Quick answer. Polyol supply allocation in 2026 is driven by propylene oxide capacity discipline in the US Gulf and Antwerp, not by raw demand growth, so allocation letters cap tonnage per quarter rather than raise price alone. The workable response is a qualified second source running in parallel at 10-20% of volume before you need it. Budget 8-12 weeks: 2 weeks for spec and TDS/SDS comparison, 3-4 weeks for lab handmix and machine trial, 4-6 weeks for line validation and REACH/customs paperwork.
Most foamers discover their allocation ceiling the week they need to fill a 40-ton order. By then the alternative supplier qualification clock — which is fixed by your own QA protocol, not by the supplier — has already run out.
This is written for purchasing and technical managers at rigid panel, spray foam, flexible slabstock and CASE elastomer plants who need a defensible backup sourcing plan on paper before the next quarter's allocation notice.
Why allocation is structural in 2026, not a spot squeeze
Polyether polyol economics run off propylene oxide (PO) and, for flexible grades, a KOH- or DMC-catalysed glycerol or sorbitol initiator. PO capacity additions since 2023 have been concentrated in Asia, while several older chlorohydrin-route PO lines in Europe have been idled or rationalised on energy cost.
The consequence for a buyer is specific: your incumbent may hold price flat and still cut your monthly nomination by 15-30%, because the constraint sits upstream of the polyol blender.
Two secondary pressures make this worse. HFC blowing agent phase-down under the US AIM Act and the EU F-Gas Regulation forced rigid formulators to requalify systems on hydrocarbons (cyclopentane, ~30-43 mL/g typical loading in a preblended system) or HFOs, which reshuffled which polyol backbones are in demand. And flame retardant substitution under REACH restriction reviews has narrowed the approved additive list for some flexible foam grades.
See the ECHA restriction and authorisation lists for the current status of substances used in PU systems at echa.europa.eu before you assume a backup grade is drop-in legal in the EU.
What an allocation letter actually restricts
Read the letter for three numbers: the baseline period used to compute your allocation percentage, the review frequency, and whether allocation is by grade or by total tonnage.
Grade-level allocation is the dangerous one. A plant running four systems can be at 100% on a commodity 3000 MW triol and at 60% on a specialty polyester polyol, which stops the same production line.
If the baseline is a 2024-2025 average and your volume has grown, the allocation percentage is applied to a number that no longer reflects your order book. Push for a rebaselining request in writing in the same week you receive the notice; it is a documentation exercise your incumbent handles routinely and almost nobody files.
Spec-matching: what has to match, and what does not
The mistake that kills a second-source trial is chasing an identical TDS. You do not need identical; you need equivalent reactivity and equivalent end-part properties. Build the comparison on the parameters below rather than on a marketing datasheet.
| Parameter | Test method | Match tolerance for drop-in | Why it moves your line |
|---|---|---|---|
| Hydroxyl value (mg KOH/g) | ASTM D4274 | ±2% of incumbent | Drives isocyanate index; ±5% shifts index enough to change compression set |
| Water content (%) | ASTM D4672 / Karl Fischer | ±0.03 absolute | CO2 generation, cream time, and core density on water-blown grades |
| Viscosity at 25 °C (mPa·s) | ASTM D4878 | ±10% | Metering pump calibration and mix quality at 1:1 or 100:105 ratio |
| Acid number (mg KOH/g) | ASTM D4662 | ≤ incumbent | High acid retards amine catalysis, lengthens gel time unpredictably |
| Functionality / initiator | Supplier declaration | Must match (e.g. glycerol f=3 vs sucrose f≈4.5) | Crosslink density; a sucrose-based rigid polyol will not substitute a glycerol triol in flex |
| Unsaturation (meq/g) | ASTM D4671 | ≤ incumbent | Monol content lowers ILD and hurts flexible foam load bearing |
| Reaction profile (cream/gel/tack-free, s) | Handmix, 500 g cup | Each within ±10% | Direct read on whether the line speed survives the change |
| Core density (kg/m³) | ISO 845 / ASTM D1622 | ±3% | Cost per board and thermal spec |
| Compressive strength (kPa) | ISO 844 / ASTM D1621 | Meet or exceed spec floor | Panel and slab-jacking acceptance |
| Lambda at 10 °C (W/m·K) | ISO 8301 / ASTM C518 | ±0.001 | Declared R-value on the finished panel |
| Flame performance | FMVSS 302 / EN 13501-1 / ASTM E84 | Must re-test, never assume | Certification is on the system, not the polyol |
Method references for the mechanical and thermal tests are published by ASTM at astm.org and by ISO at iso.org. Cite the method number in your purchase specification; suppliers who report against a different internal method are giving you numbers you cannot compare.
The three parameters buyers usually forget
Unsaturation, acid number and residual catalyst are absent from most abbreviated TDS sheets and are exactly where two nominally identical 3000 MW triols diverge. Ask for them by test method. A supplier who cannot produce ASTM D4671 unsaturation data on a flexible polyol is reselling, not manufacturing.
An 8-12 week qualification sequence that survives an audit
Weeks 1-2 — paper qualification. Collect TDS, SDS in the destination-market format, CoA from three separate production lots, REACH registration number or the importer's Only Representative details for EU delivery, and the HS code the supplier will declare. Three lots, not one, is the point: lot-to-lot hydroxyl value spread tells you more about the plant than any single CoA.
