Federal Railroad Administration Research Articles

The relationship between freight train length and the risk of derailment.

In recent years, longer and heavier trains have become more common, primarily driven by efficiency and cost-saving measures in the railroad industry. Regulation of train length is currently under consideration in the United States at both the federal and state levels, because of concerns that longer trains may have a higher risk of derailment, but the relationship between train length and risk of derailment is not yet well understood. In this study, we use data on freight train accidents during the 2013-2022 period from the Federal Railroad Administration (FRA) Rail Equipment Accident and Highway-Rail Grade Crossing Accident databases to estimate the relationship between freight train length and the risk of derailment. We determine that longer trains do have a greater risk of derailment. Based on our analysis, running 100-car trains is associated with 1.11 (95% confidence interval: 1.10-1.12) times the derailment odds of running 50-car trains (or a 11% increase), even accounting for the fact that only half as many 100-car trains would need to run. For 200-car trains, the odds increase by 24% (odds ratio 1.24, 95% confidence interval: 1.20-1.28), again accounting for the need for fewer trains. Understanding derailment risk is an important component for evaluating the overall safety of the rail system and for the future development and regulation of freight rail transportation. Given the limitations of the current data on freight train length, this study provides an important step toward such an understanding.

Railway accident causation prediction with improved transformer model based on lexical information and contextual relationships

The railway system is a prime example of a safety-critical system. Predicting the causes of railway accidents holds immense significance in enhancing railway transportation safety. Previous approaches to railway causation analysis have encountered huge challenges regarding data processing and analytical capabilities. To address this concern, this paper proposes an innovative deep model framework based on the Transformer architecture that utilizes historical data on railway equipment accidents to predict the causes behind such incidents. Firstly, this paper proposes the utilization of Convolutional Block Attention in the domain of text processing, serving as a lexical encoder to augment word semantics acquisition in accident texts. Subsequently, in order to address the deficiency of traditional Transformers that lack positional representation information, we propose incorporating a BiGRU (Bidirectional Gated Recurrent Unit) as a contextual positional information encoder to capture contextual positional information in railway accident data effectively. Finally, considering that accident data reports are discrete tabular data, this study suggests employing cue word techniques for preprocessing accident data to alleviate the model's learning burden. We applied the proposed model to the FRA (Federal Railroad Administration) dataset. The results demonstrate that our model surpasses the current state-of-the-art language models, exhibiting superior performance compared to the optimal model with a notable improvement of 3.56%, 0.42%, and 0.76% in Precision, Recall, and F1-score, respectively. Furthermore, our model accurately predicts accident categories prone to misjudgment even when trained on limited data, outperforming existing language models. The study findings will contribute to the prevention and management of railway accidents.

Use of Zero-crossings Segmentation for Track Quality Assessment

This study concerns track quality assessment of standard-gauge railways in the context of the Hungarian railway system. Data gathered by multipurpose track recording vehicles matched the EN 13,848 requirements. Track Quality Index (TQI) measurement systems (The Federal Railroad Administration (FRA), the Netherlands’, and the Chinese TQI) are considered where three types of predetermined segment techniques: separate, moving, and Zero-crossings segmentation are employed. The importance of track segmentation in quality assessment, which affects maintenance planning, is shown by key findings. For heterogeneous data, the TQIs might be deceptive, highlighting the need for alternatives. The robustness of the Zero-crossings method makes it possible to analyze deterioration factors in great detail and in some efficient way. Longer analytical segments and smoothing of the data improved precision. Based on empirical data, we advise considering a Zero-crossings strategy for precise and efficient track-quality evaluations. With the help of this study, track quality can be better evaluated for train systems.

Traffic Safety Assessment and Injury Severity Analysis for Undivided Two-Way Highway–Rail Grade Crossings

