Most refractory linings do not fail because the wrong material was chosen. They fail because nobody measured how fast the right material was disappearing. A lining that was 230 mm thick at commissioning and is 118 mm thick three campaigns later has been telling you its remaining life for years — in numbers you could have read from the outside of the shell with a thermal camera and confirmed with a depth probe the moment the vessel came down.
This guide is about building that measurement chain deliberately: what to survey while the unit is running and what only a shutdown can reveal, how each measurement method lies to you and how to stop it, how to grade the defects you find so that a crack in a burner quarl and a crack in a floor are not treated as the same event, and how to turn two surveys into a reline date with an honest confidence band around it. It closes with the acceptance numbers a new lining has to hit under API 936, because the fastest way to shorten the inspection cycle is to start with a lining that was never verified in the first place.
1. Why a Lining Outlives Its Calendar, Not Its Design Life
The default maintenance model — reline every five years, inspect in between — is wrong in both directions at once. It wastes serviceable lining on units that are lightly loaded, and it runs units with localised erosion past their safe thickness because the average condition looks fine. The two failure modes have very different costs. Relining early burns material, contractor mobilisation and production days. Relining late risks a shell breakout.
What makes the calendar model so persistent is that refractory wear is genuinely uneven. In one fired heater, the radiant section hot face may lose material at a completely different rate from a burner throat two metres away, because the drivers are different: radiant-section brick loses material to thermal cycling and alkali attack, while a burner throat loses it to high-velocity gas and entrained particulates. Published service data for petrochemical cracking units illustrates the spread well. Riser and transfer lines typically run 75–150 mm thick with abrasion losses of 5–12 cc (ASTM C704) and bulk densities of 2000–2600 kg/m³; reactor and regenerator walls and headers run the same thickness at 1150–1250 kg/m³ with cold crushing strengths of 11–20 MPa; cyclone inlet zones are only 25–50 mm thick but must hold abrasion loss to 2–5 cc. A single "average wear rate" applied across those three zones would be meaningless.
The practical consequence: a lining programme should be organised by zone, and every zone should have its own thickness history, its own minimum acceptable thickness and its own reheat-and-reline trigger.
2. The Four Questions an Inspection Has to Answer
An inspection that produces a photo album and a paragraph of prose has not answered anything. A useful survey answers four separate questions, and no single method answers all four:
- How much material is left, and where is the thinnest point? This is a geometry question and it needs a number in millimetres, referenced to a fixed survey grid.
- Is what remains still sound, or has it delaminated from the shell or from the backup layer? This is an adhesion question and it is largely answered by sound, not by sight.
- What is the chemistry doing? Has slag or alkali penetrated the hot face, has the bond phase changed, has the material densified or gone friable? This needs a sample and a laboratory.
- How fast is it changing? One survey is a condition; two surveys on a common operating-hours basis are a rate; three or more are a trend you can plan against.
The most common programme failure is to answer question 1 thoroughly and questions 2–4 not at all. A laser profile with 5 mm resolution on a lining that has already separated from its anchors produces a very precise number for a lining that is about to fall off.
3. Tiering the Survey: What You Measure Running vs. Shut Down
The methods divide cleanly by accessibility, because a lining can only be touched when the unit is down and cold, while the shell is accessible whenever the unit is running.
