Journal of Infrastructure Systems

Risk-based asset management of water pipes is important to support pipe renewal prioritization decisions. Risk is a function of the likelihood and consequence of failure. Certain gaps were observed from the literature and a practice review of the consequences of failure modeling related to lack of data used, methodologies used, and the model testing quality. This study proposes a novel fuzzy inference system (FIS)–based consequence of failure model to assess the comprehensive failure impacts of a water pipe based on economic impacts, social impacts, environmental impacts, operational characteristics, and renewal complexity and ranks pipes into a 0 = insignificant to 5 = catastrophic scale. The 20 input parameters categorized into five modules and the 381 fuzzy rules are based on published literature, secondary data, and interviews with 25 experts from large water utilities, consultancies, and pipe associations. The model results can also be visualized on a geographic information system (GIS) for each pipe segment in the water distribution and transmission network. The applicability of the proposed model was evaluated based on data from a large water utility in the US, and the sensitive parameters were also identified. The results from model validation indicated that the proposed fuzzy-based methodology was useful for accurately modeling the consequences of failure of water pipes achieving a high root mean square error (RMSE) of 0.96.

Infrastructure managers consistently monitor the condition of their assets to predict their deterioration speed and determine the optimal time to execute preventive interventions. However, despite the recent progress in more frequent and accurate monitoring of assets and storage of the related results, in practice, real-world data often contains errors and discrepancies such as missing data or faulty entries. This problem can happen owing to collection errors during routine inspections or inconsistency of data storage formats in different years. Because the quality of data plays a significant role in the accuracy of deterioration prediction and the resulting intervention programs, it is important to improve condition state predictions by imputing the values of missing information. This paper examines the efficiency of different models that use the spatial correlation of infrastructure assets in predicting the value of missing data. The models include univariate and multivariate Kriging, a hybrid artificial neural network (ANN)-Kriging model, and the bidirectional long short term memory (bi-LSTM) neural network, which can model the data with spatial correlation or a sequential relationship. The results confirm that the condition indicator values can be estimated with reasonably low levels of error.

Being an essential component of our economy, bridges play a vital role in facilitating transportation of people and goods all over the world. Bridge infrastructure systems help in developing an effective way to monitor and protect structures in all aspects. However, the bridges are exposed to various kinds of damages due to aging, heavy load of traffic, quality of construction, and so on. Hence, determining the condition of such bridges through a proper infrastructure management system is important for understanding the potential loss in the longevity of the structures. This research paper presents the development and evaluation of an autoencoder-random forest (AE-RF) model to predict the condition rating of bridges using the National Bridge Inventory (NBI) database. To demonstrate the proposed model, a case study using bridges present in the US state of Florida has been performed using historical NBI data (2011–2020). Through this research, it was identified that when one of the deep learning models, named autoencoder, is combined with random forest (RF), it results in an efficient model for determining the condition ratings of the bridge components with fewer input parameters. The developed model was about 90% accurate in predicting the bridge’s deck condition by using other rating values as input variables and about 79% accurate without the use of any other rating factors, thereby addressing the existing research gap in determining the condition rating without the use of historic condition rating. On the other hand, the model was 78% and 77% accurate in determining the condition ratings for superstructure and substructure without using historic ratings and evaluation parameters. Hence, the proposed model would be more reliable in the evaluation of condition of existing bridges across the nation.

Because infrastructure systems are highly interconnected, it is crucial to analyze their vulnerability and resilience with the consideration of interdependencies. This paper constructed a bus–metro interdependent network model based on the passenger transfer relationship and used deep learning to identify the network topology attributes. The vulnerability process of the interdependent network to different disruptions under structural and functional perspective was studied. On this basis, this paper adopted a resilience assessment framework and mainly focused on modeling and resilience analysis of interdependent networks’ recovery processes. The optimal and instructive recovery strategy was determined, and it is shown that the increase of the coupling distance cannot alleviate the vulnerability of the interdependent network effectively; after the tolerance coefficient reaches the threshold, the effect on the vulnerability of the dependent network is weakened; and a betweenness-based strategy (BBS) works best in the preferential recovery of key nodes.

Well-documented conditions of the country’s subway systems suggest that the development of a management strategy for aging infrastructure is an important asset management challenge. The purpose of this work is to address this challenge by developing a statistical data analysis approach for forecasting the aging effect on the time-dependent deterioration rate of an identified trackway cluster. The infrastructure database is the New York City subway system and the trackway that supports the delivery of service. While much focus has been given to the steel rail subcomponent of the trackway, little has been done in the way of statistical analysis on the low-level nonconformances in the support structure of those rails—namely, the invert, ties, fasteners, and plates. Though detectable, these conditions pose no risk to either safety or the serviceability of that asset, yet they do provide a certain degree of awareness that an entirely intuitive lifecycle deterioration process is underway. This paper illustrates the development of a statistical data analysis methodology using available trackway data sets for forecasting the aging effect on the system performance.

