Journal of Transportation Engineering, Part A: Systems

This paper investigates bus bunching issues encountered with a single bus line. A real-time speed control model was proposed with the objective of minimizing variations in bus headway. Three cases of a typical road infrastructure for bus lines were studied. Two main factors that influence the stability of bus service–namely, signalized intersection delays and heterogeneous roadway conditions—were studied in the modeling process. In addition, other common variables were considered, including the time required for passengers to board a bus and alight from it. Compared with findings from prior literature, that frequently assumed homogeneous roadway infrastructure conditions and ignored intersection delays. The built model output the degree of speed adjustment required in accordance with different roadway configurations and the congestion level at each road section. A case study was designed to test the performance of the proposed model, based on the data collected from 40 bus stops, on Bus route No. 600 in Xi’an, China. Results showed that the proposed model could effectively restrain the problems posed by headway deviations and reduce travel time for the passengers.

Autonomous vehicles create new opportunities for innovative intelligent traffic systems. Variable speed limit (VSL), which is a speed management system that can adjust the speed limit according to traffic conditions or a predefined speed control algorithm on different road segments, can be better implemented with the cooperation of autonomous vehicles. These compliant vehicles can automatically follow speed limits. However, noncompliant vehicles will attempt to pass the moving bottleneck created by compliant vehicles. This study builds a multiclass cell-transmission model (CTM) to represent the relation among traffic flow parameters. This model calculates flows of both compliant and noncompliant vehicles between road segments. An VSL algorithm is proposed to reduce the stop-and-go behavior of vehicles at traffic signals and to save energy consumption. Simulation is used to test the effects of VSLs on an example network. The result shows that VSLs are effective at reducing the energy consumption of the whole system.

Route choice behavior of a rail transit passenger is not directly observable and it may be affected by the route travel time to a large extent. Compared to the on-train time, travel times at stations, including walking time and waiting time, have been receiving less attention and therefore become more difficult to analyze. A common method to analyze the travel time at a rail transit station is to directly assume a distribution function and to further fit the distribution. However, most distribution functions in the prior literature were used without validation and/or conclusive decision on the best fit. In such context, this paper develops a complete approach to estimating both the walking and waiting times at stations (including origin stations, destination stations, and transfer stations) by mining automatic fare collection (AFC) and automatic train supervision (ATS) data, and their distributions are further discussed and characterized in detail. An initial case study of the Beijing subway network shows that it can deduce passengers’ walking and waiting times in sequence, and consequently obtain and depict their distributions with high performance.

Coordination of traffic signals is one of the most effective methods to improve the quality of traffic control on an artery. Properly coordinated traffic signals can effectively minimize delay and number of stops, increase travel speed, and reduce the number of collisions. MULTIBAND and AM-BAND are well known and widely used methods for arterial traffic control. The optimum solution provides the ability to travel without stops for most individual vehicles, whereas buses stopping at bus stops may experience excessive delays. This paper presents BUS-MULTIBAND and BUS-AM-BAND methods formulated as mixed-integer linear programs that solve the signal coordination problem for passenger cars and public transport vehicles simultaneously. The optimal signal coordination plans for general and transit vehicles were computed, simulated, and compared with signal plans generated for only general vehicles. Simulation results indicate that traffic signal coordination plans generated by the proposed methods can provide significant benefits for buses and, at most, small losses for general vehicles compared with plans generated by MULTIBAND and AM-BAND.

This study investigates the efficiency of parking permits in managing morning commuting considering carpooling with parking restraint. First, the morning commuting problem was revisited with and without parking restraint using bimode and multimode one-to-one networks. Parking ending time and its inverse function were formulated through analysis of morning commuting in a multimode single origin–destination (OD) pair. Second, the traffic pattern with two parking permit management policies was also evaluated. Third, the traffic pattern in a many-to-one network with carpooling behavior was investigated, and a method to compute equilibrium with parking restraint was proposed. Fourth, schemes of optimal parking permit allocation and tradable parking permits were applied to diminish the externality caused by competition. Finally, examples have been discussed, revealing the effectiveness of the policies of parking permit management in reducing the system cost by eliminating competition. In addition, the system cost in the scheme of parking permit distribution was greater than that in the scheme of tradable parking permits. This is because the scheme of tradable parking permits encourages more commuters to choose to carpool than does the policy of parking permit distribution.