| Method | Unit state | What it actually measures | Coverage | Main limitation |
|---|---|---|---|---|
| Visual inspection | Shut down, cooled | Cracks, spalling, erosion patterns, joint condition, colour change, anchor exposure | Full accessible surface | Surface only; a delamination with an intact skin is invisible |
| Hammer / tap test | Shut down | Bond and compaction — voids, delaminations, dry fill, loose brick | Full surface on a fixed grid | Requires an experienced ear; not quantifiable in millimetres |
| Mechanical depth probe | Shut down | Direct remaining thickness at the probe point | Discrete points only | Operator-dependent; misses everything between the points |
| Ultrasonic thickness gauge | Running or shut down, from the shell | Remaining thickness at accessible shell points | Discrete shell points | Sound velocity in refractory is strongly temperature-dependent, so raw readings need temperature compensation and per-material calibration |
| Infrared thermography | Running | Shell temperature field — an indirect, comparative signal of remaining insulation | Full shell surface | It reads the whole thermal resistance, not just the refractory; coating, cladding and insulation blanket all influence it |
| Laser profiling | Shut down, empty | 3D surface map of the hot face; localised thin spots and asymmetric wear | Full interior surface | Residual slag or scale on the hot face distorts the return; needs a clear line of sight |
| Core sampling + laboratory | Shut down | Density, strength, phase composition, penetration depth | Discrete cores | Destructive; each core is a hole that must be repaired |
| Embedded thermocouples | Running | Temperature at known depths behind the hot face; the wear front as it advances | Installed points only | Must be designed in at build; cannot be retrofitted to an existing lining |
Two scheduling rules follow from this table. First, put the running-state survey on a much shorter cycle than the shutdown survey, because it costs nothing but an operator's time: boiler and fired-heater practice is a thermographic survey every 2000–3000 operating hours, against a full visual and hammer inspection at each planned outage, typically every 8000–12,000 operating hours. Second, treat the running survey as a screening tool that decides where the shutdown survey concentrates its effort — not as a substitute for it.
4. Thickness Measurement: Probe, Ultrasonic and Laser Compared
The three methods are usually presented as alternatives. They are better used as a hierarchy, because they have different strengths and even a well-funded programme rarely gets all three everywhere.
- The mechanical probe is the reference measurement. It is cheap, it needs no calibration, and it is what a specialist contractor will use to check a suspect area. Its weakness is repeatability: a probe reading taken "wherever looked worst" cannot be compared with the reading taken last year at a different spot.
- The ultrasonic gauge is the only method that works from outside a running unit, and that is its entire value. Its weakness is calibration. Sound velocity through a refractory changes with temperature and with material, so a gauge calibrated on dense castable will report the wrong thickness on a lightweight insulating layer, and a reading taken on a hot shell will not match one taken cold unless the temperature at the time of measurement is recorded and compensated for.
- Laser profiling gives the full geometric picture — it will find the localised thin spot that a 1 m probe grid walks straight past. Its weakness is that it measures the surface it can see, which may be a slag layer rather than the refractory beneath it.
Whatever the method, three disciplines turn readings into data:
- Fix the measurement points and document them. Zone, elevation, circumferential location, distance from a named datum. If the grid moves between campaigns, the campaign-to-campaign comparison is worthless.
- Record the condition of each reading. Temperature at the time of measurement, surface condition, presence of slag or scale, and which operator took it.
- Log against operating hours, not calendar dates. This is the discipline most often skipped and the one that matters most, for the reason set out in section 9.
5. Shell Temperature as an Indirect Thickness Signal
Infrared thermography is the only measurement in the list that can be taken continuously on a running unit, and it works because the shell temperature is a function of the lining's total thermal resistance. The relationship is steeply non-linear, which is exactly what makes the method useful: a small loss of thickness at the end of a campaign shows up as a much larger temperature rise than the same absolute loss would have shown at the start.
The model below takes a 1250 °C hot face, a 25 °C ambient, a combined surface coefficient of 10 W/m²·K and a composite effective conductivity of 0.10 W/m·K for the lining assembly as a whole. That last figure is an assumption, not a material property: it represents a dense hot face plus an insulating backup plus a fibre board, and a real design should substitute its own layer-by-layer resistance. The point of the table is the shape of the curve, not the absolute values.
