Journal of Transportation Engineering, Part A: Systems

Although reliability centered maintenance (RCM) has been used widely in the maintenance and renewal (M&R) of railway infrastructures, M&R methods and maintenance frequencies still require considerable study to obtain information on optimum decisions. Recently, various attempts have been made to optimize the maintenance works in railway engineering, although a decision support system on selecting the maintenance or renewal works has not been suggested distinctly. Therefore, a genetic algorithm on the basis of the nondominated sorting method (NSGA) was used in this study to optimize the biobjective functions on M&R interventions and costs. The results showed that the optimum frequency of M&R could be determined using the NSGA method with various constraints. Moreover, field application of the suggested method also was considered. Future work will be concentrated on the systematic approach of the decision support model with RCM and the enterprise resource planning (ERP) system of the infrastructure manager (IM).

Traffic signal coordination has been widely used to provide smooth progression for platoons on signalized arterials to enhance performance measures such as travel times, number of stops, and delay. Considering smooth progression as the significant factor of coordination, many studies have been conducted to measure the quality of progression. However, none of these studies concentrated on measuring a continuous smooth driving pattern of each vehicle in terms of speed. In order to quantify the smoothness, this paper conducted an analysis of the speed variation of vehicles traveling along a corridor. A new measure, called the smoothness of the flow of traffic (SOFT), was introduced and evaluated for different kinds of traffic control systems. The measure can be used to evaluate how smoothly vehicles flow along a corridor based on the frequency content of vehicle speed. To better understand the impact of vehicle mode, a multimodal analysis was conducted using the SOFT measure. This study was conducted using a simulation model and a system where vehicle trajectory data are available for computing SOFT.

Traffic congestion and freeway bottlenecks continue to challenge existing transportation networks. This study presents a systematic method to evaluate freeway performance and locate and rank freeway bottlenecks while accounting for both intensity and reliability dimensions of traffic congestion. A data-driven approach is used to determine a local range of the weighting factor. Based on the vehicle probe data collected on four interstate freeways in Mecklenburg County, North Carolina, a case study is conducted to illustrate this new method. Numerical results clearly indicate that although two freeway segments have nearly identical reliability values, their intensity levels can be significantly different, and vice versa. Hence, quantifying both dimensions of traffic congestion in freeway bottleneck studies is necessary. The research results can provide insightful and objective information for decision makers and transportation professionals to systematically assess traffic conditions along freeway segments and objectively locate and rank freeway bottlenecks, competently develop congestion mitigation strategies, and thus allocate limited transportation funding in a more effective and efficient manner.

Diverging diamond interchanges (DDIs) are widely recognized to be capable of reducing conflict points, number of stops, and consequently average traffic delay. However, the design of their crossover spacing and signal offsets, which is critical to the capacity and efficient operations of DDIs, have not been addressed in most design guidelines. These two critical design components are actually interdependent in nature, because the estimated travel time between a DDI’s two subintersections for all movement paths is essential for the design of signal offsets. Also, the crossover spacing should be designed to accommodate queues comprised mostly of those vehicles not moving within the signal progression band, which is often designed with a given crossover spacing. Considering such an interdependent relation between signal offsets and the crossover spacing, this study presents a model that can concurrently optimize these two vital DDI design elements at the planning level. A case study at a DDI site with the proposed model has also been conducted to justify the necessity to perform the concurrent optimization under different operational conditions. The results of extensive numerical experiments confirm that the design with the optimized crossover spacing and offset can yield the shortest total delay and the least number of stops for vehicles over the entire network, especially under near-saturated conditions. The optimized crossover spacing can also prevent the formation of the queue spillover over the crossovers in a DDI.

This paper illustrates a newly developed methodology for passenger car unit (PCU) estimation under non-lane–based heterogeneous traffic conditions on multilane urban roads, using area occupancy as a measure of performance. The results show that an increase in the number of lanes or in carriageway width has a significant increasing effect on PCU values of vehicle categories. To examine the accuracy of the proposed method, speed–flow plots are compared on same set of axes using (1) the proposed unique (optimized) PCU values, (2) a dynamic set of PCU values (estimated using the common speed:area ratio method under non-lane–based traffic conditions), and (3) a static set of PCU values from existing guidelines, at a 5-min interval aggregation level for selected roads using real field data. The comparison shows a difference of 2.0 and 1.3% at capacity level for eight-lane divided and ten-lane divided road sections, respectively. This ensures the accuracy of the proposed simplified yet rigorous methodology of PCU estimation as well as the proposed unique set of PCU values. The soundness of the developed unique/optimized PCU values is checked on a test section of an eight-lane divided road having similar roadway and traffic conditions. At the capacity-flow level, the error is well within acceptable limits (0.3–3%), indicating the credibility of the developed matrix-based PCU method using area occupancy as a measure of performance.