This study investigates the dynamic effect of decomposed infrastructure development on foreign direct investment inflow (FDI) and export from selected South Asian countries by applying advanced econometric approaches. Our empirical results demonstrate that aggregated and disaggregated infrastructure development indicators (including the transport, energy, financial, and telecommunication sectors) significantly promote FDI and export in the long run. Human capital and quality of institutes spur export while the exchange rate impedes it. Besides, exchange rate, human capital, and institutional quality are driving factors to promote FDI inflows. Our findings are robust under different econometric approaches. Panel Granger causality found no reverse causality between infrastructure, and our two regressed both export and FDI inflows. The study offers policy insights to comprehend the long-run role of infrastructure to explain export and FDI inflows equations.

Aging buried sewer pipes are at risk of rupture, corrosion, scaling, and other damages, and they may cause urban waterlogging, road collapse, environmental pollution, and other urban safety problems. Therefore, it is very important to accurately detect and segment sewer pipe damages regularly for safety evaluation. This paper proposes an intelligent damage detection method to segment and measure five types of sewer pipe damages, namely, staggered joints, deposits, corrosion, scaling, and fracture, leveraging a fine-tuned fully convolutional network (FCN) algorithm. A case study on Zhengzhou’s sewer pipes is conducted. Then, a dataset of 3,558 images is built, and the images are collected from Zhengzhou’s sewer pipes. Subsequently, the encoder of the FCN is replaced with a pretrained VGG-19, and the upsampling part has spatial and channel squeeze and excitation (SCSE) modules added. The evaluation matrix of the pixel accuracy (PA), mean pixel accuracy (MPA), mean intersection over union (MIoU), and frequency weighted intersection over union (FWIoU) of the proposed method is 91.42%, 81.44%, 73.65%, and 87.16%, respectively. Subsequently, damage areas are measured at a pixel level to quantify the pipe damage. Compared with the existing mask region-based convolutional neural networks (RCNN) method, the proposed FCN-based method shows effective performance for multiple and similar sewer pipe damage detection under complex backgrounds. Moreover, a real-time damage detection platform for sewer pipes is proposed, realizing the data uploading, data storage, data analysis, and data visualization.

Pavements often are divided into short segments for data collection, analysis, and management in practice. However, in pavement maintenance and rehabilitation (M&R) planning, it is impractical and uneconomical to create M&R projects based on such short segments. Road agencies generally group consecutive pavement segments to facilitate larger projects. However, the grouping is not easy, especially for large networks, and inappropriate grouping might result in resource wastage and ineffective M&R plans. Thus, an advanced and effective grouping method is required for decision makers when making M&R plans. This study focused on grouping consecutive pavement segments in the context of network-level multiyear M&R planning and developed a multiobjective optimization (MOO)-based grouping method. The MOO model was established with three conflicting objectives: minimizing the total agency costs, minimizing the total road user costs, and maximizing the network pavement performance, subject to the constraints of annual budgets, individual and network pavement performance, and the minimal and maximal length required for a M&R project. The proposed MOO-based grouping method was tested and compared with the clustering-based grouping method and no grouping method using a real road network–based case study in the context of a 5-year pavement M&R plan. The results showed that the proposed method can help decision-makers effectively conduct segment grouping and generate cost-effective solutions when conducting pavement M&R planning at the multiyear network level.

With the increase in frequency and severity of extreme weather events, it is essential to incorporate the resilience of infrastructure networks into the decision-making process of resource allocation for maintenance planning. Considering that infrastructure systems are interdependent in nature, the impact of both extreme events and maintenance treatments on infrastructure resilience could be further amplified. Consequently, infrastructure interdependencies should be considered when analyzing the impact of extreme events and maintenance treatments. Additionally, uncertainties (such as uncertainties associated with the occurrence of extreme events and maintenance treatment effects) should also be into consideration. This paper proposes a resource allocation framework that incorporates these factors to optimize long-term resilience of infrastructure networks. The proposed framework capitalizes on integrating agent-based modeling with a double deep Q-network model to support decision-making in resource allocations; it allows infrastructure management agencies to maximize the long-term resilience of infrastructure networks while keeping their physical condition at an acceptable level. The results obtained from the case study show that the proposed framework is effective and can be customized to various local conditions.