Performance measures offer an essential management tool for transportation engineers to make decisions along signalized corridors. Current signal performance strategies assess coordination of a corridor using intersection-level metrics while relying on expensive infrastructure to measure vehicle arrivals. Recent crowd-sourced data collection strategies have allowed for the ubiquitous collection of individual vehicle waypoints. These trajectories can be used to replicate existing signal performance measures and improve upon current practices. This paper uses trajectory data from numerous corridors around the state of Michigan to illustrate the merit and versatility of crowd-sourced probe vehicle trajectory-based performance measures. Findings from this paper show that some current automated traffic-signal performance measures (ATSPMs), such as percent arrival on green and delay quantification, can be replicated using low-penetration-rate vehicle trajectory data. Also, the reliability-based performance metric, level of travel time reliability (LOTTR), can be improved using trajectories of vehicles known to travel the complete corridor instead of aggregating segmented probe vehicle data.

The consensus degree among vehicles is an important index to measure the cooperative driving system. First, this paper establishes a measurement method of the instantaneous consensus of an individual vehicle to evaluate the consensus degree. On this basis, the consensus degree index of the cooperative driving system on the road section is constructed; thus, a quantitative description method for the consensus of vehicle cooperative driving is formed. The simulation experiment explores the consensus phenomenon and characteristics of the vehicle cooperative driving system by means of the classic vehicle cooperative driving model. The results show that the consensus index of the vehicle cooperative driving system proposed in this paper is effective and can better describe the consensus phenomenon in traffic.

This paper investigates the effect of geometric and operational features on freeway merge capacity by analyzing prebreakdown flow rates and using nonparametric and parametric (survival) analysis and semiparametric analysis (Cox’s proportional hazard model) techniques. The probability of breakdown and capacity were quantified as a function of the number of lanes, ramp flow rate, presence of lane drop, and presence of ramp metering. The number of lanes showed a negative relationship with the per-lane average prebreakdown flow rate. Sites with ramp meters also showed higher prebreakdown flow rates than nonmetered sites by 150 to $300\mathrm{veh}/\mathrm{h}/\ln$. In survival analysis, the number of lanes and presence of ramp meters were found to have a statistically significant effect on the breakdown probability. Capacities, defined as the 15th percentile breakdown probability flow rate, were 2,043 and $\mathrm{1,952}\mathrm{veh}/\mathrm{h}/\ln$ at three-lane and four-lane unmetered sites, respectively. Capacities were $\mathrm{2,316}\mathrm{veh}/\mathrm{h}/\ln$ at three-lane and $\mathrm{2,011}\mathrm{veh}/\mathrm{h}/\ln$ at four-lane metered sites. Higher ramp flows resulted in higher breakdown probability and lower capacities at two of the four sites that were investigated, suggesting that further research is needed to evaluate this relationship.

Understanding travel behavior decisions is a fundamental aim of transportation planning. However, data from surveys or travel diaries that were traditionally used for travel mode-choice modeling are costly and have certain inaccuracies and cover limited populations. Therefore, recently, smart card data collected from automated fare collection systems have gradually become more popular for travel behavior analysis and modeling, but relatively little attention has been paid to investigating the daily variability in travel behavior decisions using more than 1-year smart card data, apart for some descriptive studies. In this study, mode-choice behaviors in public transit were investigated in Seoul using 20-month smart card data to investigate the daily variability in the ratio of the number of subway passengers depending on origin and destination. For this aim, the effects of temporal features such as weather and calendar events as well as the route information and built environments of origin and destination stations were considered on a daily basis for different time periods. To overcome the limitation that the purpose of travel cannot be identified from smart card data, this study attempted to precisely estimate subway connections and extract travel records for commuting from regular commuters’ cards. The models were trained using 1-year data and were validated using 8-month data, which verified that the selected factors explain the daily variability in mode-choice behaviors for public transportation.

