Spatialtemporal Characteristics of Tunnel Traffic Accidents in China from 2001 to Present

the


Introduction
With the booming development of transportation in China, the expressway tunnel has become an important component of the traffic system.Over the past decades, the number of tunnels in China had a massive increase, and it will continue to grow, as shown in Figure 1.According to the statistical results published at the Ministry of Transport of the People's Republic of China at the end of 2016, the number of expressway tunnels in China has reached 15181 with a total length of 14039.7 km [1].In addition, China has grown to have the largest construction scale and fastest construction speed for tunnel and underground engineering.
e operation of the tunnel can not only shorten the distance and increase the travel efficiency of vehicles but also play an important role in protecting the environment.Furthermore, the development of tunnels has resulted in great social and economic benefits.However, with the increase in the number of tunnels, the traffic accidents caused by tunnels increase correspondingly which should be paid attention [2][3][4].Tunnels as one of the underground engineering, it will face enormous risks during tunnel construction and operation under the uncertainty of the natural environment.e number of TTAs has increased year by year, which results in great economic and spiritual loss to individuals and the state [5,6].Statistics indicate that during the period of 2012-2013, there were 469 TTAs in China, and among them, 198 caused deaths.e death rate per accident was 0.42, which is 1.44 times the annual average of traffic accidents of ordinary highways [7].Some serious and fatal China TTAs in the past decades are described in Table 1 [8].
erefore, it is of great importance to investigate the causes and occurrence rate of TTAs, and it is also important to put  2 Advances in Civil Engineering forward effective countermeasures to prevent TTAs.In addition, we should pay particular attention to the location distribution of the TTAs to enhance the management of the tunnel traffic system and to improve the traffic safety of expressway tunnels [9][10][11].In recent years, many studies have explored the issue of traffic accidents in expressway tunnels.For instance, in the study of Amundsen and Ranes [2], 900 studies about tunnels in Norway were used to research the traffic safety.Existing studies have indicated that there was a correlation between the number of TTAs and the construction standard to which the expressway was constructed; the higher the construction standard is, the fewer the number of TTAs occur.We also learn that the accident rates were higher in the entrance zone of tunnels than in other zones.In addition, traffic safety issues in expressway tunnels cannot be treated the same as the traffic safety issues of an ordinary highway.Yeung and Wong [12] studied the characteristics of 608 traffic accidents occurred in expressway tunnels in Singapore during 2009-2011, and accorded to the location of traffic accidents, the location of tunnels is classified.Lu et al. [13] described the temporal and space distribution characteristics of traffic accidents occurred in river-crossing tunnels of Shanghai.Ma et al. [14] and Zhang et al. [15] explored the major distribution characteristics and rules of TTAs based on the statistical analyses of TTAs of the Shaoguan section of the Beijing-Zhuhai expressway, the causes of traffic accidents were determined, and the practical countermeasures were proposed.Zhang et al.
[16] investigated 10 expressway tunnels in the Zhejiang province and explored the relationship between traffic accident rates and tunnel traffic volume, rear end accident occurrence, fire, rollover, and tunnel wall collisions.e Gray-Markov model method was applied to create a prediction model of traffic accidents in tunnel groups by Zhan [17], and the occurrence rate of traffic accidents in the tunnel groups was determined.Zhao et al. [18] analyzed data from 208 expressway TTAs in China and presented some basic characteristics of expressway tunnel accidents.However, the TTA datasets used in the previous studies are still limited.As a result, the empirical study is still insufficient to investigate the statistic regularity or basic law of TTAs in China, and a large-scale expressway tunnel traffic accident investigation and statistical analysis have not been carried out in China either.In this study, 2703 TTAs from 2001 up to now were collected by the transportation department (AQS2010).
is study investigates the temporal and spatial distribution rules of traffic accidents, the distribution of accident types, and the distribution of accident vehicles types using a statistical analysis approach to objectively show the status quo of TTAs in China and determine the rules and characteristics of TTAs.

