How to detect creep damage

Above roughly a third to a half of its absolute melting temperature, steel under load does not simply deform elastically and stop. It keeps deforming, slowly, for as long as the load and the temperature persist. That is creep.

Three things make it difficult to inspect for:

• It removes no metal, so thickness measurement finds nothing.
• The early damage is microscopic cavities on grain boundaries, far too small for any ultrasonic method to resolve.
• Damage accelerates at the end. Most of the life passes with little to see, then the last stage goes quickly.

The stages, and what can see each one

Read the right-hand column. For the first four stages the only thing that reliably works is metallography. Conventional NDT arrives at stage E, and at stage E the decision is usually replacement rather than assessment.

Replication — the method that sees it early

The surface is ground, polished and etched in place, and the microstructure is lifted onto a plastic film for examination under a microscope. It can classify creep damage through the stages above.

The limits are real and worth stating: it examines the microstructure at the exact spot polished, on the surface only. Creep damage is often worst mid-wall or on the hotter surface, which on a heater tube is the inside. A clean replica from the outside is weak evidence about the wall behind it.

It is also slow, needs a metallurgist, and needs the right spot chosen. Which makes it a confirmatory method rather than a survey — you replicate where the other evidence says to look.

Deformation — the measurement that scales

Creep strains the metal, and on a pressurised component that shows as diametral growth. Retirement criteria for heater tubes are commonly written around a percentage of original outside or inside diameter.

Measured by laser profilometry from inside the tube, or by strapping and callipers from outside. It is a bulk measurement, it covers the whole length, and it needs no metallurgist — which is why it is the practical survey method even though it responds later than replication.

Watch for it in vessels too. Bulging of a reactor shell, or distortion of a hanger or a support in a hot service, is creep deformation and it is assessed under API 579 Part 8 as well as Part 10.

Temperature — the measurement that predicts it

This is the one worth investing in, because creep rate is exponentially sensitive to temperature. A tube 15 °C hotter than its neighbours is not 5% worse off; depending on the material and stress it can have half the life.

So the highest-value creep inspection is often not an inspection at all:

• Skin thermocouples on selected tubes, logged continuously. Their value is the accumulated history, which is exactly what a life calculation needs.
• Infrared thermography through the peepholes while the heater fires. Finds hot tubes, flame impingement and coke-restricted flow while there is still time to fix the cause.

A tube found running hot and corrected has life restored in a way no inspection can achieve after the fact.

Supporting evidence

Hardness. Long exposure at temperature spheroidises carbides and softens the steel. A hardness survey showing softening is corroboration that a component has run hotter or longer than intended.

Oxide scale thickness. On boiler tubes, the internal oxide grows at a rate governed by temperature, so measuring it ultrasonically gives an estimate of the metal temperature history — a temperature record from a tube that never had a thermocouple.

How life is calculated

Creep is assessed under API 579 Part 10, and the method is different from every other part of the standard because there is no flaw to size. Life is consumed:

1. For each period of operation, take the metal temperature and the stress.
2. Find the rupture life at those conditions — usually through a Larson-Miller parameter, which collapses time and temperature into one number so that short tests at high temperature can predict long service at lower ones.
3. Express the time actually spent as a fraction of that life.
4. Sum the fractions. When the total reaches 1.0, rupture is expected.

What actually manages creep

Inspection tells you where you are on the curve. It does not slow the mechanism. The things that do are operational:

• Keep the metal cooler. Balanced firing, clean burners, no flame impingement, decoking before restriction drives temperature up.
• Keep flow even. A restricted pass runs hotter, cokes faster and restricts further. The feedback is positive, which is why damage concentrates in a few tubes rather than spreading.
• Know the history. Every excursion counts and none of it is forgiven. A unit pushed hard for two summers has spent life that will not be visible in any measurement until much later.