AFFF to Fluorine-Free: Decontaminating the System, Not Just Changing the Foam

Changing the foam is the easy part

Draining an AFFF tank and refilling it with a fluorine-free concentrate takes an afternoon. Getting the PFAS out of the hardware that held the AFFF can take months, and that is where changeover budgets go wrong. A system that was drained, rinsed and refilled can still show PFAS in the new concentrate months later — at which point you have contaminated a fresh batch of foam and you hold a purchase order, a delivery note and a manufacturer’s declaration that all say otherwise. That is a worse position than not having started.

The mechanism has a name in the industry: rebound. Fluorosurfactant adsorbed onto pipe walls and tank surfaces, and absorbed into hose liners, gaskets, seals and bladder membranes, releases slowly back into whatever liquid is sitting against it. Your new concentrate is a surfactant solution. It is very good at picking that residue back up.

The regulatory driver, the retrofit options and the twelve-month roadmap are covered in our note on what the IMO 2026 PFOS ban means for Thai operators. This article is about the part that is routinely under-scoped: decontaminating the system you already own, proving it is clean, and re-validating that it still works on the new foam.

Why PFAS comes back after the system is drained

Rebound is not contamination arriving from outside. It is an equilibrium finding its level inside a system you thought you had emptied.

Where the residue actually sits

Two processes are at work, and they behave very differently when you try to clean them.

The first is adsorption onto rigid surfaces. The residue is a mixture of anionic, cationic and zwitterionic species, and they do not behave alike. The anionic perfluoroalkyl acids are the mobile fraction: they rinse out first and they are what a targeted analyte list looks for. The cationic and zwitterionic fluorosurfactants and precursors are the ones that sorb hard — to the oxide layer on carbon steel and aluminium, and far harder to rust, scale and tank-bottom sediment than to clean metal — and they are the fraction that rebounds longest. The loading follows the real internal area of a corroded, scaled pipe, not the nominal area on your isometric. It is stubborn, but it is reachable by aggressive flushing and by physical cleaning where you can get access.

The second is absorption into polymers, and it is the one that matters. Carried by the glycol ethers in the concentrate, the surfactant partitions into the bulk of nitrile and EPDM seals, diaphragms, rubber-lined hose, sight-glass tubing and bladder membranes. That is a volume reservoir, not a film. Removing it is a diffusion problem governed by wall thickness, not something a flushing cycle resolves. In practice you do not clean these parts to a defensible standard — you replace them.

Add physical hold-up: sludge and sediment in the tank bottom, dead legs, low points, strainer bodies, drain lines below the drain valve, and branches that never see flow between annual tests. Mark all three categories on the P&ID before you plan anything; the list is longer than most teams expect, and it includes the concentrate side of the proportioner and every small-bore run to foam chambers and rim-seal pourers.

One chemistry point that bears on the contract: the residue is more varied than the compound list on your old safety data sheet. Legacy AFFF and its fluorotelomer replacements carry precursor species that are not PFOS or PFOA today but oxidise into perfluoroalkyl acids later. That is largely the same strongly-sorbed fraction described above, which is why an early rinsate can read well while the hard material is still on the wall. A short targeted analytical method will report a comfortable number from a sample carrying a large precursor load.

Why the new concentrate makes it worse before it makes it better

Water is a poor extractant for a surfactant bound into a polymer. Fluorine-free concentrate is a good one. It is a surfactant-rich blend with its own solvent package; it wets the same surfaces, swells the same elastomers and lifts residue that rinsing left behind — into itself. The sequence follows from the chemistry: the final rinsate looks acceptable, the system is charged, and the concentrate sample taken weeks later is worse than the last rinsate was. Nothing has malfunctioned. The new foam has finished the cleaning for you.

Timing follows from that. Desorption from a surface is quick; diffusion out of the bulk of an elastomer is not, so the concentration in the new charge starts near zero, rises, and then flattens. A sample taken the day after refilling flatters the result, and a certificate based on it is worthless. If the acceptance test is “sample the new fill and confirm it passes” and nobody specified when, the contractor will sample early, pass and demobilise. Time is part of the method, and it belongs in the contract.

Why one water rinse is not a decontamination

Successive rinses behave like a dilution series against a slowly replenishing source. Each cycle removes a share of what is currently mobile, the bound phase then re-equilibrates, and you approach an asymptote that is not zero. Sample the rinsate between cycles and stop when the curve flattens, not when the number happens to look low.

Two variables do more work than the number of rinses. Contact time comes first: desorption is time-dependent, not shear-dependent, so a fill-soak-drain cycle with a long static dwell recovers more than a short high-velocity flush of the same water volume. Coverage comes second, and it is where most flushes quietly fail — water follows the path of least resistance, so dead legs, gauge and sight-glass legs, strainer bodies, pump casings and the concentrate side of the proportioner will not be swept unless you valve the flush section by section and record that each section actually took flow.