In railroads, the vertical alignment design is closely related to the cost of operation because it may affect the required energy consumptions. By considering an innovative vertical alignment design concept such as the dipped vertical alignment (DVA) profile, the required energy consumptions, namely, tractive energy and braking energy, may be reduced. As a result, the cost of train operations may be reduced. In addition, the cruising speed and coasting operations are important operational decisions. Thus, this paper proposes a multistage decision model to jointly optimize the vertical alignment profile, cruising speed, and coasting operation point. The minimum cost solution is obtained considering the costs of users and operators. A deterministic simulation model is incorporated to calculate the vehicle motion, dynamics, resistance, and other computation outputs. A case study followed by sensitivity analyses are presented to discuss how the input values (e.g., unit energy cost, passenger’s time value) affect the operational decisions (i.e., cruising speed or coasting operation).

This paper presents the results of an online survey of licensed driver comprehension of the right-turn signal displays with a focus on the flashing yellow arrow (FYA) and also including the circular green and red and red arrow. Recruitment postcards were mailed to a random sample of 9,872 residents in Oregon. The online survey yielded 399 responses. The open-ended responses were coded for comprehension and analyzed. The results suggest that FYA for right turns is well understood by Oregon drivers despite its current novelty (only two locations at the time of the research). Importantly, survey respondents were more likely to recognize the yielding requirement of the permissive movement and associate the yielding with pedestrians with the FYA over the circular green (CG) display. The research also confirmed that the expected driver response to the red arrow display for right turns is not well understood (only 52% of the respondents correctly stated the expected driver response). Binary logistic regression modeling revealed that the driver’s age and their educational level were significant factors in comprehension.

This paper presents an optimization algorithm for freeway operations at merge zones that maximizes the average speed of the segment in the presence of connected and automated vehicles (CAVs) and human-operated (i.e., conventional) vehicles. This research assumes that CAVs have the capability to communicate with each other and with the infrastructure and to execute the recommended trajectories. The proposed system receives arrival information as input and generates optimal trajectories for CAVs while predicting the behavior of conventional vehicles and accounting for deviation from expected behavior. The necessary algorithms are developed to simulate and carry out the merging operations on a two-lane freeway (one mainline and one ramp lane) and tested under a variety of scenarios considering demand level, demand splits, and CAV penetration rate. Results suggest that the proposed algorithm can efficiently manage the traffic at freeway merge zones and reduce the average total travel time (or increase average speed). The results indicate that a minimum of 25% CAV penetration rate is required to observe improvements in operational conditions.

Bus rapid transit (BRT) lanes may increase the travel time cost of other motor vehicles while improving the operating efficiency of BRT vehicles. In this paper the control strategies of the BRT lane are provided spatially and temporally. At the spatial level, the dedicated BRT lanes were divided into a certain number of fixed areas with a corresponding induction method and operation rules, and the intermittent bus lanes are allowed by other vehicles. At the temporal level, a bus priority strategy for a single intersection with an intermittent bus lane was presented, and a green wave coordination control method for the bus operation line was given. Taking the BRT in Lanzhou city, China, as an example, the spatial priority and time priority control strategies of the BRT lanes were analyzed and compared through simulation. The results showed that the control scheme of intermittent BRT lane integrating space- and time-optimized green wave coordination control performs better than other control schemes. The strategies may benefit the planning, operating, and management of BRT system.

