Meng, Qu, Yong and Lee 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

1

SOCIETAL RISK BASED IMPACT ANALYSES OF TRAFFIC VOLUME AND PROPORTION OF HAZMAT TRANSPORT IN ROAD TUNNELS Qiang Meng1, Xiaobo Qu2, Kum Thong Yong 3, Shang Pang Lee4 1 Assistant Professor, 2Research Engineer Department of Civil Engineering National University of Singapore 10 Kent Ridge Crescent Singapore 117576 3

Principal Engineer, 4Senior Engineer Systems Assurance and Integration Division Land Transport Authority of Singapore No. 1 Hampshire Road Singapore 219428 To the committee of Transportation of Hazardous Materials Committee (AT040) for review Paper submitted to the 89th Annual Meeting of Transportation Research Board to be considered for presentation and publication in Transportation Research Record

Total Number of Words Number of words in text: Number of tables: Number of figures:

0 6

Total number of words Corresponding author:

(1 ×250) (6 ×250)

= = = =

5114 words 0 words equivalent 1500 words equivalent -----------------------------6614 words equivalent

Assistant Professor Qiang Meng Department of Civil Engineering National University of Singapore 10 Kent Ridge Crescent Singapore 117576 Email: [email protected] Phone:

TRB 2010 Annual Meeting CD-ROM

+65-65165494

Paper revised from original submittal.

Meng, Qu, Yong and Lee 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

2

ABSTRACT Quantitative Risk Assessment models of road tunnels have recently been proposed for assessing the societal risk expressed by frequency / number of fatalities (F/N) curve. For a particular road tunnel, traffic volume and proportion of vehicles carrying hazardous material (hazmat) are two key parameters that influence the societal risk. To evaluate the impact of these factors on the societal risk, this paper first presents an approach that employs the QRA model to generate F/N curves for all the possible combinations of these two factors. Some combinations of these two factors may result in F/N curves which do not meet a predetermined safety target. This paper thus proposes an excessive risk index in order to quantify the magnitude of the excessive risk. The two-factor impact analysis can be illustrated by a contour chart based on excessive risk. Finally, a case study on Singapore KPE tunnel was carried out. The result shows that the maximum traffic volume is 1400 vehs per hour per lane with a maximum proportion of hazmat vehicles not more than 2%. Further discussion also shows that higher traffic volume could be acceptable if certain effective operational procedures are taken to reduce the risk of hazmat transportation. Key words: QRA; Societal risk; Road tunnel; Impact analysis