Treat cleaning as an escalating ladder, climbed component by component until the verification result holds. Each rung costs more money, more downtime and more waste than the one below it.

• Drain properly first — break flanges at true low points, clear dead legs, and remove sediment and sludge by hand or vacuum before any water is introduced. Anything you flush becomes liquid waste you then have to pay to destroy.
• Repeated water rinses with a defined dwell and a full drain between cycles, sampling after each so you can see whether you are still making progress.
• Elevated temperature or mechanical agitation, where the tank lining, coatings, gaskets and plastic components will tolerate it.
• Dedicated cleaning agents formulated to displace adsorbed fluorosurfactant, run to a written dose, temperature and contact time — and generating their own waste stream and their own rinse-out. Check availability before you build a programme around this rung: these are specialist products, regional supply is limited, and if one cannot be sourced for your system the ladder goes straight from repeated rinsing to physical cleaning and replacement. Establish that early, because it changes the cost and the schedule rather than just the method.
• Physical cleaning of what can be opened: wash-down of tank internals, sediment and scale removal, coating inspection, strainer and sight-glass renewal.
• Replacement of what cannot be verified clean: bladders, hoses, gaskets, seals, flexible connections, strainer baskets and, often, short runs of small-bore concentrate pipework.

Two things are worth noting on this ladder. Replacement is frequently the cheap option — a short run of small-bore pipe costs less to cut out and renew than to flush three times, sample three times and still argue with the laboratory report. And every rung multiplies waste volume, which is discussed below.

One further option deserves naming because operators use it and it must be agreed commercially rather than improvised: the sacrificial fill. Clean as far as is practical, charge with the new concentrate, leave it in service for an agreed period so that the foam strips the remaining residue, then drain that entire charge as PFAS-bearing waste before the final fill. It works, because it uses the best available extractant. It also means buying the concentrate twice. Decide that at budget stage, not after the first failed sample.

Where a step needs a manufacturer’s procedure, reference it by document number in the method statement. A flush invented on the day is not evidence. And plan the impairment realistically: if you need three cleaning cycles instead of one, the out-of-service period extends well past what a single-cycle plan assumed, and your insurer and your permit system both need to know that up front.

Your foam storage architecture decides the cost

Once you accept that the reservoir sits mostly in the polymers and the sediment, the cost of a changeover stops being a foam question. It becomes a question about how your system stores and meters concentrate — a decision made years ago, probably with no thought of PFAS at all.

Bladder tanks are the hardest case

In a bladder tank the concentrate sits inside a large elastomeric membrane within a pressure vessel and is displaced by firewater admitted to the annulus. The bladder is typically the largest reservoir of neat-concentrate contact in the system: a thin elastomer with a very large area relative to the volume it holds, in continuous soak contact with neat concentrate, under pressure, for the whole life of the installation. It is loaded through its thickness, not across its face.

It is also effectively un-cleanable. You cannot inspect a folded membrane without opening the vessel, you cannot scrub the inside of a bag, and cleaning aggressive enough to strip absorbed surfactant is generally aggressive enough to damage the membrane. More to the point, there is no sampling regime that proves the inside of a polymer wall is clean. The defensible route is replacement, and any proposal claiming otherwise should be challenged.

Replacement is a project, not a maintenance task: isolate and depressurise, open the manway, extract and dispose of the old bladder as bulky PFAS-bearing waste, clean the concentrate outlet and the concentrate fill connection, then reassemble to the manufacturer’s folding, seating and torque procedure with a leak and pressure check. The shell interior and the annulus pipework are water-side and are not a primary reservoir — the bladder is the boundary — but they are cleaned as well, for two specific reasons: extracting a loaded bladder contaminates them by handling, and any history of weepage or bladder failure has put concentrate on that surface. Check the vessel’s leak and top-up records before you fix that part of the scope. All of this is confined-space entry into a pressure vessel, with the permit, gas testing and rescue provision that implies, and depending on how the vessel is registered it may require re-inspection before return to service. Bladders are commonly made to order with long lead times, so the order sits ahead of the outage.

The associated proportioning hardware follows the same logic. A ratio controller, its metering orifice and its small-bore sensing lines held neat concentrate in passages no system flush will reach. They come apart and are cleaned as components, or they are replaced.

Two honest counterpoints. A new bladder is itself a complete renewal of the concentrate-wetted boundary — arguably a cleaner break than washing out a steel tank that keeps its original welds, pitting and scale — so the bladder architecture’s problem is cost, lead time and pressure-vessel entry, not verifiability. And a bladder that is already well into its service life is often at or near the end of its practical life anyway. If that is your situation, the changeover is a sensible moment to do work you were going to have to do. What you should not do is price the conversion as a foam swap and discover the bladder scope during the outage.

