Publication Notices

Notifications of New Publications Released by ERDC

Contact Us

      

  

    866.362.3732

   601.634.2355

 

ERDC Library Catalog

Not finding what you are looking for? Search the ERDC Library Catalog

Results:
Category: Publications: Geotechnical and Structures Laboratory (GSL)
Clear
  • Deep Learning Approach for Accurate Segmentation of Sand Boils in Levee Systems

    Abstract: Sand boils can contribute to the liquefaction of a portion of the levee, leading to levee failure. Accurately detecting and segmenting sand boils is crucial for effectively monitoring and maintaining levee systems. This paper presents SandBoilNet, a fully convolutional neural network with skip connections designed for accurate pixel-level classification or semantic segmentation of sand boils from images in levee systems. In this study, we explore the use of transfer learning for fast training and detecting sand boils through semantic segmentation. By utilizing a pretrained CNN model with ResNet50V2 architecture, our algorithm effectively leverages learned features for precise detection. We hypothesize that controlled feature extraction using a deeper pretrained CNN model can selectively generate the most relevant feature maps adapting to the domain, thereby improving performance. Experimental results demonstrate that SandBoilNet outperforms state-of-the-art semantic segmentation methods in accurately detecting sand boils, achieving a Balanced Accuracy (BA) of 85.52%, Macro F1-score (MaF1) of 73.12%, and an Intersection over Union (IoU) of 57.43% specifically for sand boils. This proposed approach represents a novel and effective solution for accurately detecting and segmenting sand boils from levee images toward automating the monitoring and maintenance of levee infrastructure.
  • Widened Attention-Enhanced Atrous Convolutional Network for Efficient Embedded Vision Applications under Resource Constraints

    Abstract: Onboard image analysis enables real-time autonomous capabilities for unmanned platforms including aerial, ground, and aquatic drones. Performing classification on embedded systems, rather than transmitting data, allows rapid perception and decision-making critical for time-sensitive applications such as search and rescue, hazardous environment exploration, and military operations. To fully capitalize on these systems’ potential, specialized deep learning solutions are needed that balance accuracy and computational efficiency for time-sensitive inference. This article introduces the widened attention-enhanced atrous convolution-based efficient network (WACEfNet), a new convolutional neural network designed specifically for real-time visual classification challenges using resource-constrained embedded devices. WACEfNet builds on EfficientNet and integrates innovative width-wise feature processing, atrous convolutions, and attention modules to improve representational power without excessive over-head. Extensive benchmarking confirms state-of-the-art performance from WACEfNet for aerial imaging applications while remaining suitable for embedded deployment. The improvements in accuracy and speed demonstrate the potential of customized deep learning advancements to unlock new capabilities for unmanned aerial vehicles and related embedded systems with tight size, weight, and power constraints. This research offers an optimized framework, combining widened residual learning and attention mechanisms, to meet the unique demands of high-fidelity real-time analytics across a variety of embedded perception paradigms.
  • Application of Deep Learning for Segmenting Seepages in Levee Systems

    Abstract: Seepage is a typical hydraulic factor that can initiate the breaching process in a levee system. If not identified and treated on time, seepages can be a severe problem for levees, weakening the levee structure and eventually leading to collapse. Therefore, it is essential always to be vigilant with regular monitoring procedures to identify seepages throughout these levee systems and perform adequate repairs to limit potential threats from unforeseen levee failures. This paper introduces a fully convolutional neural network to identify and segment seepage from the image in levee systems. To the best of our knowledge, this is the first work in this domain. Applying deep learning techniques for semantic segmentation tasks in real-world scenarios has its own challenges, especially the difficulty for models to effectively learn from complex backgrounds while focusing on simpler objects of interest. This challenge is particularly evident in the task of detecting seepages in levee systems, where the fault is relatively simple compared to the complex and varied background. We addressed this problem by introducing negative images and a controlled transfer learning approach for semantic segmentation for accurate seepage segmentation in levee systems.
  • Geotechnical Investigation of Mare Island Naval Cemetery

    Abstract: The Mare Island Naval Cemetery is located just outside of Vallejo, California. This historic naval cemetery was inspected in November 2022, and signs of slope instability were identified. Two follow-up inspections were conducted by geotechnical engineers and geologists from the US Army Engineer Research and Development Center. A preliminary site investigation showed that desiccation cracking was present and that seepage at the toe of the cemetery slope could contribute to long-term stability issues. Historic vegetation had also recently been cleared, exposing the soils and headstones. If left unaddressed, these factors could lead to slope instability at the site. Increased monitoring, regular surveys, seepage remediation, and reestablishment of vegetation are recommended to prevent future instabilities.
  • Development of an Inertial Profiler Specification for Airfield Pavement Construction

    Abstract: The US Army Engineer Research and Development Center (ERDC) developed a test method and specification for measuring the smoothness of newly constructed airfield pavements using the inertial profiler. The limitations inherent in the currently accepted measurement system, the California-type profilograph, are detailed in this report. The effort detailed herein draws attention not only to the superior repeatability of the inertial profiler but also to the device’s ability to report true surface profile more accurately than the California-type profilograph. Correlations were drawn between the two devices with high (greater than 0.8) goodness-of-fit, and recommendations were made pertaining to the use of inertial profilers in place of California-type profilographs. These recommendations were not only founded on the data collected and analyzed in this effort but are also consistent with the current state of practice for other federal agencies, such as the Federal Aviation Administration and the Federal Highway Administration.
  • Influence of Fines Content on the Progression of Backward Erosion Piping