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51

3

INTRODUCTION Road tunnels are considered as vital infrastructures which provide underground vehicular passageways for commuters and motorists. They contribute to the transportation systems from the viewpoints of economics and practicality because they enhance transportation system capacity and accessibility. However, safe operation of road tunnels is of significant concern due to the heavy traffic volume it carries. Any fatal accident occurred in a road tunnel may result in a catastrophic consequence. For example, in 1999, 39 people lost their lives in a fire disaster happened in Mont Blanc road tunnel from France to Italy and another disaster in Tauern road tunnel of Austria caused 12 fatalities [1]. In 2001, 8 people and 11 people were killed in Gleinalm road tunnel disaster and Saint Gotard road tunnel disaster, respectively [2]. These disasters in Europe brought about concerns on safety issues of road tunnels against various hazards such as fire and vehicle chain collision. Therefore, risk analysis has become one of the requirements under the EU directive (2004/54/EC) and Netherlands legislation on road tunnel. To evaluate various risks associated with a complex hazardous installation such as the nuclear power plant, the Quantitative Risk Assessment (QRA) methodology, including event trees, fault trees and consequence estimation models, are proposed and applied [3, 4, 5]. This is because the methodology systematically deals with all possible hazards, magnitude (severity) of possible adverse consequences and likelihood (frequency) of occurrence of each possible scenario [3, 4]. According to the QRA methodology, there are several QRA models developed by different countries for risk assessment of road tunnels; for example, TuRisMo model of Austria, TUNPRIM model of the Netherlands, Italian risk analysis model, French model, OECD/PIARC model [4, 6, 7, 8] and NUS-LTA model of Singapore [9]. All of these QRA models use the societal risk expressed by frequency vs number of fatalities (F/N) curve to measure the risk of the road tunnels. In addition, fire, flood, vehicle chain collision and vehicle explosion are considered as possible top events which can trigger road tunnel accidents. In reality, the societal risk of a road tunnel is determined by its geometries, traffic volume, vehicle composition, hazmat transport, safety provisions (E&M systems), distance between two evacuation exits etc. The road safety criterion in terms of societal risk is expressed by frequency vs number of fatalities (F/N) curve, it is based on the As Low As Reasonably Practicable (ALARP) principle [10]. Most countries have chosen the upper bound of the F/N curve as a safety target of the road tunnels [10, 11, 12, 13]. If the F/N curve generated by the QRA model is below the chosen safety target, the road tunnel is regarded as safe. Otherwise, risk reduction measures such as traffic volume control needs to be implemented [14]. This provides us an idea to determine the acceptable traffic volume and proportion of hazmat vehicles in terms of risks. It is well known that larger traffic volume of a road tunnel will results in higher traffic accident rate, traffic congestion and higher number of people at risk when an accident happens [15, 16]. Therefore, the traffic volume has a significant impact on the societal risk of a road tunnel. In addition, the risk is higher if a traffic accident involving hazmat vehicle happens in the road tunnel. The percentage of vehicle carrying hazmat out of the total number of vehicles passing through a road tunnel, which is referred to as the proportion of hazmat vehicles in this paper, is thus another vital impact factor on the societal risk [8, 17, 18]. The Land Transport Authorities therefore need to have an approach (procedure) that is able to analyze the impact of the traffic volume and the proportion of hazmat vehicles on the societal risk. The impact analysis approach not only supports the design consideration of a new tunnel by varying these two factors, but also evaluates various road tunnel traffic control schemes and hazardous transportation regulations. For example, Singapore prohibits hazmat transport in road tunnels under the Road Traffic Act. Although forbidding hazmat transport will increase the safety level of a road tunnel, but it will pose greater risk to the densely populated residents living close to or travelling along the alternative routes. More importantly, alternative paths avoiding road tunnels are generally much longer and may cause higher accident rate and severe consequence. In addition, given a combination of the traffic volume and the proportion of hazmat vehicles, the F/N curve generated by a QRA model may not fulfill a predetermined road safety target. In this case, we need an index to measure how far the societal risk represented by the F/N curve is from the safety target. This index can be used to determine the tolerable traffic volume and maximum acceptable proportion of

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

4

hazmat vehicles. Based on this index, we can plot a contour chart incorporating all the possible combinations of traffic volume and proportion of hazmat vehicles. According to this chart, traffic engineers are able to assess impact of these two parameters. In this paper, we will first propose a QRA model based procedure to analyze impact of two factors traffic volume and proportion of hazmat vehicles - on the societal risk of a road tunnel. For any combination of these two factors, an excessive risk index is defined as the average number of fatalities per year exceeding the safety target, and it reflects excessive fatalities to some extend. If the excessive risk is equal to zero for a particular combination of these two parameters, it means that the road tunnel with these two parameters fulfils the safety target. We can generate an excessive risk index based contour chart that reflects impact of various combinations of the traffic volume and the proportion of hazmat vehicles. Finally, a real road tunnel case study in Singapore is carried out to show the societal risk based impact analysis of traffic volume and proportion of hazmat vehicles. SOCIETAL RISK, EXPECTED VALUE AND SAFETY TARGET Societal risk is defined as the relationship between frequency and the number of people suffering from a specified level of harm in a given population from the realization of specified hazard [4, 19]. It can be represented graphically in the form of an F/N curve. The societal risk (F/N curve) has also been accepted in the quantitative risk assessment of road tunnels [4, 20]. A QRA model consists of event trees, fault trees and consequence estimations. A top event may trigger a number of possible scenarios associated with their frequencies and number of fatalities. The F/N curve reflects the relationship between the frequencies and the number of fatalities of all these possible scenarios on a double logarithmic scale. Let F ( N ) denoted the cumulative frequencies of all the scenarios with N or more fatalities. We thus have: n