Atmospheric concentrate tanks are the easiest case

Systems built around a water-motor-driven positive-displacement proportioner — the FireDos GEN III family works this way — store concentrate in a vented atmospheric tank and meter it volumetrically from the firewater flow. There is no pressure vessel on the concentrate side and, critically, no elastomeric bladder in contact with neat concentrate. That single architectural difference removes the hardest part of the scope.

You can open the tank and inspect the wetted surface, see where residue has accumulated, wash it down physically rather than hoping a flush reached it, and photograph the result — which matters more than it sounds when you are building an evidence file. You can drain it fully by gravity from a real low point. You can sample it in five minutes at any stage of the rebound monitoring programme, rather than mounting an operation on a pressurised vessel. And if the tank proves too corroded or too contaminated to clean, replacing an atmospheric tank is a moderate line item with no pressure-vessel recertification attached. That is where this architecture actually wins — not because it is the only one that can offer a clean break, since a new bladder renews its own boundary too, but because the work is visually verifiable, cheap to escalate rung by rung, and cheap to abandon in favour of replacement the moment cleaning stops paying.

This should not be oversold. An atmospheric tank still needs cleaning, its suction line, strainer and pump internals still held neat concentrate, and the proportioner’s concentrate-side internals still need attention. Nor is an atmospheric tank automatically polymer-free: polyethylene and GRP tanks, and steel tanks with an epoxy, rubber or glass-flake lining, are themselves polymer volumes in continuous soak contact with neat concentrate, and they are treated the way a bladder is treated — establish the tank material and lining before you assume the reservoir argument applies to you. There is also a concession to make in the other direction: a bladder system has a smaller concentrate-side parts count, with no concentrate pump, no concentrate suction line and no recirculation or test loop, all of which are neat-concentrate items on a pumped atmospheric system. What changes is that the scope is small, accessible, visually verifiable and cheap to escalate. Nor does the architecture change the downstream half: foam solution mains, risers, chambers, deluge lines, monitors and delivery hose only ever saw dilute solution, but they are the largest area in the plant and rubber-lined hose remains a known reservoir. Prioritise by contact history — the heavy reservoirs held neat concentrate continuously; the wide, light ones are the solution side.

If a bladder-to-atmospheric conversion has ever appeared on your capital plan, the changeover is the cheapest moment it will ever be, because you are already paying for the outage, the drain-down, the waste route and the recommissioning. The proportioner selection criteria themselves are set out in our note on NFPA 11 compliance and foam proportioning in Thailand; decontamination cost belongs in that comparison and is usually left out of it.

Agree how clean is clean before anyone opens a valve

The most expensive failure in a foam changeover is not a bad cleaning method. It is a scope that says “flush and refill” or “decontaminate the system” with no analyte list, no method, no matrix, no sampling points, no timing and no threshold. There is then no test the work can fail and none it can pass, and the argument happens with the system already impaired and the money already spent.

• The analyte list and the method. A short list of named acids is the cheapest option and the least informative because it ignores precursors. Specify an oxidisable-precursor screen alongside the targeted list — it captures the precursor load in the same matrix and at broadly comparable limits, which is what makes it usable as an acceptance test. Treat total organic fluorine as a screening or corroborating measurement only: its reporting limit in a surfactant-rich concentrate is far higher than the targeted method’s, so written in as the pass/fail number it is either unachievable at any sensible threshold or so insensitive that it passes almost anything.
• The matrix. Rinsate is a process check; the commercially meaningful result is PFAS measured in the new concentrate in service. Concentrate is also a difficult matrix — the dilution needed to run it raises the reporting limit — so agree the achievable limit per matrix before you agree the threshold.
• The sampling points, marked on the P&ID rather than “to be agreed on site”: concentrate store at more than one depth, proportioner outlet, the most remote branch, each identified dead leg, and a discharge device.
• A baseline of the new concentrate taken from the delivery container, plus a baseline of the in-service AFFF before anything moves. Without them you cannot separate system rebound from supply-chain background, and that argument is unwinnable after the fact.
• The delayed rebound sample, and the rule that decides when it is taken. Do not write a single dated sample into the contract; write a schedule of samples at defined intervals after charging, and define the gate as a flattening criterion — consecutive results agreeing within the laboratory’s stated reproducibility, both below the threshold. Name who sets the interval and the minimum dwell: the concentrate manufacturer, and the insurer or purchaser who will judge the result. Both answers go into the contract, and the post-clean sample is only an interim one.
• The threshold as a number, stated with its method and the laboratory reporting limit. “Non-detect” means nothing without a detection limit.
• Liability and a stop rule: a priced schedule of rates for additional cycles, who owns the waste each cycle generates, and the point at which the operator accepts a documented residual. Retention held until the rebound test passes.
• The laboratory, its accreditation, its turnaround — and whether it will accept your matrix. For water and rinsate there is accredited capacity in Thailand, with ISO/IEC 17025 laboratories in Bangkok reporting large targeted PFAS panels in potable and non-potable water, so treat sample export as a fallback rather than an assumption. Neat concentrate is the harder ask, and the constraint there is the matrix rather than the country: the dilution a surfactant-rich concentrate forces will raise the reporting limit. Capability on the foam matrix itself does exist in Thailand — MTEC has analysed firefighting foam by solid-phase extraction and LC-MS/MS at parts-per-trillion detection limits, though as research work rather than an accredited commercial service. Confirm in writing that your chosen laboratory will accept neat concentrate, and what reporting limit it can actually achieve in it, before you agree the threshold.

