Professional Documents
Culture Documents
1 s2.0 S0141029618340550 Main
1 s2.0 S0141029618340550 Main
1 s2.0 S0141029618340550 Main
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
a
College of Civil Engineering, Tongji University, 200092 Shanghai, China
b
Shanghai Municipal Maintenance & Management Co., Ltd, 200023 Shanghai, China
Keywords: Although structural robustness under different fire scenarios has been widely studied in numerous engineering
Performance-based design projects, performance-based method still needs to be further defined for better design of bridge structures under
Bridge structure fire accidents. This paper presents a practical framework for the performance-based design of bridge structures
Vehicle-induced fire accident under vehicle-induced fires. Fire scenarios, structural-thermal analysis method, fire-resistance levels, and a risk
Case study
analysis-based maintenance cost evaluation process are all defined in detail. The applicability and rationality of
this design process are illustrated through a typical case study. The results of the case study demonstrate that the
initial properties of the bridge structure can satisfy the defined fire-resistance levels properly, while additional
measures for decreasing the burning time are still needed to limit the fire maintenance cost in an acceptable
level. The proposed performance-based design process can be widely used in engineering practice.
⁎
Corresponding author.
E-mail address: a.chen@tongji.edu.cn (A. Chen).
https://doi.org/10.1016/j.engstruct.2019.109390
Received 6 December 2018; Received in revised form 8 July 2019; Accepted 8 July 2019
Available online 10 July 2019
0141-0296/ © 2019 Elsevier Ltd. All rights reserved.
R. Ma, et al. Engineering Structures 197 (2019) 109390
Fig. 1. Two fire accidents of bridge structures: (a) Oakland Highway Bridge, and (b) viaduct in Shanghai.
2
R. Ma, et al. Engineering Structures 197 (2019) 109390
simulation or experimental methods. In this case, the numerical simu- (1). Determine the potential combustibles and relative parameters,
lation method is suggested, considering its higher efficiency and lower according to the bridge site and traffic forecast results.
cost. The procedure for the numerical simulation-based analysis is (2). Determine the possible fire locations and vulnerable structural
presented in Fig. 2. components, according to the structural characteristics.
(3). Determine whether or not the wind effect should be considered,
(4). Carry out the risk analysis of maintenance cost. according to the environment of the bridge site.
(4). Determine the most representative fire scenarios based on the
Step (3) mainly focuses on the structural safety analysis under analyses results of the above steps.
certain bridge fire scenarios. In addition to this, several other factors,
such as economic loss, traffic interruption, and post-disaster repair, 2.3. Risk analysis-based maintenance cost calculation
need to be considered in the performance-based design process, as
mentioned above. All of these factors can be measured by the main- 2.3.1. Risk assessment of bridge fires
tenance cost from the bridge owner perspective. Owing to the high The main purpose of bridge fire risk assessment is to identify the
randomness of bridge fires, probability-based risk evaluation is more possible consequences as well as maintenance costs under typical fire
suitable for maintenance cost analysis. The detailed analysis method is scenarios. For quantitative calculation, bridge fire risk assessment can
discussed in Section 2.3. be defined as a product of the occurrence probability of a risk event and
its relative cost:
(5). Propose strategies to improve fire-resistance ability.
Risk = Pro· Con, (1)
If the structural responses are confirmed to exceed the performance where Risk is the risk assessment result, Pro is the occurrence prob-
targets, according to the analysis procedure in Fig. 2, certain appro- ability of a risk event, and Con is the relative cost when the risk event
priate strategies must be proposed to improve the fire-resistance ability occurs.
