Nationwide Patterns of Water Service Line Failures: Insights from 20 Years of U.S. Data

Nationwide Patterns of Water Service Line Failures: Insights from 20 Years of U.S. Data | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Short Report Nationwide Patterns of Water Service Line Failures: Insights from 20 Years of U.S. Data Sarah Wohlfahrt, Myles Meehan, Juneseok Lee This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8187953/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Key Takeaways - This study analyzes 20 years (2006-2025) of nationwide water service line failure and replacement data, collected through a unique collaboration with a home services company, covering over 76,000 real-world incidents. - Multiple survival analysis methods, including Kaplan-Meier (KM) and Weibull models and Random Survival Forest (RSF) machine learning, were used to assess failure risk across different pipe materials. - Municipalities and homeowners are encouraged to inventory existing service line materials and put measures in place to assist in managing high-risk types (e.g., polybutylene, polyethylene) to reduce failure rates, improve reliability, and lower long-term maintenance costs. Civil Engineering Environmental Engineering Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Water service lines (WSLs) are the essential link between municipal water mains and building plumbing systems, delivering potable water directly to homes, businesses, and public institutions. The material and condition of these lines directly affect water quality and public health, particularly in systems that contain lead or other high-risk materials (Lee et al., 2023 ; Lee & Meehan, 2017 ). Service lines have some of the highest failure rates within the water distribution system and often exhibit the lowest chlorine residuals, making them a critical focus for both infrastructure reliability and water quality protection (AWWA, 2024 ; NRC, 2006 ). According to the 7th Drinking Water Infrastructure Needs Survey and Assessment (DWINSA), there are an estimated 100 million service lines nationwide, spanning materials such as lead, galvanized steel, copper, and plastic ( USEPA , 2023). Failures in these lines, through cracking, bursting, leaking, or corrosion, pose growing challenges for homeowners, utilities, and municipalities. As infrastructure continues to age, understanding how long different materials last is essential for developing proactive, cost-effective replacement and maintenance strategies. This article evaluates 20 years (2006–2025) of nationwide WSL failure and replacement data, sourced through a partnership with HomeServe USA Corp. (HomeServe). The analysis examines pipe age, material type, replacement trends, and spatial distribution using multiple survival analysis techniques, including Kaplan-Meier (KM) survival curves, Weibull models, and Random Survival Forest (RSF) machine learning. This study builds upon earlier research published in the Journal of American Water Works Association by Lee and Meehan ( 2017 ), which analyzed WSL failures from 2006 to 2015. That work revealed significant variation in material longevity, for instance, copper pipes generally outperformed poly-based materials. In that study, survival analysis was performed, indicating failure rates increasing significantly after 30–60 years of service. The findings were also featured in The Washington Post as part of a broader national discussion on aging infrastructure and consumer decisions (Checkbook, 2021 ). The current study extends that analysis through 2025, offering a comprehensive 20-year view of material performance trends and failure risks across the United States. Data The dataset comprises nationwide WSL records from 2006 to 2025, collected through a collaboration with the home services company HomeServe. The original dataset included approximately 3.8 million rows of job site information related to water service line repairs and replacements. Key attributes in the dataset included: Existing Material: The type of pipe material previously in use. Built Date: The year the house was constructed and assumed pipe installation date. Job Date: The date associated with the pipe failure. Job Zip Code: Zip code of the failure location. Replacement Material: The new material used for pipe replacement. Diameter: The pipe diameter (inches). Total Cost: The cost of repairing/ replacing the pipe. The data were cleaned and organized into a workable format using R Studio. To improve accuracy for this analysis, rows with missing values, conflicting entries (e.g., multiple materials or diameters listed), or logical errors (e.g., built dates later than job dates resulting in negative pipe ages) were removed. After cleaning, the final dataset included approximately 76,000 valid records of service line repairs or replacements between 2006 and 2025. The national distribution of failed WSLs shows clear regional patterns, in part due to the areas served by HomeServe, with the highest