Introduction

In recent years, cable-stayed bridges have gradually become one common structural form for many bridge construction projects, and the substantial increase in traffic volume resulted in frequent occurrence of fire accidents on the bridge, which could cause a serious cable loss even pose a serious threat to the main span for cable-stayed bridges. Several significant cable-stayed bridge fire events occurred, reflecting that the stay cable was a very vulnerable element under fire for cable-stayed bridge, however, receiving limited attention. For example, in 2004, the Rion–Antirion Bridge in Greece had undergone a fire caused by lightning strike leading to stay cable failure. In March 2007, a traffic accident involving a tractor-trailer and two buses generated a fire at deck level bring about one stay cable failure and damage to adjacent cables to a great extent [1]. In December 2008, a fire that lasted for 4 h caused by a short circuit of wire occurred on the Hedong Bridge in Guangzhou, leading to enormous losses for ten steel cables. Especially, in October 2014, Red-Stone Bridge in Hunan China, an under-construction cable-stayed bridge was subjected to a fire started by rebar cutting operation, resulting in breakage for nine steel cables and subsidence of bridge deck, as shown in Fig. 1. Clearly, it may be seen that the steel cables of the cable-stayed bridge have been frequently susceptible to fires associated with improper operation, a short circuit or a lighting strike. Given high flammability and heat release characteristics of the HDPE sheath, two or three adjacent cables would be involved in fire, which might be one important cause for leading to more significant damage or progressive collapse for the bridge.

Fig. 1
figure 1

Diagrammatic sketch of fire scenario for Red-Stone Bridge

Full size image

Bridge fire incidents are commonly caused by crashing of vehicles and burning of gasoline in the vicinity of the bridge [2]. In the studies of steel girder bridge fires, many former researches have focused on material and mechanical response of the structural elements. Spencer Quiel [3] presented a design method based on streamlined framework for calculating the mechanical response of bridges to open-air hydrocarbon pool fires that result from leakage accident of tank truck. Venkatesh Kodur [4] proposed an approach for overcoming fire hazard in bridges with the consideration of different geometrical features, material properties, design characteristics, and traffic demands. Alos-Moya and Payá-Zaforteza [5, 6] developed numerical models to predict the fire response of a bridge on the basis of different bridge types, respectively. Extensive experimental work has been conducted on steel bridge girders exposed to fire, which solved the problem of the lack of experimental data on the governing failure limit states in steel bridge girders [7,8,9].

On the other hand, many research studies have examined characteristics of fire in cable-stayed bridges. Ian Bennetts [10] identified and characterized potential fire scenarios relevant for bridges and considered the likely impact on a bridge structure. Spencer Quiel [11] applied the streamlined framework to calculate the effects of a tanker truck fire and demonstrated the structural fire mitigation design process for a cable-stayed bridge. Sang-Hyo Kim [12] investigated the variation of the temperature distribution in the steel box girder of a cable-stayed bridge during construction. Xu Gong et al. [13] conducted simulations by Abaqus to investigate the vulnerability of cable-supported bridges during fire hazards, and two different scenarios, ship fire and truck fire, were simulated. In the previous investigations, the electric cable was always taken as the main research object, and the effects of parameters such as ignition and pyrolysis on the cable fires have received wide attention [14,15,16]. Some other scholars focused on different small-scale experiments for cable burning in laboratory to speculate fire hazards in full scale [17, 18]. An enhanced model, vertical cable tray fire, was developed by Lu Li and Martinka, to investigate the cable combustion and flame spread with different cable spacings and different materials [19, 20]. As for the flame-retardant material, the thermal degradation and combustion properties of flame-retardant composites have been studied [21, 22]. Similarly, Makhlouf et al. [23, 24] investigated the thermal stability and combustibility of the low-density polyethylene by thermogravimetric analysis, vertical combustion test. Their valuable researches have greatly promoted the scientific understanding of bridge fire and cable fire, while experimental research on fire occurred on the steel cables for cable-stayed bridges is relative less, especially for the high steel cables. Therefore, the cable-stayed bridge structure with a high probability of being subjected to a serious damage in such a cable fire, explained the necessity of research work on fire growth, fire propagation and fire control on the steel cables for cable-stayed bridges.

This work was motivated by the lack of investigating on the fire characteristics under the condition that fire occurred on the high steel cables. One fire scenario was carried out by physical experiment established according to Red-Stone Bridge fire event. Fire growth and fire propagation behaviors were analyzed, and combustion characteristics and longitudinal temperature distribution for steel cable have been obtained. Furthermore, the effect of two kinds of fire extinguishing methods, dry powder and water, was examined. This study is expected to provide some references for fire protection and fire fighting for cable-stayed bridge.

