Views: 0 Author: J-VALVES Publish Time: 2026-10-30 Origin: Site
"Internally and externally coated" is the most common coating line in a valve enquiry and one of the least useful. It does not say which mechanism the coating is meant to interrupt, what defect density will be accepted, whether the film must also survive cathodic protection, or what happens at the flange face and the seat pocket where the coating has to stop. The result is predictable: a valve arrives with a film that looks correct, passes a thickness check, and perforates in year four at a pinhole nobody tested for. This article sets out a selection method that starts from the mechanism rather than the product name, with the arithmetic that makes each decision defensible: holiday probability, anode-to-cathode area ratio, blast profile against film thickness, the finite sacrificial capacity of sprayed aluminium, and the specification clauses that decide whether a coating survives commissioning.
A valve coating is selected by the mechanism that will attack the surface, and no single coating answers more than two of the mechanisms below. Naming a product before naming the mechanism is how a 300 micron epoxy film ends up on an abrasive slurry line, or a 25 micron nickel deposit ends up on a cast body with a 70 micron blast profile.
Mechanism | Where it attacks a valve | Failure signature | Coating that answers it | Coating that only appears to |
|---|---|---|---|---|
Immersed seawater and chloride | Bore, seat pocket crevice, body-to-bonnet joint, downstream of the disc | Deep isolated pits, often under an intact film; rust bloom at 6 o'clock in a horizontal run | Thermal-spray aluminium with sealer; high-build epoxy external; duplex body where the pitting allowance cannot be afforded | Thin decorative paint; zinc-rich primer in chlorinated or alkaline water |
Marine atmosphere and corrosion under insulation | Bonnet, yoke, flange outside diameter, anything under cladding | Blister ring at the insulation lip; white or grey product; external chloride stress corrosion cracking on 300-series stainless | ISO 12944 C5-M system at 240 to 320 microns total, with the insulation detail designed to shed water | Single-coat alkyd; any system applied over a damp surface |
Abrasive and erosive slurry | Bore at the seat, and the first two diameters downstream | Score marks parallel to flow; film removed in a band; polished metal | HVOF tungsten carbide on trim; rubber or polyurethane lining on the body; hardened alloy body | Electroless nickel at 25 to 75 microns; standard epoxy |
Chemical attack | Whole wetted surface, plus every gasket crevice | Blistering, softening, film lifting from the edges inward | PTFE or PFA lining; high-phosphorus electroless nickel for mild acids; nickel-alloy body | Epoxy in strong solvent; zinc in acid or strong alkali |
Galvanic coupling, stray current, cathodic protection | Near dissimilar joints, near anodes, at damaged areas | Disbondment halo growing from a defect; hydrogen blisters | Coating qualified by cathodic disbondment, plus correctly controlled CP, plus insulating kits at dissimilar joints | Any barrier coating used as the sole protection with overprotection present |
The table also exposes the one mechanism no coating can solve alone. If the dominant attack is crevice corrosion - the gap between the seat ring and the body, the joint between a two-piece body, the thread root of a plugged tapping - then a coating helps only if it can be applied inside that gap and stay bonded there. It usually cannot, and a partially bonded film inside a crevice is worse than bare metal, because it becomes the crevice former and the steel behind it corrodes faster than it would have in the open.
A coating does not slow corrosion uniformly. It removes area from the wetted surface, and the life you buy is the time until the first defect becomes a perforation. That time is set by defect density and by the ratio of cathode area to anode area, not by film thickness alone - which is why two valves with the same 300 micron film can have service lives of three years and thirty.
Compare the two protection strategies on the same DN 200 Class 150 carbon steel gate valve in seawater. A bare body with 3 mm of corrosion allowance, corroding at a typical 0.15 mm per year, reaches its limit in about 20 years. A coated body has essentially no attack over 95 percent of its area and an attack rate at the remaining 5 percent that can be ten to a hundred times the bare rate, because a pinhole is a small anode surrounded by a large cathode.
