Duplex 2205 Valve Bodies from China: Where the Grade Gets Lost

A chemical plant in Rotterdam replaced a set of 316 ball valves because they were pitting in a chloride-rich cooling loop. The replacements were specified as duplex 2205. Eighteen months later, two of them had cracked at the weld neck under hydrotest. The material certificate said 2205. The chemistry said 2205. What was wrong was that the bodies had been cooled too slowly after solution annealing, sigma phase had formed, and the impact toughness that duplex is bought for had quietly gone away.

Nobody lied on paper. The paperwork was correct and the part was not, and this is the normal shape of duplex problems from Chinese foundries.

What duplex actually gives you

2205 (UNS S32205, EN 1.4462) is roughly 22% chromium, 3% molybdenum, 5 to 6% nickel, with a deliberately balanced two-phase microstructure — around 50% ferrite and 50% austenite — which gives you yield strength about double that of 316L (around 450 MPa minimum) plus much better resistance to chloride stress corrosion cracking. 2507 (S32750) goes further on chromium and molybdenum for more aggressive service and costs meaningfully more.

The strength is the selling point, but the austenite-ferrite balance is what you are really buying, and that balance is created by chemistry plus heat treatment, not by chemistry alone. Get the cooling wrong and you still have the chemistry on the certificate and none of the properties.

Solution annealing is the whole game

Cast or forged duplex must be solution annealed at roughly 1020 to 1100°C and then cooled fast — water quench for castings and heavy sections, forced air at minimum for thin ones. Cool it in still air or leave it stacked and you land in the 700 to 950°C window long enough for sigma and chi phases to form at the ferrite-austenite boundaries. Toughness drops, corrosion resistance drops, and the part looks and measures fine.

Ask two questions of any Chinese duplex supplier. What temperature, and how did it cool. Then ask for the furnace chart. A serious foundry prints the chart from the furnace controller with the load chart, the thermocouple readings and the timestamp. A foundry that offers “we can write you a heat treatment report” is offering to type a document.

The portable check you can do on a visit is a ferrite scope reading on a few bodies. 2205 should read roughly 35 to 55% ferrite depending on the section and where you measure. Readings under 30% suggest excessive austenite from slow cooling or nickel over-addition; readings over 65% suggest the opposite and reduce toughness.

Where the grade gets lost

The most common substitution is not 304 for 2205, which nobody thinks they can get away with. It is a lower-alloy duplex — an S31803-era chemistry sold as S32205, where the chromium and nitrogen are at the bottom of the old range and the pitting resistance equivalent number lands a couple of points low. It passes an XRF check on chromium and nickel because the difference is small. It only shows up on a full chemical analysis of nitrogen and molybdenum, and it shows up in service as early pitting in the same chloride loop the 316 failed in.

So write the PREN into your specification, not just the grade name. S32205 typically lands at PREN 34 or above. A certificate with 33.1 is technically 2205 and is not what you wanted.

Cast versus forged bodies

Ball valve bodies up to about DN100 are often cast in CF8M-duplex equivalent grades, larger ones forged or fabricated. Cast duplex is harder to get right because of section thickness variation and the cooling problem above. For anything in sour or high-chloride service, insist on RT or at minimum PT on the body casting and 100% shell and seat testing on assembly, with the test pressure and holding time on the report.

Also ask about weld repairs. Duplex castings often need repair welding, and repaired areas must be re-solution annealed or at least locally heat treated and re-tested. A factory that repairs and paints over without re-testing is the one that gives you the Rotterdam crack.

Documents that are worth the paper

EN 10204 3.1 means the certificate is issued by the manufacturer’s own inspection department with test results traceable to the heat. 3.2 means an independent authorised inspector signed it. For duplex in critical service, 3.2 on the body casting is not excessive; it costs a few percent and it removes the “paper describes a different heat” risk almost entirely.

Pair it with a retained sample: ask the foundry to keep a test coupon from each heat for two years, stamped with the heat number, and to give you the coupon photos with the shipment. If a failure turns up in year three, you have something to analyse that is not the failed part.

None of this is exotic. All of it needs to be in the PO before the metal is poured, because duplex is not a material you fix afterwards.