Sampling is where results go wrong. PFAS work has cross-contamination controls stricter than most plant teams are used to, and a contaminated sampling train costs you a cleaning cycle you did not need. Keep fluoropolymer tubing, PTFE thread tape and fluoropolymer-lined caps out of the sampling train — not because PTFE leaches the mobile surfactants you are hunting, which it does not, but because the coatings, lubricants and handling associated with fluoropolymer items are a recognised cross-contamination pathway and because the material itself registers on a total-fluorine screen. Run field and equipment blanks alongside the real samples, and maintain chain of custody. If the contractor’s method statement does not mention blanks, the results will not survive scrutiny.

This article deliberately quotes no numeric threshold. Limits are moving, they differ between principals and insurers, and a figure lifted out of context into a specification is worse than none. Get the number from the party who will judge the result, and write their number into the contract.

Re-validate proportioning on the new concentrate

A proportioner set up and proved on AFFF is not automatically correct on a fluorine-free concentrate. The general requirement to re-prove the rate after any concentrate change is already covered in our notes on NFPA 11 compliance and foam proportioning in Thailand and on what Thai operators must test before specifying a fluorine-free foam. What changes on a changeover is more specific, and the first question is not viscosity at all — it is the dose. Moving from a 1% or 1×3 AFFF to a 3% or 3×3 fluorine-free product is a hardware change rather than a trim: a different ratio-controller orifice, a different proportioner pump ratio or gearing. It is also a threefold change in stored inventory before any application-rate question is opened. Alcohol-resistant products used on polar fuels typically dose at 3% on polar duty regardless of their hydrocarbon rate, so a dual-rate product does not release you from that check.

The next point lives in the approval documentation rather than on the datasheet. Check whether your unit’s approval states its flow range as a matrix rather than as a single pair of numbers — for water-motor and positive-displacement proportioners it commonly does, with the approved minimum given against inlet pressure and against the rheology class of the concentrate, Newtonian or pseudoplastic. Inlet pressure is the first-order variable and rheology the second. On a mid-size FireDos GEN III unit, for example, the published minimum rises by roughly half across the approved inlet-pressure range — of the order of 250 lpm at 5 bar to around 380 lpm at 16 bar — with a further step of only some five to eight per cent for a pseudoplastic agent at the same pressure. So the demand check must be done at the pressure the station will actually run at, not against a headline figure, and the rheology class is an adjustment on top of that rather than the main event. Two honesty points follow. These split minimums are published by the water-motor and positive-displacement proportioner makers, which means the turndown constraint lands on the atmospheric-tank architecture recommended above, not on the bladder and ratio-controller arrangements criticised earlier. And because the certificate is a matrix, read it in both directions: the top of the approved range can narrow too. List every credible demand from the smallest upwards — one hose line, one small bund, a single foam chamber, a partial-zone deluge — and check each against the approval certificate for your unit, at your inlet pressure, with your concentrate. This is rarely discovered in commissioning; it is discovered when an auditor reads the approval document.

Viscosity is the next question, and it is usually put backwards. Fluorine-free products are often more viscous than the AFFF they replace, but not always — several are Newtonian and low-viscosity, and a Newtonian fluorine-free product can buy back turndown at the bottom of the envelope on a tight retrofit. Where a product is thickened it is normally shear-thinning, and that is what makes it meterable at all: it carries a high nominal Brookfield figure but a much lower apparent viscosity at the shear rates inside a metering orifice or a proportioner pump, so a Brookfield number on its own tells you nothing. Ask for the low-shear and the high-shear values and match them to the proportioner’s published rheology class. And if your incumbent is already an alcohol-resistant AFFF, it is already pseudoplastic and already in the high-viscosity class, so on that count the changeover costs you no turndown at all. Viscosity also varies strongly with temperature — which in a Thai foam house means the concentrate you commission on a cool morning is not the fluid being pumped at three in the afternoon in April.