F ( N ) =   Fi × δ ( xi − N ) 

24

(1)

i =1

25 26

where Fi is the yearly frequency of scenario i occurred per year; xi is the number of fatalities caused by

scenario i ; indicator function δ ( xi − N ) is defined by

1, if xi ≥ N δ ( xi − N ) =  0, otherwise

27 28 29

(2)

With the frequency shown by eqn.(1), the expected value for the number of fatalities per year (EV) can be calculated by n

EV =  ( Fi × xi )

30

(3)

i =1

31

The upper bound curve of F ( N ) is usually adopted by various countries as the safety target [3] [12]:

F (N) ≤

32

C Nk

(4)

33 34

where k and C specified the steepness and intercept point. Alternatively, eqn. (4) can also be represented by:

35

(5)

36 37 38 39 40 41 42

k log ( N ) + log ( F ( N ) ) ≤ log ( C )

It should be noted that k represents a slope, i.e. gradient of the safety target, and C denotes an intercept, i.e. constant value that determines the position of the target. Different combinations of k and C express various strictness degrees of the safety targets. As a result, different countries may propose their own safety targets. For example, the k and C values adopted by Netherlands are C=10-3 and k=2, while Switzerland adopts C=10-4 and k=1 [20, 21, 22].

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

5

QRA MODELS AND SOFTWARE TOOLS As described by Jonkman et al. [11] and Vrouwenvelder et al. [23], procedure of building a QRA model can be described as follows: Firstly, all possible hazards such as fire and flood are identified as top events. After that, fault tree and event tree for each top event are built. Event tree consists of a number of particular scenarios triggered by the top event and fault tree is used to estimate frequency of a top event that could occur. Finally, consequence estimation models are required to calculate number of fatalities for various scenarios involved in an event tree. After obtaining frequency and fatality of each scenario, the societal risk defined by eqn. (1) can be calculated. Among the existing QRA models of road tunnels, the Dutch model and PIARC/OECD/EU model (QRAM) are well recognized by researchers and the Land Transport Authorities. A spreadsheet based software was also developed to realize the latter model [24]. However, both models have their own limitations. The former model is unable to deal with non-homogeneous urban road tunnels which have different parameters for different sections of the tunnel. As for the QRAM model, it only focuses on the risk assessment of road tunnels with hazmat transportation [20]. To overcome these limitations, National University of Singapore and Land Transport Authority of Singapore have jointly developed a novel QRA model for road tunnels in 2009, called the NUS-LTA QRA model [9]. This QRA model is applicable for non-homogenous urban road tunnels which may not have the same input parameters, such as accident rate, traffic volume and tunnel configurations, for each tunnel section. The QRA model consists of seven top events (fire, flood, toxic gases generated by traffic congestion, tunnel collapse, chain collision, explosion, and spillages due to hazmat), 7 event trees, 7 fault trees and 9 consequence estimation methods. 16 categories of hazardous materials (Acrolein, Acrylonitrile, Ammonia, Bromine, Carbon monoxide, Chlorine, Ethylene oxide, Hydrogen chloride, Hydrogen fluoride, Hydrogen sulphide, Methyl bromide, Methyl isocyanate, Nitrogen dioxide, Phosgene, and Sulphur dioxide), 5883 scenarios and 19 Electrical & Mechanical systems are taken into account in the NUS-LTA QRA model. The work was conferred by the Ministry of Transportation Minister Innovation Award 2009 in Singapore. The key input parameters required by the NUS-LTA model can be categorized into 4 types: traffic parameters, tunnel user characteristics, tunnel geometries, and parameters associated with tunnel E & M systems. Traffic parameters include traffic volume, accident frequencies, vehicle composition (the different proportions of various types of vehicles), the headway distances, fraction of peak / off-peak hours, and etc. Tunnel user characteristics refer to the reaction time of tunnel users, movement speeds of tunnel users, proportion of aged tunnel users, proportion of experienced drivers, and etc. Tunnel geometries relate to the distance between two consecutive emergency exits, number of lanes, tunnel sectional area, tunnel height etc. Parameters associated with tunnel E & M systems are the functional parameters of those systems and probabilities of those systems failing to work. Note that the average vehicle speed is not an input parameter because it will dynamically change for different headway and traffic volume. To facilitate the use of the NUS-LTA QRA model, a user-friendly QRA software tool implementing the QRA model is developed by the Object-Oriented Design (OOD) approach. The QRA software tool is coded by C#, and it uses Microsoft Access database and XML files to manage the data. The NUS-LTA QRA model and QRA software tool have already been used by LTA of Singapore to perform the road tunnel risk assessment for Marina Coastal Expressway (MCE) [25]. IMPACT ANALYSIS PROCEDURE Figure 1 depicts an F/N curve diagram generated by NUS-LTA QRA software tool. The diamonds on the F/N curve are generated by the software in the case that traffic volume is 1800 vehs per hour per lane. The asterisks on the F/N curve are generated when traffic volume is 1600 vehs per hour per lane. According to the safety target shown by the straight line in Figure 1, both traffic volume scenarios are unacceptable. However, the F/N curve with diamonds is much more dangerous than the asterisks curve. In order to quantify the severity of an unacceptable scenario, we can use the sum of distances between the safety target and those F/N points which are above the safety target as the severity risk index, namely,