With use of concrete sleepers increasing for rail-transit applications in the United States, it is becoming more critical to quantify their revenue service flexural demands to improve sleeper design and maintenance practices. Rail-transit concrete sleeper bending moment field data were collected and processed to address topic areas relating to (1) overall field bending moment magnitude relative to design moments; (2) moment variation from sleeper to sleeper resulting from support conditions; and (3) seasonal variations in moments. Data from field locations on light and heavy rail-transit properties show levels of reserve flexural capacity (factors of safety) that reach as high as 6, significant sleeper-to-sleeper variability attributable to support conditions that can be as high as 100%, and seasonal variation in bending moments that is measurable but far lower than daily variability caused by temperature by a factor of 2. These data provide a valuable baseline for the future generation of mechanistic design standards for track infrastructure components.

Travel time reliability refers to the day-to-day variability of trip travel times. There is a belief that there is a cost associated with unreliability and this cost can be quantified as a function of the difference between the travel time that was experienced and the travel time that was anticipated. In the realm of public transport systems, the anticipated travel time is essentially the scheduled travel time and is therefore easy to compute. However, for personal auto modes, it is not clear how the anticipated travel time should be computed. This paper focuses on addressing the following two questions: What is the relationship between the distribution of the travel times that travelers experience and the travel times that travelers anticipate for a future trip? What effect do unusually long travel times have on this anticipated travel time? The authors explored both questions through a stated preference survey that was distributed to over 3,000 individuals. Just over 300 valid responses were received, and on the basis of the survey results, a two-stage model for estimating the anticipated travel time as a function of the experienced travel time distribution was formulated and calibrated. The proposed model can be applied to travel times obtained from simulation models or from field observations.

Despite the potential use of global positioning system (GPS) based automatic vehicle location (AVL) data in the development of reliability improvement strategies for transit systems, issues with privacy and accuracy have presented challenges to the open sharing of the AVL data. However, data driven methods of estimating the performance measures of transit vehicles on the basis of the location data are still the most prominent means of reliability studies on transit systems. This paper aims to propose a new method for developing transit performance measures, namely traffic delay, based on the stop-level location data, which does not share the issues concerning privacy and accuracy that hinder the use of GPS-based data. The paper presents a case study on route 16 of the Metro Transit system in the Twin Cities, Minnesota. The delay measures that resulted from the proposed model are at the segment level of resolution, delimited by the route timepoints, and are found to be described well by the gamma distribution. These results allow the localization of excessive delay on a segment level, and the distinction between the traffic delay and dwell time delay.

Train-mounted vertical track deflection (VTD) measurements offer new opportunities for estimating rail bending stresses over long distances. The estimations are possible because of mathematical correlations among rail deflections, rail stresses, and the loads applied to the rail. Previous numerical studies conducted by the authors resulted in a methodology that suggests the use of finite-element models to develop the correlations. These models facilitate the simulation of a stochastically varying track modulus along the track and provide a strong basis for interpreting the deflection data. In this study, data sets collected from a study site were used to validate this methodology for estimating rail bending stresses under passing train loads. The rail-mounted strain gauges and the wheel impact load detector system at the study site provided information about the rail bending strains under known applied loads. This allowed validation of the maximum bending stresses estimated using train-mounted deflection measurements. The magnitude of rail bending stresses was assessed using measurements from different seasons; stress changes over time were also investigated.

When estimating the required acceleration length of vehicles accelerating from a stopped position, a constant acceleration rate is sometimes assumed for the sake of simplicity. Nevertheless, to date there is no clear-cut evidence showing whether a constant acceleration rate is reasonable in providing appropriate acceleration lengths. In this paper, the estimation errors associated with the constant acceleration assumption were investigated. For each sample, a piecewise-constant acceleration model was used to figure out the vehicle’s speed versus location profile; accordingly, the constant acceleration rate was calculated. Using this constant acceleration rate, speeds at the predetermined locations were calculated and compared to the results that were generated by the piecewise-constant acceleration model. Based on the statistical analysis of 575 individual vehicle-accelerating trajectories, it was found that the constant acceleration model failed to reproduce the field-observed acceleration profile. Therefore, simply using a constant rate cannot accurately estimate the required acceleration lengths at various freeway design speed scenarios. This indicates that when adopting the constant acceleration assumption for each design speed scenario, it is necessary to use an appropriate acceleration rate that fits for this design speed. Finally, this paper revealed that normal acceleration rate ranges from $1.31\mathrm{m}/{\mathrm{s}}^2$ ($4.3\mathrm{ft}/{\mathrm{s}}^2$) at $64\mathrm{km}/\mathrm{h}$ (40 mph) to $1.8\mathrm{m}/{\mathrm{s}}^2$ ($5.9\mathrm{ft}/{\mathrm{s}}^2$) at $32\mathrm{km}/\mathrm{h}$ (20 mph).