Weeks 3-4 — handmix. 500 g cup shots at your production index, ambient 23 °C and at your worst-case shop floor condition. Record cream, gel, tack-free, rise height and free-rise density. Run the incumbent side by side on the same day with the same isocyanate lot — a comparison against last quarter's notebook is worthless.
Weeks 5-8 — machine trial. One shift on the real high-pressure machine at the real throughput. Watch for the failure modes that only appear at scale: pump cavitation from a higher-viscosity B-side in a cold day tank, filter blinding from particulate, and surfactant-driven cell coarsening that never shows in a cup.
Weeks 9-12 — validation and paperwork. Aged property testing (dimensional stability per ISO 2796 at -20 °C and +70 °C for rigid, 70/95 humid ageing for flex), fire re-test on the finished construction, and updated hazard communication. US plants should confirm the SDS and workplace labelling meet the OSHA Hazard Communication Standard as revised — current requirements at osha.gov/hazcom. Isocyanate exposure controls apply to the A-side, not the polyol, but the qualification file is audited as one package.
How to structure the dual-source split
Running a backup at 0% of volume is not a backup; it is a phone number. The qualification lapses, the price is untested and the supplier gives your slot to a buyer with a running order.
The split that works for mid-size foamers is 80/20 by volume with the second source taking one full grade rather than 20% of every grade. One grade at full volume gives you real lot history, real freight cost data and a supplier who treats you as an account.
Set the MOQ conversation early. A realistic ladder from a manufacturer-direct source runs sample 1-10 kg, trial order 200-500 kg, standard order 1 tonne, and full-container economics beyond that. If a quoted MOQ starts at a full container with no trial tier, that supplier is not set up for qualification work.
Where a China manufacturer-direct source fits
The relevant question is not geography but whether you are buying from the reactor owner or from a layer of traders. Manufacturer-direct gives you three things a trader cannot: the actual lot genealogy behind a CoA, formulation adjustment on the combined polyol (catalyst package, surfactant level, flame retardant loading) rather than take-it-or-leave-it drums, and a technical contact who can change the recipe when your line runs 4 °C colder in winter.
SPC Foam Material manufactures combined polyol systems for rigid panel, spray, flexible slabstock and elastomer applications, and supplies the catalyst, silicone surfactant and flame retardant components separately where a customer prefers to blend in-house. Grade-level specs and TDS are on the polyol product range, and custom system development starts from your current TDS and reaction profile rather than from a catalogue substitution.
Landed-cost math matters more than unit price under allocation. Ocean freight from East China to US East Coast or North Europe, duty at your HS classification, 30-45 days transit and one month of safety stock together add a working-capital line that a spot-market comparison hides. Model it before the trial, not after.
Due diligence that separates a plant from a broker
- Ask for the reactor capacity and the number of reactors — a real answer arrives in tonnes per year, not adjectives.
- Ask which parameters are tested in-house versus sent out, and to which method number.
- Request a video walk of the QC lab showing a Karl Fischer titrator and a viscometer in use.
- Ask for two references in your application, not in your country.
- Check that the SDS was authored for your destination market, not machine-translated from a domestic version.
Contract terms worth arguing about
Under allocation, the terms that protect you are volume commitment with a tolerance band, and a price formula tied to a published PO or propylene index rather than to quarterly negotiation. A ±10% volume band around a monthly nomination gives both sides planning room without a take-or-pay trap.
Also fix the specification in the contract by test method and limit, and state that a lot outside spec is replaceable at supplier cost including freight. Verbal spec agreement is where second-source relationships fail in month five.
Peer-reviewed work on polyurethane raw material substitution and structure-property relationships — useful when you have to justify a backbone change to a technical director — is indexed at sciencedirect.com.
FAQ
Q: How long does qualifying an alternative polyol supplier actually take?
Eight to twelve weeks for a like-for-like grade where hydroxyl value, functionality and viscosity are within the tolerances above. Add four to eight weeks if the fire classification on the finished construction has to be re-tested, since third-party fire labs are the usual bottleneck.
Q: Will a second-source polyol change my foam density?
It can, and water content is the usual reason. A 0.05% absolute difference in water on a water-blown rigid system measurably shifts free-rise density, so match water within ±0.03 and re-check core density per ISO 845 on the machine trial rather than trusting the handmix.
Q: Do I need to re-run flame testing if only the polyol changes?
Yes. Classification under EN 13501-1, ASTM E84 or FMVSS 302 is held on the tested system or assembly, not on any single component. A changed polyol backbone or flame retardant loading invalidates the existing report even when the datasheet numbers look equivalent.
Q: What paperwork does an EU importer need for a non-EU polyol?
A REACH registration covering your import tonnage band — held either by the non-EU manufacturer's Only Representative or by you as importer — plus an SDS in the language of each member state where the substance is placed on the market. Confirm which party carries the registration in writing before the first container ships.
Q: Is it cheaper to hold more inventory instead of dual-sourcing?
Rarely, for two reasons. Polyol shelf life is typically 6-12 months in sealed drums and shorter for combined systems containing amine catalyst, and stockpiling does not restore supply once your allocation percentage is cut. Inventory buys weeks; a qualified second source buys capacity.
Q: Can a supplier match my existing system without seeing my formulation?
Yes, if you provide the TDS of the current combined polyol, the reaction profile (cream, gel, tack-free), the target core density and the isocyanate index. A capable formulator works backward from reactivity and end properties; nobody needs your proprietary recipe to build an equivalent system.