How much this matters depends on architecture. Inline eductors and around-the-pump arrangements are the most exposed, because induction depends on drawing concentrate up a pickup line. Balanced-pressure and ratio-controller systems meter through an orifice sized for the original fluid: at the same differential, a more viscous product passes less flow and the mix ratio drops below specification, quietly, because nothing sounds or looks different. Correcting it means a re-sized orifice or metering insert to the manufacturer’s calculation. Positive-displacement and water-motor-driven units are far less sensitive because they displace a fixed volume per revolution, but suction-side pressure drop, strainer mesh, available NPSH and static head at low tank level all get worse and must be re-checked.

Then test it, rather than calculating it. Verify accuracy at the minimum, mid and maximum credible demand, not only at design flow, and re-run the concentrate-side hydraulic calculation with the new viscosity and specific gravity. One field detail wastes more days than it should: refractive-index and conductivity calibration curves are concentrate-specific, so the old AFFF chart will give confident, wrong numbers. Obtain the new product’s own calibration data first, and for conductivity re-calibrate against the actual firewater — a real consideration at coastal terminals on brackish supply. Inspect the strainers after the first run; residual sludge lifted by the new concentrate is exactly what blinds a strainer in the first hour.

Check seals, gaskets and linings against the new chemistry

Fluorine-free concentrates are a different chemistry, not AFFF minus the fluorine. Different surfactant packages, different solvent and glycol content, sometimes polysaccharide thickeners, often a different pH. An elastomer that was stable for two decades may swell, harden, lose plasticiser or take a permanent set on the new fluid, and the leaks appear weeks after handover rather than on the day of the test.

Build a wetted-materials inventory during the baseline survey and get written confirmation item by item: tank material and internal lining, pipe and fittings, valve seats and diaphragms, pump seals and seal faces, flange gaskets, hose liners, flexible connections, sight glasses, level float and gauge components, and strainer media. Ask specifically about galvanised steel and copper alloys, which some concentrates do not tolerate for long-term storage. Ask the concentrate manufacturer for the materials list and the equipment manufacturers which elastomer specification they support with that concentrate — where the two answers disagree, that disagreement is your finding, and it needs resolving before commissioning.

Sequence the work correctly, because it is easy to get backwards. Strip the components you intend to discard before cleaning begins, so cycles are not spent on parts destined for the skip, and fit the new elastomers only after the system is verified clean. Flushing PFAS-laden rinsate past fresh seals contaminates them and defeats the point. Keep the material certificates; they belong in the evidence pack.

One point of vocabulary that increasingly appears in tender questions: PTFE tape and fluoropolymer-lined components are polymeric fluorine, not the mobile fluorosurfactants that cause rebound. They are not the source of your contamination, though they will register on a total-fluorine screen and they genuinely do cause problems in the sampling train. Keep the engineering question separate from the analytical one and answer both explicitly. If you are still choosing between products, the comparison points are set out in what Thai operators must test before specifying a fluorine-free foam.

Old concentrate and rinsate are PFAS-bearing waste

The old concentrate, every litre of rinsate, spent cleaning solution, the removed bladder, hoses, gaskets, strainer elements, tank sediment, spent adsorbent media and contaminated PPE are all PFAS-bearing waste. This is frequently among the largest lines in the whole changeover and the one most often missing from the first budget.

The arithmetic that surprises people is the rinsate, not the concentrate. The concentrate volume is known; each additional cleaning cycle produces roughly another system volume of contaminated liquid, so the cheap-sounding answer to a failed verification — rinse it again — multiplies the waste bill every time it is used. Estimate rinsate volume during design, and put the number of included cycles and the ownership of their waste into the contract.

One damaging field error is also the most intuitive: sending rinsate to the site’s oily-water separator or effluent plant because that is where wash water normally goes. Conventional treatment does not destroy these compounds; it moves them into sludge and effluent, contaminates the treatment train, and converts a contained, characterised liquid into a plant-wide discharge-consent problem. Rinsate goes to dedicated containment on a route agreed in advance. Do not open a drain valve until the receiver is contracted, the manifest route agreed and interim bunded storage arranged in labelled, compatible containers — drums standing in a corner while somebody looks for a destination are a liability, not a plan. And never use the same pump, hose and road tanker to remove the old AFFF and then deliver the new concentrate; it undoes the whole programme in ten minutes and is invisible until the first sample.