With the increase in the complementarity of road and railway systems and a growing integration of logistical chains, there is an urgent need to coordinate the management of these two often-independent systems. Portugal has adopted a more radical approach, and with the argument of capturing synergies and reducing costs, merged the road and railway companies, creating the largest Portuguese infrastructure manager, responsible for overseeing the entire road and railway network. This research analyzes from an efficiency perspective the effects of such a policy option. The methodology used was to calculate several scores (using data envelopment analysis), and analyze the impact of a number of external factors [e.g., gross domestic product (GDP) growth, employment, financial crisis, and so on], and policy options (the merger itself), on those scores through multiple regressions. The results show that the merger had a positive impact on efficiency. The merger also improved the overall performance toward negative economic cycles, thus increasing its economic resilience.

Convolutional neural networks (CNNs) have achieved tremendous success in pavement crack segmentation. However, it is difficult for CNN-based crack segmentation methods to minimize false-negative and false-positive errors. Compared with false-positive errors, false-negative errors are more difficult to observe and reduce manually. This paper proposes a fine-tuning method for trained CNNs, called policy gradient-based focal loss (focal-PG loss). The trained CNNs will be further trained by focal-PG loss for only one epoch. The proposed focal-PG loss can be applied to reduce the false-negative errors of the trained CNNs by sacrificing their precision. The experimental results show that focal-PG loss greatly improves the crack recognition rate of the trained encoder–decoder network (EDNet). EDNet (focal-PG loss) achieves an overall precision of 96.05%, recall of 99.68%, and F1-score of 97.83% on 100 validation images. In addition, overall precision of 95.53%, recall of 99.58%, and F1-score of 97.51% are observed for the 150 testing images. U-net, LinkNet, and the feature pyramid network are also tested in the paper to validate the effectiveness of focal-PG loss. The results demonstrate that the focal-PG loss can also improve the performance of the aforementioned networks.

Abstract

Urban growth and the resulting highway congestion is driving up demand for rail transit. Rail, a significant component of transportation infrastructure, is critical to economic efficiency and is one of the least energy-intensive modes. However, the scale of operations results in high energy consumption, atmospheric pollution, and operating costs. Fortunately, some of the braking energy can be harvested and either used to power a simultaneously accelerating train or stored to power subsequent accelerations. The objective of this research was to optimize the number of locations of the energy storage devices and speed profiles. First, kinematic equations were applied to simulate energy consumption. Then, a genetic algorithm (GA) was developed to optimize the speed profiles that minimize the energy consumption with and without a wayside energy storage unit (WESS) for a rail transit line. Finally, a model was developed to optimize the WESS locations that maximized the net benefit. A case study was conducted to examine the model in a real-world setting and to demonstrate its effectiveness. The results indicate that about 980 MWh of electrical energy, or an additional 5%, could be saved by optimizing the WESS locations over only applying speed profile optimization. In addition to significant energy savings, environmental emissions could be mitigated using these methods.

Practical Applications

Excessive highway congestion and the resulting atmospheric pollution is resulting in increased demand for rail travel. Expanding service to meet these demands would result in higher overall energy consumption and increased costs. However, electric trains possess the ability to recover some of the energy dissipated as heat during the braking cycle. This recovered energy could be used to power subsequent acceleration cycles, and therefore reduce operating expenses. Due to the unevenness of rail alignment, the energy consumed and braking energy available for recovery varies on different alignment sections of equal lengths. Therefore, optimization is the key to minimizing fuel consumption and maximizing the benefit to the operator. Starting with algebraic energy equations based on Newton’s laws of motion, a model was developed to maximize the net benefit of the operator. The model contained an algorithm that dictated the position of the throttle and the locations of the energy storage devices for optimal operation. Consequently, a case study was conducted to examine the model and to demonstrate its effectiveness in a practical setting. The results indicate that, by optimizing the placements of the storage devices, approximately 5% more energy savings can be achieved than by only optimizing the throttle position.

