The urgency behind the work is stark. The researchers point to engineering data from China indicating that more than thirty bridge collapses over the past two decades were triggered by the brittle fracture of suspenders, and that the average actual replacement cycle for these components is roughly fifteen years, only half of their intended thirty-year design life. The culprit is a corrosive conspiracy: environmental agents, cyclic vehicular loading, and ambient temperature fluctuations act together on the protective sheathing of bridge cables. When the outer polyethylene layer ages and cracks, rainwater seeps in, gravity carries the moisture down to the lower anchorage zones, and the steel wires inside begin to rust in ways that are anything but uniform. Pitting corrosion, in which metal loss concentrates in discrete craters rather than spreading evenly across the surface, is far more damaging than uniform corrosion because each pit acts as a microscopic stress amplifier.

The team grounded their investigation in a real structure: the Furong Town Bridge in Xiangxi Prefecture, Hunan Province, a 302.3-meter concrete-filled steel tube arch bridge completed in 2003 and fitted with fifty suspenders, each containing bundles of 55 or 61 galvanized high-strength steel wires. When eight of those suspenders were replaced in 2014 after their protective layers failed, the naturally corroded wires recovered from the dismantled cables became the raw material for the study. The researchers descaled, cleaned, and numbered the recovered wires, then scanned them with a non-contact 3D scanner boasting a resolution of 0.01 millimeters. Following the pit identification guidelines of ISO 11463:2020, they extracted the length, width, depth, and count of individual pits from the point cloud data, reconstructing full three-dimensional solid models of the rusted wires in Geomagic Studio.

From this trove of measured geometry, the team built something the field had been missing: a probability distribution model of corrosion pits grounded in naturally corroded, not artificially corroded, steel. The statistics revealed a clear pattern. As the degree of corrosion increased from roughly 1.1 percent to 2.8 percent mass loss, the average number of pits per specimen climbed from about 88 to 177, and average pit depths grew from 0.27 to 0.33 millimeters. Regression analysis showed that pit depth follows a Gaussian distribution with fitting quality above 0.93, a departure from the Gumbel extreme-value distributions used by earlier researchers. The authors argue that the Gaussian form, while equally accurate for their data, integrates more seamlessly with the statistical moments and spatial stochastic processes needed for the next stage of modeling, eliminating the uncertainties that come with converting between distribution families. Notably, the ratio of pit length to width clustered around one to one regardless of corrosion severity.

The real innovation came in translating those statistics into simulated metal. The team wrote a Python script based on batch Boolean operations that generates corrosion pits at random locations on a digital wire, using Euclidean distance checks to guarantee that no two pits intersect, and feeds the measured probability distributions directly into the ABAQUS finite element environment. The resulting models reproduce a seven-wire spiral strand with a 5.1-millimeter center wire and 5.05-millimeter helical wires, a quarter-pitch segment 55.25 millimeters long, and realistic inter-wire contact, including a friction coefficient of 0.115 drawn from established strand-contact literature. Validation was layered and rigorous: the elastic response of the model matched the classical analytical theories of Costello and Feyrer, the elastic-plastic results aligned with prior work by Zhao, and the stress field around individual pits reproduced the characteristic pattern, peak stress at the pit bottom decaying along the axis, reported by earlier experimentalists.

With the digital framework verified, the researchers probed a question that idealized models cannot answer: does the shape of a corrosion pit matter? They compared hemispherical, conical, and cylindrical pits of identical radius and depth under identical loading. The answer was emphatically yes, though not in the way one might expect. Hemispherical pits produced a ring-shaped stress concentration band at their midsection with a distinctive V-shaped stress region at the pit edge. Conical pits generated an infinity-symbol-shaped concentration zone with two lobes of equal intensity and no V-region at all. Cylindrical pits shifted the maximum stress away from the pit floor entirely, concentrating it on the column walls instead. The helix angle of the wire added further complexity, because the classic tension-torsion coupling of helical strands amplifies stress concentration where the principal stress direction aligns with the geometric asymmetry of the pit.

Quantifying these effects with a stress concentration factor, the ratio of peak local stress to the far-field gross-section stress, the team found that most values fell between 1.0 and 2.5, with cylindrical pits reaching the highest factor of 2.7. Conical pits proved the gentlest, showing the smallest variation in stress concentration. Counterintuitively, the stress concentration factor decreased overall as the applied cross-section load increased, a reminder that local geometric amplification and global material response do not scale in lockstep. The practical implication for bridge inspectors is significant: two pits of identical depth and width can impose markedly different local stresses depending on their internal geometry, so pit shape and location deserve explicit attention in design and maintenance protocols rather than being collapsed into a single depth measurement.