The arithmetic is worth doing on a real valve. Good plant-applied fusion-bonded epoxy runs at roughly one holiday per 10 square metres when the application is well controlled. A DN 200 valve has about 0.18 square metres of bore surface and perhaps 0.10 square metres of cavity and seat pockets, so about 0.28 square metres of coated wetted area. That is a 2.8 percent chance of one holiday per valve. An eighty-valve seawater unit therefore carries two holidays before it is commissioned, and no inspection will find them unless a holiday test is run on every body.
Now size the consequence. A one millimetre pinhole is 0.8 square millimetres of exposed steel. Within a few centimetres of it sit roughly 100 square centimetres of coated surface. If that film still conducts even slightly - through a zinc-rich primer, or through the electrolyte path that cathodic protection current takes - the cathode-to-anode ratio exceeds 100 to 1. Galvanic corrosion under that ratio drives the small anode at whatever current the large cathode can supply, and ten to a hundred times the bare rate is the normal outcome. At 1.5 mm per year, a 9 mm wall with 3 mm of allowance is gone in four years, at one spot, on an otherwise perfect valve.
This reframes what you are buying. Thickness protects against permeation and mechanical damage. Adhesion protects against disbondment. Edge coverage protects the transitions where fusion pulls back. The holiday test protects against the pinhole. Specifying 500 microns instead of 300, without a holiday test, buys nothing that matters.
Electroless nickel-phosphorus is the only coating that can be deposited uniformly on the inside of a valve body, and its useful range is narrower than its reputation: mild chemical attack and light wear on machined surfaces, at 25 to 75 microns, with a hardness that depends on a heat treatment most assembled valves cannot survive.
The mechanism explains the geometry. ENP deposits by autocatalytic chemical reduction, with no applied current, so there is no current-density fall-off in a bore and no edge build-up. Thickness uniformity is typically within plus or minus 5 to 10 percent across bores, threads, seat pockets and blind cavities - exactly the places where a spray gun cannot see and an anode cannot throw. Electroplated nickel and hard chrome cannot follow that geometry, which is why they appear on stems and never on body bores.
Property | Low phosphorus (1 to 4 % P) | Medium phosphorus (6 to 9 % P) | High phosphorus (10 to 13 % P) |
|---|---|---|---|
Microstructure | Crystalline | Microcrystalline | Near-amorphous |
Hardness as plated | Highest of the three, roughly 700 HV | About 500 to 550 HV | About 450 to 500 HV |
Hardness after 350 to 400 C bake | Limited gain | About 900 HV | 850 to 1000 HV |
Neutral salt spray (ASTM B117) | Low, in the low hundreds of hours | 200 to 500 hours | 1000 hours and above |
Best used for | Wear on stems and seat retainers | General purpose bore and trim protection | Chemical and marine wetted surfaces |
Three traps decide whether the deposit performs as the table promises.
The hardening bake and the assembled valve. Heating to about 400 C for an hour precipitates nickel phosphide and takes hardness from roughly 500 HV to 850-1000 HV. A finished valve cannot be baked at 400 C with its seats, gaskets and packing installed, so you either plate and bake the trim separately, or accept as-plated hardness on the body. On an austenitic body, 400 C also sits at the edge of the sensitisation window, so a baked CF8M body in wet chloride service should come with an intergranular corrosion test result, not an assumption.
Hydrogen. Acid pickling and the plating reaction charge the steel with hydrogen. On stems, seat retainers and bolting above roughly 1000 MPa ultimate strength, require baking within a few hours of plating at 190 to 230 C, with a certificate. An ENP-coated high-strength part without a bake record is a delayed cracking risk, and the crack will appear weeks later, away from the coating.
The pore is a small anode. Nickel-phosphorus is cathodic to steel, so a pore does not merely fail to protect - it concentrates attack vertically into the substrate. Twenty-five microns is a floor, not a target. Require thickness measurement at defined points on the bore plus a porosity check on a coupon, and remember that a magnetic gauge reading over a blasted casting can be measuring the profile rather than the film.
A 25 micron coating on a surface blasted to a 60 micron profile is a coating with holes in it, and the thickness gauge will not tell you. ISO 8503 medium grade profile is 50 to 85 microns peak to valley. An ENP deposit at the bottom of the typical 25 to 50 micron range sits below the peaks, so the peaks are bare metal from the day the valve ships.