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee

6 n

S e =  max{log(F ( N i )) − log(C ) + k log( N i ), 0}

1

(6)

i =1

2 3 4 5 6

where Se is the severity risk index; Ni is the selected value of number of fatalities; n is the number of fatalities. For an acceptable F/N curve, Se takes the value of 0. Larger value of the Se reflects higher severity risk attain by the F/N curve. However, the severity risk index defined above does not provide any practical and engineering implication. Hence, an excessive risk index denoted by Re is proposed as follows: n  C   C   Re =  max  N i ×  F ( N i ) − k  −  F ( N i +1 ) − k  , Ni   N i +1  i =1  

7 8 9 10 11 12 13

 0 

(7)

In fact, the Re reflects the number of fatalities per year exceeding safety target due to dangerous events, which is part of the expected value (EV). According to the definition of F/N curve, the difference in value of y-coordinates equals the frequency of scenario with corresponding fatalities on the x-coordinates. If the frequency of scenario occurred is smaller than the corresponding value in safety target line, the scenario is regarded as safe and we will not add the number of fatalities due to these scenarios. Impact Analysis of Traffic Volume

Freqeuncy (Per Year)

1E-03

1E-04

1600 Veh/ Hour

1E-05

1800 Veh/ Hour

1E-06

1E-07 1

14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

Number of fatalities

10

Figure 1 An example of F/N curve IMPACT ANALYSIS PROCEDURE Impact analysis of the two key parameters, traffic volume and proportion of hazmat vehicles, on the societal risk for a road tunnel can be implemented easily by a QRA software such as the NUS-LTA QRA software tool. F/N curve could be generated from the software tool on a case by case basis. Figure 2 illustrates the two-factor impact analysis procedure. The QRA model based procedure works as follows. Firstly, determine the ranges of traffic volume and proportion of hazmat vehicles and discretize these ranges. Note that the values of other input parameters required by the QRA model should be acquire from the historical records or design documents of the road tunnels. Quantitative risk analysis could be performed for all the possible sets of input parameters (different combinations of traffic volume and proportion of hazmat vehicles). F/N curves and excessive risk for all the combinations of these two parameters could be generated by the QRA software tool. Finally, the excessive risk contour based chart could be drawn using curve fitting method.