| Remaining lining (mm) | % of design thickness | Total resistance (m²·K/W) | Heat flux (W/m²) | Shell temperature (°C) | Rise vs. full lining (°C) |
|---|---|---|---|---|---|
| 230 | 100% | 2.400 | 510 | 76.0 | — |
| 200 | 87% | 2.100 | 583 | 83.3 | +7.3 |
| 175 | 76% | 1.850 | 662 | 91.2 | +15.2 |
| 150 | 65% | 1.600 | 766 | 101.6 | +25.5 |
| 115 | 50% | 1.250 | 980 | 123.0 | +47.0 |
| 90 | 39% | 1.000 | 1225 | 147.5 | +71.5 |
| 80 | 35% | 0.900 | 1361 | 161.1 | +85.1 |
| 60 | 26% | 0.700 | 1750 | 200.0 | +124.0 |
The first 25 % of lining loss costs about 15 °C of shell temperature. The last 25 % costs over 75 °C. This is the single most important property of the method and also its main trap: thermography is a poor early-warning tool and an excellent late-warning tool. By the time a hot spot is unambiguous on the camera, a substantial amount of insulation has already gone, and the remaining margin is small.
Read the table with a fixed survey discipline and it becomes far more powerful. A 15 °C rise is inside the noise of a changing ambient, a change in wind, or a wet shell. A 15 °C rise at one elevation compared with the same elevation 90° around the shell, on the same day, at the same load is a signal. Always compare like with like circumferentially and by elevation, and always log the ambient temperature and load with the image.
Do not read the shell temperature as a pass/fail number either. Carbon steel is susceptible to corrosion under insulation where the metal temperature sits roughly between −12 °C and 175 °C (the range set out in API RP 583), so a normal furnace shell at 76 °C is squarely inside that window and the external coating and cladding system carries as much of the corrosion risk as the refractory does. Separately, ASTM C1055 puts the five-second contact burn threshold for metal at around 60 °C, which is why personnel-protection screening on hot surfaces is set there rather than at the shell's structural limit.
6. Hammer Testing: The Cheapest Test, Most Often Done Badly
A hammer test costs nothing and finds the failure mode that kills people — a lining that has separated but is still standing. It is also the test most often reduced to a formality, because it is slow, tedious and uncomfortable, and because a tired surveyor's ear stops distinguishing a ring from a thud after about twenty minutes on a scaffold.
The technique that makes it repeatable is a fixed grid and a defined hammer. Good practice for monolithic linings is a ball-pein machinist's hammer of about 450 g struck on a documented grid, with the grid tightened as the consequences of a miss get worse:
| Survey area | Maximum grid spacing | Why |
|---|---|---|
| Floor / bottom head | 900 mm centres | Large flat geometry; a fallen slab is a falling load but not usually an immediate shell risk |
| Vertical wall / top surface | 450 mm centres | Loss of a section above an operator is a direct hazard; geometry is harder to survey than a floor |
| Around nozzles, corners and expansion joints | 150 mm centres | These are the stress concentrators and the first places a lining pulls away |
Two details make the results usable. First, hammer each layer of a multi-layer lining after that layer has cured, before the next layer is placed — a defect buried under a sound backup layer can never be found afterwards. Second, mark anything you query with chalk or paint and photograph it in place, with a scale and a reference to the survey grid. A hammer survey whose findings are not marked on a map degrades into "the contractor said about twenty percent sounded hollow", which is not a work scope.
7. Grading the Defects: Cracks, Spalls, Delamination and Anchors
The hardest part of an inspection report is the transition from description to grading, because a crack width only becomes a repair trigger in combination with where it is, how deep it goes and what it exposes. The table below consolidates criteria that appear consistently across installation practice and industry inspection standards. Treat the numbers as screening thresholds that put a defect into a band, not as a substitute for an engineer's judgement on the specific geometry.
