Tree-based logistic regression approach for work zone casualty risk assessment.

This study presents a tree-based logistic regression approach to assessing work zone casualty risk, which is defined as the probability of a vehicle occupant being killed or injured in a work zone crash. First, a decision tree approach is employed to determine the tree structure and interacting factors. Based on the Michigan M-94\I-94\I-94BL\I-94BR highway work zone crash data, an optimal tree comprising four leaf nodes is first determined and the interacting factors are found to be airbag, occupant identity (i.e., driver, passenger), and gender. The data are then split into four groups according to the tree structure. Finally, the logistic regression analysis is separately conducted for each group. The results show that the proposed approach outperforms the pure decision tree model because the former has the capability of examining the marginal effects of risk factors. Compared with the pure logistic regression method, the proposed approach avoids the variable interaction effects so that it significantly improves the prediction accuracy.

Impact analysis of ECA policies on ship trajectories and emissions.

This study aims to investigate the impact of emission control area (ECA) measures on ship trajectories and emissions of merchant ships. Results show that a part of merchant ships sail outside the ECA water area to avoid using more expensive low-sulfur fuel. The implementation of stricter ECA policies could reduce the SOx emissions by 31.24%-42.67% for three merchant ship types. It is found that the proportion of ship voyages outside the ECA waters could increase by 4.77%-9.44% after the implementation of stricter ECA policies. Three logit regression models are established subsequently to estimate the occurrence probability of ECA evasion behavior. Results show that the evasion probabilities of merchant ships could increase with the ship size, the ship voyage distance as well as the price gaps of marine gasoline oil (MGO) with respect to high sulfur fuel oil (HSFO).

Impacts of the COVID-19 epidemic on merchant ship activity and pollution emissions in Shanghai port waters.

This study aims to evaluate impacts of the COVID-19 epidemic on merchant ship activities and corresponding atmospheric pollutant emissions in Shanghai port waters. Comparing AIS data from February 2019 and from February 2020, it is found that the merchant ship count and utilization frequency are reduced during the epidemic period. The epidemic could result in longer ship turnaround times because of more operation time for berthing and anchoring activities. Ship emission comparison results reveal that the cargo ship emissions are significantly reduced while container ships and tankers produce a slightly decreased emissions resulting by strict COVID-19 quarantine measures. In addition, the unit ship emission intensity is greatly reduced for ships which are under the normal cruising status while berthing and anchoring operations are associated with increased ship emissions. This implies that it is urgent to promote the use of shore power equipment for merchant ships during the epidemic period.

Exploring the effectiveness of ECA policies in reducing pollutant emissions from merchant ships in Shanghai port waters.

Using the AIS data in 2017, this study aims to investigate the effectiveness of five ECA policies on pollutant emissions from merchant ships in Shanghai Port waters. Results show that the estimated annual emissions from merchant ships including cargo ships, container ships and tankers are 3.4029 × 104 tons for NOx, 2.1037 × 104 tons for SO2, 2.291 × 103 tons for PM2.5, and 2.921 × 103 tons for PM10 in 2017, respectively. Impact analysis results highlight the fact that effects of each ECA policy vary significantly among different merchant ship types and different water areas. The amount of pollutant emissions from cargo ships (e.g., SO2 and PM2.5) is most affected by the ECA policy. However, the NOx emissions are not significantly changed under different ECA policies. Results also show that future ECA policies could cause a much greater decrease of pollutant emissions in water areas of Yangshan and Wusong.

Time-varying mixed logit model for vehicle merging behavior in work zone merging areas.

This study aims to develop a time-varying mixed logit model for the vehicle merging behavior in work zone merging areas during the merging implementation period from the time of starting a merging maneuver to that of completing the maneuver. From the safety perspective, vehicle crash probability and severity between the merging vehicle and its surrounding vehicles are regarded as major factors influencing vehicle merging decisions. Model results show that the model with the use of vehicle crash risk probability and severity could provide higher prediction accuracy than previous models with the use of vehicle speeds and gap sizes. It is found that lead vehicle type, through lead vehicle type, through lag vehicle type, crash probability of the merging vehicle with respect to the through lag vehicle, crash severities of the merging vehicle with respect to the through lead and lag vehicles could exhibit time-varying effects on the merging behavior. One important finding is that the merging vehicle could become more and more aggressive in order to complete the merging maneuver as quickly as possible over the elapsed time, even if it has high vehicle crash risk with respect to the through lead and lag vehicles.

Investigation of work zone crash casualty patterns using association rules.