The study then escalated to the strand level, simulating what happens when corrosion strikes neighboring wires versus wires on opposite sides of the seven-wire bundle. In the adjacent-corrosion case, wires two and three, both riddled with ten random cylindrical pits, progressively lost stiffness as their net cross-sections shrank, generating pre-stress relaxation and lateral displacements of 0.5 to 0.9 millimeters along the X-axis as the helix angle converted axial tension into torsional wander. Stress concentrations at individual pits eventually linked up into continuous damage paths, and the two corroded wires fractured while the remaining wires carried on. In the opposite-wire case, wires two and five failed in a similar pit-to-pit cascade, but the whole bundle fractured sooner under the same displacement load, and lateral displacement was smaller, peaking at 0.55 millimeters. Relative corrosion, the team concluded, pushes failure preferentially along the tensile axis and makes the strand more prone to outright breakage.

Finally, the researchers confronted their simulations with physical reality. They fabricated 15.2-millimeter, 1,100-millimeter seven-wire strands and used electrochemical corrosion, driven by a 2-ampere direct current through a sodium chloride electrolyte, to imprint pits at the exact three-dimensional coordinates mapped from the digital models, with Faraday’s law of electrolysis controlling the degree of corrosion to within 3.5 percent of the theoretical mass loss. Static tensile tests at room temperature then delivered a dramatic spectacle: uncorroded strands broke cleanly across all wires with flat fracture surfaces, while the corroded specimens displayed a birdcage-like deformation, and their pitted wires snapped almost instantaneously, within one to two seconds of the first fracture, at wedge-shaped breaks sitting squarely in the densest pit clusters with no visible necking. Load-strain and load-displacement curves from the tests matched the finite element predictions closely, even after the team corrected for grip slippage with a nonlinear spring element, a correction that shifted ultimate strength predictions by less than 1.2 percent.

The take-home message is both sobering and empowering. Within the corrosion range studied, weight loss below about 10 percent, random pitting leaves the yield strength and elastic modulus of these wires largely intact but steadily eats away at ultimate strength and fracture strain, and the precise spatial choreography of pits, whether on adjacent or opposing wires, determines how and how fast the strand dies. The authors acknowledge that their idealized pit geometries and their assumption of non-overlapping pits may underestimate the worst-case local stress concentrations at high corrosion rates, and they plan to incorporate pit fusion and overlap algorithms, along with X-ray micro-computed tomography of real pit topography, in future work. For the engineers responsible for the world’s aging cable-supported bridges, however, the study already offers a powerful new toolkit: a statistically faithful, experimentally validated way to see, in silico, how the random scars of corrosion decide which wire breaks first, and when the whole cable will follow.

Subject of Research: Tensile strength degradation of spiral high-strength steel wires in bridge cables due to random pitting corrosion

Article Title: Numerical and experimental study of tensile strength of spiral-high-strength steel-wires considering random pitting coupling

Article References: Xiao, X., You, C., Xiao, K., Luo, Y., Chen, F., & Zhang, H. (2026). Numerical and experimental study of tensile strength of spiral-high-strength steel-wires considering random pitting coupling. Case Studies in Construction Materials, 25, Article e06590. https://doi.org/10.1016/j.cscm.2026.e06590

Image Credits: AI Generated

DOI: 10.1016/j.cscm.2026.e06590

Keywords: corrosion, pitting corrosion, high-strength steel wires, bridge cables, stress concentration, finite element analysis, 3D scanning, electrochemical corrosion, tensile testing, probabilistic modeling, structural health, ABAQUS

The team focused on linear low-density polyethylene, or LLD-PE, the workhorse polymer of artificial turf, chosen by manufacturers for its flexibility, chemical and thermal resistance, and low cost. Their source material came from a third-generation pitch with 50-millimetre pile length, installed in 2006 and used for football around 40 hours per week, year-round, until its removal in 2023. Over its lifetime the surface endured 1,825 hours of annual sunshine with 8.63 watts per square metre of UVA radiation, a mean temperature of 9.5 degrees Celsius, 74 percent mean humidity, and 920 millimetres of yearly precipitation. Because the average lifespan of such pitches is only 10 to 15 years, the 19-year service life represents an exceptionally high-exposure case, making it a stringent test of the polymer’s resilience.

Recovering the fibres was itself a technical challenge. The dismantled carpet was cleaned of its elastomeric performance infill and stabilising sand by tapping, shaking and vacuum cleaning, after which the polyethylene fibres were sheared from the backing and separated from sand and supporting polyester yarn by density separation in a water bath. Multiple cold-water washing cycles without surfactants removed residual mineral particles: washed fibres showed a residual mass of 4.3 percent after thermal degradation, compared with 6.9 percent for unwashed material, and a density of 0.98 grams per cubic centimetre versus 1.19 for the unwashed fibres. The cleaned fibres were then regranulated in a twin-screw extruder, pelletised, dried, and injection-moulded into standardised test specimens alongside two benchmarks: fibres from a new successor product from the same manufacturer, and a compound replicating the original material recipe with commercial LLD-PE grades and a masterbatch of antioxidants, UV stabilisers and pigments.

The first question was how badly the fibres had aged in place. Light and electron microscopy revealed unmistakable surface damage: cracks running along the fibre axis, pronounced curling, and discolouration. Energy-dispersive spectroscopy line scans across fibre cross-sections showed oxygen penetration up to 65 micrometres into the used fibres, compared with only 12 micrometres in new ones, indicating oxidative damage roughly five times deeper. Fourier-transform infrared spectroscopy confirmed surface oxidation, detecting hydroxyl stretching above 3,000 inverse centimetres, carbonyl bands near 1,714 and ether bands around 1,031, alongside signals from inorganic sand contamination below 600. On the face of it, the material looked tired and chemically battered.

