Carbon fiber reinforced plastic fails through distributed damage accumulation, not a single crack tip, and the composite shows no traditional fatigue limit even beyond 10^9 cycles under ultrasonic loading at 20 kHz [S1].
Conventional metals such as carbon steel and alloy steel follow linear-elastic fracture mechanics: a single dominant crack propagates per cycle, and the da/dN vs ΔK curve is fit by Paris law until final fracture [S3]. The two damage philosophies drive completely different inspection, repair, and material-selection logic for any structure exposed to cyclic loading.
Mechanism: Distributed Damage vs Single Crack Tip
CFRP damage initiates at the intersection of fiber bundles as matrix microcracks, progresses to near-bundle matrix cavities parallel to fibers, and finally culminates in matrix penetration and fiber breakage; this three-stage morphology is observed under 20 kHz ultrasonic VHCF testing with liquid nitrogen cooling [S1]. Post-fatigue bending modulus of CFRP drops significantly relative to pre-fatigue values, which is why stiffness-based health monitoring flags damage long before final fracture [S1].
Metallic structures follow the opposite pattern. Corrosion produces discrete rust pits that act as equivalent initial semi-elliptical surface cracks; the stress intensity factor at the crack tip governs the entire fatigue life [S3]. Increasing the weight loss rate of the corroded steel, the initial crack size, or the adhesive thickness accelerates crack growth, whereas raising CFRP patch stiffness or applying prestress reduces da/dN and extends life [S3]. For a process engineer, the operational consequence is simple: a metal part is either cracked or not, while a composite part is quietly accumulating stiffness loss the entire time it is in service.
Threshold and Long-Life Behaviour
CFRP exhibits a characteristic step in the S-N curve at the transition between high cycle fatigue (HCF) and very high cycle fatigue (VHCF, defined as 10^7-10^11 cycles), with no endurance limit, and damage still accumulates under low-amplitude long-duration loading [S1]. Modern aircraft routinely exceed 10^7 cycles per part over service life, so VHCF is a design constraint, not an academic curiosity [S1].
Adhesive-bonded CFRP laminates have been shown to significantly prolong the fatigue life of corroded steel specimens, with the smaller the initial crack size, the more sensitive the propagation life to equivalent initial crack size variation [S3]. The two material families therefore converge on one engineering reality: as defects grow, the gap between a metal and a composite part narrows because the metal behaves more and more like a damage-tolerant laminate.
Selection Criteria: Who It Is For vs Not For
Specifying CFRP makes sense for weight-sensitive, corrosion-exposed, or high-cycle structures where the load spectrum is well-bounded and stiffness retention can be monitored, and for retrofit strengthening of corroded steel members where the equivalent initial crack size is small [S3]. Examples: aerospace primary structure, wind turbine blades, automotive SMC-R body panels, and CFRP-patch retrofits of bridge girders and offshore platforms [S1][S3][S6].
Specifying metal remains correct for high-stress, high-temperature, impact-loaded, or damage-tolerant-by-inspection applications: pressure vessel shells, cast iron pump housings, alloy steel shafts, and any component where a single through-thickness crack must be detectable by NDT before catastrophic failure. CFRP is not appropriate where the matrix temperature exceeds the polymer Tg, where out-of-plane impact is routine, or where the designer cannot instrument stiffness change, because damage in a composite is invisible to a visual inspection even when the residual strength has dropped 30-40% [S5].
Decision Matrix: CFRP vs Metal on Four Criteria
On specific strength per unit weight CFRP wins, typically 5-10x that of structural steel, which is why aircraft skins and carbon fiber wind blades are dominated by composites [S1]. On corrosion resistance, CFRP is essentially inert, while carbon steel requires coatings, cathodic protection, or scheduled replacement in marine or industrial atmospheres [S3]. On inspection and damage detection, metals win because a single growing crack shows up on dye-penetrant, magnetic-particle, or phased-array UT, while CFRP damage is distributed and requires ultrasonic C-scan, thermography, or stiffness tracking to quantify [S1][S5]. On repairability, both can be patched, but CFRP patching of corroded steel can reduce crack growth rate and increase fatigue life significantly when CFRP stiffness and prestress are tuned correctly [S3].
Failure Modes and Limits You Cannot Ignore
For metals, the dominant failure mode is unstable fracture once ΔK exceeds K_IC; for corroded steel plate strengthened with CFRP plates, the influence of adhesive thickness on fatigue life is limited and convergent, so adding glue beyond an optimum does not help and can shift the failure mode to adhesive shear [S3]. CFRP fatigue is further complicated by loading mode: transverse-compression (T-C) fatigue accelerates out-of-plane damage propagation compared to tension-tension (T-T) fatigue at the same nominal stress, because of additional micro-buckling and shear-driven delamination mechanisms [S5].
Frequency effects also diverge. Metals are largely insensitive to cyclic frequency in the 0.5-200 Hz range, but CFRP heating under high-frequency loading softens the matrix and lowers the effective fatigue strength; ultrasonic 20 kHz testing requires liquid nitrogen cooling specifically to extract the heat and isolate the mechanical damage response [S1]. Material damping in CFRP releases heat under high-frequency cyclic loading through increased molecular friction, which the S4 study identifies as a primary driver of stiffness loss at high cycle counts (2025-08). Designers should derate CFRP fatigue strength when the part sees more than a few Hz and is not actively cooled.
Standards, Modelling, and Sourcing Notes
CFRP fatigue modelling is dominated by residual stiffness degradation curves and S-N fits, because Paris law does not apply when there is no single crack tip; VHCF characterisation requires an evolution threshold introduced from the S-N curve to separate HCF and VHCF regimes [S1]. For metals, the equivalent initial crack from a corrosion pit is modelled as a semi-elliptical surface flaw, and SIF at the crack tip drives da/dN, with CFRP patch stiffness and prestress as the primary retrofit variables [S3]. Composites fatigue data quality varies widely; reviewers note the damage mechanism is "complicated" by nature and any CFRP S-N curve should be cross-checked against the layup, frequency, and cooling conditions used to generate it [S2].
For sourcing, request VHCF data when the part will see more than 10^7 cycles, ask for stiffness-degradation curves not just S-N points, and confirm the matrix Tg versus service temperature when the application runs hot. For corroded steel retrofits, demand the equivalent initial crack size assumption, the SIF calculation method, and the assumed CFRP prestress level before accepting a life-extension claim [S3]. A retrofit with sub-optimum adhesive thickness or zero prestress can still extend life, but the utilisation rate of the CFRP material and the fatigue strengthening effect both fall off sharply [S3].
Two trackable signals to watch: the ASTM D7615/D7791 damage tolerance test method revisions for polymer-matrix composites, and any extension of the two-stage CFRP-patch crack-propagation model [S3] to fatigue of aluminum-alloy airframe skins, where weight loss rate and pit geometry play the same role they do in steel. For more on layup-level material selection, see the 3K vs 12K vs 24K tow spec guide.