Data Basis and Statistics
2.1.Accident Causes.TTAs are caused by coordination imbalance between human, vehicle, road (tunnel), and environment.erefore, it is important to analyze how these factors affect TTAs.e statistics of TTAs in China is analyzed, it can be shown that overspeeding, icy pavement, vehicles stopping in front of other vehicles, vehicle fault, tunnel collapse, pavement debris, and other less common factors can result in TTAs. Figure 2 shows the statistical results of the major causes of TTAs.It can be seen that overspeeding and icy pavement are the primary causes of TTAs, which account for 60.1% of all reasons.
is is consistent with the statistical results of the literature of Cao et al. [19].Additionally, the traffic accidents of expressway tunnels are little affected by the weather, but accidents may occur at the entrance and exit of the tunnel as the friction coefficient of pavement is low on rainy and snowy days.Many rear end collisions are a result of a vehicle stopping in front of another vehicle without warning.According to the statistics, many traffic accidents are caused by the negligent operation of drivers.In the three factors of the accident, human, vehicles and road (tunnel), and environment, the human factor is the main factor that causes expressway tunnel traffic accident.

Accident Types.
Several typical TTAs types are shown in Figure 3. Specifically, in August 10th at 11 pm, a motorbus was involved in a bad traffic accident which occurred at the Qinling No.1 tunnel, killing 36 people.According to the statistics result of cases of our research, rear end accidents and tunnel wall collisions are the primary TTA types, accounting for 60% and 20% of all TTAs, respectively, and 17% are the result of vehicle rollover.Accidents resulting from fire and other causes (such as goods falling out of a vehicle) are relatively few, accounting for 1% and 2%, as shown in Figure 4. e results of this research are in line with the results of some scholars in China [20][21][22][23][24].During emergency situations or when drivers engage in excessive speeds, drivers cannot brake in time or lack of space is the main cause of rear end collision in the tunnel.

Accidents versus Vehicle Types.
e vehicle types involved in TTAs include minibuses, minivans, motorbuses, and large trucks.Several accident datasets from four tunnels were examined to study the vehicle types involved in TTAs, the Shaoguan section of the Beijing-Zhuhai expressway, 510 traffic accidents in the Zhongliangshan tunnel of Chengdu-Chongqing expressway, traffic accidents in 31 tunnels of the Ningbo-Taizhou-Wenzhou expressway in the Zhejiang province, and 2703 TTAs in China [14,17,25].e vehicle types involved in TTAs and vehicle type proportion are shown in Figure 5.It can be seen from the figure that minibuses and large trucks are involved in the most traffic accidents.In the traffic accident involving the minibuses, most of the sedans are made in China, and most of the vans are light chassis.ose traffic accidents involving minivans and minitrucks are most involved.Among the traffic accidents involving large trucks, heavy trucks account for the majority; in addition, traffic accidents caused by overloading account for the majority.

Spatial Distributions of Traffic Accidents
3.1.Accident Quantity and Tunnel Lengths.According to "JTG/T D70-2010, Guidelines for Design of Highway Tunnel," 156 tunnels are divided into 4 types: short tunnels, medium tunnels, long tunnels, and super-long tunnels [26].e number of TTAs and the length of the tunnel involved are shown in Table 2. e datasets indicate that TTAs occur more frequently in longer tunnels.e reason for this is that drivers easily get fatigue in a long tunnel with monotonous environment for a long time.Moreover, with the increase in the tunnel length, the traffic volume also increases in the tunnel which leads to the increase of smoke concentration and visibility reduction.Advances in Civil Engineering

