The tank is too big to heat treat, and that decides the plate

A molten salt storage tank for a concentrating solar plant is typically 20 m across and 12 m high. Nothing that size goes into a furnace, and no fabricator will build a temporary one on site. Whatever the welding does to the heat affected zone stays for thirty years. That constraint, more than the 565 C operating temperature, drives the plate specification.

What the salt actually is

The mixture in most commercial plants is a eutectic of sodium nitrate and potassium nitrate, roughly 60/40 by weight, which melts at about 220 C. The cold tank is held near 290 C, not because the process needs it but because the margin above the freeze point has to survive a cold night and a sluggish trace heating circuit. The hot tank runs at 565 C. Between the two, the salt is the plant's entire storage capacity: a 20 m tank filled to 10 m holds about 3,100 m3, near 6,000 tonnes at 1.9 t/m3.

Freeze protection is therefore a piping problem before it is a metallurgy problem: every line that can hold salt is traced and insulated, and every pocket that can trap it is a future plug. The metal minimum temperature is set by the salt, not by ambient.

Ordinary stainless is the wrong first instinct

The instinct is to reach for 316L because it is the plant standard, or 304 because 565 C is below the creep range. Both miss the point. The salt is not pure. Industrial nitrate carries chloride, perchlorate and moisture, and those impurities concentrate at the hot wall as water boils off and salt decomposes. Chloride plus tensile stress plus temperature is the mechanism behind chloride stress corrosion cracking, and a welded tank shell is full of tensile stress.

That is why the salt certificate is a material selection document: its limits on chloride, perchlorate and moisture are usually tighter than anything in the metal specification.

347H and the anneal that cannot happen

347 is stabilised with columbium and tantalum, which hold carbon more tightly than chromium does. The carbides precipitate inside the grains rather than at the boundaries, so the chromium stays in solution. The grade is recommended for intermittent heating between 427 C and 899 C, or for joints that cannot be given a post-weld anneal afterwards - an exact description of a field-erected tank.

The H suffix is a carbon range, not a strength claim. 347H seamless pipe carries 0.04 to 0.10 per cent carbon, 17 to 19 per cent chromium and 9 to 13 per cent nickel, with 515 MPa tensile and 205 MPa yield, supplied annealed and pickled. The higher carbon is what holds creep strength at 565 C; the stabiliser is what allows the weld to be left alone.

For the salt piping, that combination is the whole argument. A 347 seamless pipe to A213 or SA312 can be welded in the field, at night, in a wind, without a heat treatment tent, and still pass the corrosion review afterwards.

321H is the other answer, not a substitute

Titanium does the same job as columbium in a different way. 321/321H plate is titanium stabilised and rated for service up to 1,650 F, with better creep and stress rupture behaviour than the unstabilised grades. The plate route is A240, thickness from 4 to 100 mm, widths to 3,500 mm and lengths to 6,000 mm, with proof stress quoted at 200 to 300 MPa and tensile at 500 to 700 MPa.

The choice is usually made by the fabricator rather than the metallurgist: titanium stabilisation needs a higher solution temperature, so the weld procedure has to be qualified accordingly. Where the shop already has a 347 procedure, 347H is cheaper even if the plate costs more.

The receiver is a different problem

Receiver tubing sees concentrated flux on one side and salt on the other, at the top of the salt range. 310/310S pipe is the usual starting point: 24 to 26 per cent chromium with 19 to 22 per cent nickel, good for continuous service at 2,000 F and intermittent service at 1,900 F, with a low coefficient of expansion, which limits warping in hot service. One condition on the page is worth repeating: the oxidation resistance assumes no reducing sulphur gases.

Thin tube and thick plate do not belong in the same line of the enquiry, even when both are 300 series.

The cold end and the salt mains

Below 400 C the argument changes. The cold tank, the drain tank and most of the salt mains can be 304/304L pipe or 316L, in 3 to 2,200 mm OD with 0.08 to 170 mm wall. Where the header size goes beyond that, large diameter thick wall stainless pipe runs to 2,500 mm.

Expansion in the mains is taken by geometry rather than by anchors: the line is routed with bends and offsets so that the growth between cold and hot is absorbed by the pipe flexing, and the supports are set to allow it. A 12 m shell growing from 290 C to 565 C rises about 60 mm at 18 x 10 to the minus 6 per K, and every nozzle, tapping and support shoe has to allow for it.

Fabrication is where the specification is won

Tank plate arrives flat and leaves as a shell. Cutting has to leave a weld preparation that can be closed without a wide gap, bending has to produce a consistent radius across the course, and welding has to be done to a procedure qualified on the same grade and thickness. There is no post-weld heat treatment to hide behind, so fit-up and cleanliness carry the quality.

After welding, the surface is pickled and passivated rather than left as welded, because a heat tinted surface is a depleted surface and the salt will find it. Insulation follows, and the cladding is ordinary galvanized steel: Z275 coil, 0.12 to 3 mm thick, or a prepainted product where the plant is near the coast.

The tank sits on a foundation and is restrained by structural steel that never sees salt. Q345 plate in thicknesses from 0.3 to 200 mm for base plates and embedments, A36 plate where the design is American, and A29M H-sections or rolled H-beams from 5 to 34 mm for the support frame. A 20 m tank carrying 6,000 tonnes puts a bearing pressure on the ring beam that the civil engineer, not the plate mill, has to close.

What goes on the enquiry

The salt specification first, with its chloride, perchlorate and moisture limits, then the hot and cold tank temperatures and the thermal cycle count. After that the sizes: shell diameter and course height, plate thickness, nozzle schedule, and the length of salt main with its bend count. Finally the fabrication scope - who welds, to which qualified procedure, and whether pickling is done in the shop or on site.

A salt tank is bought as plate and judged by its welds. The grade that works is the one chosen for the anneal that will never happen.