Toll plazas are commonly described as a potential bottleneck due to the collection process at specific points on the highway. In a highway concession model, tolls are used to cover the majority of infrastructure costs. However, not all existing road concessions have toll plaza–related key performance indicators (KPIs), resulting in less efficient operations for transport managers and road users. To address this issue, this paper presents a discrete-event simulation (DES) model for policy analysis and the model for the planning, management, and evaluation of toll plazas using an extensible simulation package with animation capabilities, built by selecting modules that contain process features. The DES describes situations in which entities (vehicles) arrive, wait in the queue (if necessary), and then are attended to before leaving the system, either directly or indirectly. As a result, the simulation is based on discrete events (arrivals and departures) and statistical process data. This paper examines data from Brazil’s first public–private partnership (PPP) project in the area of road infrastructure and analyzes the toll plaza in Itaúna/MG using a case study methodology. The DES model was used in this case study, and the results show that the toll plaza was in good operating condition during the peak hour analyzed because the vehicular flow was lower than the system capacity. The willingness to pay the toll rate through the concept of open road tolling or a free-flow system, on the other hand, would provide significant benefits to road users because it would significantly reduce the time it takes to cross the toll plaza. The developed DES model assists toll plaza managers in making decisions while keeping in mind the need to comply with the operational efficiency measures agreed upon in the concession contracts.

Based on internal estimates, the US Air Force (USAF) has a deferred infrastructure maintenance backlog for its real property portfolio of over $33 billion (about 10% is airfield pavements). Without an increase in funding, divestiture of excess infrastructure, or change in strategy, this backlog is estimated to climb to over $50 billion by 2030. Reversing the growing infrastructure backlog trend requires new methods and strategies to rethink how the USAF invests in its infrastructure. This paper studied three areas aimed at improving network-level budget allocation. The algorithm behavioral and economic airfield simulation tool (BEAST), developed herein, creates a simulation tool capable of modeling behaviors and decisions of 109 organizations managing a global network of airfield pavements over 30 years. The BEAST is used to forecast outcomes of USAF investment decisions utilizing its current management strategies and historical behaviors. Analysis findings are compared to outcomes using a lowest life cycle cost tool, known as rapid asset modeling of pavement sustainment strategies (RAMPSS). The USAF’s current strategy is unsustainable; however, switching to recommendations from BEAST and RAMPSS provides a potentially significant course correction.

Flooding is and has historically been the most frequent natural disaster in the globe. There is strong evidence that the frequency of flooding is increasing. Pavements, however, are currently designed based on historic climatic conditions assuming a stationary climate, which no longer seems to be a good proxy for future conditions. The goal of this study is to characterize the performance of flooded pavement and quantify the impact of flooding on pavement deterioration and service life. To achieve this goal, the study analyzes the floods occurring in Colorado, United States, in 2013, and uses empirical data on pavement conditions to (1) quantify whether flooding impacts pavement deterioration, (2) define a conceptual model to characterize the deterioration of flooded pavements, and (3) quantify the loss of service life derived from flooding. To address these inquiries, the study used statistical analysis, stochastic Markov deterioration modeling, and Monte Carlo simulations. The study found that flooding accelerates pavement deterioration. Specifically, flooding induces a sudden drop in condition, followed by an accelerated long-term (i.e., multiyear) deterioration that reduces the pavement service life. The better the condition before flooding, the higher the loss of pavement service life. This new understanding of the impact of flooding on pavement conditions will help transportation agencies in the design of resilience programs and postflood strategies. Further research is recommended to analyze other flood events using similar methodologies to understand the influence of factors such as flood characteristics, location, and climate conditions in this phenomenon.

The structural capacity of concrete foundations supporting, for example, telecommunications and electrical transmission towers depends, among other things, upon the thickness of the concrete foundation. Current methodology for investigating the thickness of concrete foundations nondestructively has largely been focused upon deep foundations, such as caissons, and in situations where the top of the foundation is exposed. For example, the parallel seismic (PS) method is a borehole method that uses the arrival times of waves imparted in a foundation at a sensor next to a buried foundation to detect the thickness of the foundation but is typically used for deep caisson type foundations. However, in many instances, these foundations are pad and pier type footings that can be buried deep below ground and can be relatively thin, for example, less than 45 cm in thickness. The objective of the present work is to present a nondestructive method for determining the thickness of concrete pads with no direct access. The dry PS method was developed to determine the thickness of thin, buried concrete foundations. The method was first used to verify the as-built thickness, as mentioned in the engineering drawings of a guyed tower dead-man block. Second, it was used to determine the thickness of the pad foundations of five telecommunication towers with pad and pier concrete foundations in South Asia. These towers had foundation thicknesses ranging from 16 to 36 cm, as verified through excavation. Analysis of the data collected during dry PS testing resulted in a mean absolute error (MAE) of 18 mm, which equates to approximately 6% error. Thus, this paper demonstrates that the dry PS method overcomes the limitations of the current methodology and is capable of accurately determining the thickness of buried concrete foundations.