Investigation of the casualty crash characteristics and contributory factors is one of the high-priority issues in traffic safety analysis. In this paper, we propose a method based on association rules to analyze the characteristics and contributory factors of work zone crash casualties. A case study is conducted using the Michigan M-94/I-94/I-94BL/I-94BR work zone crash data from 2004 to 2008. The obtained association rules are divided into two parts including rules with high-lift, and rules with high-support for the further analysis. The results show that almost all the high-lift rules contain either environmental or occupant characteristics. The majority of association rules are centered on specific characteristics, such as drinking driving, the highway with more than 4 lanes, speed-limit over 40mph and not use of traffic control devices. It should be pointed out that some stronger associated rules were found in the high-support part. With the network visualization, the association rule method can provide more understandable results for investigating the patterns of work zone crash casualties.

A hybrid finite mixture model for exploring heterogeneous ordering patterns of driver injury severity.

Debates on the ordering patterns of crash injury severity are ongoing in the literature. Models without proper econometrical structures for accommodating the complex ordering patterns of injury severity could result in biased estimations and misinterpretations of factors. This study proposes a hybrid finite mixture (HFM) model aiming to capture heterogeneous ordering patterns of driver injury severity while enhancing modeling flexibility. It attempts to probabilistically partition samples into two groups in which one group represents an unordered/nominal data-generating process while the other represents an ordered data-generating process. Conceptually, the newly developed model offers flexible coefficient settings for mining additional information from crash data, and more importantly it allows the coexistence of multiple ordering patterns for the dependent variable. A thorough modeling performance comparison is conducted between the HFM model, and the multinomial logit (MNL), ordered logit (OL), finite mixture multinomial logit (FMMNL) and finite mixture ordered logit (FMOL) models. According to the empirical results, the HFM model presents a strong ability to extract information from the data, and more importantly to uncover heterogeneous ordering relationships between factors and driver injury severity. In addition, the estimated weight parameter associated with the MNL component in the HFM model is greater than the one associated with the OL component, which indicates a larger likelihood of the unordered pattern than the ordered pattern for driver injury severity.

Modeling the probability of freeway lane-changing collision occurrence considering intervehicle interaction.

OBJECTIVE: To better capture the relationships between lane-changing collisions and explanatory variables, a microscopic model is developed for freeway lane-changing collisions based on the interactions between lane-changing vehicles. METHODS: The model applies an intervehicle interaction structure to account for the occurrence mechanism of lane-changing collisions. The occurrence mechanism can be described as the failure of a vehicle driver of an adjacent lane in avoiding the lane-changing vehicle, which disturbs the smooth movement of the adjacent lane vehicle and requires the driver's brake action to avoid an angle collision. This model is examined using data collected from freeways in Washington State during 2010 to 2011 and validated using lane-changing collision data for the SR 520 freeway. RESULTS: The findings of this study show that generalized truck percentage has a significant decreasing effect on lane-changing collision risk, whereas average spacing and several roadway characteristics have significant increasing effects. The frequency of slight collisions during peak hours is higher than that during off-peak hours. Young female drivers are more likely to be involved in collisions during lane-changing than young male drivers, but the result for senior drivers is opposite, with older male drivers having a higher probability of lane-changing collisions than female drivers in the same age group. CONCLUSION: The process of lane-changing collisions is a complicated maneuver. Truck percentage, average spacing, and good roadway characteristics, such as straight and level segment, in the target lane have a significant effect on the occurrence of lane-changing collisions. Age and gender are also 2 important factors contributing to the relationship between lane-changing collisions and explanatory variables.

Investigation of shipping accident injury severity and mortality.

Shipping movements are operated in a complex and high-risk environment. Fatal shipping accidents are the nightmares of seafarers. With ten years' worldwide ship accident data, this study develops a binary logistic regression model and a zero-truncated binomial regression model to predict the probability of fatal shipping accidents and corresponding mortalities. The model results show that both the probability of fatal accidents and mortalities are greater for collision, fire/explosion, contact, grounding, sinking accidents occurred in adverse weather conditions and darkness conditions. Sinking has the largest effects on the increment of fatal accident probability and mortalities. The results also show that the bigger number of mortalities is associated with shipping accidents occurred far away from the coastal area/harbor/port. In addition, cruise ships are found to have more mortalities than non-cruise ships. The results of this study are beneficial for policy-makers in proposing efficient strategies to prevent fatal shipping accidents.

How Does the Driver's Perception Reaction Time Affect the Performances of Crash Surrogate Measures?