This is why ENP belongs on machined surfaces - a bore honed to Ra 0.4 microns takes a 25 micron deposit as a continuous film - and not on a blasted casting. If a casting must be plated, specify a finer profile, a thicker deposit in the 50 to 75 micron range, and verification by sectioned coupon under a microscope rather than by a magnetic gauge alone.
The same logic condemns thin "epoxy painted" valve bores. Forty to sixty microns of liquid epoxy over a 75 micron profile leaves 30 to 50 percent of the area bare or nearly bare, while the average of the thickness readings looks acceptable. The rule that prevents both errors is simple: the film must exceed the profile by at least three to one for a barrier coating. Three hundred microns of fusion-bonded epoxy over a 75 micron profile is a comfortable four to one. Twenty-five microns of nickel over the same profile is not a coating at all.
FBE is a powdered thermoset fused onto a preheated surface at 180 to 250 C, typically 150 to 300 microns, with no measurable porosity when the fusion is complete. It is the default internal coating for water, wastewater and buried-service valves, and it deserves that position - for the conditions it was designed for.
The service ceiling is the first limit. FBE formulations run to roughly 90 to 120 C depending on chemistry and on whether the exposure is immersion or dry. Above that, water permeates the film, vapour pressure builds at the interface and the coating blisters from behind. A "FBE lined" valve on a 140 C line is not a protected valve; it is a blister waiting for its first shutdown.
The second limit is disbondment. The test that predicts life in soil and seawater is cathodic disbondment: the radius of coating lost around a deliberate defect after a set time at elevated temperature in a specified electrolyte. A disbonding coating fails invisibly - the film stays intact and the steel beneath it pits - which is why the failure is so often discovered by a perforation rather than by an inspection.
The third limit is edges. Fusion pulls the film back from a sharp edge, so a 90 degree machined transition can end up with a fraction of the nominal thickness exactly where the fluid impinges. Require radiused or chamfered bore transitions, a stripe coat or second pass at welds and edges, and edge thickness measured rather than assumed.
The fourth limit is repair. FBE is not field-repairable to original quality. Patch compounds are epoxy putties with different adhesion and no fusion bond to the substrate or to the surrounding film. On a valve that will be opened for maintenance, a patched bore is the future failure site, and recoating means full disassembly and a return to the oven. Holiday testing is normally done at 1500 to 2000 V DC for films in the 300 to 500 micron range; the voltage must be set from the measured thickness, because too low misses pinholes and too high burns good film.
Arc-sprayed aluminium at 200 to 300 microns is porous by nature, typically 2 to 8 percent, and is sealed with an epoxy or silicone sealer. What makes it structurally different from every coating above is that it protects galvanically: a holiday is protected by the surrounding metal, which sacrifices itself. A barrier coating fails at its first holiday. A sacrificial coating consumes itself at its first holiday and keeps the steel intact while it does so.
That difference makes thickness a design variable rather than a habit. Aluminium delivers about 2980 ampere-hours per kilogram. A 250 micron layer weighs about 0.675 kg per square metre and therefore carries roughly 2000 ampere-hours per square metre. At a protection current demand of 10 milliamperes per square metre, that is about 200,000 hours, or 23 years of theoretical capacity; at a realistic 60 percent utilisation it is 13 to 14 years. Halve the thickness and you halve the life. Ask for that calculation instead of an ISO 12944 system number, because the standard describes a paint system, not an anode.