Experimental

Cable specimen

Actually, the original length of the stay cable generally is up to tens of meters, and one complete cable experiment is not realistic. A specified length of steel cable was intercepted as an experimental specimen, which consists of a bundle of steel strands that are encased by HDPE (high-density polyethylene) sheath with 8 mm thickness, and each strand is sheathed by a thin HDPE sleeve with 3 mm thickness, as shown in Fig. 2. In order to better imitate the realistic fire scenario according to Red-Stone Bridge fire event, fire source was assumed to be located at the steel strands without sheath interior of pylon, and the steel strands for specimen were enwrapped by HDPE sheath incompletely. Figure 3 provides the three-dimensional layout for the steel cable specimen; the length of the part with HDPE sheath is 1200 mm, and the length of the bare steel strands is 600 mm, which could meet our requirements to observe the cable combustion and fire propagation phenomenon on the both sides, respectively.

Fig. 2
figure 2

Experimental specimen. a steel cable and internal steel strands, b outer HDPE sheath, c steel strands

Full size image
Fig. 3
figure 3

Schematic diagram of dimension for steel cable specimen

Full size image

Material properties

The internal sleeve for each strand and external protective sheath for the bundle of steel strands, the major combustible in the cable-stayed bridge fire event, consist of high-density polyethylene. In this work, a series of experiments on material property was conducted, and the combustion parameters and flame-retardant properties are shown in Table 1 and 2. As is shown, the HDPE material has a relatively small oxygen index value of 18.5 and a pretty high calorific value 4.63 × 107 J kg−1, indicating that the HDPE material is flammable, and shows high fire fatalness.

Table 1 Combustion parameters of material
Full size table
Table 2 Flame-retardant properties of material
Full size table

Test equipment

Figure 4 shows the testing device and thermocouples arrangement for the experimental model. The purpose of installment of the experiment equipment, two layers, was to observe the process of fire spread between different cables. The establishment of wall was aimed at imitating the role of the bridge pylon, to get a similar fire obstacle effect when the fire spread out of the pylon, the size of the wall was 2250 mm in height, 1000 mm in width, 200 mm in thickness, and the size of the reserved hole was 300 mm × 300 mm. The height of the fixed bracket could be adjusted, whose purpose was to maintain a certain degree of inclination for the cable specimen. Hence, slanting angle of the cable in the experiment was set to 18° according to the one representative cable involved in fire in Red-Stone Bridge fire event. Temperature data were collected by 48 thermocouples, which were distributed on both sides of the wall, in the inside part and upper part of the stay cable, as illustrated in Fig. 4.

Fig. 4
figure 4

Schematic of the experimental device and thermocouples arrangement

Full size image

In this work, the fire propagation and the temperature distribution of the cable were the research objects and the influence of the mechanical properties is ignored. This experiment mainly took the steel strand quantity as a variable, and the quantity for two test conditions was, respectively, set to be 73 and 55.

Results and discussion

Fire growth and propagation behaviors

To clearly show the complete and detailed process of fire spread on the single cable, flame status of eight typical moments is shown in Fig. 5. When internal steel strands were ignited, the polyethylene material on its surface would melt and drop at high temperature, causing lateral and longitudinal flow for burning drops. The most violet flame is at the central region for the cable cross section, and also fire began to spread along the cable at the central region, partly because a large amount of liquid molten substance was produced by the burning of the HDPE sleeve, and an aggregation flow of molten materials was formed in this region, driving flame spread downwardly along the steel cable. It took 20 minutes for flames to pass through the wall shown in Fig. 5c–e,which clearly prove that the wall has an obstacle effect for flame propagation.

Fig. 5
figure 5

Process of fire spread for single steel cable (73-wire case)