With the merit on representing traffic conflict through examining the crash mechanism and causality proactively, crash surrogate measures have long been proposed and applied to evaluate the traffic safety. However, the driver's Perception-Reaction Time (PRT), an important variable in crash mechanism, has not been considered widely into surrogate measures. In this regard, it is important to know how the PRT affects the performances of surrogate indicators. To this end, three widely used surrogate measures are firstly modified by involving the PRT into their crash mechanisms. Then, in order to examine the difference caused by the PRT, a comparative study is carried out on a freeway section of the Pacific Motorway, Australia. This result suggests that the surrogate indicators' performances in representing rear-end crash risks are improved with the incorporating of the PRT for the investigated section.

In-depth analysis of drivers' merging behavior and rear-end crash risks in work zone merging areas.

This study investigates the drivers' merging behavior and the rear-end crash risk in work zone merging areas during the entire merging implementation period from the time of starting a merging maneuver to that of completing the maneuver. With the merging traffic data from a work zone site in Singapore, a mixed probit model is developed to describe the merging behavior, and two surrogate safety measures including the time to collision (TTC) and deceleration rate to avoid the crash (DRAC) are adopted to compute the rear-end crash risk between the merging vehicle and its neighboring vehicles. Results show that the merging vehicle has a bigger probability of completing a merging maneuver quickly under one of the following situations: (i) the merging vehicle moves relatively fast; (ii) the merging lead vehicle is a heavy vehicle; and (iii) there is a sizable gap in the adjacent through lane. Results indicate that the rear-end crash risk does not monotonically increase as the merging vehicle speed increases. The merging vehicle's rear-end crash risk is also affected by the vehicle type. There is a biggest increment of rear-end crash risk if the merging lead vehicle belongs to a heavy vehicle. Although the reduced remaining distance to work zone could urge the merging vehicle to complete a merging maneuver quickly, it might lead to an increased rear-end crash risk. Interestingly, it is found that the rear-end crash risk could be generally increased over the elapsed time after the merging maneuver being triggered.

Analysis of work zone rear-end crash risk for different vehicle-following patterns.

This study evaluates rear-end crash risk associated with work zone operations for four different vehicle-following patterns: car-car, car-truck, truck-car and truck-truck. The deceleration rate to avoid the crash (DRAC) is adopted to measure work zone rear-end crash risk. Results show that the car-truck following pattern has the largest rear-end crash risk, followed by truck-truck, truck-car and car-car patterns. This implies that it is more likely for a car which is following a truck to be involved in a rear-end crash accident. The statistical test results further confirm that rear-end crash risk is statistically different between any two of the four patterns. We therefore develop a rear-end crash risk model for each vehicle-following pattern in order to examine the relationship between rear-end crash risk and its influencing factors, including lane position, the heavy vehicle percentage, lane traffic flow and work intensity which can be characterized by the number of lane reductions, the number of workers and the amount of equipment at the work zone site. The model results show that, for each pattern, there will be a greater rear-end crash risk in the following situations: (i) heavy work intensity; (ii) the lane adjacent to work zone; (iii) a higher proportion of heavy vehicles and (iv) greater traffic flow. However, the effects of these factors on rear-end crash risk are found to vary according to the vehicle-following patterns. Compared with the car-car pattern, lane position has less effect on rear-end crash risk in the car-truck pattern. The effect of work intensity on rear-end crash risk is also reduced in the truck-car pattern.

Development of a subway operation incident delay model using accelerated failure time approaches.

This study aims to develop a subway operational incident delay model using the parametric accelerated time failure (AFT) approach. Six parametric AFT models including the log-logistic, lognormal and Weibull models, with fixed and random parameters are built based on the Hong Kong subway operation incident data from 2005 to 2012, respectively. In addition, the Weibull model with gamma heterogeneity is also considered to compare the model performance. The goodness-of-fit test results show that the log-logistic AFT model with random parameters is most suitable for estimating the subway incident delay. First, the results show that a longer subway operation incident delay is highly correlated with the following factors: power cable failure, signal cable failure, turnout communication disruption and crashes involving a casualty. Vehicle failure makes the least impact on the increment of subway operation incident delay. According to these results, several possible measures, such as the use of short-distance and wireless communication technology (e.g., Wifi and Zigbee) are suggested to shorten the delay caused by subway operation incidents. Finally, the temporal transferability test results show that the developed log-logistic AFT model with random parameters is stable over time.

Uncertainty analysis of accident notification time and emergency medical service response time in work zone traffic accidents.

