mm sits on 100 mm of insulation. What is the footprint?

95 mm. FP = 0.65 × 100 + 30. The probe is the same; what grew is the geometry. And since the minimum defect scales with the footprint volume, resolution falls much faster than that number.

06 — Averaging

Why small defects are undersized

The instrument averages: a defect smaller than the averaging area is always reported shallower than it really is.

The instrument does not read the point directly under the center of the probe. It reads an average over a whole zone. That zone is called the averaging area, and it measures 1.8 times the footprint.

That averaging produces two behaviors that look like instrument faults in the field, and are not:

  • Undersizing. A defect smaller than the AA is averaged with the surrounding sound metal. The instrument reports less loss than there really is. It is not miscalibrated: it is doing exactly what the physics allows.
  • Mass effect. Any metal structure within an AA (a flange, a support, a nozzle, another pipe) adds its electromagnetic contribution to the reading. That is why a zone next to a flange reads "thicker" without being so.
Correct sizing the defect fills the averaging area average WT Undersizing nearby sound metal dilutes it
Fig. 5 · The red line is what the instrument reports. On the right the defect is just as deep, but because it occupies little of the averaging area the reading barely departs from nominal. It is not a calibration error: it is averaging.
Field rule

If the defect you are looking for is smaller than the averaging area, the reading is a minimum value, not a measurement. It must be confirmed with ultrasonics before making an integrity decision.

A real defect at 50 % wall is reported as 85 %. Is the instrument out of calibration?

Not necessarily. Most likely the defect is smaller than the averaging area and was averaged with the surrounding sound metal. It is undersizing, not a calibration error. Confirm with ultrasonics before touching the calibration.

07 — Calibration

Everything depends on the reference zone

The entire inspection is measured against the reference. Choose it badly and the whole map shifts.

PEC does not give millimeters: it gives percentages. And it gives them against a zone the inspector declared sound. If that zone is badly chosen, the whole scan is shifted.

The formula in section 03 explains why. The instrument calculates thickness as √(τ/τref), and that τref comes from the reference zone. If that value is wrong, the whole map is shifted.

valid reference support mass within the AA corroded zone already lost wall flange mass effect Every reading in the scan is expressed against this zone. An error here scales the entire map.
Fig. 6 · The reference must be more than one averaging area away from any metal mass, on sound material of the same nominal thickness. It is the only point in the method where an inspector error contaminates every other result.
RequirementWhy
Same material and heatμ and σ enter τ just like thickness does. A different material is a different scale.
Known nominal thicknessIt is the 100 % against which everything else is reported.
Free of corrosionIf the reference has already lost wall, real corrosion is underestimated across the whole scan.
Away from metal massesMore than one AA from flanges, supports and nozzles, or the mass effect contaminates the zero.
About the same liftoffAlthough τ does not depend on liftoff, the signal-to-noise ratio does; a reference with much less insulation does not represent the rest.
No strong thermal gradientTemperature variations on the order of 20 °C across the component generate false indications.
08 — Scope

The capability envelope

The maximum thickness and maximum liftoff on a datasheet do not happen at the same time.

A datasheet says 100 mm of wall and 300 mm of insulation. They look like two separate numbers. They are not.

The instrument does not do both at once. On top of that is the scope of the standard, which is considerably narrower than what the instrument can measure.

The figure that almost never appears on the datasheet

The 305 mm of liftoff applies to single-element probes. Array probes, the high-productivity ones, allow 50 mm. You gain speed and lose reach, and that is the real decision when planning the job: scan little insulation fast, or a lot of insulation slowly.

Thickness versus liftoff

Interactive
PredictPredict: an 80 mm wall under 100 mm of insulation — is it inside or outside the scope of the standard? Decide before moving the point.
Within ISO 20669
Expected τ

The inner box is the scope of the standard. Outside it the measurement may still be possible, but it is no longer covered, and that is what gets signed on a report.

09 — Worked example

One case, start to finish

Everything above, applied to a real line, in the order decisions are made in the field.

The job: 8 in steam line (219 mm outside diameter), 10 mm nominal wall in carbon steel, 50 mm of insulation with a 0.5 mm aluminum jacket. The client wants to know if there is corrosion under insulation. Available probe: FP₀ = 40 mm.

Total liftoff

Insulation plus jacket: 50 + 0.5 = 50.5 mm. The aluminum jacket is not ferromagnetic, so it only adds distance.

Footprint

FP = 0.65 × 50.5 + 40 = 72.8 mm. In practice, the 40 mm probe sees a 73 mm circle.

Averaging area

AA = 1.8 × 72.8 = 131 mm. Any metal within 131 mm enters the reading: supports, flanges, nozzles.