Tra c Accidents in Tunnel
Zones.Measures can be taken to e ectively prevent tra c accidents and to improve the tra c safety situations according to the understanding of the tra c accidents distribution at various tunnel sections and the analysis of the characteristics of TTAs' spatial distributions [27,28].e tra c accident characteristics for each of the various tunnel sections were analyzed in this study.e division of tunnel zones is shown in Figure 6.Zone 1 consists of both the interior and exterior areas within 200 m of the tunnel portal; Zone 2 consists of the area from 200 m to 400 m inside the tunnel; and Zone 3 composes of the remainder of the tunnel length.It is worth noting that tunnels with a length of less than 200 m only include Zone 1.
Figure 7 shows the spatial distributions of TTAs.It can be seen from the statistical results that Zone 1 is the most accident-prone section, and the accident rate accounts for 58% of all tunnel accidents.e number of tra c accidents occurred in Zone 2 and Zone 3 is less, as each accounts for 21%.Additionally, the statistical data of the literature of Amundsen and Ranes [2], which includes data of 587 tunnels in Norway from 1992 to 1996, show that accidents 50 m away from the tunnel entrance accounted for 25% of all the tunnel tra c accidents and those within 150 m of the tunnel entrance accounted for 50% of the total tra c accidents.e statistical results of this study regarding tunnel accidents in China are basically consistent with those of Norway; that is, the probability of tra c accidents near the tunnel entrance is higher than elsewhere in the tunnel because of the special environmental characteristics of the tunnel and the e ects of ray aberration, inside and outside the tunnel.On a sunny day, when vehicles enter the tunnel from the outside to the inside, the light suddenly changes from bright to dark, which generates the so-called "black hole e ect."Conversely, when vehicles drive away from the inside of the tunnel, the light changes from dark to bright, generating "white hole e ect."Both of these e ects reduce the driver's emergency response time and may lead to accidents [29][30][31].
Great di erences exist in tra c environment inside and outside the tunnel, especially in illumination and brightness, which seriously a ects the extraction, analysis, and judgment of tra c information by driver visual organs.erefore, it has a serious impact on safe driving in the tunnel [32].Drivers would have a series of visual problems in the process of approaching, entering, passing, and driving out of tunnels from bright environment, as shown in Figure 8. at is, (1) Visual problems before entering the tunnel in the day: "black hole e ect" for the long tunnel and "black frame e ect" for the short tunnel, which are mainly due to the excessive brightness di erence.(2) Visual problems immediately after entering the tunnel in the day: dark adaptation, which mainly refers to the lagging phenomenon of adaptation.(3) Visual Problems inside the tunnel in the daytime and at night: smoke from exhaust gases, which reduces the visibility of the tunnel.( 4) Visual problems at the tunnel exit: "white hole e ect" for the long tunnel in the daytime, which induces the driver glare, and "black hole e ect" at night, which causes the alignment of the external road and the obstacles on the road to be indistinguishable.
During the driving process, the information of change of road tra c environment can be transmitted to the human brain decision-making system through senses, and the proportion of visual information is about 80-90%.When approaching the transitional section of tunnel entrance and exit, the tra c information and brightness around the tunnel change.Driver physiological load tends to approach the critical value, and they are prone to panic, vertigo, and physical disharmony.Meanwhile, vehicle lights have a certain impact on the brightness of the environment in the transitional zone of the tunnel entrance and exit, which will a ect the driver handling stability, thus a ecting the light and shade adaptation characteristics of the driver [33].e purpose of tunnel lighting is to reduce or eliminate the visual di erences caused by the light and shade of the road inside and outside the tunnel and to ensure that vehicles traveling in the daytime and at night can safely approach, cross, and pass through the tunnel at the designed speed.Moreover, the safety and comfort of drivers should be no less than that of the open zone adjacent to the tunnel.e illumination system of highway tunnel includes interior zone illumination, entrance zone illumination, transition zone illumination, exit zone illumination, access zone dimming facilities, emergency illumination, and approach illumination outside the tunnel.Highway tunnel lighting area can mainly be divided into ve zones (Figure 9): access zone, entrance zone, transition zone, interior zone, and exit zone.
To reduce the rate of tunnel tra c accidents, the highway tunnel illumination system should not only have su cient brightness but also meet the requirements of quality.In  Advances in Civil Engineering recent years, the tunnel illumination system has gradually developed from automation to intellectualization [34].For the intelligent control method, advanced technologies such as artificial intelligence and neural network are adopted to simulate the brightness inside and outside the tunnel by collecting data.After constant debugging and testing, an efficient, accurate, and stable system is established, and dynamic dimming is carried out according to the actual demand of illumination brightness in the highway tunnel so as to realize the purpose of lighting according to local conditions and on demand.