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

7

Figure 2 Flow chart of impact analysis CASE STUDY The Kallang / Paya Lebar Expressway (KPE) of Singapore, shown in Figure 3, is 12km in total length and 9km is built underground as a road tunnel, which was built to serve the growing traffic demands of the north-eastern sector of Singapore. It is also the longest road tunnel in the South East Asia. The KPE road tunnel is a dual 3-lane, 9km underground passageway and has nine entry slip roads, eight exit slip roads and six ventilation buildings. The accident frequency of the road tunnel is 560 per year according to the historical records. The distance between two emergency exits is 100 meters. The tunnel air velocity when tunnel ventilation works normally is 4 m/s. There is a 24-hrs manned Operation Control Centre (OCC) at one ventilation building and an unmanned hot standby OCC located in another ventilation building. The major E&M Systems of the KPE tunnel include Tunnel Ventilation and Environmental Control System, Fire Protection System, Electrical System, Integrated Traffic and Plant Management System and Communications System. The functionality and working profiles of the E & M systems can be obtained from their instruction manuals. The values of the vehicle profiles are obtainable

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee 1 2

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

8

from the OCC. The NUS-LTA QRA model and software tool are being utilized to do the impact analysis for this case study.

Figure 3 KPE road tunnels in Singapore Traffic volume impact analysis In order to obtain the maximum tolerable value of tunnel traffic volume, impact analysis of traffic volume is performed with the assumption that the proportion of hazmat vehicles is 0%. All the other input parameters (default values) are based on operational data collected from KPE road tunnel of Singapore. The traffic volume varies from 1000 to 1800 vehs per hour per lane. The F/N curves associated with different traffic volumes are shown in Figure 4. The safety target of (10-3 / N2) is applied in this case study and this is similar to the safety target used in Netherlands. Figure 4(a) depicts the F/N curves of traffic volumes varying from 1000 vehs per hour per lane to 1400 vehs per hour per lane. In these three scenarios, the KPE tunnel can be considered as safe according to the selected safety target. An interesting finding is that only some frequencies of F/N values increase significantly with an increase of traffic volume. These F/N points could be regarded as the bottleneck events for improving the tolerable traffic volume in terms of risk. Tunnel operators can refer to the respective event trees to find out the details of the corresponding scenarios and implement appropriate mitigating measures to reduce the frequencies and/or consequences of those scenarios so as to increase the tunnel traffic volume without exceeding the safety target. Figure 4(b) shows the F/N curves of traffic volumes varying from 1600 to 1800 vehs per hour per lane. It can be seen that these two scenarios are not acceptable in accordance to the selected safety target. Therefore, it can be concluded that the maximum tolerable traffic volume of the KPE road tunnel should not exceed 1600 vehs per hour per lane. Vehicle speed and traffic volume are interrelated, as traffic volume increases the vehicle speed will decrease. This explains why there is no significant difference between the two curves in Figure 4(b). Although the increase in traffic volume reduces the safety level of the tunnel, the resultant decrease in vehicle speed will reduce the accident frequencies and severities such that there is no evident change in the safety level.

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee

9

Impact Analys is of Traffic Volume 1E-03

Freqeuncy (Per Year)

Safety Target 1E-04

1000 Ve h/ Hour 1200 Ve h/ Hour 1400 Ve h/ Hour

1E-05

1E-06

1E-07 1

Numbe r of fatalitie s

10

(a)

Impact Analysis of Traffic Volume 1E-03

Freqeuncy (Per Year)

Safety Target 1E-04

1600 Veh/ Hour

1E-05

1800 Veh/ Hour

1E-06

1E-07 1

Number of fatalities

10

(b)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Figure 4 Two F/N curve diagrams of KPE road tunnel Impact analyses on proportion of hazmat vehicle Currently, hazmat vehicle is not allowed in all road tunnels in Singapore according to the Road Traffic Act. The impact analysis done helps to demonstrate the necessity for this implementation. The proportion of hazmat vehicles ranges from 0 to 10%. Meanwhile, the traffic volume takes value of 1200 to 1800 vehs per hour per lane. The F/N curves of various scenarios are as follows, where Figure 5(a), Figure 5(b), Figure 5(c), and Figure 5(d) represent the scenarios with traffic volume of 1200, 1400, 1600, and 1800 vehs per hour per lane, respectively. From figure 5(a), it shows that if the traffic volume is relatively low, even the proportion of hazmat vehicles is 10%, the F/N curve is able to satisfy the safety target. However, if the KPE road tunnel has a traffic volume around 1400 vehs per hour per lane due to the densely populated nature of Singapore, even 2% of hazmat vehicles could impose a significant threat to the tunnel users – some of the F/N points are very close to the safety target (Figure 5(b)).