Full size image

Compared with the burning phenomenon of the bundle of steel strands without sheath on the right side, violent burning phenomenon of HDPE sheath behaved with a large mass of molten drops and a prolonged flame. Moreover, the fire spread rate for the bottom side of the cable is faster than that for the top side owing to inclined angle for the steel cable, as illustrated in Fig. 5f, g. When the bottom side of the sheath has been burnt through entirely, the flame for the top side still lay close to the wall, reflecting that the central and top sides of the cable were in a state of pyrolysis during this period. Besides, the flow of the liquid melts produced by HDPE sleeve combustion along the bottom of a cable took partial responsibility for priority fire spread of the bottom side. Another important experimental phenomenon observed was the formation of a large number of gaseous matters, which was due to the special pyrolysis mechanism and irregular chain breaking mechanism of the HDPE. From 95 min to 110 min, the HDPE sheath was consumed quickly due to the expansion of burning surface area, as shown in Fig. 5g. Then the internal steel strands went through the prosperous burning stage after the complete consumption of outer sheath. In the circumstance of slanting angle for 18°, with limits, of course, the ignition of the sheath would be earlier than that for the internal steel strands. However, a different fire spread phenomenon was appeared with the increase in the slanting angle, exhibiting that the sheath was ignited in a way of burning through from interior, as observed in one burning cable with the angle for 28° in the actual fire of Red-Stone Bridge, obviously the cause of this was the quickening flow of liquid melts and gather in a region inside of the sheath.

In conclusion, the HDPE sheath was poorly suited to resist fire, once ignited, enlarging the scale of fire, increasing the severity of damage and accelerating the flame propagation. Actually, on a burnt cable with high inclination, the internal steel strands, as the main load-carrying element for the cable, were more vulnerable to fire damage earlier, causing quicker fracture for the cable.

Drop ignition, one important fire spread phenomenon between two adjacent cables, is shown in Fig. 6, exhibiting the process of molten drop ignition. As illustrated from Fig. 6b–d, when the HDPE sheath on the left side was ignited, a large mass of molten substance with violent flame was produced and fell down, thus resulting in the accumulation of burning molten materials on the under-layer cable. Therefore, the top surface for the under-layer cable was ignited (see Fig. 6d), then the fire spread to both sides for the underneath cable, as illustrated in Fig. 6a. However, as for the bundle of steel strands without sheath on the right side, steel strands of the under-layer cable could not be ignited (see Fig. 6g), which was largely due to that the flames for the most of molten drops went out during the falling process and these molten drops lay scattered on the under-layer cable. The contrast shows that the HDPE sheath, if ignited, would play a more important role in advancing the fire development to the underneath cable.

Fig. 6
figure 6

Drop ignition behaviors between two adjacent cables (73-wire case)

Full size image

Herein, to explain more clearly the fire growth and propagation behaviors during the fire spread process on one single cable or between two adjacent cables, a schematic diagram was provided, as shown in Fig. 7, illustrating that three major kinds of fire spread modes were summed up, namely, upward propagation (see Fig. 7a), downward propagation (see Fig. 7b) and drop propagation (see Fig. 7c). As we know, heat transfer characteristics were different under the different fire spread modes, and upward propagation was largely influenced by a stronger convective heat transfer due to that high-temperature gas flowed through the surface of the unburned part of the steel strands. As a result, the upward sleeve along the steel strand was decomposed and volatiles were released, then leading to the upward propagation of the flame. The maximum height of flame was not higher than 100 cm and lasted for a very short time, combing with the spacing between two adjacent cables, which would not ignite the cable above the fire source. Unlike conventional spreading rules, the speed of downward propagation was not significantly less than that of upward propagation, mainly because of the flow of the liquid melts produced by HDPE sleeve combustion along the bottom of a cable and decomposition of the sheath, as illustrated in Fig. 7 (b). Moreover, the phenomenon of igniting internal steel strands by flow of liquid melts produced by HDPE sleeve was observed in this experiment, which took partial responsibility for priority fire spread of the bottom side, as shown in Fig. 6b. In addition, when the HDPE sheath was of high burning intensity, this kind of flaming molten drops was extremely easy to ignite the underneath adjacent steel cable, as illustrated in Fig. 6d. A large number of high-temperature melts fell on the surface of the under-layer HDPE sheath, resulting in softening the sheath and caving in, and the molten material accumulation in the depressed area was mainly responsible for the ignition of the under-layer cable.

Fig. 7
figure 7

Diagrammatic sketch for three primary kinds of spread modes

Full size image

Through comparative analysis, it can be concluded that the drop ignition for the outer sheath was more likely to lead to the burning of the under-layer cable than that for the internal steel strands. Hence one can see that, caught by surprise, the outer HDPE sheath originally as a protection element has become major part of the fire load, promoting fire development to the underneath adjacent cable. And it can be argued that its high flammability and heat release characteristics were fundamentally to blame.

