Hidden Cavities Drive Uneven Heating in Large Bridge Cables Under Fire

The steel cables that hold up long-span bridges are far more complicated on the inside than their smooth outer surface suggests. Parallel wire cables, bundles of high-strength steel wires serving as main cables, hangers and stay cables in cable-supported bridges, contain internal cavities between the bundled wires, and those gaps turn out to matter enormously when fire strikes. A new study published in Advances in Bridge Engineering reports temperature rise tests on large-diameter parallel wire cables ranging from 100 to 340 millimetres across, together with a high-fidelity numerical model that explicitly represents the cavity structures inside the cross-section. The work, led by Bo Geng, Wei Chen, Ruili Shen, Xinghua Chen, Pei Yuan and Zhi Zheng, shows that ignoring these internal cavities, as simplified homogeneous round steel models do, can misrepresent both how unevenly a cable heats and how hot it gets overall.

The motivation comes from a string of real bridge fires. The authors recount the 2005 fire in the stay cable lighting system of Greece’s Rio-Antirio Bridge, which caused stay cable rupture just 40 minutes after ignition. In 2014, an electric welding fire at Hunan’s Chishi Grand Bridge ruptured nine stay cables and produced more than two metres of deflection in the main girder on the severely affected side, along with extensive cracking of the concrete girder. A 2019 oil tanker fire at South Korea’s Ulsan Port Bridge burned the polyethylene anti-corrosion layer off multiple hangers, and a 2023 fire from lighting wires at Hebei’s Hutuo River Grand Bridge led to the rupture of one stay cable, which fell onto the bridge deck. Because main cables are irreplaceable primary load-bearing components whose fire damage can trigger catastrophic collapse of suspension bridges, and because hangers and stay cables lose bearing capacity under fire, the cross-sectional temperature of a cable is the key quantity governing fire-induced strength degradation.

Previous research on cable temperature fields has concentrated on small-scale components. Earlier experimental and numerical studies examined cable strands of roughly 15 to 77 millimetres in diameter under the ISO834 standard fire, 37-wire cables of about 35 millimetres, and 19-wire strands of roughly 15 millimetres, and a review cited in the paper noted that most cable temperature field studies involve components with diameters of 80 millimetres or less. Numerically, because temperature gradients are relatively small in such small cross-sections, researchers have commonly used simplified homogeneous round steel models that ignore the bundling characteristics of real cables. For large-scale cables, the authors argue, neglecting internal cavity effects undermines the reliability of such calculation models, and experimental and refined numerical studies on parallel wire cables that account for cavity influences have been scarce.

To close that gap, the team fabricated cable models from 127-wire galvanized parallel wire strands, using 6.0-millimetre-diameter high-strength steel wires with a tensile strength of 1960 megapascals. Constrained by the furnace chamber, they tested three specifications with diameters of 100, 200 and 340 millimetres, each 1.0 metre long, with porosity maintained at 18 to 20 percent, comparable to actual cable components. The tests were conducted in an electric resistance furnace with an inner diameter of 500 millimetres, a height of 600 millimetres and a rated temperature of 800 degrees Celsius. The furnace was gradually raised to 700 degrees Celsius and held constant, with temperature data logged every two minutes for 120 minutes. Each condition was repeated three times and averaged. The authors are candid about the limitation: the furnace provides a stable and uniform heat source that is convenient for analysing heat transfer mechanisms, but its temperature variation differs significantly from actual bridge fires, so real fire tests will be needed in future work.

Measuring points were embedded inside the cables during fabrication, with thermocouple probes bound to high-strength wires using fine iron wire. The three cable specifications carried 10, 19 and 31 measuring points respectively, arranged in layers from the surface to the centre, with four, four and five layers for the three cables. The K-type thermocouples had a measurement range of 0 to 1100 degrees Celsius and an accuracy of plus or minus 3 degrees. Stability checks showed that temperature rise laws among measuring points within the same layer were basically consistent, with maximum temperature difference rates between points of 7, 12 and 15 percent for the three cables, and maximum differences among the three repeated test runs not exceeding 50 degrees Celsius or 10 percent. Inside the furnace, the point near the heat source reached 760 degrees Celsius, exceeding the rated temperature due to equipment aging and controller overshoot, while points near the cable surface peaked around 730 degrees, with circumferential distribution relatively uniform.