Destruction routes are limited, and licensed capacity is often the binding constraint on the programme. High-temperature incineration is the usual accepted route; adsorption on activated carbon or ion-exchange media reduces the volume needing destruction but transfers the PFAS to a spent medium that itself requires disposal. Put the question of what the receiver will actually do with the material in writing rather than assuming.

On the Thai regulatory side, keep the framing honest. PFOS and related substances are controlled internationally as persistent organic pollutants, and operators are increasingly asked — by insurers, corporate parents and international customers — to handle foam waste through the licensed hazardous-waste route with full manifest documentation, whether or not a PFAS-specific rule is cited on the paperwork. Exactly how that maps onto the Hazardous Substances Act framework and the Department of Industrial Works notifications for your specific stream is a question for your licensed contractor and your environmental adviser, before the shutdown rather than during it.

One loop to close: the first live discharges of the cleaned system may themselves be PFAS-bearing, particularly if the bund or retention basin holds AFFF residue of its own. Plan containment for commissioning discharges in the same waste plan as the rinsate.

Fire performance does not transfer automatically

A clean system that doses correctly can still fail its design intent. AFFF worked partly by spreading a thin aqueous film ahead of the blanket; fluorine-free foams do not form that film, so suppression rests wholly on the blanket — its expansion, drainage, mechanical stability and burnback resistance. Design assumptions inherited from the AFFF basis need re-deriving rather than transcribing.

• Application rate and duration, taken from the listing or approval of the specific new concentrate with the specific discharge devices installed — not from the AFFF design basis and not from a generic code minimum.
• The inventory consequence, which has two drivers rather than one. The dosing percentage comes first: a move from a 1% to a 3% product multiplies stored volume before anything else is considered. The application rate comes second — if the rate rises and the duration holds, required storage rises again with it. Real consumption on alcohol-resistant grades also commonly runs well above the like-for-like AFFF figure, so the tank that comfortably held the AFFF supply frequently does not hold the new one. Check this early; a tank change is a different project from a foam change.
• Discharge device compatibility. Many fluorine-free concentrates depend far more on aspiration. Non-aspirating deluge heads and monitors that performed acceptably on AFFF may not be listed with the new product, which can turn a foam swap into a nozzle and chamber replacement.
• Subsurface and semi-subsurface injection, which needs written confirmation from the manufacturer that the product is listed for injection on your fuel. Do not assume a topside listing carries over; if it does not, you have a system redesign, and you need to know at concept stage.
• Fuel slate and water. Polar solvents and ethanol-blended fuels need alcohol-resistant grades, usually at different rates. Verify expansion and drainage with the water you will actually use, at the temperature you will actually see — results on fresh water in temperate conditions do not transfer to a coastal terminal on brackish supply.
• Hydraulic consequences. A higher application rate pulls on solution demand, pump capacity and water supply duration. Re-run the calculation rather than assuming headroom.
• Third-party evidence. Alongside the listing, operators increasingly ask for large-scale results against the LASTFIRE protocols, particularly for storage-tank applications. Where the risk justifies it, ask for test data with your discharge device, not only the general listing.

If this section changes your discharge devices or your water demand, that is not a failure of the project. It is the project doing its job before an incident does it for you.

A practical sequence for a defensible changeover

1. Baseline the system. Mark up the P&ID with every wetted component, its material and its contact history; sample the in-service concentrate and the identified dead legs; pull the pressure-vessel records, bladder age and existing approval documents. This is your only chance to know what you started with.
2. Select the new concentrate and re-derive the design. Dosing percentage, viscosity and rheology class, elastomer compatibility, application rate, duration, inventory and tank capacity all follow from that choice, and every credible demand must be checked against the approved minimum flow at the inlet pressure the station will actually run at. This step tells you whether you are doing a foam swap or a system rebuild — and it comes before the cleaning scope is written.
3. Decide clean-or-replace component by component, not for the system as a whole, and settle the storage architecture question now.
4. Write the acceptance criteria into the contract: analyte list, method, matrix-specific reporting limits, sampling points, threshold, the rebound sampling schedule and its flattening rule, the stop rule, and who re-cleans at whose cost.
5. Contract the waste route in writing, sized on rinsate as well as concentrate. Nothing is drained before this is in place.
6. Plan the impairment: interim cover, permits spanning the whole outage including the rebound dwell, and notification to the insurer and the authority having jurisdiction, with at least one repeat cleaning cycle allowed for.
7. Drain and recover the old concentrate to dedicated containment, using transfer equipment that will never touch the new product.
8. Strip the sacrificial components — bladder, hoses, gaskets, seals, flexibles, strainer elements, sight glasses — before cleaning starts.
9. Clean iteratively to the manufacturer’s written procedure, recording volumes, temperatures, dwell times, agents and results for every cycle, escalating a rung when the rinsate curve flattens above target.
10. Refit with new elastomers of the confirmed grade once the system is verified clean, then charge with the new concentrate, retaining a sample from the delivery container first.
11. Run the rebound dwell and sample the concentrate in service to the agreed schedule, until consecutive results flatten below the threshold. That flattening — not the post-clean result, and not a single dated sample — is the acceptance gate.
12. Recommission: proportioning accuracy at minimum, mid and maximum demand using the new concentrate’s own calibration data, foam quality at the discharge devices, hydraulic verification, and strainer inspection after the first run.
13. Update the paperwork and keep sampling. Design basis, hydraulic calculation, ITM procedures, SDS, pre-fire plan and operator training, plus a scheduled follow-up sampling programme through the first year of service with a written decision rule for a failing result.