Temporal Distributions of Traffic Accidents
e temporal distribution of traffic accidents is defined as the statistical characteristics of traffic accidents over time.
rough analysis of the characteristics of the temporal distribution of TTAs, the traffic accidents trends can be 6 Advances in Civil Engineering revealed, and a basis can be provided to understand the causes of the TTAs.For convenience of study, the characteristics of the temporal distribution of TTAs can be grouped into distribution characteristics of month, week, and hour.

Monthly Distribution Characteristics of
TTAs. e monthly distribution characteristics of traffic accidents in expressway tunnel sections are affected by numerous factors, including changes in geographic position, weather conditions, traffic volume, and human behavioral characteristics [35][36][37][38].us, the monthly distribution for traffic accidents can be analyzed to make the necessary monitoring of tunnel traffic safety during different months, conduct effective control over traffic accidents, reduce accident rates, and guarantee the safe operation of the tunnel.
e monthly accident distributions of a certain tunnel in Hebei province, the Beijing-Zhuhai expressway tunnel section, the Ningbo-Wenzhou expressway tunnel section in Zhejiang province, and the monthly accident distributions of the 2703 traffic accidents in various expressway tunnels in China are shown in Figure 10 [17,19,21].It can be seen from the statistical results of the 2703 TTAs that the monthly distribution trend of TTAs in various regions of China generally shows that most accidents occur in January, February, April, May, and July, and fewer accidents occur in March, June, August, September, October, November, and December.is trend shows that the number of tunnel traffic accidents dropped sharply from January to March, increased again from April to July, and remained stable from August to December.Traffic volume is generally high during the several months before and after Spring Festival, which includes January, February, and December, which proved that accident rates are directly proportional to traffic volume.However, there are fewer accidents in December than in January and February, which is related to the mental state of the drivers and an increase in the traffic department's management efforts.After March, the weather gradually warmed up, the number of holiday travel trips increased, the traffic volume of expressway increased significantly, and the traffic accidents showed an upward trend after March.at is to say human factors are the main cause of traffic accidents [39][40][41][42].

Weekly Distribution Characteristics of TTAs.
Influenced by the pattern of people's lives, traffic volume varies within a week on the road; therefore, the weekly distribution corresponding to traffic accidents varies with number of trips that people take.e distribution of TTAs by weekday in China is shown in Figure 11.It is obvious from the figure that the weekday distribution of TTAs is significantly different than the weekend distribution.Saturday and Sunday account for 43.24% of traffic accidents, while other days of the week show little fluctuation in the number of accidents per day which is because of the traffic volume from Monday to Friday is less and the volume is consistent.Conversely, the gross traffic volume sharply increases on the weekends, and the probability of traffic accidents correspondingly increases.
e traffic structure of expressway also changes as weekend travelers generally use sedans, while large vans can make up a large component on weekday traffic.Sedan speed is far higher than that of large vans, which intensifies the occurrence of traffic accidents on the weekend.

Hourly Distribution Characteristics of TTAs.
e hourly distribution of traffic accidents in expressway tunnels is shown in Figure 12.Influenced by climate, the environment, and the pattern of people's lives, there is an obvious difference in the hourly distribution of TTAs throughout the day.From the figure, we can observe that there are three peak periods of traffic accidents, namely, 9:00-10:00, 11:00-12:00, and 13:00-15:00.
e number of traffic accidents occurring in these three periods account for 48% of the total TTAs.e traffic accidents that happened in 9:00-10:00 are primarily the result of speeding.During the period from 11: 00 to 12:00, the main cause of tunnel traffic accidents is the increase of traffic volume, especially due to an increase in the number of sedans which increases the complexity of the Advances in Civil Engineering traffic structure.From 13:00-15:00, many drivers' physiological function decreases after noon, and they are easily susceptible to somnolence.In addition, this is also the period during the day with most sunshine, which results in a large luminance difference between the inside and the outside of the tunnel.As a result, the ability of the driver to adapt to the transition from brightness to darkness at the entrance is reduced, and thus accidents occur.