System | Typical thickness | How it protects | Real limits on a valve |
|---|---|---|---|
Three-coat paint (zinc-rich primer, epoxy intermediate, polyurethane topcoat) | 240 to 320 microns | Barrier, with limited galvanic action from the primer | Needs dry, well prepared steel; damage is permanent; not for immersion at temperature |
Fusion-bonded epoxy, internal | 150 to 300 microns | Barrier | Ceiling near 90 to 120 C; not repairable; edge coverage must be engineered |
Electroless nickel-phosphorus | 25 to 75 microns | Metallic barrier, pore is a small anode | Requires a smooth substrate; thin for abrasion; cannot be baked after assembly |
Thermal-spray aluminium with sealer | 200 to 300 microns | Sacrificial anode plus barrier | Line of sight only; pH roughly 4 to 9; never a sealing surface |
Thermal-spray zinc or zinc-aluminium | 150 to 300 microns | Sacrificial anode | Attacked above pH 12.5 and in chlorinated water; not for sour or amine service |
PTFE or PFA lining | 2 to 4 mm | Thick barrier | Permeation and vacuum limits; a separate discipline from coating |
Aluminium's passive film is stable across roughly pH 4 to 9. Chloride is fine - that is precisely why thermal spray works offshore - but strong alkali and strong acid both attack the metal, and hydrofluoric acid destroys it. Because the process is line of sight, a DN 50 bore cannot be coated more than a few diameters deep, and because the as-sprayed surface is 20 to 40 microns Ra, it is never a sealing surface. Its home on a valve is the body and bonnet externals in the splash zone, buried seawater valves, and any location where mechanical damage during installation is close to certain.
Most coating failures on valves are not coating failures at all. They are failures of masking - the film was applied where the geometry needed bare metal, and the resulting leak or seizure gets blamed on the product.
Location | What the film does there | What to do instead |
|---|---|---|
Flange gasket faces | Adds a compressible layer inside the joint, so bolt preload relaxes and the joint creeps; also hides a damaged face | Mask; terminate the coating 3 to 5 mm outside the raised face; coat the flange outside diameter only |
Seat and sealing faces | Changes geometry by more than the flatness tolerance; polymer films creep under seat load and take a permanent set | Mask with a pressed plug; inspect bare metal before assembly |
Threaded ends (NPT, BSP, API line pipe) | Fills the thread roots, changes pitch diameter, causes galling and false torque readings | Cap or plug; use a thread compound, not a coating |
Stem inside the packing zone | Wears, and the debris packs into the braid; changes friction and can break a qualified emission class | Coat above the gland only, or use a sleeved stem with the sleeve as the sealing surface |
Bolt holes and stud threads | Reduces engagement and preload; a hidden cause of joint leaks after maintenance | Mask; use anti-seize on the fastener instead |
Inside a crevice (seat pocket, body joint, two-piece body split) | Disbonds, then acts as the crevice former; the steel behind corrodes faster than if it were bare | Leave bare and control by material, sealant injection, or geometry |
Within 50 to 100 mm of a field weld | The coating burns back, and the field joint becomes the first failure point | Cut back, coat after welding with a heat-shrink sleeve or liquid epoxy |
Oxygen service | Any organic film is fuel in the presence of oxygen and an ignition source | Oxygen-cleaned bare metal, or a qualified non-organic finish, with a cleaning certificate |
On valves, the external coating fails at a predictable place, and it is rarely the body. Carbon steel is at risk of corrosion under insulation across roughly 50 to 175 C, and 300-series austenitic stainless is at risk of external chloride stress corrosion cracking across roughly 50 to 150 C. Both risks are driven by water, not by temperature alone, and the valve is where water collects: the insulation stops at the flanges, rain runs down the cladding, and the bonnet, yoke and stem are usually left bare and wet. That is how a valve arrives at overhaul with a perfect body and a sponge for a bonnet.
ISO 12944 gives the system backbone. Corrosivity category C3 takes about 200 microns, C4 about 240 to 280, C5-M about 240 to 320 as a three-coat system - typically 60 to 80 microns of zinc-rich primer, 80 of micaceous iron oxide epoxy, and 80 of aliphatic polyurethane - and CX or offshore specifications go to 320 to 400 and beyond. Durability bands are low at 7 to 15 years, medium at 15 to 25 and high above 25.
Detail beats thickness at every one of those numbers. Sealant at the insulation lip, a drip ring or deflector above the bonnet, no crevice at the yoke, coating carried onto the first 50 mm of the mating pipe, and the nameplate mounted on a bracket rather than through the film - each of these does more for a twenty-year life than another 80 microns of topcoat.