The evidence pack an auditor or insurer will ask for

When the changeover is challenged — by an insurer’s engineer, a client audit, a regulator, or your own management after an incident — the question will not be whether you cleaned the system. It will be whether you can show it. Assemble this as the work proceeds; reconstructing it afterwards is expensive and unconvincing.

• Marked-up P&ID with the wetted-component inventory and each item’s disposition — cleaned, replaced or retained, with the reason
• Baseline analysis of the old concentrate and system residues, with method, reporting limits and chain of custody
• The signed decontamination specification: threshold, analyte list, method, sampling points, timing
• Cleaning records as executed: cycles, volumes, contact times, temperatures, agents and the procedure document referenced
• Verification results for every cycle and every round, with field and equipment blanks
• The delayed rebound test reports, from the same accredited laboratory, with the sampling schedule and the flattening criterion the acceptance gate was judged against
• Component replacement records with material certificates, and the manufacturer’s written compatibility statement
• New concentrate documentation: listing and approval certificates, batch certificate of analysis, SDS, retained delivery sample, and the fluorine-free declaration with its stated detection limit
• Waste documentation: characterisation, receiver licence, written acceptance, manifests and destruction certificates for every stream including the removed bladder
• Re-derived application rate, duration and inventory calculation with the tank capacity check at the new dosing percentage
• The approval certificate showing the flow range against inlet pressure and rheology class, and the demand list checked against it at the station’s actual inlet pressure
• Proportioning accuracy records at minimum and maximum credible demand, naming the calibration data used, plus foam quality results at the discharge devices
• Impairment correspondence with the insurer and the authority having jurisdiction, and its formal closure
• Revised ITM procedures, operator training records, and the follow-up sampling schedule with results to date

One note on the manufacturer’s fluorine-free declaration. It certifies the product as it left the factory, at a stated detection limit. It says nothing about what that product will contain after months in your system. Those are two different claims, and conflating them is how a well-documented project becomes undocumented at exactly the point it matters. If a proposal you are evaluating does not produce this pack, it is not a changeover project — it is a foam delivery with a flush attached, and the residual risk stays with you.

Where to start

Two things are worth doing first and cost very little. Take a baseline analysis of what is in your system now, while you still can. And establish honestly which storage architecture you have — because that single fact decides whether decontamination is a tank-cleaning exercise or a pressure-vessel project with a bladder lead time in front of it. If the concentrate sits in a bladder, price replacement now and put it in the capital request. If it sits in an atmospheric tank, price a new dedicated tank and treat a clean break as the base case rather than the upgrade.

Working with SATU on a changeover

A changeover is usually quoted as three separate things: the proportioning hardware, the concentrate, and the cleaning. Splitting it that way is a large part of why the decontamination scope ends up under-specified — the party who writes that scope is rarely the party who has to live with the verification result. SATU works the three as one problem.

Decontamination. SATU is Engiva’s preferred partner in Thailand for PFAS decontamination services. That arrangement covers decontamination of PFAS-impacted firefighting systems, vehicles, tanks and pipework; engineering and advisory support on PFAS transition and waste handling; and training and knowledge transfer for the people who will operate the system afterwards. In practice it means the wetted-component survey and the clean-or-replace decision, the cleaning method statement and its escalation ladder, the verification protocol and sampling schedule, and the waste routing — settled before work starts rather than argued about once the system is already impaired.

Hardware and concentrate. SATU is a FireDos Authorized Distributor in Thailand and supplies SOLBERG fluorine-free foam concentrate in Thailand. The most useful early work on this side is the re-derived design: checking your existing proportioner’s approved flow range at the inlet pressure your station actually runs, against the rheology class of the concentrate you are considering, and telling you before you tender whether your concentrate tank is still big enough at the new dosing percentage.

What we will not do is verify our own work. The analysis should come from an independent accredited laboratory appointed by you, and your contract should say so. A contractor who cleans the system and then certifies it clean has produced a document, not evidence. That separation is what makes the result worth anything when an insurer, a client auditor or a regulator asks.