The refined numerical model was built on the ABAQUS heat transfer platform. Heat from the source reaches the surface wires through convection and radiation, and then moves inward through heat conduction and cavity radiation between wires, with ABAQUS’s cavity radiation function simulating the radiative exchange within gaps enclosed by three or four wires. After a mesh sensitivity study, a mesh size of 0.1 millimetres was adopted, yielding 239,776 elements for the 200-millimetre cable cross-section. Thermal conductivity and specific heat capacity followed Eurocode 3 temperature-dependent expressions, with density taken as 7850 kilograms per cubic metre. A parameter study on the 200-millimetre cable identified a convective heat transfer coefficient of 25 watts per square metre per degree Celsius and a radiative heat transfer coefficient of 0.7, for both the surface and the cavities, as the values that best matched the experiments in the later stage of fire exposure.

The agreement between model and test defines the model’s validated accuracy. Numerical and experimental temperature rise histories were relatively consistent across the layers of all three cables, with maximum discrepancies of no more than 35 degrees Celsius within the first 60 minutes of fire exposure and less than 10 percent thereafter. For the largest cable, the maximum temperature difference was about 35 degrees for surface wires, a 16 percent difference rate, and about 34 degrees for central wires, a 32 percent rate, with the largest gaps occurring in the first 50 minutes while the furnace temperature was still rising rapidly. Surface wire agreement was better than central wire agreement, which the authors attribute to inevitable wire dislocations in the physical models that the ideally arranged numerical model cannot capture. Maximum surface wire temperatures fell with diameter, at 661, 650 and 603 degrees Celsius for the 100, 200 and 340 millimetre cables respectively, since larger cables heat more slowly.

Because the furnace temperature capped at around 700 to 800 degrees Celsius, well below real bridge fire temperatures, the team extended the validated model to three standard fire curves: ISO834, HC and ASTM E119, all exceeding 950 degrees Celsius at their peaks. For a 200-millimetre cable uniformly exposed for 120 minutes, the maximum surface wire temperatures were 982, 1035 and 937 degrees Celsius, central wire temperatures 727, 778 and 711 degrees, and maximum cross-sectional temperature gradients 254, 445 and 225 degrees Celsius. The HC fire, an open hydrocarbon fire suited to oil tanker fire scenarios on bridge decks, produced the highest temperatures and largest gradients, and the authors recommend it as the fire source model for fire resistance analysis of cable components. Peak values under HC fire were reached at 48 minutes for temperature and 12 minutes for gradient.

Porosity and diameter proved to be decisive internal factors. In numerical models of 127-wire cross-sections of roughly 70 millimetres equivalent diameter with porosities of 12, 15, 18 and 21 percent under HC fire, surface wire temperatures were essentially unaffected, reaching about 1080 degrees Celsius after 30 minutes, but central wire temperatures differed markedly within the first 60 minutes: the higher the porosity, the more delayed the internal temperature rise, because larger cavities mean less contact area and less efficient radiative transfer compared with conduction between solid wires. Larger porosity also produced greater cross-sectional temperature gradients. At main cable scale, models with diameters of 500, 700, 900 and 1100 millimetres showed maximum refined-model gradients of 768, 800, 867 and 902 degrees Celsius, against 366, 470, 550 and 612 degrees for homogeneous round steel, differences reaching 402, 330, 317 and 290 degrees. Under HC fire the homogeneous model differed from the actual cable by 357 degrees in internal wire temperature and 402 degrees in gradient, meaning the simplified model underestimates the non-uniformity of the temperature field while overestimating the overall temperature level, which could lead to conservative fire protection design. The authors note their model’s limitation is low computational efficiency, and plan high-precision equivalent models and thermomechanical coupling analysis of real bridge components under fire as future work.

Subject of Research: Heat transfer and internal cavity effects in large-diameter parallel wire cables of cable-supported bridges under fire

Article Title: Quantifying internal cavity effects on heat transfer in large-diameter parallel wire cables: tests and high-fidelity simulations

Article References: Geng, B., Chen, W., Shen, R., Chen, X., Yuan, P., & Zheng, Z. (2026). Quantifying internal cavity effects on heat transfer in large-diameter parallel wire cables: tests and high-fidelity simulations. Advances in Bridge Engineering, 7(1), Article 23. https://doi.org/10.1186/s43251-026-00211-w

Image Credits: AI Generated

DOI: 10.1186/s43251-026-00211-w

Keywords: parallel wire cables, bridge fires, heat transfer, internal cavities, temperature field, fire resistance, HC standard fire, cable porosity, ABAQUS simulation, temperature gradient, suspension bridge main cables, stay cables