Existing studies show that rainfall has a significant impact on bus ridership, and few studies exist on the impact of rainfall on urban rail transit (URT) ridership. Based on the daily and hourly URT ridership and rainfall data collected in Guangzhou, China during continuous thirteen months, this study explores the effects of rainfall on URT ridership and proposes a prediction approach of URT ridership under rainfall conditions, which is calculated by the sum of background ridership and rainfall influenced ridership. First, the Seasonal Autoregressive Integrated Moving Average model is employed to predict background ridership. Next, the rainfall impact factor is proposed and estimated using the Support Vector Regression model. Finally, the last month of data are applied to validate the performance of the proposed approach. The results show that the proposed approach performs well in both daily and hourly URT ridership prediction and, thus, provides a novel solution for quantifying the impact of rainfall on URT ridership, enabling URT managers to better understand the ridership variations under rainfall conditions and react well to it.

Urban rail transit transfer stations are key facilities for gathering and switching passenger flow. High passenger demand coupled with nonrecurrent events (e.g., train delays) may cause extreme passenger flow and density within the station, resulting in high risks of disastrous consequences (e.g., stampedes). This paper proposes multiline coordinated train skip-stop strategies to prevent extreme passenger flow and densities at transfer stations. A binary programming model is developed and solved by the genetic algorithm (GA) to best plan the multiline skip-stop strategy. Case studies using real passenger demand and train data are performed to verify the model and the algorithm. Results indicate that the application of the proposed skip-stop strategy can significantly reduce the total passenger waiting and delay time at transfer stations, therefore lowering the risk of disastrous consequences.

This study was motivated by the work of researchers who conducted field experiments on roadways with no bottlenecks. The ring roads are circular roadways where passing was restricted. All drivers who took part in these field studies promised to maintain the same constant speed. They were unable to do so. The field tests showed the importance in the variability in speed, i.e., traffic noise. Like the field test, traffic noise plays a prominent role in this simulation study. First, the focus is on triggering traffic breakdown and queue formation. Stochastic car-following models are developed. They feature Brownian motion models that help explain the connection between driver behavior and traffic noise. Second, the focus is on preventing traffic breakdown and queuing. State-space modeling and feedback control using Kalman filtering are featured. Real-time noisy data are collected and recursively passed through a system that tracks a target that is a function of time. All simulations in this study imitate the instructions given to the drivers in the ring-road field experiments. The drivers start from rest, accelerate to the same constant speed, and then attempt to hold that speed through the duration of the experiment. In the first part, the drivers fail. In the second part, they succeed. To demonstrate the effectiveness of the feedback control approach, a worse-case scenario is investigated. A ring road is assumed to operate at capacity; thus, queuing and breakdown are expected. Testing and implementing the feedback control system in a smart city environment and its fate are discussed.

The archived data from traffic sensors are used in a wide range of traffic management applications. However, missing or invalid data are becoming an important concern. This study proposes a systematic approach to identify and characterize data error patterns to facilitate large-scale loop detector troubleshooting. Data were collected from loop detectors in Phoenix. A set of quality control criteria was applied on daily 20-s data to find the error percentage for each loop detector. A fuzzy $c$-means clustering method was implemented on the data quality check results and preliminary clusters were identified. To discover the most frequent rules within the clusters, an association rule mining method was applied to the clusters’ data subsets. Loop detector stations with different error patterns were visited in the field to verify the clustering and association rule mining results, identify potential causes, and recommend appropriate solutions. The analysis identified four key patterns, indicating that the proposed approach successfully found the relationships in the data errors. The findings of this study help traffic engineers to more easily diagnose and troubleshoot large-scale loop detector errors.