Temperature distribution

To reflect longitudinal temperature distribution between two adjacent cables, thermocouples were mounted along the vertical plane. Figures 8–11 show temperature curves of four measuring points at different heights above the upper-layer cable for different positions under two conditions. As shown in Figs. 8 and 10, fire for the steel strands on the right side (see Fig. 6) first propagated along the lateral direction and then along the vertical direction, and temperature for measuring points above the steel strands increased rapidly in the vertical spread period, illustrating that there exists a positive correlation between temperature and flame height. In contrast, as shown in Figs. 9 and 11, temperature curves for the measuring points on the left-side cable (see Fig. 6) ran as a different variation tendency, periodical fluctuation for 73-wire case and attenuation fluctuation for 55-wire case, mainly as a consequence the unstable flame in the process of fire spreading to the internal steel strands. In addition, the maximum temperature for P5 was less than that of P1, indicating that the protective sheath has a hindering effect on the combustion of the internal steel strands. By comparison with two conditions, the affection of combustion intensity and flame height from quantity of steel strands was acquired, appearing as that the temperature peak and the flame height decrease with the decrease in the number of steel strands. As shown in Figs. 8 and 9, only temperatures at P1and P2 exceeded the ignition temperature for HDPE sheath in a period of time, in combination with cable spacing, inferring that the adjacent cable above the fire source could not be ignited.

Fig. 8
figure 8

Temperature curves measured above the right-side cable (73-wire case)

Full size image
Fig. 9
figure 9

Temperature curves measured above the left-side cable (73-wire case)

Full size image
Fig. 10
figure 10

Temperature curves measured above the right-side cable (55-wire case)

Full size image
Fig. 11
figure 11

Temperature curves measured above the left-side cable (55-wire case)

Full size image

Figures 12 and 14 show temperature curves of three different areas, P9 (top), P10 (middle) and P11 (bottom), measured inside the strands bundle under two conditions. It could be seen that in the initial stage, temperature at P9 increases sooner than those at P10 and P11 because the incendiary source was located on the upper surface. While in the rising stage, temperatures at P10 and P11 increase successively illustrating that the vertical fire spreading rate was accelerated, the rate of turning to 3.5 cm min−1 from 1.9 cm min−1 for 73-wire case and the rate of turning to 2.4 cm min−1 from 1.38 cm min−1 for 55-wire case, which was due to the increase in molten drops produced by combustion of the sleeve for steel strands. Approximately, the prosperity stage lasts 20 min, and temperature at P11 decreases sooner than those at P9 and P10, partly because of the fall of a large amount of molten material formed by combustion of the bottom side of the cable.

Fig. 12
figure 12

Temperature curves measured inside the strands on the right side (73-wire case)

Full size image
Fig. 13
figure 13

Temperature curves measured inside the HDPE sheath on the left side (73-wire case)

Full size image
Fig. 14
figure 14

Temperature curves measured inside the strands on the right side (55-wire case)

Full size image

In contrast, temperature curves of three typical positions P12 (top), P13 (middle) and P14 (bottom), measured inside the HDPE sheath show a different trend, as illustrated in Figs. 13 and 15. The fall of large lump of molten material produced by the burning of outer sheath and lack of oxygen inside the sheath led to a depressed burning for the inner steel strands, expressing as that temperature values for P13 and P14 were much lower than that for P12. On the left side, when the outer sheath was burnt out completely, the internal steel strands began to burn violently and temperature ascend velocity increased gradually, 120th minute to 165th minute for 73-wire case and 117th minute to 135th minute for 55-wire case, as shown in Figs. 13 and 15. The reason for the small increase in temperature at P14 in the 18th minute was the flow of liquid melts in 73-wire case; dissimilarly, the temperature curve for P12 rises in 55-wire case earlier and faster due to that the top side of sheath was first ignited.

Fig. 15
figure 15

Temperature curves measured inside the HDPE sheath on the left side (55-wire case)

Full size image

Through comparison of two working conditions for 73-wire case and 55-wire case, it is shown that the number of inner steel strands has a certain influence on the temperature distribution. These influences could be summarized from two aspects: the right side without sheath and the left side with sheath. By changing the steel strand quantity, for the right side, the temperature curve change laws of two conditions is similar; however, the temperature peak decrease with the decrease in the number of steel strands. In contrast, the influence of the decrease in the number of steel strand on the temperature for the measuring points on the left side was greater than that on the right side, appearing as temperature curve change laws rather than temperature peak at P12. Consequently, the influence of the HDPE sheath on temperature value is concluded that it can effectively reduce the maximum temperature of the cable in combustion, illustrating that the HDPE sheath can to some extent prevent from more strenuous burning for the internal strands although it plays an important role in the process of fire spread.