    Abstract: Backward erosion piping is a form of internal erosion that endangers the structural stability of levees and dams. Understanding the factors that influence this form of erosion can result in improved risk assessment and more appropriate modifications to new and existing structures. Historically, it has been assumed that the presence of silt size particles would reduce the gradient required for erosion. This study investigated the influence of fines content on backward erosion piping through a series of laboratory experiments on silty sands. Laboratory results show that as the fines content increased in the samples, so too did the gradient required to produce and progress piping to failure. The results indicate that a new factor is needed to properly account for silt content in backward erosion piping (BEP) risk assessment of silty sands.
  • Unified Facilities Criteria and Unified Facilities Guide Specifications for Sustainable Military Construction : Concrete, Asphalt, Wood, and Life-Cycle Assessment Perspectives

    Abstract: Construction materials such as concrete, asphalt, and wood are essential components for Department of Defense (DoD) Military Construction (MILCON) and construction for contingency operations around the world. From housing facilities, to airfields, to magazines and hardened structures, each of these materials fulfill numerous Army building applications. However, greenhouse gas (GHG) emissions stemming from the manufacturing, application, maintenance, and disposal of concrete and steel exact a significant climate burden. Thus, due to their pervasive use and commodity status, the advancement of sustainable concrete, asphalt, and wood materials are a critical driver for GHG mitigation. This report communicates a first step toward decarbonization-focused updates to UFC and UFGS by outlining major specifications related to concrete, asphalt, and wood with near- and long-term strategies to facilitate modernization. The Engineer Research and Development Center (ERDC) is poised to make a significant impact on the identification and integration of sustainable materials to meet regulatory goals for the re-duction of GHG emissions in MILCON. New guidance will be integrated into UFC and UFGS by leveraging unique re-search, development, test, and evaluation (RDT&E) capabilities in materials science, life-cycle assessment, and federal relationships with discipline working groups
  • Seismic analysis for Pohnpei Island, Federated States of Micronesia

    Abstract: The purpose of this study was to determine the seismic hazards that can affect Pohnpei Island and provide estimates of the ground motion parameters. The primary parameters are peak ground acceleration (PGA), pseudo-spectral acceleration (PSA), and velocity. These values were determined both probabilistically and deterministically to illustrate the overall seismic hazard to Pohnpei Island. A review was conducted of the technical literature to determine geologic studies that have been performed for characterization of the island’s volcanism, stratigraphy, and tectonism. This report is a desktop study that examines the tectonism of the region, the geology of the island, its geologic history, and its seismic record. No liquefaction areas or tsunami hazards were identified by researchers on the island.
  • Full-Scale Evaluation of Saltwater Concrete for Airfield Pavement Construction and Repair

    Abstract: The US Navy has a need to rapidly construct concrete facilities onshore to support contingency operations. Mixing water for concrete is typically specified to be freshwater; however, in many scenarios there are limited amounts of freshwater available for construction. Thus, use of readily available saltwater would be advantageous. This project’s objective was to evaluate the suitability of saltwater as a replacement for freshwater for producing concrete airfield pavement under relevant operational scenarios. Three full-scale test sections were constructed, and performance was evaluated in the context of relatively short design life requirements. First, direct comparison slabs of freshwater and saltwater concrete were constructed and exposed to ambient environmental conditions for one year; periodic concrete strength measurements were made. Next, 8 in. thick and 11 in. thick saltwater concrete pavements were constructed then subjected to P-8 aircraft accelerated loading. Finally, four airfield damage repair techniques were executed using saltwater and subjected to accelerated P-8 aircraft loading. Saltwater concrete performance was found to be similar to freshwater concrete for all scenarios investigated. The overall conclusion was that saltwater can be used in place of freshwater for concrete airfield pavement construction and repair for short- to medium-term use (1–2 yr) with no meaningful impact to mission requirements.
  • A Geospatial Model for Identifying Stream Infrastructure Locations

    Abstract: Management of hydraulic infrastructure for flood control, hydropower, navigation, and water supply is a critical component of the Army Dams and Transportation Infrastructure Program (ADTIP). This project provides a tool to locate stream infrastructure using a one-dimensional approach supplemented with geospatial filtering that only needs digital elevation model (DEM) files as primary input. The regions in and around Forts Liberty, Sill, and Cavazos were selected as study areas, and stream networks with corresponding stream elevation profiles were created and searched for elevation changes that met vertical threshold and search window criteria. Recall, Fβ, and a ratio of under to overprediction were used to evaluate performance. The search algorithm generally overpredicts the number of stream infrastructure locations and especially so for large search windows (20 or 25 cells) and small vertical threshold values (5 or 10 m). Overall, it was found that midrange vertical threshold values (2 or 2.5 m with long search windows (20 or 25 cells) with the land cover classification (LCC) check applied yielded results that minimized false negatives and overpredictions. The significance of this tool is that it may reduce costly field investigations, or at least aid in the prioritization of site visits for hydraulic infrastructure managers.