of the structures or components and the fire safety analysis should be In the performance-based design process, the bridge damage and its
carried out again. relative maintenance cost can be classified into different levels. Thus,
for a given bridge fire scenario, the risk assessment result is expressed
(6). Determine the final fire-resistance projects. as:
n
2.2. Fire scenarios definition Riskk = Proi ·Coni ,
i=1 (2)
The definition of fire scenarios is one of the most important steps in
where Riskk is the assessment result of risk event k , Proi is the
fire safety and maintenance cost analysis. Several standard temperature
curves, such as the ISO 834 [35], ASTM-E119 [36], HC [37], and RWS
[38] curves, have generally been adopted in traditional fire-resistance Fire hazard identification
analysis. However, these curves were mainly established for tunnels or
Environmental parameters,
high-rise buildings, and may be not suitable for bridge structures in Define bridge
Fire scenarios e.g. location, wind speed
open-air environments. Therefore, typical bridge fire scenarios need to
be redefined to reflect actual bridge fire situations. Little recognized Thermodynamic parameters,
research to bridge fire scenarios definition can be found at present. e.g. heat release rate
Considering the source of bridge fire is quite similar to tunnel fire,
tunnel fire researches and standards can be adopted as a reference for Fire scenarios CFD-based fire scenario
bridge fire definition. These definitions can be found in part of our analysis model
previous work [39]. Table 1 presents definitions of the fire scenario
parameters. It should be noticed that tunnel fire and bridge fire have
Structural-thermal model
some obvious differences as is shown in [12]. Thus, among these de-
finitions, the maximum heat release rate can be properly determined by Heat transfer
analysis Boundary conditions
tunnel fire research because they are both induced by vehicles. While
the burning time in bridge fire is usually shorter than that of the tunnel Proposed Temperature parameters
fire due to the sufficient oxygen supply, however, the bridge fire acci- measures
dents which last more than 6000 s were also reported in previous stu- Load condition of structure
Structural
dies [40–42], hence, the definition of burning time is suitable for en-
response Structural parameters
gineering design to obtain a conservative result. The most noticeable
limitation for the definition is the bridge fire size. In a bridge fire ac-
cident, fuel may spread over a significant distance and the fire size may Multi-level evaluation
be larger than tunnel fire. However, the spread of the fuel can be in- Fire-resistance standards
evaluation
fluenced by many environmental factors, e.g. the wind speed, wind Maintenance cost
direction, longitudinal slope of bridge deck, transverse slope of bridge
deck, road roughness, and so on, which makes the real size determi-
nation become more complicated. Besides, it is hard to say the larger Exceed
fire size will be favourable or unfavourable for engineering design, limitation?
Yes
since with the increase of fire size, the heat release rate per square
meter will be smaller and the maximum temperature of fire will also be No
lower. These limitations should be clearly noticed and further research
should be carried out to determine the real fire size considering spread End
effect.
The general procedure for the definition of bridge fire scenarios is Fig. 2. Performance-based design procedure under vehicle-induced fire acci-
detailed as follows. dents.
3
R. Ma, et al. Engineering Structures 197 (2019) 109390
Table 1
Parameter definitions of typical fire scenarios.
Fire scenarios Fire model Fire increasing modulus (kW/s2 ) Burning time (s) Maximum heat release rate (×103 kW) Fire size (m2)
where RiskF is the final risk assessment result. RiskFD = RiskFD, K + RiskFD, M , (6)
where RiskFD is the entire direct loss.
2.3.2. Maintenance cost analysis
As both the structural safety and other complex factors need to be 2.4. Definition of evaluation standards
considered in the overall performance-based design process, the bridge
maintenance cost can be divided into direct and indirect losses. In this The key point of the definition of evaluation standards in perfor-
case, direct loss refers to the investments required for component repair mance-based design under fire accidents is determining several dif-
and replacement, while indirect loss consists of other complicated ferent fire-resistance levels of the structure and then selecting the
factors, such as vehicle restrictions, traffic blockades, environmental proper level for different fire scenarios, according to the detailed ac-
pollution, and public opinion. Thus, the quantitative calculation of in- ceptable degrees. Thus, defining the explicit fire-resistance levels is the
direct loss may be extremely different according to the bridge site, first step towards the definition of evaluation standards.
traffic volume, maintenance time, and social popularity of the bridge.
However, the direct loss of the bridge can be approximately estimated 2.4.1. Fire-resistance levels
based on the following method. The fire-resistance levels of a bridge structure can be determined
For the same bridge, the direct losses of different components may according to the following aspects:
vary significantly, based on their importance to the structure.
According to the importance to the entire bridge, the components can (1). the influence on the traffic functions; for example, traffic restric-
be divided into three categories: key, main, and secondary components. tions and bridge blockades;
Key components refer to the key load-bearing bridge members, such as (2). the damage degrees of the components, which can be divided into
the towers, piers, main beams, and main cables, which may directly five levels: no damage, slight damage, medium damage, serious
determine the overall stability of the structure. Main components damage, and total damage;
generally refer to force-transmitting members, such as cables, hangers, (3). the elapsed time for complete repair of the bridge.
and supports. Secondary components include the bridge accessory
members, e.g., crash barriers, damping devices and the lighting system, Based on the above principles, the fire-resistance levels can be
which usually make little contribution to the carrying capacity of the classified as displayed in Table 2.
overall structure. In general, damage to several secondary components It should be noted that Table 2 only provides a general qualitative
has little influence on the initial bridge function, while damage to key description for the different levels. For the performance-based design of
components may lead to the destruction of the entire structure. a specific bridge, certain quantitative structural or economic standards
Two different situations of the corresponding results of key com- are surely required in practice.
