Thermal Fatigue in Cyclic High-Temperature Pipe Fittings: Mechanism, Crack Morphology, and Design Mitigation
Thermal fatigue is a fatigue failure mechanism caused by cyclic thermal stresses arising from repeated temperature changes in a constrained structure. It is distinct from mechanical fatigue (caused by cyclic applied load) and from creep (caused by sustained load at high temperature) — though in practice, thermal fatigue and creep interact in high-temperature pipe fittings operating in intermittently loaded plant such as peaking power stations, steam turbine bypass systems, and process plant with frequent startups and shutdowns.
The Thermal Fatigue Mechanism
When a pipe fitting is heated or cooled, the thermal gradient through the wall creates differential thermal expansion between the hot inner bore and the cooler outer surface (during heating) or vice versa (during cooling). The inner bore wants to expand more than the outer surface during rapid heating — but the two are mechanically constrained by being part of the same fitting. The constraint means the inner bore is put into compression (by the outer wall preventing it from expanding freely) and the outer surface is put into tension. During cooling, the stresses reverse: the inner bore goes into tension and the outer surface into compression. If this thermal stress cycle is repeated — every time the fitting is heated and cooled in service — the alternating tensile and compressive stress at the inner bore accumulates fatigue damage, eventually initiating a crack at the bore surface. The crack propagates radially outward through the wall with each subsequent thermal cycle, ultimately reaching a critical size for fracture or leakage.
Crack Morphology and Inspection
Thermal fatigue cracks in pipe fittings have characteristic morphology: they initiate at the bore (inner surface) because the highest thermal stress gradient occurs at the inner surface during rapid temperature transients; they are typically multiple, parallel, and regularly spaced — a "crazing" or "fire cracking" pattern — reflecting the uniform thermal stress field; they propagate radially (perpendicular to the bore surface, i.e., through the wall thickness); and they often branch or have a broad, blunt crack tip, reflecting the low-stress-concentration nature of thermally induced fatigue compared to mechanically induced fatigue (which typically produces sharper cracks). Detection of thermal fatigue cracks from the outside is challenging because the cracks initiate at the bore — external visual inspection and MT/PT are ineffective for inner bore cracking. Phased array ultrasonic testing (PAUT) from the outside, or internal visual inspection using a borescope, are the primary methods for detecting thermal fatigue cracking in service.
Fitting Geometry Effects
The stress concentration geometry of fittings significantly affects thermal fatigue life relative to straight pipe. At a weld toe on the outer surface of a pipe fitting, the stress concentration (Kt ~1.5–3.0) amplifies both the thermal stress amplitude and the mean stress — cracks can also initiate at the weld toe under thermal fatigue loading, progressing into the fitting wall from the outer surface. Elbow intrados (inside of the bend) is thinner than the nominal wall after forming, and the stress from thermal cycling is superimposed on the hoop stress from pressure — the combination accelerates crack initiation. Tee branch crotch corners have a very high local stress concentration under thermal transients because the temperature gradient is highest at the inside corner of the branch-run junction. These locations must be included in any thermal fatigue life assessment and are the priority inspection sites on fittings with a known thermal cycling history.
Temperature Ramp Rate: The Primary Control Variable
The magnitude of the thermal stress during a heating or cooling transient is directly proportional to the temperature difference between the inner and outer bore surface at any instant — which depends on both the rate of temperature change (ramp rate, °C/min) and the wall thickness. For a given wall thickness, thermal stress scales approximately as: σ_thermal = E × α × ΔT / (2(1−ν)), where E is Young's modulus, α is the coefficient of thermal expansion, ΔT is the through-wall temperature difference at the instant of peak gradient, and ν is Poisson's ratio. Reducing the ramp rate reduces ΔT and hence the thermal stress — halving the ramp rate approximately halves the peak thermal stress and extends thermal fatigue life by approximately a factor of 8 (since fatigue life scales approximately as stress^(−3) for most metals in the high-cycle regime). For P91 pipe fittings, typical startup ramp rate limits are 2–4°C/min for wall thicknesses above 50 mm — these limits are calculated from a thermal fatigue life assessment and must be specified in the operating procedure, not just in the design document.
Material Selection for Thermal Fatigue Resistance
Thermal fatigue resistance in metals correlates with: low elastic modulus (lower E reduces thermal stress for the same ΔT); low coefficient of thermal expansion (lower α reduces thermal strain); high thermal conductivity (reduces ΔT for the same ramp rate by spreading heat faster through the wall); and high fatigue ductility (higher ductility at operating temperature allows more plastic strain per cycle before crack initiation). Austenitic stainless steels (304L, 316L) have high thermal expansion (~16 µm/m·°C) and low thermal conductivity (~15 W/m·K) — a poor combination for thermal fatigue. Ferritic-martensitic steels (P91, P92) have lower thermal expansion (~11 µm/m·°C) and higher conductivity (~28 W/m·K) — significantly better thermal fatigue resistance. This is one reason P91 is preferred over austenitic stainless in high-cycle thermal fatigue applications such as steam turbine bypass piping, despite austenitic stainless having comparable high-temperature strength.