Coating and CP are usually specified by different people and then installed as if they were independent. They are not. The coating exists to reduce the current the CP system has to deliver; the CP system exists to protect the holidays the coating will always have.
The numbers make the coupling explicit. Bare steel in flowing seawater draws roughly 100 to 150 milliamperes per square metre. A good coating brings initial demand down to 10 to 20. As the coating ages and accumulates damage, demand climbs - toward 30 to 50 after a decade of service. A CP system sized on year-zero demand is therefore undersized for the second half of the design life, which is exactly when the coating can no longer carry the load alone.
The protection window runs from about -0.80 V to -1.05 V against a silver/silver chloride seawater reference. Push past roughly -1.1 V and hydrogen evolves at the metal-coating interface: the film disbonds from behind, invisibly, and on high-strength steel the same hydrogen feeds cracking. Overprotection is a coating failure mode, and a thin barrier coating is the wrong choice for a surface that will be protected at high current density. Either specify a coating qualified by cathodic disbondment at the design temperature, or use a sacrificial spray that is designed to be consumed.
A 1000 hour neutral salt spray result is quoted in almost every coating datasheet and answers a question nobody asked. ASTM B117 runs at 35 C in a continuous 5 percent sodium chloride fog, with no ultraviolet light, no wet and dry cycling, no flow, no differential pressure and no stress. One thousand hours is 42 days - about 0.6 percent of a twenty-year life, under conditions that no valve ever sees.
What 500 hours against 1000 hours actually discriminates is process control: bath chemistry, cure schedule, substrate cleanliness. It is a useful factory gate and a useless life prediction. Ask instead for four things that do predict behaviour: cathodic disbondment at the design temperature, adhesion by pull-off after immersion, flexibility or impact at the service temperature, and a reference installation in the same medium at the same temperature. Then ask for the holiday test report on your own valve serial numbers, because that is the only document that describes the valve you are about to install.
Coating is the cheapest protection per square metre and the most expensive protection per year when it is applied to the wrong mechanism. The comparison below is indicative only - relative indices for a DN 100 to DN 200 valve at Class 150 or 300, with shop pricing varying by roughly 40 percent between regions and years - but the ranking is stable and the "loses when" column is the one that decides projects.
Option | Relative cost index | Wins when | Loses when |
|---|---|---|---|
Carbon steel with corrosion allowance | 1.0 | Corrosion rate is known and low; no crevice; the body is accessible for inspection | Crevice corrosion dominates; wall thickness is fixed by the standard rather than by the calculation |
Carbon steel, three-coat external system | 1.3 to 1.6 | Atmospheric external exposure, C4 to C5-M, dry internal | Immersed, abrasive, or above the coating's temperature ceiling |
Carbon steel, FBE internal | 1.5 to 2.0 | Water, wastewater, buried service, up to 90 to 120 C | Hot service, abrasion, or a valve that will be repaired in the bore |
Carbon steel, ENP on bore and trim | 1.4 to 2.0 | Mild chemical attack, light wear, machined surfaces | Slurry, strong acid, or a coarse cast surface |
Carbon steel, thermal-spray aluminium external | 1.8 to 2.5 | Splash zone, immersed seawater, damage during installation expected | Outside pH 4 to 9; small bores; sealing faces |
Duplex or super duplex body | 4 to 7 | Chloride stress corrosion cracking, crevice corrosion, 20 years with no maintenance access | Dry, low-chloride service where a coating would have done |
Nickel alloy or fully lined | 8 to 15 | Strong acid, hydrofluoric acid, wet chlorine | Where duplex or a qualified coating already meets the life target |
The rule that survives contact with projects: coating is the right answer when the attack is uniform and the surface can be prepared, inspected and repaired. It is the wrong answer when the attack is crevice corrosion or erosion-corrosion, because a coating cannot reach a crevice and cannot survive a particle jet. In those two cases the money belongs in material or geometry, not in a thicker film - the same conclusion that governs duplex and super duplex selection for seawater, where the alloy choice, not the coating, decides the life.