The safety and reliability of undivided two-way highway–rail grade crossings (HRGCs) are of paramount importance in transportation systems. Utilizing crash data from the Federal Railroad Administration between 2020 and 2021, this study aims to predict crash injury severity outcomes and investigate various factors influencing injury severities. The χ2 test was first used to select variables that were significantly associated with injury outcomes. By employing the eXtreme Gradient Boosting (XGBoost) model and interpretable SHapley Additive exPlanations (SHAP), a cross-category safety assessment that offers an evidence-based hierarchy and statistical inference of risk factors associated with crashes, crossings, vehicles, drivers, and environment was provided for killed, injured, and uninjured outcomes. Some significant predictors overlapped between the killed and injured models, such as old driver, driver was in vehicle, main track, went around the gate, adverse crossing surface, and truck, while the other different significant factors revealed that the model could distinguish between different severity levels. Additionally, the results suggested that the model has varying performances in predicting different injury severities, with the killed model having the highest accuracy of 93.36%. The SHAP dependency plots for the top three features also ensure reliable predictions and inform potential interventions aimed at strengthening traffic safety and risk management practices, such as enhanced warning systems and targeted educational campaigns for older drivers.

Railroad safety: A systematic analysis of Twitter data

Railroad safety is a universal goal all over the world, which is continually facing challenges. For example, the U.S. railroad trespasser casualties excluding highway-rail incidents, but including suicide incidents, increased by 65% from 765 in 2011 to 1261 in 2020, according to the data from the Federal Railroad Administration in U.S. To improve railroad safety of the public, efforts are needed from all stakeholders that are related to railroad incidents, including the general public, government, public safety agencies, railroad companies, and first responders, etc. Social media usage of those stakeholders shed light on their perceptions, beliefs, and behaviors toward rail safety. However, few studies across the world have conducted social media analysis for railroad safety, e.g., railroad trespassing issues. To fill the knowledge gaps, in this paper, tweets related to railroad safety in the past 5 years (93,239 tweets) were collected and analyzed via user analysis, retweet analysis, BERTopic modeling techniques, sentiment analysis, emotion analysis, and hashtag and mention analysis. The user analysis showed that stakeholders related to railroad safety used Twitter in our dataset, however, there are significantly inconsistent efforts among those stakeholders in terms of using Twitter for delivering messages on railroad safety. Retweet analysis, sentiment and emotion analysis all showed that Twitter users used tweets to inform the public of railroad safety knowledge and express their concerns. The major rail safety topics (with at least 100 tweets per topic) discussed on Twitter were related to safety culture, teaching children railroad safety, removing headphones, walking illegal tracks, and emergency notification. The contributions of this paper lie into identifying benefits and potential of using social media for informing general public of railroad safety knowledge, understanding general public’s concerns, beliefs and attitudes toward railroad safety, and demonstrating the existing status of railroad related stakeholders in terms of their usage of social media for railroad safety. The identified inconsistent efforts from different stakeholders may suggest the barriers to the use of social media for railroad safety are related to government or organizational policies and social media strategies or missions. We believe that leadership support from government agencies and organizations, and the partnership between different stakeholders is the key to promoting social media for railroad safety.

Development of track component health indices using image-based railway track inspection data

The primary role of the US Department of Transportation (USDOT) Federal Railroad Administration (FRA) is ensuring the safe operation of railway rolling stock and infrastructure by way of regulatory oversight. FRA regulations require US railroads to conduct visual track inspections as often as twice per week depending on a specific track segment’s FRA track class, which also governs maximum train operating speed. Such inspections are often subjective due to the inherent limitations of human visual inspection and cognition. Additionally, human visual inspections require some level of risk given the need for inspectors to be on track while also consuming valuable network capacity. As a result, and the desire to collect objective data to improve both safety and maintenance planning, railroads are pursuing new means and methods to assess track condition and evaluate track component health. This paper presents a numerical method to define track component health using field data collected on the High Tonnage Loop (HTL) at the Transportation Technology Center (TTC) in Pueblo, Colorado, USA. Line scan laser and image data of the track were captured using a 3D Laser Triangulation system and were subsequently processed using Deep Convolutional Neural Networks (DCNNs). The track heath quantification method proposed establishes benchmarks that were developed based on the understanding of railway track mechanics, high axle load (HAL) railroad engineering instructions, and FRA regulations. The novel metrics presented are referred to as Track Component Heath Indices (TCHIs) and are quantitative values that objectively assess track condition and provide a means to monitor condition change with time and tonnage. These data can be used in conjunction with traditional track geometry and other forms of track heath data (e.g. GPR and rail profile) to more holistically assess the condition of the track structure and its components and ultimately predict its future state.