Holiday Distribution Characteristics of TTAs.
In July 24, 2012, China implemented the official holiday highway free passage policy, which led to the growth of passenger flow in holidays, and the traffic accidents were becoming more and more serious.ese official holidays include the Spring Festival, Tomb-sweeping Day, May Day, and National Day.Tomb-sweeping Day and May Day are the most concentrated periods of the annual passenger flow before and after Spring Festival, but also is in a high rate of the traffic accident.According to the data analysis of traffic accidents during the past three years, the traffic accidents are mainly concentrated in the fourth and sixth days of the Spring Festival holidays, as shown in Figure 13.In order to prevent the occurrence of traffic accidents during the festival, government sector should make full use of television, radio, newspapers, internet, and other major media to strengthen traffic control measures and real-time traffic information release, suggesting that the masses consciously abide by traffic laws and regulations and reasonably choose traveling routes.At the same time, drivers must carefully check the vehicle before the trip and ensure good condition.During the peak period of vehicle centralized travel of holidays, travelers should be called upon to abide by traffic regulations and travel civilly.Traffic departments should strengthen management and prohibit drivers from drunk driving, fatigue driving, and illegal overspeed.

Impact and Assessment on Casualties
. By collecting the data of 2703 expressway traffic accidents from 2001 to present, the temporal and spatial distributions of TTAs, the distribution of accident types, and the distribution of accident vehicle types are studied.Considering the different casualties caused by accidents, according to the Road Traffic Safety Law, the casualties can be divided into four levels: slight, general, major, and extraordinarily large.Accordingly, the objective impact factors on casualties are calculated.Since there are no casualties in some traffic accidents, the calculation is divided into five accident grades and assigned as follows: no casualties (1 point), 1-2 injuries (2 points), 3-10 injuries (3 points), 1-2 deaths (4 points), and more than 3 deaths (5 points).e proportions of accidents and casualties caused by external factors at various levels are multiplied by the corresponding grades scores, presented in the following equation: where K i is the influence weight, P i is the proportion of casualties caused by external factors, and N i is the accident grade.Advances in Civil Engineering By calculating the data of 2703 TTAs, the temporal and spatial distributions of traffic accidents, the distribution of accident types, and the influence weight of accident vehicle types on casualties are obtained as detailed in Table 3.Moreover, the impact of external factors such as accident type, tunnel length, and time and place of tunnel accident on casualties is pointed out.In tunnel traffic accidents, the most important external factor affecting casualties is Zone 1 with the influence weight of 1.98.at is, the casualties caused by accidents are more serious when vehicles are driven in Zone 1.Secondly, the influence weight of super-long tunnel is 1.38; that is, accidents in super-long tunnels are more likely to cause serious casualties.Additionally, the impact weight of large truck is 1.25; that is, accidents caused by large trucks can also cause serious casualties.