concentrations in the Northeast, Midwest, and parts of California (Fig. 1 ). Dense clusters are visible around legacy urban centers such as New York City, Chicago, and Philadelphia, reflecting the prevalence of older infrastructure. Additional hotspots in the Southeast and Southwest suggest emerging risks in rapidly developing areas, likely driven by material degradation under environmental stress. In contrast, failures are less frequently reported in the Mountain West and Great Plains, which may reflect both lower population density and limited data coverage. A descriptive statistical analysis was conducted to characterize the dataset. The majority of replacement costs ranged from $ 0 to $ 5,000, with the mean around $ 1,000. Pipe diameters spanned from 0.25 to 6 inches, with 0.75 inches being the most common size. The most frequently encountered existing pipe material was copper, followed by galvanized pipe, while the most common replacement material was copper, followed by cross-linked polyethylene (PEX). Of the 76,000 service line repairs analyzed, 2,711 involved the replacement of lead pipes, representing approximately 3.6% of all documented failures. While this may appear to be a relatively small proportion, the presence of lead carries outsized public health implications, as even intact lead lines pose contamination risks. The concentration of these replacements in older urban areas, such as New York City, Boston, Chicago, and Columbus (Fig. 2 ), reflects historical installation practices and highlights regional equity and infrastructure aging challenges. These findings underscore the continued urgency of identifying and replacing lead service lines, not solely based on failure likelihood, but also on regulatory compliance and urgent health risk mitigation. To further understand the factors driving WSL failures, including material performance over time, this study applied a combination of statistical and machine-learning techniques. The following section outlines the analytical methods used to evaluate failure timing, model survival behavior, and identify the most influential variables across the full dataset of 76,000 service line replacements. Methods: Survival Analysis and Machine Learning This study applied survival analysis to evaluate the longevity of WSLs, where the event of interest was failure, defined as the need for repair or replacement. Survival analysis models estimate the probability that an asset will continue functioning over time. Because all data points in this study represent actual failures ( i.e. , no censored data), traditional survival models may introduce selection bias. However, these methods still provide valuable insights into relative material performance and failure timing (Lee & Meehan, 2017 ). The following techniques were used to characterize failure behavior across pipe materials: Kaplan-Meier (KM) Survival Curves : A nonparametric method to estimate survival probability over time. Weibull Models : Parametric models useful for identifying failure patterns and estimating time-to-failure distributions. Random Survival Forests (RSF) : A machine learning approach for modeling survival without assuming specific functional forms. KM and Weibull models were applied to estimate survival over time. KM, a nonparametric method, captures the proportion of assets remaining in service at each time point and is useful for comparing materials with different longevity profiles. Weibull modeling, on the other hand, introduces a parametric framework that quantifies how failure risk evolves with age, offering interpretable parameters and the ability to forecast future performance (Jackson, 2016 ; Kleinbaum & Klein, 2012 ; Therneau, 2023 ). Lastly, machine learning was introduced through the Random Survival Forest (RSF) model, which can capture complex, nonlinear relationships and interactions among predictor variables. Unlike traditional models, RSF does not rely on prior assumptions about data distribution and can handle high-dimensional data effectively (Ishwaran et al., 2008 ). RSF provides several analytical advantages over traditional survival models: No parametric assumptions: RSF is fully nonparametric, improving flexibility and robustness. Nonlinear interaction detection: RSF identifies interactions among variables (e.g., pipe material, diameter, installation year) that KM and Weibull models cannot detect. Variable importance ranking: RSF quantifies the relative influence of each input variable on survival outcomes, even if those variables are not statistically significant in parametric models. This combination of statistical and machine-learning approaches enables a more comprehensive assessment of pipe material performance and failure risk over time. Results Kaplan-Meier Survival Analysis. The KM survival curves illustrate the probability that a given pipe material will continue functioning over time since installation (Figure 3). Steeper declines and curves that drop closer to the y-axis indicate