Corrosion resistance

A material is never corrosion resistant on its own. It is resistant to a particular medium, at a particular concentration and temperature, and the British Stainless Steel Association is explicit about what follows from that: assessing corrosion resistance in any particular environment usually involves considering specific corrosion mechanisms. There is no general answer to give.

Stainless steel carries a chromium-rich oxide film that forms on its own in air, and that film is the corrosion resistance. When it cannot re-form, the metal corrodes. Passivation covers that mechanism and what restores the film after welding or machining.

The trap is assuming a liquid's aggressiveness matches how unpleasant it sounds. Grundfos warns that pumping demineralised water with a conductivity below 2 μS/cm through a pump with a silicon carbide against silicon carbide shaft seal brings an increased risk of electrochemical corrosion, and recommends a silicon carbide against carbon or silicon carbide against tungsten carbide seal instead. Water that has had everything taken out of it is a corrosion problem. So the test is not how nasty the medium sounds. It is what the medium does to the specific surface in front of it.

Ranking alloys by corrosion resistance implies a universality that does not exist, and it is the single most misleading thing this page could do. BSSA states the consequence better than we could: choosing an inappropriate grade can result in failure by perforation or catastrophic cracking failure in shorter times than failure by uniform corrosion in lower resistant alloys such as ordinary, plain carbon steel.

Read that twice. A higher-alloy stainless can fail faster than plain carbon steel in the wrong duty. A ranking cannot express that, which is why a ranking is the wrong shape for the information.

The nearest thing to a comparative number is the pitting resistance equivalent, and its own author hedges it. BSSA gives the common formula as PREN = Cr + 3.3Mo

• 16N, with a variant that accounts for tungsten, and calls the numbers useful for ranking and comparing the different grades. It then adds that they cannot be used to predict whether a particular grade will be suitable for a given application, where pitting corrosion may be a hazard. PREN also only speaks to pitting. It says nothing about acid attack, nothing about stress corrosion cracking, and nothing about a galvanic pair.

What does generalise is mechanism, not ranking. BSSA notes that corrosion resistance relies on a good supply of oxygen, and that higher levels of chromium, nickel, molybdenum and nitrogen increase resistance to localised corrosion. That tells you which lever to pull once you know what is attacking the part. It does not tell you which grade wins.

Corrosion is not one process, and the failure mode decides what you look for and what you do about it. BSSA lists the principal mechanisms for stainless steel as crevice corrosion, pitting, intercrystalline corrosion, stress corrosion cracking and galvanic corrosion, with erosion-corrosion and corrosion fatigue as related mechanisms, and describes all of them as a localised breakdown of the passive layer. The table below sets out the five that turn up most often on a pump duty.

General attack is the one you mostly do not get. BSSA points out that in contrast to other steels, stainless steels in the passive state are normally protected against corrosion affecting large areas of the surface. That is not reassurance. It means the failure you do get is localised and small, so a wetted part can look sound and still be perforated.

Pitting and crevice attack are the common pair. BSSA associates localised corrosion with chloride ions in aqueous environments, and adds that acidic conditions and increases in temperature both contribute. It puts crevice corrosion as rare at chloride levels below 200 and 1000 ppm respectively for 304 and 316 types. Crevices are usually designed in rather than corroded in: bolted joints, beneath flanges or between flanges and gaskets, valve seats, and welds with incomplete penetration. BSSA's prevention list is mostly geometry. Keep junctions as wide open as possible, get full weld penetration with smooth rounded beads, and clean out tanks where deposits settle.

Stress corrosion cracking is the one that catches people out. It needs a tensile stress, applied or residual, on top of the chloride. BSSA reports crack growth rates as particularly high around 100°C and cracking as less common below 60°C, while noting it has been observed as low as 25°C, and that there is no limit for chloride because it concentrates inside a crevice well above the bulk level. The cracks are transgranular and branched. Nothing thins, so a wall thickness check finds nothing.

Galvanic attack is a joint problem, not a material problem. BSSA is precise: it can only occur when two dissimilar metals are in electrical contact and are bridged by an electrically conductive liquid, and if the metals are dry it cannot occur at all. Area ratio decides the severity, because a large cathode against a small anode accelerates the attack on the anode. That is why the fastener is where it usually shows up.

Corrosion resistance is a property of everything the medium touches, not of the pump. That path is the casing, the impeller, the shaft or containment shell, the gaskets, the O-rings and the seal faces, and a grade stamped on a casing covers none of the others.

The elastomers usually go first, because an O-ring is a thin section of polymer in constant contact with the medium. The three common ones cover three different parts of the chemistry, and no two of them agree. EPDM takes water, steam and caustic cleaning chemistry and swells in oils. FKM, sold as Viton, runs the other way. PTFE covers the broadest range and gives up resilience to do it. Check each one against the medium separately.

The other half of the path is fabrication. A weld and a machined face are where the passive film has been disturbed, so they are where localised attack starts, and passivation is the treatment that puts the surface back. Where the medium is beyond what stainless will take at all, the wetted parts move to a thermoplastic, a lining, or a high-nickel alloy such as Hastelloy.

Materials come fourth in the order we work through a duty, after flow rate, the pipework and viscosity. Temperature follows immediately, because temperature and the medium together drive the sealing arrangement, and the seal is next after that. Corrosion resistance is decided across those three questions, against your medium. It is never decided in the abstract.

Grundfos splits one range in two on exactly this question, which makes it a clean worked example. The CR and CRI are stated as suitable for non-corrosive liquids, for transfer, circulation and pressure boosting of clean cold or hot water. The CRN is stated as suitable for industrial liquids, for systems where all parts in contact with the liquid must be high-grade stainless steel. The wetted parts move from EN 10088 1.4301, roughly AISI 304, to EN 10088 1.4401, roughly AISI 316.

Same hydraulics, same curve, same duty point. The metal changes, and the medium is what changes it. If your enquiry arrives with a flow and a head and nothing about the liquid, that is the question we will come back with.

The medium, its concentration and its temperature. If you have them, add the chloride content and the pH, because two of Grundfos's four deciding factors are exactly those.

Then two things enquiries almost never include, both of which change the answer. The first is what the pipework, flanges and fasteners around the pump are made of, because a galvanic pair is created at a joint and a small anode next to a large cathode is the bad arrangement. The second is whether the wetted path was welded or modified on site, because that is where the passive film is disturbed and where crevices get built in.

Treat any resistance table you have been handed as a starting point. Grundfos attaches the caveat to its own list of pumped liquids: it is a general guide only and cannot replace actual testing of the pumped liquids and pump materials under specific working conditions, with concentration, liquid temperature and pressure all capable of moving the answer.

If the honest answer turns out to be that your grade is fine and the failure is a crevice under a gasket, we will tell you that and you will not need a new pump. Where the medium is genuinely aggressive, start at pumps for corrosive media.