With the combined analysis of temperature curve and experimental phenomena, fire spread speed and maximum flame height for the upper-layer cable are tabulated in Table 3. On the right side, the occurrence of lateral spread was earlier than the vertical spread, and the lateral spread speed was lower than the vertical spread speed, the reason of which is that the molten drops were prone to run downward, which also proves that these drops play an important role in fire spread. In contrast, only the lateral spread speed could be obtained due to that the fire spread from around to center for the cross section on the left side. Meanwhile, the lateral spread speed of fire and maximum flame height on the left side decreased significantly compared with the fire spread on the right side, which is probably because of the existence of the outer HDPE sheath. After burnt out of sheath, the burning of internal steel strands became more intense. Though these results are achieved under one fire scenario, specified tangent angle, designated steel strand quantity and appointed fire source position, and cable-stayed bridge fire characteristics are also highly affected by bridge height, wind speed and cable spacing, etc., but the conclusions on temperature and fire spread in this work might provide effective data support for actual fire. In a summary, the two experimental cases show similar fire spread law; however, the fire spread speed and maximum flame heights for the 55-wire case were slightly less than that of 73-wire case. And the molten drops produced by the burning of HDPE sheath and sleeve determined the fire spread speed on the steel cable.

Table 3 Fire spread speed and maximum flame height
Full size table

Re-ignition

HDPE (high-density polyethylene) is a thermoplastic polyolefin, and the burning of HDPE has the characteristics of fierce fire and difficulty in extinguishing. Aiming at such fires, dry powder is one common method used in fire extinguishing according to national standard; however, water is one method that violates standard. In this experiments, two methods were chosen to examine the effects of fire extinguish aimed at cable fire, respectively. As shown in the dry powder case, it can be found in the fire extinguishing process that the flame disappeared firstly; however, the cable would be reignited after a period of time. On the contrary, the phenomenon of re-ignition was not appeared under the case of water extinguishing. Figure 16 provides the schematic diagram of re-ignition principle in the process of fire extinguishing. It can be seen that the dry powder fire extinguishing cannot reduce the temperature of the steel observably, and the temperature of the steel was still higher than the ignition point of the HDPE material in a short time after the fire extinguishing, which led to the re-ignition of the cable. While the water contacted the surface of the high-temperature steel strand, it immediately evaporated a lot of heat, which cooled down the high-temperature steel, bringing about the effective fire extinguishing. Through the comparison of two kinds of fire extinguishing methods, the effect of water extinguishing is better than that of dry powder extinguishing.

Fig. 16
figure 16

Schematic diagram of re-ignition principle in the process of fire extinguishing

Full size image

Conclusions

In this article, a set of experiments was conducted to investigate fire growth, propagation behaviors and temperature distribution around the cable under one special fire scenario on the high steel cables. These experimental results indicate that complex form and special layout of stay cables could bring about some interesting phenomenon, such as flow of liquid melts and drop ignition, which seriously influenced fire propagation characteristics for steel cables. Moreover, re-ignition behavior is observed in the fire extinguishing test for steel cables.

Beyond common expectation, the HDPE sheath originally as an anti-corrosion component, once ignited, actually becomes principal fire load, dramatically promoting the fire development to the underneath cable due to a large number of burning molten drops produced by the HDPE sheath combustion. Meanwhile, the behaviors of melting drop also lead to a rapid fire spread speed in the vertical direction. Average lateral spread speed is, respectively, 1.41 cm min−1 for 73-wire case and 1.22 cm min−1 for 55-wire case. Time required to ignite the under-layer cable is, respectively, 85 min and 100 min.

From the tests with two kinds of fire extinguishing methods, re-ignition phenomenon arises in the test condition of dry powder, mainly because the inner steel remains at a high temperature. Contrarily, re-ignition phenomenon is not observed in the test condition of water due to a notable reduction in the temperature of steel. It is noteworthy that, as to the fire on the steel cable, water, as one method deprecated in standard, actually has better performance on fire extinguishing than that with dry powder, one traditional method recommended by NFPA.

Though the tests were not conducted by full-scale field experiment, there exists a certain difference from the actual fire, such as the simulated bridge tower and neglected mechanical properties. However, the experiments in this work find out that special material composition for steel cable produces a great impact on fire occurrence and fire spreading, which are expected to be helpful for the fire protection and fire fighting for cable-stayed bridge. Actually, real cable fires are much more complex, and multiple factors will affect fire behavior such as bridge height, wind speed, cable spacing, fire position and inclination of cable which should be investigated further.