ponent damage need to be considered, namely partial damage and total
damage. The partial damage cost mainly contains the investments of 2.4.2. Evaluation standards for different fire scenarios
component replacement, while the total damage refers to the loss of the During the operation period of the bridge structure, different fire
entire structural collapse. The cost of main component damage mainly scenarios may arise with varying occurrence probabilities and cause
includes the investments for component repair or replacement. The cost different damage to the structure. Therefore, in the performance-based
of secondary components is generally so small that it can be neglected design process, different fire scenarios can’t be treated equally, and
in most of the design processes. multi-level standards for different fire scenarios should be set appro-
Thus, the direct loss of key components can be calculated as follows: priately. The damage degrees of different fire scenarios are mainly
m n
determined by the maximum heat release rate, according to the defi-
RiskFD, K = Proi, j ·Coni, j + max (Pron, j )·ConT , nitions in Table 1. Table 3 presents the defined evaluation standards
j=1 i=1
j
(4) under different fire scenarios on the basis of Table 1.
where RiskFD, K is the direct loss of the key components; m is the number 2.4.3. Risk analysis-based criterion
of key components; n is the number of damage levels; Proi, j is the oc- In addition to the evaluation standards defined in Section 2.4.2 for
currence probability of damage level i in component j ; Coni, j is the structural safety analysis, the risk analysis-based criterion should be
direct loss of level i in component j ; max(Pron, j ) is the maximum oc- formulated to clarify the acceptable degree of the maintenance cost.
j
currence probability of all possible structural collapse types cause by The risk analysis-based failure criterion for bridge structures can be
the damage of key components; and ConT is the cost for reconstructing a expressed by the annual failure probability Pf or reliability index .
new bridge. Table 4 lists the reliability index values in different Chinese codes.
The direct loss of the main components can be calculated by: In general, the bridge structures can be regarded as the key
4
R. Ma, et al. Engineering Structures 197 (2019) 109390
3. Case study
No repair is needed
(1). A vehicle-induced bridge deck fire can arise anywhere along the
Parts of the main components are slightly damaged, while none of the key
(2). The wind load on the bridge deck can arise in all directions with
performance is significantly affected and traffic must be blockaded
functions
pavement. Previous study has shown that the deck fire has little influ-
ence on the main girder [46]. In this engineering project, the towers are
constructed by reinforced concrete with higher fire-resistance abilities,
while the cables are constructed using high-strength steel wires, which
Slight damage
Total damage
are easily affected by high temperatures. When a fire occurs near the
No damage
Medium
cable, it may lead to cable fracture and then cause an obvious impact on
Damage
damage
damage
Serious
degree
(V)
(II)
(I)
5
R. Ma, et al. Engineering Structures 197 (2019) 109390
Table 3
Definition of evaluation standards for different fire scenarios.
Maximum heat release rate (×103 kW) Typical fire scenarios Evaluation standards
Table 4
Suggested reliability index values in different codes.
Code index Key structure ( /Pf ) General structure ( /Pf ) Secondary structure ( /Pf )
−8 −6
GB/T 50283-1999 [43] 5.2/1.0 × 10 4.7/1.3 × 10 4.2/1.3 × 10−5
GB 50158-2010 [44] 4.5/3.4 × 10−6 4.0/3.2 × 10−5 3.5/2.3 × 10−4
GB 50068-2001 [45] 4.2/1.3 × 10−5 3.7/1.1 × 10−4 3.2/6.9 × 10−4
Table 5 Table 7
Fire locations and relative cable characteristics along span direction. Maximum heat release rate distribution of bus fire.
Fire locations A B, C, F D, E G H Maximum heat release rate (×103 kW) 20 30
Horizontal spacing of cables (m) 7.5 15 15 7.5 7.5 Probability 0.5 0.5
Horizontal dips of cables (°) 31 49 27 60 45
Table 6
Maximum heat release rate distribution of car fire. deck fire at these four positions are all assumed to be approximately
Maximum heat release rate (×103 kW) 1.5 10 0.25.
6
R. Ma, et al. Engineering Structures 197 (2019) 109390
Table 8
Maximum heat release rate distribution of truck fire.
Maximum heat release rate (×103 kW) 30 60 80 100 125 150 175 200
Table 9
Maximum heat release rate distribution of tanker fire.
