Fibre reliability and ageing: strength, fatigue, hydrogen, radiation, lifetime

Glass fibre is stronger than steel by cross-section and, kept dry and unstressed, practically does not age. In a real plant it is bent, pulled, wetted, frozen, exposed to hydrogen from corroding metal and sometimes to radiation, and the parts around it — coatings, gels, connectors, closures — age faster than the glass. This page collects what limits the life of a fibre link: the strength and fatigue physics behind the bend and tension rules, the ageing mechanisms and their signatures, the realistic service life of each part, what to expect from fibre of a given era and how to design margins for ageing.

Mechanical strength

Quantity	Value	Meaning
Theoretical strength of silica	~14 GPa	flawless glass; never reached in practice
Measured strength of short pristine lengths	5–6 GPa (700–800 kpsi)	a few metres tested in the lab
Proof stress	0.69 GPa (100 kpsi), ≈ 1 % strain	every metre of telecom fibre survived it for ≥ 1 s (How fibre is made)
Young's modulus	72 GPa	strain = stress / modulus: 0.69 GPa ↔ 0.96 %
Strength of long lengths	set by the rarest flaw	a 10 km length contains more chance of a weak point than 1 m (Weibull statistics) — which is why proof testing screens the whole length
Long-term installed strain limit	≤ 20 % of proof (≤ 0.2 %)	common TIA/IEC practice; used for aerial sag-tension and duct pulls
Short-term installation limit	≤ 60 % of proof (≤ 0.6 %)	during pulling and blowing only (Installation)
Excess fibre length in loose tubes	0.1–0.5 %	cable strain does not reach the fibre until this "strain window" is used up (Cable construction)

Static fatigue: why stressed fibre breaks later

Under tension in the presence of moisture, surface flaws grow slowly until the fibre fails — stress corrosion, or static fatigue. Crack growth scales with stress to the power n, the fatigue parameter: n ≈ 20 for standard acrylate-coated fibre (IEC 60793-1-33 requires n ≥ 18), above 100 for hermetic carbon-coated fibre. Because of that power law, halving the stress multiplies the time to failure by about 2²⁰ ≈ 10⁶, which is why the 20 % rule gives an effectively infinite life while a fibre held at 60 % of proof may fail in months. The lifetime model is in IEC TR 62048.

Bending is tension on the outside of the bend: strain ε = r / R with r = 62.5 µm (the glass radius).

Bend radius R	Strain on the outer surface	Long-term verdict for standard fibre
30 mm	0.21 %	unlimited turns for life — the classic minimum
15 mm	0.42 %	acceptable for a few turns (G.657.A1 territory); watch optical loss on G.652
10 mm	0.63 %	G.657.A2/B2 optically; mechanically fine for a few turns over 20 years per IEC 62048
7.5 mm	0.83 %	G.657.B3 optically; mechanically a handful of turns, not a coil
5 mm	1.25 %	above proof strain — failure within years
Kink at a staple or a closed tray lid	several %	breaks in days to weeks