Scan grid

A pitch of 73 mm (one FP) gives 50 % overlap between neighboring points. Tighter adds no information; wider leaves gaps.

Minimum detectable defect

Footprint volume: π × (72.8/2)² × 10 = 41,700 mm³. 15 % of that is 6,250 mm³. For a loss 5 mm deep, that equals a diameter of ≈ 40 mm.

Acquisition window

With μr ≈ 200 and σ ≈ 5 MS/m, τ ≈ 12.7 ms. The window must cover at least 3τ ≈ 38 ms to see the straight portion of the decay.

Standard scope

10 mm wall (between 3 and 65), 50.5 mm coating (between 0 and 200), 219 mm diameter (greater than 50). Within ISO 20669.

Top view · everything to the same scale (2.2 px per mm) grid pitch 73 mm FP 72.8 min Ø 40 mm AA 131 mm Neighboring footprints touch at half height: the 50 % overlap set by the grid.
Fig. 7 · The section's calculation, drawn. The dark disc is the smallest defect this setup can detect: it fits comfortably inside the footprint, which is why a 20 mm pit does not show up.
Verdict

Applicable, with one condition. For general corrosion under insulation the case is ideal. But if they are looking for an isolated pit 10 or 20 mm in diameter, this setup will not detect it: it is well below the 40 mm minimum diameter. That limitation must be declared during planning, not in the final report.

Your turn: the same line, but with 150 mm of insulation. What changes?

Liftoff 150 mm → footprint FP = 137.5 mm, averaging area 248 mm, grid of 138 mm. The footprint volume rises to 148,500 mm³, so 15 % is 22,300 mm³ and the minimum 5 mm deep defect now needs a 75 mm diameter: almost double.

τ does not change: it stays at 12.7 ms, because the thickness is the same. And it is still within ISO 20669, because 150 mm is below the 200 mm coating limit. The instrument measures just as well; what was lost is resolution.

10 — Standards

What the standard covers

Instrument capability and standard scope are not the same thing, and on a report the standard rules.

Three documents recognize PEC. For each one, what matters is where its scope ends.

DocumentScope
ISO 20669:2017Thickness measurement by PEC on ferromagnetic components. Carbon and low-alloy steel, 3 to 65 mm wall, 0 to 200 mm coating, pipe no smaller than 50 mm in diameter, −100 to 500 °C surface temperature. Excludes cracks and localized pitting loss.
ASME BPVC Sec. V, Art. 21Incorporates PEC examination into the boiler and pressure vessel code.
API RP 583Corrosion under insulation and under fireproofing. Lists PEC as an applicable inspection method.
The distinction to keep clear

Instrument capability and standard scope are not the same thing. An instrument can measure above 65 mm of wall; ISO 20669 does not cover that measurement. If the client asks for a standard-compliant report, the envelope that rules is the standard's.

80 mm wall under 100 mm of insulation. The instrument can measure it. Is it covered by ISO 20669?

No. The standard goes up to 65 mm of wall. The measurement may be correct, but it falls outside the document's scope, and that has to be declared in the report.

11 — Operation

Field constraints

Temperature shifts μ and σ, and both of them are inside τ.

  • Temperature. It is not one limit, it is three, and they are often confused. The steel component allows −150 to 500 °C. The metal jacket must not exceed 70 °C. And the instrument only operates between 0 and 40 °C, which in a plant under the sun is the constraint that shows up first. For hot walls a ceramic standoff is placed in between.
  • Minimum readable wall. The instrument does not report below 15 % of nominal thickness. A larger loss saturates the scale: you know it is very thin, but not how thin.
  • Thermal gradient. Differences on the order of 20 °C along the component produce false indications, because temperature shifts μ and σ, and both of them are inside τ.
  • Minimum diameter. Below a certain diameter the curvature is no longer negligible compared with the footprint and the averaging becomes unrepresentative. On small-diameter pipe a counterintuitive effect appears: up to about 60 mm of insulation, detection of a defect on the far side can improve with more liftoff, because the footprint starts to wrap around the pipe. It does not contradict the above: the footprint keeps growing, it just covers more of the circumference.
  • Metal jacket. Each material has its maximum thickness: stainless steel up to 1.5 mm, aluminum up to 1 mm and galvanized up to 0.5 mm. Galvanized is the hard case because it is ferromagnetic and takes part in the signal; with a dedicated probe you can reach 1 mm.
  • Geometry. Elbows, reducers, welds and reinforcements bring mass into the AA. They can be inspected, but they are interpreted with different criteria than a straight run.
12 — Limitations

What PEC cannot do

None of these limitations is a manufacturer's fault: they all come from the equation in section 03.

