In-Situ Metallography: Identifying In-Service Degradation and Early Warnings

Paresh Haribhakti
4 days ago
11 min read

Updated: 3 days ago

Steel announces its failures years in advance. The announcement is written in the microstructure, and you do not have to cut the component to read it.

AT A GLANCE
The problem	Creep, spheroidisation, sensitisation, graphitisation and hydrogen attack all mature quietly inside the microstructure long before there is any wall loss or crack for conventional NDT to find. By the time a flaw is detectable, most of the component's life is already spent.
The technique	In-situ metallography, metallographic replication per ASTM E1351: prepare a window of roughly 25 mm × 25 mm on the installed component, lift the polished and etched microstructure onto a cellulose acetate replica, and read it at 100–1000X optically and under SEM. Surface removal stays below 0.5 mm. No cutting, no weld repair, no PWHT.
The early warning	Isolated creep cavities appear while a header still has years of safe life. Classified A to D on the Neubauer and Wedel scale, each class carries a defined re-inspection interval and action, from routine observation to immediate repair.
The decision	Replica evidence anchors remaining life assessment, feeds Level 2 and Level 3 creep assessment under API 579-1/ASME FFS-1 Part 10, and sets inspection density per location in AiOM deployments.
The bench	TCR Advanced has practised and promoted this technique in India since 1999: 26+ years of process-industry integrity work, 10,000+ failure investigations, and a microstructure database of 1,50,000+ micrographs behind every interpretation.

Why it matters

In-Situ Metallography : Two headers of the same grade, built to the same code, run the same number of hours. One is fit for another decade; the other is one start-up away from a catastrophic release. Nothing in their thickness surveys separates them. Hours in service do not measure damage. Temperature history, stress and microstructural response do, and the only place that response is recorded is in the metal itself.

Design codes size creep-range components for a nominal life of the order of 100,000 hours at design metal temperature. Real life diverges sharply from nominal, because creep is brutally sensitive to temperature: the working rule is that a sustained 10–15°C rise above design metal temperature can halve the creep life of a low-alloy steel. A decade of small excursions, invisible in the operating log's averages, quietly re-writes the retirement date. Conventional NDT cannot see any of this. Ultrasonics, radiography and magnetic particle testing find geometry: thinning, voids large enough to scatter sound, cracks that have already formed. The damage that precedes all of them, cavities of a micron or two strung along grain boundaries, carbides changing shape and chemistry, grain boundaries losing their chromium, sits far below every conventional detection threshold.

That is the gap in-situ metallography closes. Across 10,000+ failure investigations in the TCR Advanced archive, the pattern repeats with uncomfortable regularity: the microstructure carried the warning years before the failure, and nobody had read it. The technique exists so that the reading happens before the event, not after it, in the failure analysis report.

The technique: laboratory metallography, brought to the metal

In-situ metallography, also called field metallography or metallographic replication, reproduces the full laboratory sequence of grinding, polishing and etching on the surface of the installed component, then lifts a faithful copy of the revealed microstructure onto a replica film. The governing document is ASTM E1351, Standard Practice for Production and Evaluation of Field Metallographic Replicas, with surface preparation and etching practice drawn from ASTM E3 and ASTM E407.

The in-situ metallography workflow per ASTM E1351. The full laboratory sequence is executed on the installed component; only the replica travels back to the laboratory.

he discipline in each step decides whether the replica is evidence or artefact. Grinding proceeds through successively finer grits until at least 0.3–0.5 mm of surface has been removed, because the outermost layer, decarburised, oxidised and cold-worked by service and by the grinding itself, belongs to nobody's microstructure. Polishing runs down to 1 µm diamond, or to a mirror by portable electrolytic polishing where the geometry allows. Etching is grade-specific per ASTM E407: nital for ferritic steels, Vilella's reagent for the 9 to 12 per cent chromium martensitics, electrolytic oxalic acid for the austenitics. Then a cellulose acetate film, softened in acetone, is pressed onto the etched window. As the solvent flashes off, the film hardens into a negative relief of the surface, faithful to well below a micron. A properly taken replica examined under SEM resolves individual creep cavities of 0.5–1 µm, which is precisely the scale at which creep damage begins.

The component does not care that any of this happened. The prepared window needs no weld build-up, no post-weld heat treatment, no repair of any kind; it is simply a locally smoother patch of the same pressure boundary. The replica is verified on a portable microscope at the job face at 100–400X, so a suspect finding can be re-prepared and re-replicated in the same visit, and then examined in the laboratory at up to 1000X optically and under SEM with EDS where cavitation counting or precipitate chemistry is needed. Each replica is logged with location, orientation, etchant and magnification, because a micrograph that cannot be traced to a position on the equipment is an anecdote, not evidence.

What the microstructure reveals: a damage atlas from the field

The chart below is drawn from TCR Advanced's field replication archive. Every frame is a real microstructure read on installed plant, organised by material class, and together they make the central argument of this article: almost every damage mechanism that matters in a process plant has a distinct, readable microstructural signature long before it has a detectable crack.