Crash severity outcomes are random and are influenced by the interactions of vehicle, driver, crash, road, and environmental factors. Limited research has attempted to assess the direct and latent influence of road factors like pavement and shoulder conditions, crash and vehicle factors, along with weather and driver factors on crash severity outcomes. This work attempts to develop crash severity prediction models considering the direct and indirect influence of road factors measured with pavement distress conditions, shoulder type and condition, crash factors measured with crash type and collision partners, human and weather factors measured with crash time (for visibility and traffic condition), season of crash occurrence, and driver age and gender. The work models crash severity outcomes using the commonly used Ordered Probit model, which recognizes the inherent ordered nature of crash severity outcomes, and also using Structured Equation Modeling (SEM), which not only considers the direct influences of the predictor variables but also their unobserved or latent influence. The Ordered Probit model was developed to assess discrete change probabilities for each factor for each severity level outcomes. The direct and indirect influence of individual factors was analyzed in detail using the calibrated SEM model. It could be observed that pavement distress condition, shoulder type and condition, crash type, and collision partners play an important role in the determination of the severity level outcomes of crashes. The overall model fit for Ordered Probit was not significant, but the SEM calibrated model was significant, indicating that the SEM model can be calibrated reasonably with smaller crash datasets.

Traffic incidents occurring in freeway work zones mainly include rear-end accidents and roadside accidents, as well as collisions between vehicles and traffic facilities and construction personnel. Speeding is the key factor causing all these incidents. Therefore, in this study, to ensure the construction and traffic safety in the process of freeway reconstruction and expansion, the setting of speed limit signs in a traffic-free flow state upstream of the work zones is investigated. Based on the analysis of the braking and visual characteristics of drivers, two new approaches are proposed for individually calculating the speed limit values and spacing among the gradient speed limit signs (GSLS) upstream of the work zones. Subsequently, these approaches are applied to the reconstruction and expansion engineering project from Jilin to Longjia Airport on the Huiwu Freeway. The research results show that when the design speed is $120\mathrm{km}/\mathrm{h}$, the maximum daytime speed limit upstream of the freeway reconstruction and expansion work zones is $100\mathrm{km}/\mathrm{h}$, and it is appropriate to set the speed limit in decrements of $20\mathrm{km}/\mathrm{h}$; also, the maximum nighttime speed limit is $110\mathrm{km}/\mathrm{h}$, and it is appropriate to set the speed limit in decrements of $10\mathrm{km}/\mathrm{h}$. A positive linear relationship exists between the spacing and speed limit values, and the spacing among the nighttime speed limit signs is greater than that for the daytime speed limit signs.

The Oregon Department of Transportation (ODOT) has been using survey-grade mobile Light Detection and Ranging (LiDAR) technology to scan and create three-dimensional (3D) representations of its more than 12,880 km of roadways and gradually replacing traditional surveying when possible since 2015. The application of such technology to create a repository of images of all roadways within a state is in the early adoption phase; the direct impact of such an endeavor on the public is minimal, as it mainly affects the state agency. This article presents a method to quantify the benefits and cost savings of this endeavor using a case study approach, a present value index, and a breakeven analysis. The present value index showed over a $2 saving for every $1 invested over the technology’s 5-year useful life. The breakeven analysis showed that ODOT will breakeven with its annualized cost when mobile mapping is used to survey 79.2 km of roadways and 14 projects annually, regardless of the amount of funding received.

This paper aims to model the behavior of merging acceleration/deceleration when cars are running in a congested weaving section on a freeway during the merging implementation period by using a data-driven method called gradient-boosting decision tree (GBDT). Different from other black-box machine learning techniques, GBDT can provide abundant information about the nonlinear effects for independent variables by drawing the partial effects. Noise-filtered vehicle trajectory data collected on US Highway 101 are investigated in this study. The partial dependence plots show that the influence of independent variables on merging acceleration/deceleration is nonlinear and complicated and thus is different from the car-following behavior, which indicates that the adoption of traditional car-following models to merging execution behavior cannot reflect the distinctive behavior of merging vehicles. Evaluation of the performances in comparison with other state-of-the-art methods indicates that the proposed method can obtain more accurate results and thus is practical for simulating the merging execution behavior.