Rail integrity control system with closed loop approach

Rail systems have been used for transportation and carriage purposes since the beginning of the 19th century and are still an important means of transportation today. This transportation method is actively used in densely populated Asian countries, especially in America and European countries. According to data from the National Safety Council (NSC) in the United States, rail travel is indeed one of the safest modes of transportation, second only to air travel. NSC reported that in 2019 rail fatality rates were 0.31 deaths per billion passenger miles, while the death rate for air travel was only 0.061 deaths per billion passenger miles. However, this does not mean that serious accidents do not occur. Moreover, the cause of the majority of these accidents is derailment caused by line deterioration and breaking of rails. According to technical data from the Federal Railroad Administration (FRA), in 2020 there were 1,601 accidents caused by track issues, including rail cracks and deterioration. These statistics highlight the importance of regular road maintenance and inspection to ensure the integrity of the railway and prevent accidents caused by road problems. To date, many studies have been carried out to control and ensure the integrity of the rails. In our study, the main aim is to provide a system that will verify that there is no crack in the rails, and otherwise inform the machinist. Furthermore, this study does this verification by dividing the rail structure into small blocks and checking whether the circuit is completed with the electric current it provides over the train. Thanks to this easy-to-apply system, the deformations on the rails can be easily detected pointwise and the number of fatal accidents can be reduced.

Injury severity analysis of highway-rail grade crossing crashes in non-divided two-way traffic scenarios: A random parameters logit model

Highway-rail grade crossing (HRGC) crashes in non-divided two-way traffic scenarios have caused numerous fatalities and injuries over the years. Although crucial to the safety of multimodal transportation systems, these crossings have received little attention and previous studies did not fully account for the unobserved heterogeneity and its potential interactive effects. To bridge these gaps, the HRGC crashes occurring between 2019 and 2020 in the United States were collected from the Federal Railroad Administration's Office of Safety Analysis System. A random parameters logit model with heterogeneity in means was developed to investigate the impact of multiple factors associated with crossings, crashes, drivers, vehicles, and the environment. The present study indicates that did not stop behavior generates the random parameter with heterogeneity in means that is influenced by the dark and land with commercial power indicators. Furthermore, the findings show that factors such as estimated vehicle speed > 25 MPH, train speed > 45 MPH, going around the gate, old driver, female driver, motorcycle, and the driver was in vehicle indicators would increase the likelihood of more severe injury outcomes in HRGC crashes. Notably, the adverse crossing surface and truck indicators demonstrate unexpected marginal effects by reducing the likelihood of severe injury outcomes at non-divided two-way traffic HRGCs. This study emphasizes the importance of considering unobserved heterogeneity in the context of HRGC crashes. The findings can serve as a foundation for developing targeted interventions aimed at enhancing road and railway safety.

Quantification analysis of potential risk in railway accidents: A new random walk based approach

Safety and efficiency are the fundamental goals to improve the transportation capacity and service level of railway system. Moreover, the accurate description and in-depth exploration of potential risks are essential in reducing the railway accidents. However, there have been limited publications that focus on quantitatively analyzing the potential risk of railway accidents by using the random walk method. In this paper, a new random walk method named Comprehensive-Biased Random Walk with Different Restart (CBDRWR) is proposed in analyzing the potential risk of railway accident generation process. Combined with the established accident causation network, we give each node a different restart probability and comprehensively improve the biased transition probabilities. According to the solution algorithm of CBDRWR equation, we propose four evaluation indexes for quantifying and analyzing the potential risk of accident occurrence. In the case study, the proposed method is used to analyze the Federal Railroad Administration (FRA) dataset. The experimental results verify that the proposed method can effectively quantify the potential risk and quickly locate the key risk sources. Resultantly, this research can provide an accurate, scientific and reasonable basis for railway accident prevention and rescue.