If you are budgeting a changeover, drafting the TOR for one, or reviewing a proposal that says flush and refill, we are happy to review the specification against the points above.

Frequently asked questions

We drained the system and refilled with fluorine-free foam. Why did the concentrate test positive for PFAS months later?

Because draining removes the bulk liquid, not the residue. Fluorosurfactant adsorbs onto pipe and tank surfaces and absorbs into the bulk of hose liners, seals, gaskets and bladder membranes, then desorbs slowly back into whatever liquid is against it. The new concentrate makes this worse before it makes it better: it is a surfactant- and solvent-rich blend and a far more effective extractant than rinse water, so it strips residue that flushing left behind, into itself. The concentration rises and then plateaus, which is why a sample taken on the day of charging reads low and one taken later does not. Acceptance testing must therefore include delayed samples of the new concentrate in service, taken to an agreed schedule and judged on whether consecutive results have flattened below the threshold.

Can a bladder tank be cleaned instead of having the bladder replaced?

In practice, no — not to a standard you can defend. The bladder has a very large area for the volume it holds, has been in soak contact with neat concentrate under pressure for years, and is loaded through its full thickness rather than across its face. Removing absorbed material from the bulk of a polymer is a diffusion problem governed by wall thickness rather than something a flushing cycle resolves, you cannot inspect or scrub a folded membrane without a pressure-vessel entry, and cleaning aggressive enough to strip the surfactant is usually aggressive enough to damage the bladder. Budget for replacement, order early because bladders are commonly made to order with long lead times, and plan the work as confined-space entry with reassembly and pressure checks to the vessel manufacturer’s procedure.

Why is an atmospheric concentrate tank cheaper to decontaminate than a bladder tank?

Mainly because the work is visible and cheap to escalate. There is usually no elastomeric bladder in neat-concentrate contact, so the largest reservoir of neat-concentrate contact does not exist — though a polyethylene or GRP tank, or a lined or coated steel one, is itself a polymer reservoir and is treated like a bladder, so establish the material first. You can open the tank, inspect it, wash it down and photograph the result rather than relying on sampling alone to prove a closed vessel is clean. And if it turns out too corroded or too contaminated to clean, replacing an atmospheric tank is a realistic option with no pressure-vessel recertification attached, so the cleaning can be abandoned for a clean break at moderate cost. A new bladder also delivers a clean break on the storage boundary — the bladder architecture’s difficulty is cost, lead time and pressure-vessel entry, not verifiability. The downstream solution pipework is identical in both architectures and still needs a plan.

Do we have to re-test proportioning accuracy if we keep the same proportioner?

Yes, and two things change even when the hardware does not. First, check whether the new product is used at the same percentage as the old one: a move from 1% to 3% is a metering-hardware change and a threefold inventory change, not a trim. Second, pull the approval certificate for your unit. On water-motor and positive-displacement proportioners the approved flow range is commonly stated as a matrix, against inlet pressure and against the rheology class of the concentrate, and the approved minimum rises substantially with inlet pressure and by a smaller step again for a pseudoplastic agent — so a small demand that was acceptable on AFFF may fall below the approved minimum at the pressure your station actually runs at. Check every credible demand from the smallest upwards, then test rather than calculate: accuracy at minimum, mid and maximum credible demand, using the new concentrate’s own calibration data instead of the old AFFF chart.

What is the most common contractual failure on an AFFF changeover?

Not defining how clean is clean before work starts. A scope that says flush and refill, with no analyte list, no method, no matrix-specific reporting limits, no sampling points, no delayed rebound test and no allocation of responsibility for further cycles, leaves no test the work can pass or fail — and the argument happens with the system already impaired. Fix the analyte list including an oxidisable-precursor screen alongside targeted analysis, the sampling points on a P&ID, the rebound sampling schedule and the flattening rule that closes the acceptance gate, the threshold as a number with its method and reporting limit, a priced schedule of rates for additional cycles, and the point at which a documented residual is accepted.

What do we do with the old AFFF and the rinse water?

Treat all of it as PFAS-bearing waste: concentrate, rinsate, spent cleaning solution, the removed bladder, hoses, gaskets, strainer elements and tank sediment. None of it goes to the site drain, the oily-water separator or the wastewater plant — conventional treatment does not destroy these compounds and spreads them into sludge and effluent. Size the estimate on rinsate volume, which grows with every cleaning cycle and is usually far larger than the concentrate volume, and secure a licensed contractor, a named destruction route and written waste acceptance before anything is drained. Operators are increasingly asked to handle this stream through the licensed hazardous-waste route with full manifests; confirm the classification and forms for your specific stream with your contractor and environmental adviser.

Talk to SATU about your foam changeover