What the instrument sees · dashed red trace: nominal calibration · black trace: measurement V t WT 60 % External loss V t WT 60 % Internal loss V t WT 99 % Isolated pit V t WT 99 % Perforation
Fig. 8 · Curves calculated with the model from section 03, not drawn by hand. The first two lose the same amount and their traces fall identically: the instrument does not know which side the metal is missing from. The last two stay glued to the nominal curve — 99 % and 98.8 % — because their volume is negligible compared with the footprint.

None of these failures is a manufacturer's fault. They all come from the formula in section 03.

LimitationOrigin
Relative measurement√(τ/τref) is recovered. Without a reference there is no number.
Ferromagnetic onlyμ is in τ. Without high μ there is no usable contrast: carbon steel and cast iron.
Cannot tell inside from outsideτ depends on total thickness, not on where the metal is missing.
Does not detect pitting or perforationA defect with small volume compared with the footprint does not move the average.
Sensitive to property variationμ and σ enter τ with the same weight as thickness.
Undersizes small defectsAveraging over the AA.
Mass effect near metalEverything conductive within the AA contributes.

Several manufacturers sell algorithms that correct undersizing and help separate false positives. They work, and that is why an instrument costs what it costs. But they reduce the problem; they do not remove it.

13 — Selection

When not to use PEC

Knowing when the technique does not apply is worth as much as knowing how to operate it.

Knowing when it does not apply is worth as much as knowing how to operate it. These cases call for another method.

SituationAppropriate method
Isolated pittingStraight-beam or phased array UT; digital radiography on small-diameter pipe
Absolute thickness neededUltrasonic thickness gauge on a clean spot
Stainless steel or non-ferromagnetic materialUT, conventional eddy current, radiography
Wall below 3 mmHigh-frequency UT, digital radiography
Crack detectionAngle-beam UT, magnetic particle, liquid penetrant
Very small diameter or complex geometryProfile radiography, UT with a dedicated probe
Long screening of buried or insulated pipeGuided waves to locate, then PEC or UT to size
What are you looking for? Generalwall loss Pittingor cracks Absolute thicknessat a point Carbon steelunder coating? yesno PEC UT UT · RTMagnetic particle UT on a cleansurface
Fig. 9 · PEC occupies a single branch of the tree, and it is a wide branch: general wall loss in carbon steel under coating. Outside of that, there is a better method.
Quick rule for choosing

Large area → PEC. When you need to scan a lot of surface for wall loss and open as little as possible.

Small area → conventional eddy current. When sensitivity to something localized and superficial matters, and you can work in contact with the metal.

How it fits in the bigger picture

PEC is a screening method: it covers surface fast and without preparation to tell you where to look closely. The integrity decision is made with the method that sizes. Used in that order they complement each other; used alone, PEC promises a precision its physics cannot deliver.

14 — The instrument

What the instrument does inside

The instrument does not measure thickness: it measures τ and then translates it. Everything you touch on the screen is there to make that translation reliable.

Up to here the guide has talked about the technique. This section is about the box: what it calculates, what it asks of the inspector, and why.

The τ curve is real, not a metaphor

Modern instruments include a view called Tau-scan. Its formal definition is that it is inversely proportional to the first time derivative of the log-linear representation of the A-scan. Translated: if the signal decays as e−t/τ, then the derivative of the logarithm is −1/τ, and its inverse is exactly τ.

Why it matters

The screen is plotting, point by point, the instantaneous time constant. It is not a teaching analogy: the τ from section 03 is literally the number the instrument calculates and draws.

And that is where its practical value comes from. A true wall loss shifts the curve cleanly; a permeability variation distorts it in a different way. With the A-scan alone you only see a lower signal; with the τ view you can tell the cause apart.

Acquisition · scan zone 01 A-scan · log V versus t — nominal— measured τ view · time constant red line =calibration C-scan · remaining thickness 100 %< 38 % cursor · min 58 %
Fig. 10 · The three views the instrument shows at once. The A-scan is the raw signal; the τ view is that same signal transformed so the time constant can be read directly; the C-scan is the map that gets delivered. The map here is simulated with the model from section 03 — it is for learning to read the scale, not field data.

The three setup routines

What appears on the instrument as a single calibration button is, internally, three different routines. They run with the probe over the nominal zone, and it is worth knowing what each one does because they fail for different reasons.

Automatic pulse adjustment

Sets gain, pulse duration, time gates and filters. It starts from two values derived from the declared component: the jacket delay and the characteristic decay constant. If the component was declared wrong, this routine starts from the wrong place.

Thickness calibration

Sets the 100 %. It is done on nominal wall or on a known thickness. It is the step from section 07, and the only one that turns percentages into millimeters.