Polymer lining sits outside this comparison and should not be treated as a heavy coating. A 3 to 4 mm anchored liner is a structural element with its own permeation, vacuum and thermal expansion limits, and its failure modes are different in kind - the analysis of PTFE and PFA liner failure covers that boundary in detail.
A removed valve is a test report. The pattern of the damage identifies the mechanism, and the mechanism identifies what to change - which is usually not the coating brand.
Signature on the removed valve | Mechanism | What to change |
|---|---|---|
Pits in a tight cluster, bright metal at the base, film intact around them | Holiday with cathodic current concentration | Holiday test every body at a voltage set from thickness; consider sacrificial spray |
Film intact, steel deeply pitted beneath, disbonded halo around the pit | Cathodic disbondment front running ahead of visible damage | Specify a disbondment-tested system; reduce protection current density |
Blisters concentrated at 12 o'clock on a horizontal run | Water permeation and vapour pressure under the film | Coating qualified for immersion rather than atmosphere; dry substrate before application |
Blister ring exactly at the insulation lip | Corrosion under insulation | The insulation detail, not the coating: seal, drain, and keep water out |
Score marks parallel to flow, film removed in a band | Erosion-corrosion or abrasive slurry | Material or geometry; no coating survives a particle jet at velocity |
Mud-cracking or a crazed, brittle film | Film far above nominal, or over-cured | Enforce the maximum dry film thickness, not only the minimum |
White or grey product at a sprayed aluminium edge, steel still bright | Sealer breakdown while the anode works as intended | Re-seal at the next shutdown; the system is behaving correctly |
Black film, nickel layer lifted, in sour service | Sulphide attack on the nickel-phosphorus deposit | Nickel alloy or lining; ENP is not a sour-service coating |
Coating perfect, threads galled and joint leaking | Film left in a thread | Masking procedure and a final inspection item |
Nearly every failure described above is procured rather than manufactured, because the purchase order named a product family and left the rest to the shop. These clauses close the gap, and each one corresponds to a specific failure mode.
Clause | Wording that works | Failure it prevents |
|---|---|---|
Surface preparation | Sa 2.5 (SSPC-SP10) with a profile band stated numerically, for example 50 to 85 microns, verified by comparator | Film thinner than the profile |
System definition | Each coat named by family, with minimum and maximum dry film thickness and the reference standard | "Epoxy coated" with no measurable meaning |
Thickness measurement | Per ISO 19840: 5 percent of readings may fall below nominal, none below 80 percent of nominal, maximum 3 times nominal | Both under-thickness and mud-cracking |
Holiday test | Method, voltage set from measured thickness, zero defects accepted, report per valve serial number | The pinhole that perforates in year four |
Adhesion | Pull-off per ISO 4624, 5 MPa minimum, on a coupon or a non-critical area | Disbondment in service |
Edges and welds | Radius or chamfer all bore transitions; stripe coat welds and edges before the full coat | Thin film at the highest-velocity location |
No-coat zones | List gasket faces, sealing faces, threads, packing zone and bolt holes explicitly, with the masking method and a verification step | Leaks and seizures blamed on the coating |
Repair | Repair procedure approved before production; patch area limited; no repair permitted within the seat region | A patched bore becoming the next failure site |
Test sequence | Hydrostatic test before internal coating, or full drying and re-inspection after it; no coating over standing test water | Blistering initiated at the factory |
Heat treatment | Bake certificate for ENP on parts above 1000 MPa ultimate strength, per ASTM B850 | Delayed hydrogen cracking |
Cathodic protection interface | Disbondment test result at the design temperature where CP will be applied | Invisible disbondment under an intact film |
Traceability | Coating batch, applicator, date, and the operator's qualification record, filed against the valve serial number | Everything above, argued about after the fact |
The same discipline applies to the pressure test that follows: a coated body still has to pass its shell and seat tests, and the API 598 pressure testing sequence is where a coating applied before the test water is drained usually declares itself, as a blister found during the post-test inspection. Coated bodies are also why the paint specification and the test sequence have to be written by the same person.