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee

10

1E-03

1E-03

1E-04

0 2% 4%

1E-05

6% 8% 10%

1E-06

Frequency (Per Year)

Frequency (Per Year)

TRB 2010 Annual Meeting CD-ROM

Impact Analysis of hazmat transport proportion

Impact Analysis of hazmat transport proportion

1E-07

1E-04

0 2% 4%

1E-05

6% 8% 10%

1E-06

1E-07 1

Number of Fatalities

10

1

Impact Analysis of hazmat transport proportion

10

Impact Analysis of hazmat transport proportion

1E-03

1E-03

1E-04

0 2% 4%

1E-05

6% 8% 10%

1E-06

Paper revised from original submittal.

1E-07

Frequency (Per Year)

Frequency (Per Year)

Number of Fatalities

1E-04

0 2% 4%

1E-05

6% 8% 10%

1E-06

1E-07 1

Number of Fatalities

10

1

Number of Fatalities

Figure 5 Four F/N curve diagrams of KPE road tunnel

10

Meng, Qu, Yong and Lee

11

Excessive risk index contour chart The above impact analysis indicates that traffic volume and the proportion of hazmat vehicles have significant impact to the tunnel operational risk. Therefore both parameters should be taken into consideration during impact analysis. For each combination of these two parameters, the excessive risk index can be calculated by using the NUS-LTA QRA software tool. The excessive risk index based contour chart can be drawn by varying traffic volume ranging from 1000 to 1800 vehs per hour per lane with an interval of 200 and proportion of hazmat vehicles ranging from 0% to 10%. After obtaining the excessive risk value of all the possible combinations (30 combinations) of the two variables, the risk contour chart can be drawn using curve fitting method. The contour chart is shown in Figure 6. According to this figure, it can be observed that the region 1 is considered as safe region. The excessive risk index will become bigger as the traffic volume and proportion of hazmat vehicles increases. This contour chart can assist policy makers to decide the most appropriate combination of traffic volume and proportion of hazmat vehicles for any given safety target. As the population of Singapore increases, there will be more road tunnels being built due to the need for more efficient use of land. Transportation of hazmat through road tunnels may be allowed in the future due to limited alternatives. Considering the urban nature of Singapore road tunnels, the traffic volume would tend to be on the higher end of the range that we have considered in our impact analyses. This means that the road tunnels may be operating near risk contour line 0 which is the maximum allowable traffic volume based on the selected safety target. If transportation of hazmat is allowed in road tunnels, then the risk index may increase nearer to risk contour line 10-5 which does not satisfy the safety target. However, upon close examination of the plotted F/N curves, the excessive risk on the risk contour line 10-5 is not significant as only some points of the F/N curve exceed the selected safety target by a small magnitude (10-5 fatality more per year). In addition, according to the official LTA website, the percentage of Very Heavy Goods Vehicles (VHGVs) is only about 1.4%. Vehicles with license to carry hazmat will be even smaller than this percentage. If effective operational procedures are implemented, it will helps to mitigate the risk involved in the transportation of hazmat in road tunnels, then this would effectively shift the excessive risk index from risk contour line 10-5 to risk contour line 0 which could then satisfy the selected safety target. Operational procedures to be considered in reducing the risk may include the followings: 1) Transportation of the hazmat is allowed only during off peak hours so as to reduce the risk exposure to other motorists. In Singapore, the details (route, time, content) of hazmat transport are under the supervision of Singapore Civil Defense Force (SCDF). The risk index contour chart could be considered as a reference to decide whether to approve the route and time of the vehicle carrying hazmat or not. 2) To ensure a safe distance of more than 100m or at least the braking distance depending on the speed limit of the road tunnel from the vehicle carrying hazmat to the next vehicle to avoid accidents which is one of the main causes of fire. 3) Electronic road pricing approach and / or ramp metering method can be applied to limit the traffic flow [26]. The risk index contour chart can examine the efficiency of road pricing strategy from the viewpoint of risk reduction.