Investigating Highway–Rail Grade Crossing Inventory Data Quality’s Role in Crash Model Estimation and Crash Prediction

The highway–rail grade crossings (HRGCs) crash frequency models used in the US are based on the Federal Railroad Administration’s (FRA) database for highway–rail crossing inventory. Inaccuracies or missing values within this database directly impact the estimated parameters of the crash models and subsequent crash predictions. Utilizing a set of 560 HRGCs in Nebraska, this research demonstrates variations in crash predictions estimated by the FRA’s 2020 Accident Prediction (AP) model under two scenarios: firstly, employing the unchanged, original FRA HRGCs inventory dataset as the input, and secondly, utilizing a field-validated inventory dataset for the same 560 HRGCs as input to the FRA’s 2020 Accident Prediction (AP) model. The findings indicated a significant statistical disparity in the predictions made with the two input datasets. Furthermore, two new Zero-inflated Negative Binomial (ZINB) models were estimated by employing 5-year reported HRGCs crashes and the two inventory datasets for the 560 HRGCs. These models facilitated the comparison of model parameter estimates and estimated marginal values. The results indicated that errors and missing values in the original FRA HRGCs inventory dataset resulted in crash predictions that statistically differed from those made using the more accurate and complete (validated in the field) HRGCs inventory dataset. Furthermore, the crash prediction model estimated upon the corrected inventory data demonstrated enhanced prediction performance, as measured by the statistical fitness criteria. The findings emphasize the importance of collecting complete and accurate inventory data when developing HRGCs crash frequency models. This will enhance models’ precision, improve their predictive capabilities to aid in better resource allocation, and ultimately reduce HRGCs crashes.

Evaluation of Rail Freight Line Reactivations in Small and Mid-Size Communities Using Data from Federal Railroad Administration’s Grade Crossing Inventory

This study explores the characteristics and performance of projects that have rehabilitated dormant rail to allow for the restoration of freight service to communities with populations of 500 or more. Drawing on data from Federal Railroad Administration’s Grade Crossing Inventory, it assesses the locational and operational characteristics of twenty-seven lines identified by the study team as having been reactivated between 2004 and 2022 after three or more years of inactivity. The results show that sixty-one communities regained service, and more than half are on lines that have attracted enough traffic for public funding to sustain daily train operations. In most instances, the number of train movements per day is two or more, although a few have seen a decline in the number of movements, and timetable speeds are commonly limited to 10 or 20 mph. The large number of instances in which operations are restricted to the bottom end of this speed range and the thirteen accidents that occurred at grade crossings in the communities over the final 6 years of the study period suggest that more investment is needed before higher-density operations become feasible. Moreover, a variety of factors may result in the hazard indices used by state governments to prioritize grade crossing investment understating the risks of train activity on reactivated lines. Nearly three-fourths of grade crossings have only passive warning equipment.

Propagation of structure-borne noise in building adjacent to subway lines

Today, the need for development of metro lines in crowded urban environments is inevitable. Construction and operation of metro generally have adverse side effects (such as noise and vibration) on residents of buildings adjacent to the metros. Based on various published reports, train-induced Structural Borne Noise is mostly annoying, causing people discomfort. Addressing this issue, probabilistic assessment of train-induced annoyance level in buildings adjacent to subway lines was made in this study. For this purpose, a numerical model of typical buildings in vicinity of a subway line was developed. The model was validated by field tests performed in this study. Nine different sites along metro lines were selected to obtain vibration characteristics of train-induced accelerations received by the metro adjacent building foundations. Various possible vibration time histories were applied to the foundation of the model, using an Incremental Vibration Analysis. At each step of the analysis, the Sound Pressure Level in the building stories were obtained and compared with the limitations proposed by the Federal Railroad Administration. Using the results obtained, the probability of the Sound Pressure Level exceeding the allowable limits was derived. In this regard, a reliability index was obtained. Based on the reliability index, the propagation level of the impermissible Structure-Borne Noise in different stories of buildings was presented as a function of number of the building stories and the peak ground accelerations. Accuracy of the proposed function was illustrated by applying it to eight different sites, indicating the accuracy of more than 83%.