Name the mechanism first. Uniform chemical attack, crevice corrosion, erosion, atmospheric exposure, galvanic coupling - the mechanism selects the family, and the family selects the thickness.
Buy defect control, not thickness. A holiday test on every body is worth more than another 150 microns of film.
Match the film to the profile. Below a three-to-one ratio of film to profile, a barrier coating is perforated by its own substrate.
Use ENP for bores and trim, never for slurry. It reaches where nothing else can, and it is far too thin for particles.
Use FBE for water, and respect its ceiling. Below roughly 90 to 120 C, in immersion, with engineered edges and no expectation of field repair.
Use sprayed aluminium where damage is certain. A sacrificial coating is the only one that keeps working after it is breached, and its thickness is an anode calculation.
Leave bosses bare where geometry needs bare metal. Gasket faces, sealing faces, threads and packing zones are not missed areas; they are specified exclusions.
A coating is a manufacturing process, not a material purchase. The valve that reaches twenty years is the one whose purchase order named the mechanism, the profile, the film, the test and the exclusions - and whose supplier could show a holiday test report against its serial number. The gate valve range and the wider technical article library carry the same information for the specific body styles and services discussed here.
Can I just specify "epoxy coated inside and out" for a seawater valve?
Not usefully. A seawater valve has two different problems on two different surfaces: the bore sees immersed chloride with a crevice at the seat pocket, and the exterior sees marine atmosphere plus water running down the insulation. "Epoxy coated" does not say which epoxy, at what thickness, tested how, or what was masked. Write the system instead: for the external, a C5-M three-coat system at 240 to 320 microns total; for the bore, either a fusion-bonded epoxy at 300 microns with a holiday test, or nothing at all if the body is duplex. Then add the exclusions - gasket faces, sealing faces, threads, packing zone. That single change removes most of the arguments that happen after delivery.
Is a thicker coating always better?
No, and in three common cases it is worse. A zinc-rich primer above roughly 150 microns loses its conductive network and mud-cracks as curing stresses accumulate; measured pull-off adhesion can fall from above 5 MPa to under 3. Fusion-bonded epoxy far above nominal becomes brittle and cracks at impact rather than absorbing it. Electroless nickel above roughly 75 microns builds residual stress and can crack, particularly if it is later baked. ISO 19840 already caps the acceptable maximum at three times nominal for this reason, and that maximum belongs in the specification as firmly as the minimum.
Why did an ENP-coated gate fail in 18 months when a painted one lasted five years?
Usually because the substrate was never suitable for a thin metallic coating. If the body was blasted to a 60 to 85 micron profile and plated to 25 microns, the peaks of the profile were bare from day one, and a magnetic thickness gauge averaged them into an acceptable reading. The second common cause is mechanism: nickel-phosphorus is a barrier with a cathodic pore, so in an abrasive slurry it is removed in a band, and in sour service the sulphide attacks the nickel directly. ENP is the right answer for a machined bore in mild chemical service, and the wrong answer for a cast surface in a particle-laden stream.
Can fusion-bonded epoxy be repaired on site after mechanical damage?
It can be patched, but not restored. Field patch compounds are epoxy putties that bond adhesively instead of fusing, so the patch has different adhesion, different permeation and no bond to the surrounding film at its edge - which is exactly where disbondment starts. For damage outside the seat region on a low-risk line, a documented patch is acceptable. For damage in the bore of a valve that will be opened repeatedly, the honest options are replacement or a return to the applicator for stripping and re-fusing. Do not accept a patch inside the seat region under any circumstances.
What does a 1000-hour salt spray result actually prove?
That the applicator controlled the process on the day the coupon was made. ASTM B117 runs at 35 C in continuous neutral salt fog with no ultraviolet light, no wet and dry cycling, no flow and no applied stress, so 1000 hours is 42 days against a design life measured in decades. It is a factory gate, not a life prediction, and comparing 500 against 1000 hours between suppliers tells you about bath chemistry rather than about your plant. Ask for cathodic disbondment at the design temperature, adhesion after immersion, impact or flexibility at service temperature, a reference installation in the same medium, and the holiday test report for your own valve serial numbers.