Repeatability optimization

This is the interesting part: the instrument offers a list of averaging settings and the inspector chooses one. More averaging, more repeatability, more time per point. Repeatability is not a property of the instrument: it is something you buy with time, and the inspector decides how much to buy.

Jacket delay

The metal jacket does not just attenuate: it delays the signal. The instrument shows that delay as a number, and comparing expected delay against measured delay is a free diagnostic. A difference between the two means the component was declared wrong or there is a jacket overlap under the probe. It is worth checking before blaming the wall.

Decisions to make before scanning

  • Grid resolution. The instrument offers "minimum for full coverage" and "ready for compensation". The second is tighter. If you scan with the first and later want to compensate for undersizing, you cannot: you have to scan again.
  • Acquisition mode. Grid mode, point by point, each reading in under a second. Or dynamic with an encoder, up to 15 readings per second but with a travel speed limit of about 75 mm/s. Going faster does not speed up the inspection: it invalidates it, and the instrument warns you.
  • Calibration area with arrays. With array probes, the reference zone needs at least 5 points on the index axis and 3 on the scan axis. A single point is not enough.
  • Array probe wheels. Some lift the probe about 12.7 mm off the surface. That liftoff is already accounted for and must not be added to the insulation when calculating the footprint.
What the instrument delivers in the end

The color map can show four different things, and mixing them up changes the report: remaining thickness in percent, remaining thickness in millimeters, wall loss in percent and compensated thickness. The first three are the raw measurement expressed three ways. The fourth has already gone through the algorithm that corrects undersizing, and only exists if the grid was planned for it.

15 — Misconceptions

The nine most common misconceptions

These nine misunderstandings come up again and again in the field, with any instrument and any brand.

None of these misunderstandings is trivial. They all come from applying the intuition of ultrasonics to PEC, since ultrasonics is the method almost all of us learned to measure thickness with.

PEC gives me the thickness in millimeters.

No. It gives a percentage against the reference zone. To convert it to millimeters you need to calibrate at a point on the object itself, usually with ultrasonics. Without that point, the values are only comparative.

If the instrument repeats the same reading, the measurement is accurate.

Repeatability and accuracy are different parameters, and here they differ by a factor of ten. Typical accuracy is around ±0.5 mm, but measuring the same point again repeats within ±0.05 mm. That is why PEC is very well suited to monitoring (measuring the same point at intervals and following the trend) and barely adequate for a one-off measurement.

It missed the pit, so the instrument must be faulty.

It is working. Field rule, with 30 to 120 mm of insulation: the minimum detectable defect has a diameter of roughly 50 % of the insulation thickness. With 100 mm of insulation, no defect smaller than about 50 mm in diameter. Isolated pits are not detected, and that is not an instrument fault.

A thin zone showed up where there is no corrosion.

It is probably material property variation, not wall loss. Permeability differences within the same part produce false variations on the order of 10 %. They show up mostly on vessels. Before reporting, check whether the zone coincides with a plate change, a weld or a heat treatment.

I can measure anywhere on the pipe.

You need about 50 mm (2 in) of clearance around the probe. You cannot measure under steam tracing, next to supports or on tight elbows. And those are precisely the places where corrosion under insulation shows up most. It is worth declaring that before quoting the coverage.

The scale has to be cleaned off before measuring.

No. You can measure with the corrosion product in place, even with 20 mm of scale. It is one of the big advantages of the method: nothing is touched, and on in-service equipment removing it is often not allowed.

The instrument only handles 120 °C.

There are three different limits, and that is neither of the two that usually get in the way. The component allows −150 to 500 °C; the metal jacket, 70 °C; the instrument, 0 to 40 °C. With a ceramic standoff, walls up to 550 °C have been inspected. And at any temperature, gradients matter: a 20 °C difference across the part already generates false indications.

I need a Level II at the plant all day.

Data collection can be done by a Level I. The analysis is done afterwards by a Level II or III from the file, and they do not need to be on site or present during collection. Separating the two is what makes the job profitable, especially at remote plants.

Galvanized jacketing makes measurement impossible.

It is possible, but with lower performance than aluminum or stainless steel, because galvanized steel is ferromagnetic and takes part in the signal. There are dedicated probes for that case. Aluminum and stainless steel are no problem.

ULTRASONICS 8.24 mm contact · one point · absolute value PEC averaged area 74 % no contact · an area · relative value
Fig. 11 · The same wall, both methods. Ultrasonics returns a number in millimeters from a clean point; PEC returns a percentage averaged over an area, without touching anything. Almost every misconception in this section comes from asking the second for what the first provides.
The common thread

Ultrasonics measures a point and gives absolute millimeters. PEC measures an area and gives a relative percentage. Almost all the misconceptions above are the same mistake: expecting from the second what the first provides.