How to find sulfidation before it finds you

Sulfidation — spelled sulphidation in half the literature, the same thing either way — is the reaction of sulphur compounds in a hydrocarbon stream with the steel containing it, at temperature. The product is an iron sulphide scale. The scale is not protective in the way an oxide is, the metal underneath keeps reacting, and the wall keeps going.

It starts to matter somewhere above about 230 to 260 °C and the rate climbs steeply with temperature from there. Crude and vacuum units, hydrotreater feed circuits, coker heaters, transfer lines: anywhere hot and sour.

Two different problems with one name

The chemistry splits by whether there is hydrogen present, and the two behave differently enough that the corrosion rate correlations for them are separate bodies of work.

• Without hydrogen — crude and vacuum service, atmospheric heaters, transfer lines. Rate depends on temperature and the total sulphur content of the stream.
• With hydrogen — hydroprocessing, where H₂S and hydrogen together attack considerably harder than the sulphur content alone would suggest.

Both are made slower by chromium. Carbon steel is the exposed material; 5Cr and 9Cr are markedly better; the 300-series stainless steels better again. Most of the design decisions in a hot sour circuit are about where to put that boundary.

The silicon problem, which is the whole article

Here is what makes sulfidation different from every other thinning mechanism in the guide.

The corrosion rate of carbon steel in sulfidation service depends strongly on the silicon content of that particular piece of steel. Steel with silicon below roughly 0.10% corrodes several times faster than steel above it. Silicon is a deoxidiser, not a strength or corrosion element, and it was never a controlled variable in the older pipe specifications — a seamless pipe spec can be met comfortably by a low-silicon heat and by a high-silicon heat alike.

This is not a theoretical concern. In August 2012 a fire at the Chevron Richmond refinery in California started when a carbon steel elbow in a crude unit sidecut line ruptured. The elbow had thinned by sulfidation far faster than the components around it because it was low in silicon, and it had not been individually measured. The investigation that followed is the reason component-level silicon variation is now discussed in every sulfidation programme, and the reason PMI shows up in inspection plans for a mechanism that has nothing to do with getting the alloy wrong.

Why a normal CML grid does not protect you

The standard approach to general thinning is a grid of condition monitoring locations, measured on an interval, trended to a rate, extrapolated to a retirement date. It works because general thinning is, by definition, general — a reading at one point represents the metal around it.

Sulfidation breaks that assumption, not because the thinning within a component is patchy, but because the rate changes at every weld. A CML on the spool tells you nothing about the elbow. Extrapolating the spool's comfortable rate across the circuit is exactly the error that has to be avoided, and it is an easy error to make because every number in the trend looks reassuring.

What actually works

• Measure every component, not every circuit. In susceptible hot sour piping the unit of inspection is the individual elbow, tee, reducer and spool. This is more work than a grid and there is no way around it.
• PMI for silicon, not just for alloy. Portable XRF is normally used to confirm a component is the chrome it claims to be. Here it is used on carbon steel to find the low-silicon components so they can be prioritised or replaced. Confirm the analyser can actually read silicon — not all field instruments do it well.
• Corrosion mapping rather than spot readings where access allows, so the minimum in a component is found rather than the average.
• Profile radiography for small bore, where getting a probe onto every fitting is not realistic.
• Watch the temperature. A heater outlet running 20 °C hotter than design, or a tube with a hot spot, is corroding faster than the rate you trended.

The short version

Sulfidation is easy to detect and easy to miss. Any thickness method finds it once you point the probe at the right component; the difficulty is entirely in knowing which component that is. Treat a hot sour circuit of carbon steel as a population of individuals rather than a length of pipe, and the mechanism becomes manageable. Treat it as one circuit with a representative rate and it stays invisible until it is not.