Damage mechanisms identified in the microstructure by in-situ metallography. TCR Advanced field replication archive, magnifications 100X to 3500X.

Carbon and low-alloy steels tell the story of time at temperature. Against the normalised baseline of ferrite and lamellar pearlite, prolonged service above roughly 450°C drives the pearlite to spheroidise: the lamellar carbides break up and coarsen into spheroids, and the steel trades strength for the change. A Widmanstätten structure betrays improper heat treatment or a severe overheating event. Decarburisation and graphitisation each record their own thermal and environmental histories. Every one of these states is invisible to ultrasonics and visible on a replica.

Austenitic stainless steels record chemistry as much as temperature. Exposure between 425°C and 815°C precipitates chromium carbides on the grain boundaries, the sensitised condition that opens the door to intergranular corrosion and intergranular stress corrosion cracking. Sigma phase forming in long service embrittles the steel at ambient temperature. When cracking does start, the replica separates the transgranular, branched path of chloride stress corrosion cracking from the intergranular path that follows sensitised boundaries, and that distinction changes the entire corrective action.

Reformer tubes and hot manifolds in HK40 and HP-modified grades develop creep damage as aligned interdendritic voids and grain boundary cracks, and the cast structure makes interpretation a specialist's task: an inter-granular crack network in HK40 reads very differently from the interdendritic cracking of a solidification problem. Replication at the hottest tube elevations, correlated with diametral growth measurement, is how a reformer's re-tube decision gets an evidence base.

Weldments and rotating equipment contribute the mechanisms that concentrate where geometry and metallurgy change: Type IV cracking in the fine-grained heat-affected zone of P22 and P91 weldments, quench cracks, caustic stress corrosion in carbon steel, thermal and corrosion fatigue, high temperature hydrogen attack in C-½Mo, and high temperature sulphidation read at magnifications up to 3500X. Each has a signature. Each signature carries a different remaining-life calculation and a different fix.

Early warnings: the creep damage ladder

Creep is where in-situ metallography earns its keep, because creep damage is progressive, classifiable and slow. It begins as isolated cavities on grain boundaries, of the order of a micron; the cavities multiply and align on boundaries normal to the maximum principal stress; aligned cavities link into microcracks; microcracks join into macrocracks; the macrocrack grows to failure. The widely used classification of these stages comes from Neubauer and Wedel, and it maps each microstructural state to an action:

Class	Microstructural state	Typical action
A	Isolated creep cavities on grain boundaries	Continue operation; re-inspect at the normal interval, typically 3–5 years
B	Oriented, aligned cavitation	Continue with reduced interval, typically 1.5–3 years
C	Linked cavities forming microcracks	Limited operation; plan repair or replacement within about a year
D	Macrocracks	Immediate repair or replacement

The intervals are stated here as the ranges established plant practice uses; in an assessment they are set against the specific stress, temperature and consequence context of the location, never applied blind. The decision rule is the point: a replica finding is not a photograph for the report, it is a classified damage state with a defined consequence for the operating and inspection plan.

Creep voids at the grain boundaries, 400X. At this stage the component is years from failure, and no conventional NDT method can see the damage. This is the early warning.

Read the table against the capability of conventional NDT and the value of replication becomes arithmetic. Classes A and B, the stages with years of remaining life and the full menu of corrective options, are entirely invisible to ultrasonics, radiography and surface methods. The first conventional detection typically happens somewhere in Class C or D, when options have narrowed to repair scope and outage planning. In-situ metallography moves the discovery point two classes earlier, and those two classes are where run-repair-replace decisions are cheap.

From replica to decision: hardness, remaining life and fitness-for-service

A replica never travels alone. At every window TCR Advanced pairs the microstructure with in-situ hardness measurement and positive material identification, because the three together over-determine the diagnosis: a spheroidised structure should read soft, a suspected hardened HAZ should read hard, and an unexpected alloy explains many mysteries. On boiler tubes, steam-side oxide scale thickness measured ultrasonically gives an independent estimate of effective metal temperature, which the microstructural state must corroborate. When the evidence lines agree, the assessment stands on rock.

The findings then flow into the engineering. A creep damage class and a corroborated metal temperature anchor the remaining life assessment, whether by Larson-Miller parameter methods or by the more advanced approaches of API 579-1/ASME FFS-1 Part 10 and its Omega method, where microstructural evidence is what justifies the material state assumed in the calculation. Where a crack-like flaw has been found, the replica tells the assessor which mechanism produced it, and the mechanism decides the assessment route and the re-inspection logic. In AiOM, TCR Advanced's asset integrity platform, each replication window is a recorded location with a damage class and a date, so inspection density concentrates where the microstructure says the life is being spent, and the next turnaround's scope is written by evidence instead of habit.