| Defect | Accept / monitor | Plan a repair | Act immediately |
|---|---|---|---|
| Crack width | Random crazing and shrinkage cracks up to about 1.5 mm — expected, and largely closes at operating temperature | Greater than 3 mm and extending up to half the layer thickness; more than 6 mm must be repaired in any case | Widening cracks, or a crack that visibly penetrates more than half the layer, in a quarl, arch, lintel or other load-bearing element |
| Crack pattern | Isolated cracks with sound material on both sides | Cracks within about 300 mm of each other forming blocks, or network cracking, regardless of individual width | Blocks that have already begun to move, drum or ring hollow |
| Spalling depth | Shallow surface flaking under about 20 mm | Local loss deeper than 20 mm, or any loss exposing the backup layer | Deep loss exposing the shell, or progressive spalling across a wide area |
| Delamination (hammer) | No hollow areas on the survey grid | An isolated hollow area larger than about 150 × 150 mm; a void or inclusion larger than 25 mm or one quarter of the layer thickness | Total defective area exceeding roughly 25 % of the lining surface — this is normally a full reline rather than a patch |
| Soft / dry fill | Locally soft surface that does not reduce effective thickness | Compromised material that reduces effective lining thickness by more than a quarter of the original, or is deeper than about 13 mm | Material that is visibly disengaging in service |
| Anchors | Fully embedded, with the normal cover over the anchor tip intact | Surface rust or minor displacement discovered during a shutdown | Any visible anchor exposure. Anchors are normally covered by at least 25 mm of material, so an exposed anchor is itself a wear measurement |
| Shell temperature | Within the zone's own established baseline envelope | A sustained rise of roughly 30 °C above an elevation-matched baseline at the same load | A hot spot that is growing, or any local spot above the level at which the shell material's own design temperature is approached |
Colour is a better indicator than most reports give it credit for. A local patch of castable that has turned white or light grey against the normal grey-brown of the surrounding material is a local overheating signature, and it usually appears before the thermographic hot spot becomes unambiguous. Photograph it and record its extent on the same grid as the thickness readings.
8. Core Sampling and What the Laboratory Can Tell You That the Field Cannot
When a zone is degrading faster than its history predicts, the field methods have run out of information. They can tell you how much is gone; they cannot tell you why. That is the point at which a core earns its cost.
A core is drilled at a location chosen to capture the full section, in the zone where the mechanism is suspected — not in the convenient place with the best access. Laboratory analysis then proceeds in a defined order:
- Chemical analysis by XRF across the section — original material, penetrated layer, hot-face skin — to establish the penetration depth and concentration of slag, alkali or process species. This is the measurement that tells you whether the problem is chemical or mechanical.
- Phase analysis by XRD to identify what has formed and what has decomposed. Detection of phases such as nepheline or potassium nepheline in a mullite- or corundum-bearing lining is direct evidence of alkali attack; decomposition of the original bond phase tells you the hot face has been over its service temperature.
- Microstructure by SEM with EDS to see pore structure, crack propagation paths, grain-boundary changes and where the penetrating species actually sit. This is where a mechanism stops being a hypothesis.
- Physical properties on the as-received core: bulk density and apparent porosity (ASTM C20 or C830), cold crushing strength (ASTM C133) and, where abrasion is the suspect, abrasion resistance (ASTM C704). Compare these against the original compliance data sheet, not against a generic table.
- Rebound hardness on the in-situ surface as a fast, cheap relative indicator of surface sintering, softening or friability — useful for comparing zones or tracking a zone over time, but it is not a strength measurement and should not be reported as one.
The output of this sequence is a mechanism statement, and the mechanism decides the repair. A lining that has lost thickness to abrasion is fixed by changing the hot-face material or reducing the velocity in front of it. A lining that has lost thickness to alkali penetration is not fixed by making it thicker; it is fixed by reducing the alkali reaching it or changing the chemistry it sees.
9. Wear-Rate Arithmetic: From Two Surveys to a Reline Date
The arithmetic itself is trivial and worth stating precisely, because almost every error in lining forecasting is an error in the inputs rather than the calculation. For each survey point:
- Wear rate = (original thickness − current thickness) ÷ operating hours since the previous survey
- Remaining life = (current thickness − minimum acceptable thickness) ÷ wear rate
Worked example, dense castable hearth: 200 mm at commissioning, 170 mm after 8000 operating hours, minimum acceptable thickness set at 100 mm.