Ageing and failure mechanisms

Mechanism	Cause	Sign	Prevention
Mechanical fracture (static fatigue)	tight bends, residual tension, pinches in trays, staples, twisted pigtails	a break weeks or years after installation, always at a bend or pinch point	bend radius, no tension on bare fibre in closures, proper slack storage
Hydrogen ageing	H₂ diffuses into the glass; at defects it forms OH (permanent loss at 1383 and 1240 nm and a broad rise) or dissolves (reversible peaks); sources: galvanic corrosion of metal armour and strength members, decomposing gels, submarine repeaters	attenuation slope rising, worst at 1383 nm and above 1550 nm	modern G.652.D passes the IEC hydrogen ageing test; old phosphorus-doped fibre (pre-1990) is very sensitive; dielectric cables, no mixed metals
Water and ice	water in tubes or closures; ice formation squeezes fibres → microbending; long-term coating swelling; accelerated fatigue	seasonal loss rise in winter at 1550/1625 nm, wet closures, water-blocking tape swollen	sealed closures, dry water-blocking, no low points in ducts, drainage
Temperature	cable materials contract when cold → microbending; heat degrades gels and coatings	+0.05–0.2 dB/km at −40 °C on some cables; loss steps at cabinet hot spots	use the cabled temperature specification; outdoor-rated cable outdoors
Radiation-induced attenuation	ionising radiation creates colour centres; germanium- and phosphorus-doped cores darken, partly recovering afterwards	dB/km growing with dose; worst at short wavelengths	pure-silica-core fluorine-doped fibre (1–2 dB/km per kGy instead of tens), shielding, pre-irradiation
UV and weather	sunlight embrittles jackets; aerial jackets chalk and crack	cracked sheath, water ingress	UV-stabilised black polyethylene outdoors
Coating degradation	incompatible gels, solvents, wrong cleaning fluids, overheating	strip force changes, microbend loss, delamination	approved cleaners, compatible gel, cable temperature limits
Fibre fuse	above roughly 1–1.5 W in single-mode fibre a contaminant or a bad connector ignites a plasma that runs back toward the source at about 1 m/s	fibre destroyed with a periodic chain of voids; instant loss of the span	power limits, clean high-power connectors, fuse detectors in Raman systems (Non-linear effects)
Rodents, termites, lightning, vehicles, shotguns on aerial cable	external	breaks and crushes	armour, dielectric design, grounding, clearance (Maintenance & restoration)
Connector wear	every mating abrades the endface; debris scratches	rising loss and reflectance after hundreds of matings	500–1 000 matings per connector; replace cords and cassettes, inspect before mating
Splice protector and closure ageing	heat-shrink sleeves last decades; mechanical splice gel dries in 10–20 years; closure gaskets harden	reflectance at mechanical splices, water in closures	fusion for permanent joints, reseal closures at inspection
Gel migration	gel drips in vertical indoor runs, contaminates trays	messy closures, coating damage	dry cables indoors and in risers

Service life of the parts

Part	Typical service life	Limiting factor
Glass fibre	25–40 years and more; 1980s fibre still carries traffic	flaws under stress, hydrogen in old fibre
Coating	25–30 years	chemicals, heat
Outdoor cable in duct	25–40 years	jacket, water blocking
Direct-buried cable	25–30 years	soil movement, rodents, dig-ups
Aerial cable	20–25 years	ice and wind fatigue, UV, vehicle strikes
Indoor cable	25+ years	renovations rather than ageing
Connectors	by matings and contamination, not by years	500–1 000 matings
Splice closures	20–25 years with resealing	gaskets, water
Mechanical splices	10–20 years	gel drying
Splitters, WDM filters	20+ years	epoxy and adhesives at temperature extremes
Patch cords	5–10 years	handling, bends, endface wear
Documentation	until the first unrecorded change	process, not physics (Documentation & labelling)

Fibre by era: what to expect

Installed	Typical fibre	Limits for today's services
Before 1985	50 and 62.5 µm multimode, early single-mode with 0.5–1 dB/km and high OH	1310 nm only; 100 Mbit/s–1G
1985–1995	G.652.A/B; PMD 0.5–2 ps/√km; high water peak; some phosphorus-doped, hydrogen-sensitive	10G to 25–100 km depending on PMD; no 40G NRZ; no E-band; coherent works (Fibre characterization)
1995–2005	G.652.B/C; G.653 dispersion-shifted on some long-haul routes; G.655 on new long-haul; 62.5 µm OM1 everywhere indoors	G.653 breaks C-band DWDM (four-wave mixing) — use L-band or unequal spacing; OM1 gives 33 m at 10G
2005–2015	G.652.D low water peak; OM3/OM4; G.657 in FTTH	fit for 100G/400G coherent and 10G CWDM at all 18 wavelengths
2015 onward	OM5, G.654.E for terrestrial coherent, 200 µm coatings, rollable ribbons, bend-insensitive everywhere	800G coherent, 400G PAM4 on OS2

Designing for ageing

Design measure	Typical allowance
Ageing allowance in the link budget	1 dB per link, or 0.05 dB/km on long routes (Link budget)
Repair margin	0.1 dB per expected future splice, 2–4 repairs per 100 km over the life
Connector degradation	0.2 dB per pair over the life
Spare fibres	20–50 % of the count, or the next standard count up
Dielectric cable	in lightning-prone areas, along power lines and where metals would corrode
Dry cables indoors and in risers	no gel migration
Baseline and trend	commissioning OTDR and DDM, annual comparison (Maintenance & restoration)

In CodingBox

The other thing that ages on a link is the laser. A rising bias current at constant output power is the module's end-of-life signature; CodingBox reads and trends Tx bias. A slow decline of received power with the far end's Tx and bias both stable is the plant ageing — connectors, closures, water — not the module (Tx bias & laser ageing, DDM in the app).