Quality decides everything: why the replica must be beyond argument

A replica is forensic evidence, and it is only as good as the discipline behind it. Over-etching manufactures grain boundary relief that reads as cavitation; under-polishing leaves comet tails and smeared metal that hide it. Both errors are expensive in opposite directions: one condemns a healthy header, the other clears a dying one. The craft is knowing the difference at 400X in the field, at height, on a curved surface, and that knowledge does not come from a procedure document alone. It comes from having read enough microstructures that the artefacts announce themselves.

TCR Advanced has been the pioneer in generating awareness of in-situ metallography as an integrity tool in Indian industry, practising, teaching and publishing the technique since 1999, and the replication quality the TCR group produces stands second to none in the world. We hold it there deliberately, because our replicas are cross-examined: in regulatory submissions, in insurance claims, in disputes where a re-tube decision worth crores rests on whether a cavity is real. The practice is governed by an in-house in-situ metallography manual so the method survives any individual operator, every interpretation is made against a proprietary microstructure database of 1,50,000+ micrographs built over 26+ years, and the laboratory behind the field team works under the accreditations in the footer of this page. When a client's engineer asks how we know the difference between damage and artefact, the answer is that we have seen both, at scale, for decades.

Where replication belongs in your inspection programme

Choose locations by physics, not by access. The hottest tube elevations, weldments and their heat-affected zones, dissimilar metal joints, cold-formed bends, attachment welds and known stress concentrators are where damage concentrates. A replica taken at a convenient location answers a question nobody asked.

Make it a time series. The single most valuable replica is the baseline taken before degradation starts, because creep classification is strongest as a trend. A component replicated at every major turnaround carries its own damage history, and the interval between classes measured on the same location is a direct, plant-specific damage rate.

Scope it into the turnaround, not around it. Replication needs surface access and metal below roughly 50°C, and a multi-point campaign on a header or reformer runs in parallel with other turnaround work. The marginal cost of adding windows while insulation is already off is small; the cost of wishing you had, three years later, is not.

Pair it with the decision, not just the report. Ask for the damage class, the re-inspection interval and the FFS implication at every window, not a photo album. Microstructure without a decision rule is expensive photography.

"In 10,000 investigations I have rarely seen a component fail without the microstructure announcing it first. In-situ metallography is how we take the metal's statement before the failure, instead of after it, in the failure analysis report."

Paresh Haribhakti, Managing Director, TCR Advanced Engineering

Frequently asked questions

What is in-situ metallography, and how does metallographic replication work?

In-situ metallography, also called field metallography, reproduces the laboratory metallographic sequence of grinding, polishing and etching directly on an installed component, then lifts the revealed microstructure onto a cellulose acetate replica per ASTM E1351. The replica, a faithful negative of the surface, is examined at 100 to 1000X optically and under SEM, so grain structure, carbides, creep cavities and cracking can be assessed without cutting a sample from the equipment.

Is in-situ metallography truly non-destructive?

Yes. Preparation removes less than about 0.5 mm from a window of roughly 25 mm by 25 mm, well within normal corrosion allowances, and leaves a locally polished patch of the same pressure boundary. No weld build-up, post-weld heat treatment or repair is needed, and the component returns to service immediately.

Which damage mechanisms can in-situ metallography detect?

Creep cavitation, spheroidisation of pearlite, graphitisation, decarburisation, sensitisation and sigma phase in austenitic stainless steels, intergranular and transgranular stress corrosion cracking, thermal and corrosion fatigue cracking, Type IV cracking at weldments, high temperature hydrogen attack and overheating signatures such as Widmanstätten structure or untempered martensite. Most of these exist for years before any crack that conventional NDT could find.

What resolution can a field replica achieve?

A well-executed acetate replica reproduces surface detail finer than a micron. Under SEM examination it resolves individual creep cavities of 0.5 to 1 micron, the scale at which creep damage first appears, and optical examination to 1000X covers grain structure, carbide morphology and crack path characterisation.

How are replica findings converted into remaining life and fitness-for-service decisions?

Creep damage is classified on the Neubauer and Wedel A to D scale, and each class carries a re-inspection interval, from a normal 3 to 5 year cycle at Class A down to immediate repair at Class D, set against the location's stress, temperature and consequence context. The classified state, corroborated by in-situ hardness and oxide-scale temperature estimation, anchors remaining life assessment and Level 2 and Level 3 creep assessment under API 579-1/ASME FFS-1 Part 10.

When should a plant include in-situ metallography in its inspection scope?

At every major turnaround for equipment operating in the creep range or in sensitising or embrittling regimes: fired heater and reformer tubes, headers, main steam piping, hot manifolds and critical weldments. The strongest programmes take a baseline replica early in life and re-replicate the same locations each turnaround, turning classification into a measured, plant-specific damage trend.

About the author

Paresh Haribhakti is Managing Director of TCR Advanced Engineering Pvt. Ltd., Vadodara. A metallurgical engineer from M.S. University of Baroda, he is co-author of an ASM International monograph on boiler tube failure analysis (2018), a contributor to ASM Handbook Volume 11A (2021), and the architect of the AiOM asset integrity platform. He has led 9,000+ industrial failure and integrity investigations.

Related TCR Advanced Resources