| Case | Original (mm) | Measured (mm) | Operating hours | Wear (mm / 1000 h) | mm per year at 6000 h/yr | Minimum (mm) | Remaining (h) | Remaining (yr at 6000 h/yr) |
|---|---|---|---|---|---|---|---|---|
| Furnace hearth, dense castable | 200 | 170 | 8000 | 3.750 | 22.5 | 100 | 18,667 | 3.11 |
| Transfer line, erosion castable | 100 | 78 | 8000 | 2.750 | 16.5 | 50 | 10,182 | 1.70 |
| Boiler burner throat | 150 | 118 | 12,000 | 2.667 | 16.0 | 75 | 16,125 | 2.69 |
| Kiln burning zone (brick) | 230 | 200 | 16,000 | 1.875 | 11.25 | 115 | 45,333 | 7.56 |
Two things worth noticing in that table. The transfer line has the largest remaining thickness in absolute terms (78 mm against a 50 mm floor) and the shortest remaining life, because its wear rate is high. And the kiln brick, which has lost the most material (30 mm, the same as the hearth), has more than twice the hearth's remaining life, because it has already run twice as long to lose it.
That second observation leads to the discipline that matters more than any formula: normalise to operating hours. Take the same 30 mm of loss and measure it over three different intervals with the same 100 mm floor and 170 mm reading:
| Interval over which the 30 mm was measured | Implied wear rate (mm / 1000 h) | Implied remaining life (h) | Years at 6000 h/yr |
|---|---|---|---|
| 4000 h | 7.500 | 9333 | 1.6 |
| 8000 h | 3.750 | 18,667 | 3.1 |
| 16,000 h | 1.875 | 37,333 | 6.2 |
Same lining, same measured loss, same minimum thickness, and a projected reline date that moves by a factor of nearly four. The difference is not statistics — it is real, because a wall that lost 30 mm in 4000 hours is genuinely wearing faster than one that took 16,000 hours to lose it. The lesson is that a measured difference is not a rate until it has been divided by the hours that produced it, and that comparing wear rates between zones or between campaigns without that division is comparing nothing.
10. Confidence: Why One Interval Is Not a Trend
A single interval produces a straight line and the average wear rate of that interval. It cannot tell you whether wear is accelerating, decelerating or steady, and it cannot tell you how much of the apparent loss is measurement scatter rather than material loss. That matters because a projected reline date built on a straight line has no error bar, and a plan with no error bar gets ignored.
| Scatter between repeated readings (± mm) | Projected life range (h) | Range in years at 6000 h/yr | What the data supports |
|---|---|---|---|
| ±5 (tight, same operator, same method, same grid) | 16,000 – 22,400 | 2.7 – 3.7 | Schedule the reline inside the band and tie it to a planned outage |
| ±10 (routine scatter) | 14,000 – 28,000 | 2.3 – 4.7 | Increase survey frequency as the zone approaches the low end |
| ±20 (mixed operators, mixed methods, or a slag-covered hot face) | 11,200 – 56,000 | 1.9 – 9.3 | Not a basis for a reline decision. Tighten the method first |
Widen the band rather than collapsing it. The temptation with three scattered readings is to average them and quote a single confident date. That number will be wrong and the plan built on it will fail. A range of 2.3–4.7 years is a more useful deliverable than "we expect to reline in 3.2 years", because it tells the planner how much contingency the outage schedule needs.
A defensible minimum for a trend is three consistent measurements on the same grid using the same method, at intervals short enough that the zone has not gone from measurable to critical between two of them.
11. API 936 As-Installed Limits: the Numbers a New Lining Must Hit
All the forecasting above assumes a lining that started at the specified thickness with the specified properties. That assumption is worth verifying, because installation is where most linings lose the campaign before the unit ever starts. API 936 (Refractory Installation Quality Control) sets out the sampling, testing and acceptance framework, and its as-installed acceptance table is the one to write into a purchase specification.