shorter service life and earlier failures. Materials such as PEX, polybutylene, and blue polyethylene (poly) show faster declines, indicating more rapid degradation and shorter lifespans. In contrast, lead and galvanized steel exhibit higher survival probabilities for longer durations, suggesting greater longevity. Weibull Survival Function Analysis . The Weibull survival curves provide insight into how quickly different pipe materials deteriorate over time (Figure 4). Steep declines in the curves closer to the x-axis indicate shorter lifespans and earlier failures, while initially flatter curves with delayed declines suggest greater durability and longer service life. The overall plot shows a sharp decline in survival probability within the 100 years, indicating that most service lines fail within this timeframe. The survival curves are closely grouped, suggesting broadly similar failure patterns across materials, though subtle differences are evident. The failure dynamics of different pipe materials can be described through the Weibull survival model by estimating the shape and scale of deterioration. The Weibull shape parameter ( β) clarifies how each pipe material’s failure risk changes with age. Values below 1 indicate a high incidence of early (infant‑mortality) failures that taper off, whereas values above 1 signal an aging‑related hazard that grows over time. In this dataset, PEX ( β = 0.53), brass (0.68), and PVC (0.75) exhibit the strongest early‑failure tendencies; they are more likely to fail soon after installation but become comparatively stable thereafter. Materials such as copper (0.82), polybutylene (0.80), blue poly (0.89), and black poly (0.84) show milder early‑failure behavior that likewise improves with age. By contrast, lead (1.46), galvanized (1.10), and steel (1.08) display β > 1, meaning their failure risk increases as they age, consistent with corrosion and long‑term degradation processes, while iron (0.99) sits near the exponential boundary, suggesting a roughly constant hazard throughout its life cycle. The scale parameter ( λ ) of the Weibull survival model provides insight into the typical time frame over which failures are expected to occur, effectively reflecting the material's average service life. Higher λ values indicate longer-lasting performance, while lower values suggest that failures are likely to occur earlier in the material's lifespan. In this dataset, lead ( λ = 4.64) and steel (4.26) exhibit the highest scale values, implying these materials generally persist longer before failure, despite aging-related risk increases. Copper (4.07) and iron (4.28) also show relatively long service lives, consistent with their known durability. In contrast, PEX (3.67), polybutylene (3.68), and blue poly (3.65) have the lowest scale values, indicating shorter average lifespans and a higher concentration of early failures. These scale differences help contextualize the failure timing observed across materials and complement the shape parameter ( β ) in characterizing both how and when different pipe types are likely to fail (Table 1). Table 1. Pipe Material Failure Trends Described by Weibull Model Parameters Material β (Shape Parameter) Λ (Scale Parameter) Expected Failure Trend Copper 0.82 4.07 Moderate early failure risk, but generally long-lasting performance Brass 0.68 3.91 Pronounced early failures, shorter average service life Galvanized 1.10 4.35 Aging-related failures increase over time, relatively long service life Iron 0.99 4.28 Constant failure risk, moderate-to-long expected lifespan Steel 1.08 4.26 Mild aging effect, stable performance over moderate duration PEX 0.53 3.67 High infant-mortality rate, short lifespan unless early failures avoided PVC 0.75 3.77 Early failures present, but relatively stable afterward Polybutylene 0.80 3.68 Early failure trend, shorter average lifespan Lead 1.46 4.64 Strong aging effect, long physical life but increasing failure risk Blue Poly 0.89 3.65 Early failure is common, but risk declines with age Black Poly 0.84 3.84 Similar to blue poly, with early failures tapering over time Random Survival Forest (RSF) Analysis. The RSF model was used to evaluate the factors influencing WSL longevity. A variable importance analysis was performed to find the parameters with the strongest influence on failure likelihood. With a higher importance value, the analysis revealed that pipe material was the most significant predictor of survival, followed by pipe diameter as a minor predictor (Figure 5). Replacement cost has a negative importance value, and therefore, has no influence on failure likelihood. The RSF survival curves illustrate the probability of each material type remaining functional over time, with survival probability on the y-axis and years since installation on the x-axis (Figure 6). Galvanized and lead exhibited the slowest decline in survival probability, indicating superior durability and longer service life. Lead, in particular, maintained survival