Stress Distribution in a Railroad Track at the Crosstie–Ballast Interface

Excessive crosstie wear and abrasion and ballast wear and fouling are fundamental problems contributing to inadequate railroad track performance. This adversely affects the attainment and long-term maintenance of desired track geometric requirements. The magnitudes and distribution of the stresses at the crosstie–ballast (CT-B) interface must be known to determine the stress distribution on and within the ballast. However, the track design recommendations to determine these pressures, which are largely based on a methodology from the 1980s, are currently valid for modern-day railroad applications for multiple reasons discussed in this study. This study analyzed CT-B interfacial pressure data measured on an active freight mainline in Mascot, Tennessee. Dynamic contact pressures at the CT-B interface were measured using hydraulic earth pressure cells for various wheel loads and train speeds. The test train was a Federal Railroad Administration (FRA) test train consisting of a diesel electric locomotive, a test car that had different wheel loads based on the deployable axle load, and an inspection car. Although the maximum train speed was limited to 64 km/h, this research found that speed variation has a minimal effect on the CT-B interfacial pressures. From the measured data, a Gaussian stress distribution equation is proposed to determine longitudinal pressure distribution transmitted to the CT-B interface for static conditions. In addition, the stress distribution along the length of a crosstie was investigated via laboratory experimentation using a half-length crosstie. As a result of the experimentation, a dimensionless trilinear approximation was developed to estimate the stress distribution along the length of the crosstie. In general, this research recommends that the longitudinal and lateral stress distributions be considered together to design a better railroad track.

Statistical Analysis of Train Derailment Severity for Unit Trains Versus Manifest Trains

Using freight trains is a cost-effective and safe method for transporting numerous products over long distances. However, freight train derailments have significant consequences. Derailment severity, measured by the number of cars derailed per derailment, is an important risk factor. Based on prior literature and Federal Railroad Administration data, this research analyzes and estimates derailment severity of freight railroad transportation using a statistical method: the Truncated Geometric (TG) model. The methodology accounts for three train types: manifest train, loaded unit train, and empty unit train. Train length, speed, and gross tonnage per car are the input features, and the output variable is the number of cars derailed in a derailment incident. The TG model quantifies the influence of these factors contributing to derailment severity and estimates train derailment severity with a performance of the mean square error (MSE) of 68.91 and mean absolute error (MAE) of 6.05. When data outliers such as abnormally high or low severity are excluded, the MSE drops to 36.82 and the MAE drops to 4.22. Overall, the results indicate that the train derailment severity estimation performance based on the given factors is satisfactory. With all other factors being equal, a loaded unit train is likely to derail more cars than a manifest train and an empty unit train. When data outliers are excluded, there is no significant difference between manifest trains and empty unit trains as regards derailment severity, and both are less likely to derail than a loaded unit train.

A railway accident prevention method based on reinforcement learning – Active preventive strategy by multi-modal data

Railway systems are entering an era of highly intelligent automation where stability and safety are becoming increasingly important. However, there is still a lack of intelligent and effective ways for railway accident prevention, especially active accident prevention strategies. This paper presents a railway accident prevention method based on the reinforcement learning model and multi-modal data to achieve active railway accident prevention strategies. Three metrics are designed to show the performance of active prevention methods. Based on the three metrics and the data from Federal Railroad Administration, the effectiveness of the proposed method is verified in the case study by introducing two methods as baselines. The results also show that nearly 30% of accidents can be effectively prevented through active preventive measures with the proposed method. Finally, this paper analyzes the influence of personal skills on the proposed model and advises on improving railway safety based on the analysis of the results.

Railroad Crossing Surveillance and Foreground Extraction Network: Weakly Supervised Artificial-Intelligence Approach

According to the report to Congress National Strategy to Prevent Trespassing on Railroad Property issued by the Federal Railroad Administration, trespassing is currently the number one cause of all railroad-related deaths. The number of fatalities resulting from trespassing, including both illegally entering and remaining in the railroad right-of-way, is even higher than the number of deaths caused by collisions between vehicles and trains. Accurately and effectively detecting objects at rail crossings is critical for improving railway safety. The convolutional neural network (CNN) is one of the most potent tools in image processing and video surveillance. However, the current CNNs under supervised learning rely on a substantial amount of manually labeled images, which is labor-intensive and time-consuming. This study proposes a weakly supervised learning technique for training foreground-segmentation networks and develops a tailored CNN network, RC-SAFE (railroad crossing surveillance and foreground extraction), to accurately and effectively detect and track both still and moving objects in the railroad crossing area. The weakly supervised learning significantly reduces the load for image labeling. The experimental results on the public CDnet 2014 dataset show that the weakly supervised network outperforms the supervised network and some other state-of-the-art foreground-segmentation methods. Railroad crossing monitoring videos are collected at different railroad crossings to validate the proposed RC-SAFE network. The results confirm the robustness of the proposed RC-SAFE network, and that RC-SAFE can detect any object that does not belong to the railroad crossing, which outperforms object-detection-based models, such as Mask RCNN (a region-based CNN).

