Click on illustration for a larger image.
9. Cyclic service. If a vessel will experience an unusual number of thermal or pressure cycles over its design life, this could result in premature fatigue failure (usually at a weld) unless preemptive measures are taken. Fatigue is cumulative material damage that manifests as a small crack and progressively worsens (sometimes to failure) as the material is repeatedly cycled. A 1985 survey showed that fatigue was the second most prevalent cause of failure in industry (25%), closely behind corrosion (29%); in the airline industry, it predominates (61%) .
It’s up to the purchaser to instruct the fabricator what design/fabrication practices to follow to avoid fatigue. Cyclic service is usually associated with batch processes and ASME  provides the following rules:
Design for fatigue if N1 + N2 + N3 + N4 ≥400 for non-integral (fillet weld) construction and ≥1,000 for integral construction (i.e., no load-bearing fillet welds), or 60 and 350, respectively, in the knuckle region of formed heads, where N1 is the number of full startup/shutdown cycles; N2 is the number of cycles where pressure swings 15% (non-integral) or 20% (integral); N3 is the number of thermal cycles with a temperature differential (ΔT) exceeding 50°F between two adjacent points no more than 2.5 (Rt)½ apart (where R is inside radius of vessel and t is thickness of vessel under consideration) —apply a two-times factor if ΔT exceeds 100°F, a four times factor if more than 150°F, and see Div. 2 for more than 250°F; and N4 is the number of thermal cycles for welds attaching dissimilar materials in which (α1-α2)ΔT (where α is the thermal expansion coefficient) exceeds 0.00034, or for carbon steel welded to stainless steel, the number of cycles where 2ΔT exceeds 340.
Equipment and piping in continuous processes also can experience fatigue due to the relentless mechanical loading/unloading of reciprocating compressors, piston pumps, bin vibrators or from vibration, etc., from any type of mis-aligned rotating equipment.
Fatigue failures in welded equipment most commonly occur in fillet welds where there’s an abrupt change in equipment geometry. Division 2 of the ASME Code designs around fatigue cracking in nozzles by limiting the use of fillet welds. However, fillet welds and sharp corners are ubiquitous in Div. 1 designs and can’t be avoided without cost.
Figure 2 -- Blend grinding of fillet
Crack initiation usually begins at the surface due to small microcracks. Therefore, surface smoothness is a good defense. Polished surfaces have four times the fatigue resistance  but polishing generally can’t be justified for fatigue life alone. Shot peening imparts compressive stresses into the metal surface that impede crack initiation but, again, only high-end applications can economically justify peening. For mid- to small-size process vessels, good weld quality often is the most economical defense against fatigue; so, state requirements in the equipment specifications. Because fatigue cracks often initiate at the toe or root of fillet welds, grinding the face to gently blend the weld into the base metal with a generous radius remarkably reduces stress risers (Figure 2). Another method to reduce stress risers is to TIG (tungsten-inert-gas) wash a weld toe to improve smoothness and remove microcracks. Initially target welds where cyclic loading is occurring. Experience has shown that most fatigue problems occur due to inadequately supported attachments or where saddles/supports lacked wear pads or rounded corners.