Maximum heat release rate (×103 kW) 100 125 150 175 200 225 250 275 300
Probability 0.062 0.125 0.125 0.125 0.125 0.125 0.125 0.125 0.062
(1). The wind direction is parallel to the span direction and opposite to
Table 10 the cable direction (denoted AR, see Fig. 6a)).
Wind speed probability distribution. (2). The wind direction is parallel to the span direction and the same as
Wind speed 0 m/s 1 m/s 2 m/s 3 m/s 4 m/s 5 m/s
the cable direction (denoted AS, see Fig. 6b)).
(3). The wind direction is parallel to the cross-section direction and
Probability 0.061 0.703 0.226 0.009 0.001 0.000 towards to the cable plane (denoted TR, see Fig. 6c)).
(4). The wind direction is parallel to the cross-section direction and
away from the cable plane (denoted TS, see Fig. 6d)).
Fig. 6. Diagram of relationship between cables and wind direction: (a) AR condition, (b) AS condition, (c) TR condition, and (d) TS condition.
7
R. Ma, et al. Engineering Structures 197 (2019) 109390
Table 11
Definition of quantitative fire-resistance levels.
Maximum heat release rate (×103 kW) Resistance level Typical fire scenarios Damage degree Limited value of deflection-to-span ratio
Table 12
Thermodynamic parameters of cables.
Parameter Density (kg/m3) Thermal conductivity (W/(m· C)) Specific heat C (J/(kg· C)) Emissivity Heat transfer coefficient h (W/(m2· C))
Fig. 8. Simulation results of 300 MW tanker fire scenarios with wind speed: (a) 1 m/s, (b) 2 m/s, (c) 3 m/s, (d) 4 m/s, and (e) 5 m/s.
8
R. Ma, et al. Engineering Structures 197 (2019) 109390
900 900
750 750
Temperature (°C)
Temperature (°C)
600 600
450 450
D-I-0 m/s D-I-0 m/s
D-I-AR-3 m/s D-I-AR-3 m/s
300 D-I-AR-5 m/s 300 D-I-AR-5 m/s
D-I-TR-3 m/s D-I-TR-3 m/s
150 D-I-TR-5 m/s 150 D-I-TR-5 m/s
0 0
0 1500 3000 4500 6000 7500 9000 0 1500 3000 4500 6000 7500 9000
Time (s) Time (s)
a) Cable 1 b) Cable 2
900 900
D-I-0 m/s
750 D-I-AR-3 m/s 750
D-I-AR-5 m/s
D-I-0 m/s
D-I-TR-3 m/s
Temperature (°C)
D-I-AR-3 m/s
Temperature (°C)
600 600
D-I-TR-5 m/s
D-I-AR-5 m/s
D-I-TR-3 m/s
450 450
D-I-TR-5 m/s
300 300
150 150
0 0
0 1500 3000 4500 6000 7500 9000 0 1500 3000 4500 6000 7500 9000
Time (s) Time (s)
c) Cable 3 d) Cable 4
Fig. 9. Temperatures variations with burning times and different wind effects in position D of: (a) cable 1, (b) cable 2, (c) cable 3, and (d) cable 4.
speed and direction, needs to be calculated and compared carefully to simulation techniques.
obtain the most unfavourable conditions. The symbol of fire conditions (2). The structural-thermal analysis is carried out in ANSYS based on
is defined as follows: fire position along the span direction, fire position the fire temperature calculated by step (1) to obtain the variation
in the cross-section direction, wind direction, and wind speed; when the of cable temperature.
wind speed is equal to zero, the wind direction is ignored. For example, (3). The fractured cables are removed from the finite element model,
A-I-AS-3 m/s means the fire position is located in A (Fig. 3) and I and the structure performance is evaluated according to the stan-
(Fig. 4), the wind direction is parallel to the span direction and the same dards in Table 11.
as the cable direction (Fig. 6b)), and the wind speed is 3 m/s. If the
wind speed is equal to zero, it can be expressed as A-I-0 m/s. Noting that all of the following calculations and Figures in this
The whole process of structural safety analysis can be divided into section are presented using the 300 MW tanker fire scenario as an ex-
the following three steps: ample. The remaining three fire scenarios will only demonstrate the
final results directly at the end of this section. Fig. 8 plots the simulation
(1). The time-dependent fire temperature laws are simulated in FDS diagrams of the fire scenarios with different wind speeds. The simula-
based on computational fluid dynamics theory and large eddy tion volume for 300 MW tanker fire is 27 m × 27 m × 45 m and the
9
R. Ma, et al. Engineering Structures 197 (2019) 109390
Fig. 11. Failure analyses of different cables under the 300 MW tanker fire with conditions of: (a) A-I-0 m/s and (b) D-I-0 m/s.