But surface appearance proved deceptive. Because oxygen and water diffuse only slowly into polyethylene, ageing concentrates in the outer layers, and surface analysis alone can misrepresent the state of the bulk polymer. When the researchers measured wetting behaviour on injection-moulded plates, the recycled used turf showed a total surface free energy of 29.7 millinewtons per metre, barely below the 31.1 and 31.5 of the reference and new-turf materials, with virtually identical polar contributions. Reprocessing had effectively erased the polarity signature of oxidation. Differential scanning calorimetry revealed the characteristic double melting peak of LLD-PE in all materials, with the used material’s higher-temperature maximum shifted down by only about 2 to 3 degrees Celsius, and crystallinity of 40.2 percent against 43.0 for new turf and 45.2 for the reference, contrary to the increase expected from ageing-induced chain scission. Thermogravimetric analysis found decomposition temperatures essentially unchanged, with the only notable difference being a higher mineral residue in the used material, attributable to fine sand.

Mechanical testing delivered perhaps the most striking result. Tensile strengths of the recycled used turf, the new turf and the reference compound were statistically indistinguishable at 16.5, 16.9 and 16.7 megapascals respectively, with a p-value of 0.65. The used material did show a slightly higher tensile modulus, 318.7 versus 300.9 megapascals for new turf, but the researchers attribute this minor stiffening to residual mineral contamination restricting chain mobility rather than to structural degradation, noting it would be negligible in industrial compounding where mineral fillers are standard. Remarkably, the recycled material exhibited the smallest scatter in properties, defying the common perception that recyclates behave unpredictably. Melt flow rates of 3.3 to 3.7 grams per ten minutes across all materials confirmed that processability was preserved, with the used material’s slightly lower flow likewise explained by steric hindrance from sand rather than molecular damage.

High-temperature gel permeation chromatography added molecular-level nuance. The number-average molar mass of the recycled used fibres was 25,358 daltons, compared with 29,978 for recycled new turf and 34,391 for the reference compound, and the dispersity rose to 7.41 from 5.25 in the reference. Crucially, the reduction of about 4,620 daltons between recycled new and recycled used material was of the same magnitude as the 4,413-dalton drop caused by reprocessing alone. In other words, shredding, compounding and extrusion did as much molecular damage as nearly two decades of outdoor service. The broader distribution and lower average chain length had no measurable consequence for thermal, mechanical or rheological performance.

The spectroscopic extracts told a story of well-designed protection. Reference spectra identified the phenolic antioxidant Irganox 1010 and the hindered amine light stabilisers Chimasorb 2020 and Chimasorb 944 in the masterbatch, and traces of these stabilisers remained detectable in extracts of the used granules even after 19 years. No low-molecular-weight polymer fragments or degradation products appeared in the extract, indicating that the additive package had largely prevented bulk degradation throughout service. The authors conclude that the lifetime-limiting factor for high-quality artificial turf is not oxidative ageing of the polymer but mechanical wear, abrasion and bending from intensive use, and that a carefully balanced formulation can preserve performance almost indefinitely.

The implications reach well beyond football pitches. Europe hosted some 30,000 full-size and 70,000 small-size artificial turf pitches in 2021, covering more than 300 million square metres, and around 12,000 tonnes of the annual end-of-life waste stream is polyethylene. The study argues that from a materials standpoint there is no obstacle to closing the loop for several hundred thousand tonnes of this polymer; the real barriers are logistics, the availability of clean material streams, and the scarcity of advanced recycling facilities capable of separating the multi-layered turf construction and removing mineral contamination. It also cautions that the FTIR surface analysis routine in the recycling industry is insufficient to judge material condition, recommending supplementary differential scanning calorimetry to probe the bulk. With life cycle assessments showing mechanical recycling preferable to incineration, the authors call for political incentives to create demand for high-quality recyclates. If supported by better sorting, melt filtration and re-stabilisation, aged turf fibres could travel from field to feedstock, turning one of recycling’s most awkward waste streams into a genuine circular resource.

Subject of Research: Mechanical and chemical characterisation of aged linear low-density polyethylene fibres recovered from end-of-life artificial turf for high-value recycling.

Article Title: From field to feedstock: Mechanical and chemical behaviour of aged polyethylene fibres from artificial turf

Article References: From field to feedstock: Mechanical and chemical behaviour of aged polyethylene fibres from artificial turf. (n.d.). https://doi.org/10.1016/j.clet.2026.101321

Image Credits: AI Generated

DOI: 10.1016/j.clet.2026.101321

Keywords: artificial turf, polyethylene recycling, LLD-PE, polymer ageing, mechanical recycling, UV stabilisers, DSC, gel permeation chromatography, FTIR spectroscopy, tensile testing, circular economy, plastic waste