Correlation Analysis of External Factors.
rough research and statistics, the influence degree of different time, place, vehicle type, accident type, accident reason on traffic accident, and the influence and evaluation of external factors on casualty situation can be obtained.On this basis, the potential relationships between accident types and other factors causing traffic accidents are explored, so as to facilitate the safety education and management of travelers.
e research results show that there is a great relationship between the accident type and tunnel length, tunnel section, vehicle type, and accident cause.erefore, it is necessary to calculate the confidence level of 2703 cases of expressway traffic accidents, as shown in equation (2).To facilitate expression, the above factors are represented by specific letters, as follows: Accident type: rear end accident (A 1 ), tunnel wall collision (A 2 ), rollover (A 3 ), fire (A 4 ), and others (A 5 ) Tunnel length: short tunnel (B 1 ), medium tunnel (B 2 ), long tunnel (B 3 ), and super-long tunnel (B 4 ) Tunnel zone: Zone 1 (C 1 ), Zone 2 (C 2 ), and Zone 3 (C 3 ) Vehicle type: minibus (D 1 ), minivan (D 2 ), motorbus (D 3 ), and large truck (D 4 ) Accident cause: overspeeding (E 1 ), icy pavement (E 2 ), front vehicles stop (E 3 ), vehicle fault (E 4 ), tunnel collapse (E 5 ), pavement debris (E 6 ), and others ( According to the correlation between accident types and other external factors, the attention should be paid when the confidence level exceeds 60%.As shown in Table 4, for the condition of the collision tunnel wall in super-long tunnel, the confidence level of the association rule A 2 �>B 4 is as high as 69%; for the condition of tunnel fire in Zone 3, the confidence level of the association rule A 4 �>C 3 is as high as 77%; and for the condition of rear end accident caused by overspeeding, the confidence level of the association rule A 1 �>E 1 is as high as 72%.Advances in Civil Engineering summarized as follows: fatigue driving, failure to keep a sufficient safety distance between vehicles, speeding, vehicle overloading, improper emergency measures, illegal lane change or lane occupation, and low speed.At the same time, the effect of vehicle type, tunnel factors, weather, and other environmental factors affecting traffic accidents also cannot be ignored.

Conclusions
(3) Zone 1 is a highly accident-prone area.Traffic accidents in Zone 1 account for 58% of the total number of accidents.In addition, the probability of a traffic accident occurring will increase as the tunnel length increases.e traffic accidents are mainly concentrated in the fourth and sixth days of the Chinese Spring Festival holidays.(4) TTAs of expressway are distributed into three peak periods, 9:00-10:00, 11:00-12:00, and 13:00-15:00.Accidents during these three periods account for 48% of the total traffic number of traffic accidents.Additionally, TTA rates on the weekend account for 43.24% of the total traffic accidents, while the number of traffic accidents during weekdays shows little fluctuation.Also, more accidents occur during the months of January, February, April, May, and July.e number of traffic accidents occurred from January to March shows a law of sharp decline; from April to July, the occurrence of traffic accidents showed an upward trend; and monthly number of accidents from August-December is relatively steady.(5) Based on the impact and evaluation of external factors on casualties, the most disadvantageous combination of road tunnel traffic accidents is super-long tunnel + Zone 1 + large truck + overspeeding + rear end accident.After the statistics of external factors and unfavorable combination, the potential relationship between accident types and other factors causing traffic accidents can be explored, and the probability of unfavorable combination can also be calculated according to the confidence level.For the disadvantageous combination with high confidence level, effective measures should be taken to prevent and control it.

Data Availability
In this article, 2703 cases from 2001 up to now were collected by the transportation department (AQS2010).

Figure 1 :
Figure 1: (a) Number and length of expressway tunnels and (b) tunnel traffic accident locations.

Figure 2 :
Figure 2: Reasons for traffic accidents in expressway tunnels.

Figure 9 :
Figure 9: Illumination system of a highway tunnel.

( 1 )
Rear end accidents are the main form of TTAs, accounting for 60% of all accidents.Minibuses and large trucks are the most common vehicle types involved in traffic accidents.Additionally, the main causes of accidents are illegal speeding and icy pavement, which account for 60.1% of the total TTAs.(2) Human factors are some of the most important factors in TTAs.ese human factors can be

Figure 13 :
Figure 13: Traffic accident distributions during Chinese Spring Festival.

Table 1 :
Part of the fatal and serious traffic accidents in China expressway tunnels.

Table 2 :
Number of accidents and tunnel lengths.

Table 3 :
Influence weight of objective factors on casualties.

Table 4 :
Association rules with A�>B, A�>C, A�>D, and A�>E.