Under the fifth edition the as-installed acceptance is expressed as a range around the manufacturer's own compliance data sheet value, tested on specimens formed from the material actually being installed:
| Property | Acceptance against the manufacturer's data sheet (fifth edition) | Test basis |
|---|---|---|
| Abrasion loss (erosion service) | Not more than 110 % of the data sheet value — no lower limit | ASTM C704, specimens fired to 815 °C, tested on an original free (not trowelled, moulded or cut) surface |
| Cold crushing strength | Not less than 90 % of the data sheet value — no upper limit | ASTM C133, cubes fired to 815 °C |
| Density | Within ±80 kg/m³ (±5 lb/ft³) of the data sheet value | Determined at room temperature on dried or fired specimens |
| Permanent linear change | Not less than zero shrinkage; a product with positive PLC after firing is not to be used unless the owner approves it | ASTM C113, measured along the 230 mm specimen dimension on all four edges |
| Workability index (plastic refractories only) | Not less than the minimum the manufacturer states as suitable for the installation method | ASTM C181, five specimens per sample |
The framework around those numbers is as much of the specification as the numbers themselves:
- Sampling frequency. For erosion service, one sample per pallet or partial pallet from each production run. For other service, one sample per three pallets or fewer. Material is also qualified before it ships from the works, on the same frequency basis, with a single permitted retest before rejection.
- Specimen sizes. Abrasion plates 114 × 114 × 25 mm; permanent-linear-change bars 50 × 50 × 230 mm; crushing cubes 50 × 50 × 50 mm.
- Specimen preparation. Specimens must be formed by the same method as the installation — cast the same way, vibrated if the lining is vibrated, hand-packed if the lining is hand-packed. Cube loading faces must be parallel to within ±0.8 mm and square to within 1°. The firing cycle runs at roughly 170 °C per hour to 815 °C with a five-hour hold.
- Curing before testing. Hydraulic-bonded castables cure for at least 24 hours with the exposed surface covered or sealed; air-setting phosphate-bonded castables cure uncovered for at least 24 hours protected from moisture; heat-setting plastics air-dry for at least 24 hours before oven drying.
- Anchors. Anchor alloy is verified by 100 % positive material identification before the lining goes over it. An anchor whose alloy is wrong is a defect that can only be found before it is buried.
- Material storage. Held between about 5 °C and 38 °C, moisture-proof, on a ventilated platform, with each pallet uniquely coded so that a failed as-installed test can be traced back to a batch and forward to the area of lining it represents.
- Rejected areas. A failed sample rejects the area of lining that sample represents. Where there is disagreement, cores are taken from the questionable area and retested against the same criteria; a second failure means replacement, and the contractor must map the rejected and replaced areas.
Documents worth pairing with it for a non-petrochemical duty are ASTM C401 for castable classification, ASTM C862 and C865 for casting and firing practice, ASTM C27 for fireclay brick and ASTM C155 for insulating firebrick, together with ISO 1927-1 for monolithic refractory classification where an ISO-based specification is being used.
12. Repair or Reline: Decision Rules and How to Patch Properly
Once a defect has been graded, there are three possible actions and the criteria should be written before the outage, not argued about during it.
Patch where the defect is local and the surrounding lining is sound: spalling shallower than about 20 mm, individual cracks, or a hollow area smaller than roughly 150 × 150 mm. Replace the section where the defect is deep, where the defect penetrates more than half the layer, or where a hollow area is large but the rest of the lining is still good. Reline where the total defective area exceeds about a quarter of the lining surface, where damage is distributed rather than local, or where the anchor system itself has been compromised over a wide area.
A patch is a small piece of refractory construction and it fails for the same reasons a new lining fails. The sequence that works:
- Remove the full thickness of the defective layer over an area that extends to at least three or four anchors, with the periphery of the cut located mid-way between anchors rather than on one. Cut at a shallow angle to the shell where the geometry allows.
- Clean back to sound material — no loose material, no debris, no dust on the interface.
- Wet the adjacent sound material thoroughly before placing new material. A dry interface draws water out of the new castable and produces a weak, porous joint at exactly the place the repair will be judged.
- Restore the anchor pattern. If the repair is wide, add auxiliary anchors to bring the longitudinal and transverse pitch back to roughly the original spacing — around 200 mm is a common design pitch — because the original anchors at the centre of the repair area have been lost.