OBJECTIVE: Taking into account the uncertainty caused by exogenous factors, the accident notification time (ANT) and emergency medical service (EMS) response time were modeled as 2 random variables following the lognormal distribution. METHOD: Their mean values and standard deviations were respectively formulated as the functions of environmental variables including crash time, road type, weekend, holiday, light condition, weather, and work zone type. Work zone traffic accident data from the Fatality Analysis Report System between 2002 and 2009 were utilized to determine the distributions of the ANT and the EMS arrival time in the United States. A mixed logistic regression model, taking into account the uncertainty associated with the ANT and the EMS response time, was developed to estimate the risk of death. RESULTS: The results showed that the uncertainty of the ANT was primarily influenced by crash time and road type, whereas the uncertainty of EMS response time is greatly affected by road type, weather, and light conditions. In addition, work zone accidents occurring during a holiday and in poor light conditions were found to be statistically associated with a longer mean ANT and longer EMS response time. The results also show that shortening the ANT was a more effective approach in reducing the risk of death than the EMS response time in work zones. CONCLUSIONS: To shorten the ANT and the EMS response time, work zone activities are suggested to be undertaken during non-holidays, during the daytime, and in good weather and light conditions.

Analysis of driver casualty risk for different work zone types.

Using driver casualty data from the Fatality Analysis Report System, this study examines driver casualty risk and investigates the risk contributing factors in the construction, maintenance and utility work zones. The multiple t-tests results show that the driver casualty risk is statistically different depending on the work zone type. Moreover, construction work zones have the largest driver casualty risk, followed by maintenance and utility work zones. Three separate logistic regression models are developed to predict driver casualty risk for the three work zone types because of their unique features. Finally, the effects of risk factors on driver casualty risk for each work zone type are examined and compared. For all three work zone types, five significant risk factors including road alignment, truck involvement, most harmful event, vehicle age and notification time are associated with increased driver casualty risk while traffic control devices and restraint use are associated with reduced driver casualty risk. However, one finding is that three risk factors (light condition, gender and day of week) exhibit opposing effects on the driver casualty risk in different types of work zones. This may largely be due to different work zone features and driver behavior in different types of work zones.

Evaluation of rear-end crash risk at work zone using work zone traffic data.

This paper aims to evaluate the rear-end crash risk at work zone activity area and merging area, as well as analyze the impacts of contributing factors by using work zone traffic data. Here, the rear-end crash risk is referred to as the probability that a vehicle is involved in a rear-end crash accident. The deceleration rate to avoid the crash (DRAC) is used in measuring rear-end crash risk. Based on work zone traffic data in Singapore, three rear-end crash risk models are developed to examine the relationship between rear-end crash risk at activity area and its contributing factors. The fourth rear-end crash risk model is developed to examine the effects of merging behavior on crash risk at merging area. The ANOVA results show that the rear-end crash risk at work zone activity area is statistically different from lane positions. Model results indicate that rear-end crash risk at work zone activity area increases with heavy vehicle percentage and lane traffic flow rate. An interesting finding is that the lane closer to work zone is strongly associated with higher rear-end crash risk. A truck has much higher probability involving in a rear-end accident than a car. Further, the expressway work zone activity area is found to have much larger crash risk than arterial work zone activity area. The merging choice has the dominated effect on risk reduction, suggesting that encouraging vehicles to merge early may be the most effective method to reduce rear-end crash risk at work zone merging area.

A probabilistic quantitative risk assessment model for the long-term work zone crashes.

Work zones especially long-term work zones increase traffic conflicts and cause safety problems. Proper casualty risk assessment for a work zone is of importance for both traffic safety engineers and travelers. This paper develops a novel probabilistic quantitative risk assessment (QRA) model to evaluate the casualty risk combining frequency and consequence of all accident scenarios triggered by long-term work zone crashes. The casualty risk is measured by the individual risk and societal risk. The individual risk can be interpreted as the frequency of a driver/passenger being killed or injured, and the societal risk describes the relation between frequency and the number of casualties. The proposed probabilistic QRA model consists of the estimation of work zone crash frequency, an event tree and consequence estimation models. There are seven intermediate events--age (A), crash unit (CU), vehicle type (VT), alcohol (AL), light condition (LC), crash type (CT) and severity (S)--in the event tree. Since the estimated value of probability for some intermediate event may have large uncertainty, the uncertainty can thus be characterized by a random variable. The consequence estimation model takes into account the combination effects of speed and emergency medical service response time (ERT) on the consequence of work zone crash. Finally, a numerical example based on the Southeast Michigan work zone crash data is carried out. The numerical results show that there will be a 62% decrease of individual fatality risk and 44% reduction of individual injury risk if the mean travel speed is slowed down by 20%. In addition, there will be a 5% reduction of individual fatality risk and 0.05% reduction of individual injury risk if ERT is reduced by 20%. In other words, slowing down speed is more effective than reducing ERT in the casualty risk mitigation.