mesh size is chosen to be 0.6 m × 0.6 m × 0.6 m to keep a better bal- dependent temperature laws of fire are obtained for the further struc-
ance between the simulation accuracy and simulation time [52]. It can tural-thermal analyses.
be observed that the maximum fire temperature is above 1100 °C, and In Fig. 6, the AS and TS conditions are both favourable to the
with the increase in wind speed, the flame tilt angle also increases, structural performance, as the AS condition causes the same tilt direc-
while the maximum temperature area decreases obviously. The time- tion between the cables and the flame itself, thereby affecting fewer
10
R. Ma, et al. Engineering Structures 197 (2019) 109390
0.6 compared to A-I-0 m/s, even if fewer cables are fractured in D-I-0 m/s.
A-I-0 m/s
D-I-0 m/s Using similar method, the maximum deflection-to-span ratios can be
0.4 calculated, as displayed in Table 13 for all fire resistance levels. Thus, it
can be determined that all of the values can satisfy the fire-resistance
demands defined in Table 11.
Vertical displacement(m)
0.2
Table 13
Calculation results of maximum value of deflection-to-span ratio with different fire-resistance levels.
Typical fire scenario Fire-resistance level Limited value of the deflection-to-span ratio Maximum value of deflection-to-span ratio
11
R. Ma, et al. Engineering Structures 197 (2019) 109390
900 900
Cable 1 Cable 1
Cable 2 Cable 2
750 750
Temperature (°C) Cable 3 Cable 3
Cable 4 Cable 4
Temperature (°C)
600 Cable 5 600 Cable 5
450 450
300 300
150 150
0 0
0 1500 3000 4500 6000 7500 9000 0 1500 3000 4500 6000 7500 9000
Time (s) Time (s)
a) A-I-0 m/s b) A-II-0 m/s
900
Cable 1 900
Cable 1
Cable 2 Cable 2
750 750
Cable 3 Cable 3
Cable 4
Temperature (°C)
Temperature (°C)
Cable 4
600 Cable 5 600 Cable 5
450 450
300 300
150 150
0 0
0 1500 3000 4500 6000 7500 9000 0 1500 3000 4500 6000 7500 9000
Time (s) Time (s)
c) A-III-0 m/s d) A-IV-0 m/s
Fig. 15. Temperature curves of different affected cables under the 300 MW tanker fire when fire scenarios are: (a) A-I-0 m/s, (b) A-II-0 m/s, (c) A-III-0 m/s, and (d) A-
IV-0 m/s.
Table 14
Probability calculation results of different fire types.
Number of fractured cables 0 1 2 3
Table 15 The occurrence probabilities of a car fire, bus fire, truck fire, and tanker
Maintenance cost of one cable in different positions. fire were determined to be 0.0684, 0.0405, 0.142, and 0.023 times per year
Position along span A B C D E F G H
in our previous work by means of the analytic hierarchy process and fuzzy
direction comprehensive evaluation [54]. Thus, the annual maintenance cost can be
calculated as RiskF = 30000 RMBs . The acceptable maintenance cost can
Maintenance cost of 0.675 0.212 0.212 0.738 0.738 0.212 0.357 0.524 be calculated as [Con]T × 10 6 = 2500 RMBs . It can be determined that
one cable
(million RMBs)
RiskF > [Con]T × 10 6 , thus some additional fire-resistance measures must
be drafted.
Fig. 11 shows that the burning time strongly influences the damage
degree of cables. Thus, fire extinguishers and fire hydrant systems can
Table 16 be considered to set up on the bridge. On the other hand, bridge fire
Probability calculation results of different fire types after adopting suggested management centre and fire monitoring system also should be estab-
measures. lished, further more, fire stations can be set up in the vicinity of bridge
Number of fractured cables 0 1 2 3 site, etc. The aim for these measures is to ensure that the burning time
would not exceed 45 min regardless of any fire scenarios. By adopting
Probability Car fire 1 0 0 0 these measures, the fracture probabilities under different fire types are
Bus fire 1 0 0 0
recalculated in Table 16. Thus, the annual maintenance cost can be
Truck fire 9.994 × 10−1 4.594 × 10−4 1.411 × 10−4 0
Tanker fire 9.872 × 10−1 1.126 × 10−2 1.565 × 10−3 0 calculated as RiskF = 200 RMBs . It can be judged that
RiskF < [Con]T × 10 6 , the suggested measures can well fit the design
requirement.
12
R. Ma, et al. Engineering Structures 197 (2019) 109390
13