- Use the same material and the same installation method as the original lining. Substituting a different castable because it was in the store produces a repair with a different thermal expansion, a different modulus and a different dry-out requirement from the lining around it, and it will fail at the interface.
- Form, cure and dry out. Cure per the material's own requirement (typically at least 24 hours), then dry out on the same controlled schedule that the original lining needed. Assuming the repair can skip the dry-out because it is small is the single most common cause of a patch falling off in the first week of service.
For a single isolated crack, the equivalent treatment is to remove the full layer thickness over a strip roughly 120–150 mm either side of the crack and parallel to it, then reinstate as above.
Everything above requires the unit to be down, isolated and cooled — comfortably below 60 °C before anyone works on the hot face. Ceramic-fibre and flexible mortars can seal some surface cracking on accessible external areas while a unit runs, but nothing that involves removing material, replacing anchors or casting new refractory belongs in a running unit. An "emergency hot patch" applied to keep a unit online is a decision to defer a repair, not a repair.
Planning needs lead time, and the lead time is set by the material rather than by the labour. Dense castables and plastics typically need 4–8 weeks from order to site; ceramic-fibre modules 2–4 weeks; specialised burner-block materials 10–14 weeks. A reline date derived from a wear-rate projection is only usable if the material is ordered off the back of it.
13. Writing It Into the Specification and the Dossier
The durable output of a lining programme is not a report; it is a specification plus a history. For each zone, the specification should state the design thickness, the minimum acceptable thickness and the basis for it, the hot-face and backup materials with their compliance data sheets, the anchor alloy and pattern pitch, the survey grid with its datum, the measurement method to be used at each point, and the defect grading table with the actions attached to each grade.
The history should hold, per zone and per inspection: date and operating hours, method and equipment with calibration status, the raw readings against the fixed grid, temperature and conditions at the time of measurement, photographs with scale, hammer findings with the map, core analyses where taken, and the wear rate and remaining-life projection calculated from them. Trend the rate, not just the thickness, and treat a rising rate as the signal that a mechanism has changed.
Two practical notes make the difference between a dossier that gets used and one that gets shelved. Store the raw readings, not only the summary — a rate cannot be recalculated from an average. And keep the same measurement points across campaigns even after a partial repair, because a repaired zone still needs a comparable history, and a grid that changes whenever the lining changes gives you nothing to compare.
14. Sourcing Lining Materials Against a Real Specification
An inspection programme is only useful if the replacement material is available with the properties the survey says are needed and the documentation to prove them. For zones where the survey has identified a specific mechanism, the material choice follows from it: high-temperature aluminium-silicate ceramic fibre board for furnace backup and hot-face insulation where the driver is thermal, mullite insulating firebrick where the driver is a hot face that has to carry load at temperature, alumina ceramic fibre board for kiln zones where the duty temperature is higher, and dense high-alumina brick where abrasion and chemical attack at the hot face are the governing mechanism. Each of those carries a different classification, a different density and conductivity package, and a different set of acceptance tests.
Ask for the compliance data sheet with the material, confirm which ASTM or ISO method produced each number, and confirm that the batch is traceable. The as-installed acceptance limits in section 11 are expressed relative to that data sheet, so a lining installed against a vague data sheet cannot be verified at all.
If you are scoping a reline, send us the zone, the duty temperature, the atmosphere and the mechanism your survey identified, and we will come back with a material recommendation, the classification and property data that supports it, and the test set that should be specified for acceptance.
Related reading: how the materials are compared and selected in Refractory Materials Selection; why the dry-out after any repair matters as much as the material in Castable Mixing, Curing and Dry-Out; how anchors and expansion joints are detailed in Refractory Anchor Systems; grade selection for lightweight linings in Insulating Firebrick Grade Selection; and zone-by-zone brick selection in High-Alumina Bricks for Cement Kilns. Product pages: aluminium-silicate ceramic fibre board for furnaces, mullite insulating firebrick, 1500 °C alumina ceramic fibre board for kilns and the full fireproof materials range.
Need help selecting FRP products?
Our engineers answer technical questions free of charge.
Ask an Engineer on WhatsApp