Effect of Critical Factors Influencing Longitudinal Track Resistance Leveraging Laboratory Track Panel Pull Test Experimentation

There are, on average, 12.5 Federal Railroad Administration reportable derailments per year on U.S. mainlines and sidings caused by “defective or missing spikes or rail fasteners.” Because fastener failures are most commonly caused by a combination of vertical, lateral, and longitudinal loads, it is important to quantify all loads placed on the fasteners to reduce the number of failed fasteners and increase rail safety. Multiple researchers have developed analytical models that leverage longitudinal track resistance and stiffness to quantify the fastener demands. Therefore, to support the refinement of these analytical models that leverage longitudinal track resistance and stiffness, track panel pull tests (TPPTs) were executed in the laboratory to expand on the values within the available literature. These TPPTs quantified the effect of sleeper type (i.e., timber versus concrete), given that 88% of previous studies have focused on concrete. Further, these novel tests quantified the effect of the fastening system, crib ballast height, shoulder width, and ballast condition on the panel’s longitudinal resistance and stiffness. From these experiments and the resulting analysis of data, multiple conclusions were drawn. For example, concrete sleeper panels exhibit 20% higher resistance than timber sleeper panels, disturbing ballast reduced the longitudinal resistance by 13% and stiffness by approximately 80%, and the crib, shoulder, and bottom ballast provide approximately 65%, 5%, and 30% of the total longitudinal resistance, respectively.

Wheel Out-of-Roundness Detection Using an Envelope Spectrum Analysis.

This paper aims to detect railway vehicle wheel flats and polygonized wheels using an envelope spectrum analysis. First, a brief explanation of railway vehicle wheel problems is presented, focusing particularly on wheel flats and polygonal wheels. Then, three types of wheel flat profiles and three periodic out-of-roundness (OOR) harmonic order ranges for the polygonal wheels are evaluated in the simulations, along with analyses implemented using only healthy wheels for comparison. Moreover, the simulation implements track irregularity profiles modelled based on the US Federal Railroad Administration (FRA). From the numerical calculations, the dynamic responses of several strain gauges (SGs) and accelerometer sensors located on the rail between sleepers are evaluated. Regarding defective wheels, only the right wheel of the first wheelset is considered as a defective wheel, but the detection methodology works for various damaged wheels located in any position. The results from the application of the methodology show that the envelope spectrum analysis successfully distinguishes a healthy wheel from a defective one.

Yük Vagonu-Hat Dinamik Analizi

In this study, a six-degrees-of-freedom half-car model for a freight wagon is used to study the ride motions of the vehicle under random rail inputs. As a rail input, the track profiles taken from Federal Railroad-Administration (FRA) international standard are used. A coupled vertical vehicle-track system is formed as a polytopic function of the track roughness parameter and vehicle speed and their influence on the vehicle dynamics is discussed. Then an optimization problem with a single objective function is formulated for a range of track roughness parameters and vehicle speeds. H_∞ control focuses on obtaining a robustness concerning the uncertainty of the system. So, we used H_∞ optimization to obtain the solutions while maximizing the trade-off between the respective performance indices. Later, a controller with a fixed speed value is synthesized and the results are compared by using the frequency response plots and the root-mean-square values of the car body accelerations and secondary and primary suspension deflections. The simulation results demonstrate that the active system is effective in improving ride comfort while keeping the rail holding within allowable limits.

Identifying Risk Factors for Autos and Trucks on Highway-Railroad Grade Crossings Based on Mixed Logit Model.

This study aimed to determine different influencing factors associated with the injury outcomes of heavy vehicle and automobile drivers at highway-rail grade crossings (HRGCs). A mixed logit model was adopted using the Federal Railroad Administration (FRA) dataset (n = 194,385 for 2011-2020). The results show that drivers' injury severities at HRGCs are enormously different between automobile and truck/truck-trailer drivers. It was found that vehicle speed and train speed significantly affect the injury severity in automobile and truck drivers. Driver characteristics such as gender and driver actions significantly impact the injury severity in automobile drivers, while HRGC attributes such as open space, rural areas, and type of warning device become significant factors in truck models. This study gives us a better understanding of the differences in the types of determinants between automobiles and trucks and their implications on differentiated policies for car and truck drivers.