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee

12

Excessive Risk Contour Chart 10% 8%

Hazmat Proportion

6% 4% 2% 0% 800

1000

1200

1400

1600

1800

Traffic Volume (vehs per hour per lane) Figure 6 Risk contour chart in terms of excessive risk index CONCLUSIONS This paper has developed a QRA model based impact analysis procedure to determine two critical parameters: the maximum tolerable traffic volume and the acceptable proportion of hazmat vehicles. In addition, the excessive risk index is proposed to quantify the severities of unacceptable scenarios relating to road tunnel operations which do not meet a predetermined safety target. A contour chart based on the excessive risk index can be plotted once obtaining all the possible combinations of the two critical parameters. NUS-LTA QRA software tool is applied to generate the F/N curves and values of excessive risk index for the KPE road tunnel case study. The case study shows that the maximum tolerable traffic volume is 1400 vehs per hour per lane and the maximum acceptable hazmat vehicle proportion of not more than 2% of total traffic volume. If the traffic volume is already near or at the maximum limit based on the selected target, transportation of hazmat may still be feasible if the excessive risk is minimize through effective operational procedures and the proportion of hazmat vehicles is kept at less than 2 percent out of the overall traffic volumes. ACKNOWLEDGMENTS This paper is supported by the innovation fund of Land Transport Authority of Singapore (Contract No: ER 253). The work is conferred the Ministry of Transportation Minister’s Innovation Award. Special thanks will also be expressed to Mr. Leong Kwok Weng, Dr. Samuel Chan and Mr. Yap Kwee Seng from LTA for their supports on data collection for this research. REFERENCES [1]. F. Vuilleumier, A. Weatherill, B. Crausaz. Safety Aspects of Railway and Road Tunnel: Example of the Lotschberg Railway Tunnel and Mont-Blanc Road Tunnel, Tunn. Undergr. Sp. Tech., Vol. 17 (2002) 153–158. [2]. A. Leitner, The Fire Catastrophe in the Tauern Tunnel: Experience and Conclusions for the Austrian Guidelines. Tunn. Undergr. Sp. Tech., Vol. 16 (2001) 217–223. [3]. D. Jones, Nomenclature for Hazard and Risk Assessment in the Process Industries, Institution of Chemical Engineers, Rugby, 1992. [4]. PIARC Technical Committee C3.3 Road tunnel operation, Risk Analysis for Road Tunnels, May 2008. http://publications.piarc.org/ressources/publications_files/4/2234,TM2008R02-WEB.pdf. Accessed 19th July 2008.

