1 June 2020 in Journal of Infrastructure Systems
Past studies developed pavement performance models using data from all or multiple states across the United States. This study hypothesized that due to variation in agency practices and work activity profiles, individual pavement performance models should be estimated for each state, using data from its own roadway network, for use in its pavement management system. To test this hypothesis, this study used condition data of Interstate flexible pavements from eight Midwestern states to estimate three models: fixed-parameters regression, random-parameters regression, and artificial neural networks (ANNs). The ANNs model was found to statistically outperform the regression counterparts when estimating pavement roughness across all states. In contrast, the random-parameters model was statistically superior to the ANNs model in some cases when exploring the performance of these models for individual states. The statistical performance of models did not have a consistent trend across all states. Therefore, the application of models, based on data from multiple jurisdictions, could lead to erroneous/nonoptimal maintenance and rehabilitation decisions. Highway agencies are recommended to rely on their own jurisdictional data when developing their pavement performance models.
1 May 2017 in Journal of Transportation Engineering, Part A: Systems
A significant portion of highway infrastructure life-cycle costs can be attributed to maintenance expenditures, particularly in the later years of an infrastructure’s life. These expenditures are a key input for agency functions including cost allocation, financial planning and needs assessment, and investment evaluation. Thus, agencies seek improved techniques for forecasting future maintenance expenditures. Evidence from empirical data suggests that maintenance expenditure data exhibits significant spatial and temporal variations. If this is not addressed, the analysis could be impaired by disturbance correlation and yield potentially inefficient model estimates. In addressing this issue this paper introduces and validates (for highway pavement sections) a random-effects panel-data methodology to account for variations in routine highway maintenance expenditures at different locations and years. The results suggest that the annual maintenance expense (AMEX) values are: $250–$1,120/lane-km [$400–$1,800/lane-mile(mi)] for interstates, $125–$650/lane-km ($200–$1,000/lane-mi) for U.S. routes, and $125–$310/lane-km ($200–$500/lane-mi) for state roads, depending on the pavement age, surface type, climate, and traffic.
1 August 2017 in Journal of Transportation Engineering, Part A: Systems
Vehicle miles of travel (VMT) data have a wide range of applications in highway agency business processes. However, at all administrative levels, highway agencies continue to be stymied by the poor reliability of local-road VMT estimates resulting from the inadequacy of local-road traffic counts. This paper presents a methodology to address this problem. The methodology first clusters local roads and then imputes traffic-volume data for segments within each cluster by applying spatial interpolation techniques and sparse traffic-volume data. The proposed methodology uses geographic information system (GIS)-enabled spatial interpolation algorithms, including Kriging, inverse distance weighting (IDW), natural neighbor (NN), and trend techniques. The accuracy, in terms of prediction error, of each technique was validated using actual traffic counts. Spatial interpolation techniques can yield efficient imputations of absent traffic data and therefore can produce reliable estimates of local-road VMT. The results in this paper suggest that the use of spatial interpolation for local-road VMT estimation is cost-effective because it makes use of the available traffic-count data from existing road segments and therefore does not require additional data collection efforts. Also, a comparison was made of the relative efficacies of the alternative spatial interpolation techniques for purposes of imputing missing traffic data at certain links and ultimately for VMT estimation or prediction. The methodology can be updated easily with new traffic-count data and can be used by any highway agency for local-road VMT estimation. An essential prerequisite is a comprehensive inventory of the local roads for which the VMT is sought.
7 January 2017 in Road Materials and Pavement Design
Traditional methods of evaluating the quality of pavement subbase and subgrade compaction require considerable time and resources. Therefore, agencies continue to seek and evaluate safe, reliable, rapid, and cost-effective alternatives for quality assurance of the field compaction of unbound pavement layers. The light weight deflectometer (LWD) is a promising alternative that is increasingly gaining attention. This study investigated the feasibility of developing statistical limits for the compaction of specified combinations of subbase and subgrade materials in terms of their LWD maximum allowable deflections (MAD), to replace the need for site-specific LWD limits derived from the onsite test sections. Statistical limits were developed for six common subgrade and subgrade–subbase combinations used for highway pavement construction: lime-modified, cement-modified, and natural subgrade and the first six-inch lift of Number 53 crushed stone subbase overlaying these subgrades. Two sources of data, test sections and acceptance tests, were used. For a given material type, the data from the former exhibited less variability between projects. The test section data yielded MADs that did not vary significantly between projects for cement- and lime-modified subgrade, non-modified subgrade, and six inches of Nr.53 crushed stone over lime-modified subgrade.
31 July 2020 in Transportmetrica A: Transport Science
Subway station dwell time (Subway SDT) is used by transit agencies in determining scheduling, capacity, emissions, and cost. Currently, most agencies and researchers use a single subway SDT value that is not responsive to changes in demand, weather, location, and users’ actions. The current research addressed these shortcomings by estimating a regression model that quantifies the impact that these variables have on subway SDT. User-induced delays of greater than 2 seconds occur when riders intentionally hold the doors open and delay the departure. This results on an average SDT increase of 3.7 seconds. However, the average SDT is not statistically affected when riders hold the door for less than 2 seconds. These short delays are typically unintentional and caused by users unable to fully board the subway car due to crowding or other issues, and reflect an SDT that would otherwise be inadequate for all passenger movements. To further investigate user-induced SDT delays, a binary Probit model was estimated that quantifies the factors that influence the likelihood that a subway train will experience user-induced delay at a given station. A decrease in passenger arrival rate, a decrease in time between train stop and doors open, an increase in boarding/alighting time, and a decrease in outdoor air temperature, all increased the likelihood that a subway would experience user-induced delay.
1 June 2021 in Journal of Infrastructure Systems
Accurate forecasts of public transportation ridership are key components to a transit agency’s management program. These forecasts are used to develop revenue projections that are then imputed into short-term and long-term maintenance, rehabilitation, and capital investment programming. The forecasts are complicated by the fact that many of the causes of the underlying variability, such as socioeconomic and land-use factors, are not constant over the network. Failure to account for these spatial processes will yield biased, inefficient, and inconsistent demand estimates. To overcome these challenges, the current paper presents a spatial Durbin model for analyzing the change in the New York City subway ridership between 2011 and 2016. The results indicate that subway stations experienced a greater increase in ridership over the study period when the stations served more train lines, were located in areas comprised of census tracts with a greater number of tax units (residential, commercial, etc.), or served lower mean household incomes. Furthermore, the subway stations located in areas surrounded by census tracts with more commercial property or higher median family income are also expected to have a greater increase in ridership. Lastly, ridership at a given station decreases due to an increase in ridership at neighboring stations. This may indicate that a change in ridership at a station is due, in part, to riders in a region changing which station they use instead of riders shifting from alternative modes of transportation.