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee

13

[5]. G. Antonioni, G. Spadoni, and V. Cozzani, A Methodology for the Quantitative Risk Assessment of Major Accidents Triggered by Seismic Events, J. Hazard. Mater., Vol. 147 (2007) 48 – 59. [6]. L. A. Brussaard, M. M. Kruiskamp, and M.P.O Essink. The Dutch Model for the Quantitative Risk Analysis of Road Tunnels, Presented at European Safety & Reliability International Conference ESREL 2001 – towards a Safer World, Torino, Italy, September 16-20, 2001. [7]. H. Knoflacher, Quantitative Risk Analysis Model for Transport of Dangerous Goods through Tunnels. 2002. www.ivv.tuwien.ac.at/fileadmin/mediapool-verkehrsp lanung/Bilder/Fors-chung/2002_ 3.pdf . Accessed 2 January 2008. [8]. OECD/PIARC, ERS2, Transport of Dangerous Goods Through Road Tunnels. 1997. http://www.piarc.org/library/aipcr/3/465FXt072588e5301sj89OVv.pdf. Accessed 19th Aug. 2008. [9]. Q. Meng, T. F. Fwa, X. Qu, X. Wang, K. T. Yong, S. P. Lee, Y. H. Wong, and Y. Vivi, Project Report on Quantitative Risk Assessment Models for Road Tunnels. Department of Civil Engineering, National University of Singapore and Systems Assurances and Integration Division, Land Transport Authority of Singapore. [10]. T. Aven, On the Ethical Justification for the Use of Risk Acceptance Criteria, Risk Anal., Vol. 27 (2009) 303-312. [11]. S. N. Jonkman, P. H. A. J. M. Gelder, and J. K. Vrijling, An Overview of Quantitative Risk Measures for Loss of Life and Economic Damage, J. Hazard. Mater., Vol. 99 (2003) 1~30. [12]. L. Kpanake, B. Chauvin, and E. Mullet, Societal Risk Perception among African Villagers without Access to Media, Risk Anal., Vol. 28 (2008) 193-202. [13]. D. J. Ball, Boehmer-Christiansen S., Societal Concerns and Risk Decisions, J. Hazard. Mater., Vol. 144 (2007) 556- 563. [14]. N. Paltrinieri, G. Landucci, M. Molag, S. Bonvicini, G. Spadoni, and V. Cozzani, Risk Reduction in Road and Rail LPG Transportation by Passive Fire Protection, J. Hazard. Mater., Vol. 167 (2009) 332-344. [15]. G. A. Davis, Accident Reduction Factors and Casual Inference in Traffic Safety Studies: A Review, Accid. Anal. Prev., Vol. 32 (2000) 95 – 105. [16]. Abdel-Aty M. and A. Pande., Crash Data Analysis: Collective vs. Individual Crash Level Approach. J. Saf. Res., Vol. 38 (2007) 581 – 587. [17]. V. Cozzani, S. Bonvicini, G. Spadoni, S. Zanelli, Hazmat Transport: A methodological Framework for the Risk Analysis of Marshalling Yards, J. Hazard. Mater., Vol. 147 (2007) 412 – 423. [18]. B. Fabiano, F. Curro, E. Palazzi, and P. Pastorino, A Framework for Risk Assessment and Decision-making Strategies in Dangerous Good Transportation, J. Hazard. Mater., Vol. 93 (2002) 1-15. [19]. V. M. Trbojevic, Risk Criteria in EU, undated, http://www.risk-support.co.uk/B26P2-Trbojevic-final.pdf. Accessed 2 Jan. 2009. [20]. Q. Meng, X. Wang, X. QU, K. T. Yong. S. P. Lee, and S. C. Wong, Quantitative Risk Assessment Models of Road Tunnels – State of the Art and Their Implications for Singapore’s Road Tunnels. Rresented at the 2nd International Tunnel Safety Forum for Road and Rail. Lyon, France. April 2009. [21]. P.J.M. Stallen, R. Geerts, H. K. Vrijling, Three Conceptions of Quantitative Societal Risk. Risk Anal., Vol. 16 (1996) 635~644. [22]. P.H. Botterlberghs, Risk Analysis and Safety Policy Developments in the Netherlands. J. Hazard. Mater., Vol. 71 (2000) 59~84. [23]. T. Vrouwenvelder, R. Lovegrove, M. Holicky, P. Tanner, and G. Canisius, Risk assessment and risk communication in civil engineering, Safety, risk and reliability - trends in engineering, 2001. http://www.bouwweb.nl/pdf/riskmalta2001.pdf. Accessed 4 Jan. 2009. [24]. Organization for Economic Cooperation and Development. Safety in Tunnels Transport of Dangerous Goods through Road Tunnels. 2001. http://www.oecd.org/document/9/0,3343,en_2649_34351_2071369_1_1_1_1,00.html. Accessed 10 Jan. 2009.

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.

Meng, Qu, Yong and Lee

14

[25]. C. Thangasamy, Y. H. Wong, S. P. Lee, Q. Meng, and X. Qu; Quantitative Risk Analysis for Marina Coastal Expressway Using Software Based Approach; Presented at the 2nd World Road Congress; 26-28 Oct. 2009, Singapore. [26]. M. Abdel-Aty, A. Dhindsa., and V. Gayah, Considering Various ALINEA Ramp Metering Strategies for Crash Risk Mitigation on Freeways under Congested Regime, Transport. Res. C-Emer, Vol. 15 (2007) 113 – 134.

TRB 2010 Annual Meeting CD-ROM

Paper revised from original submittal.