probabilities above 50% beyond 75 years. In contrast, polybutylene, blue poly, and brass showed the steepest declines. Their survival probabilities dropped rapidly within the first 50 to 75 years, suggesting significantly shorter service lives. Black poly and blue poly approached zero survival probability well before the 100-year mark. The middle-ground materials include copper, steel, and iron showing high probabilities of survival (>85%) through the first 30 years and declining to approach zero around 100 years. Interestingly, while KM results suggest that copper consistently outperforms PEX, the Weibull model indicates that PEX may demonstrate longer survival under certain installation or environmental conditions. These differences underscore the value of using multiple modeling approaches . The Weibull model’s ability to describe the changes in risk trends over time allows predicting pipe longevity behavior while considering additional factors to pipe material. Additionally, the RSF model’s ability to account for complex, nonlinear interactions make it particularly well-suited for context-specific asset management, rather than relying on uniform material replacement strategies. Implications The findings from this 20-year analysis offer critical guidance for infrastructure planning, material selection, maintenance scheduling, and public health policy. Materials with longer survival times, such as copper, should be prioritized for long-term installations, while poly-based materials, which exhibit earlier and more frequent failures, may require more frequent monitoring and earlier replacement. Homeowners should be aware of the material composition of their service lines and the associated risks. Proactive replacement or having measures in place to help manage replacement of high-risk materials can reduce long-term costs, mitigate failure rate, and prevent property damage. However, material selection is rarely based on performance data alone. Factors such as health concerns, taste and odor, corrosion resistance, installer recommendations, insurance requirements, and even marketing play significant roles in decision-making. Prior research has shown that homeowners value health and water quality above cost or installation convenience (Lee et al., 2013). Utilities and engineers should adopt data-driven, predictive maintenance strategies, leveraging survival models and RSF techniques to anticipate failures before they occur. This shift from reactive to proactive asset management enables more strategic resource allocation, minimizes emergency repair costs, and improves service reliability. Finally, the implications extend beyond engineering into public health. While materials like lead may exhibit structural longevity, their continued use poses significant health risks. As of January 2025, an estimated 9.2 million lead service lines remain in use across the U.S. (Keller et al., 2025). In response, the EPA’s Lead and Copper Rule Improvements mandate the full replacement of lead pipes by 2035. Although the average replacement cost is approximately $6,000 per line, federal, state, and local programs offer funding support and incentives to assist homeowners and utilities with this critical transition (Keller et al., 2025; Rhyan et al., 2023; US EPA, 2019). Limitations This study relies on a dataset composed exclusively of failed water service lines. While the absence of censored data ( i.e., WSL still in service ) limits the application of traditional survival models, the large volume of failure records, 76,000 cases over 20 years, still enables meaningful insights into material performance, temporal patterns, and regional variation in failure behavior. Apparent differences in the results from KM, Weibull, and RSF models are not contradictions but reflections of each method’s underlying assumptions and analytical perspective. KM and Weibull provide aggregate survival insights, with KM capturing nonparametric survival patterns and Weibull enabling parametric forecasting. In contrast, RSF captures complex, nonlinear interactions among multiple variables, such as geography, pipe diameter, and installation era. These differences underscore the importance of using multiple modeling approaches to capture diverse failure dynamics and highlight the need for utility-specific modeling when planning replacement and maintenance strategies. Conclusions This study is based on a failure-only dataset, where all observations represent WSLs that were replaced due to failure. While this limits the ability to estimate unbiased survival probabilities, since censored data are not available, it still provides valuable insights for comparative analysis. With more than 76,000 real-world failure records collected over 20 years across a wide range of geographic and climatic conditions, the dataset enables a robust evaluation of material-specific degradation trends, failure timing, and regional risk patterns. Similar to the Lee & Meehan 2017 study based on data from 2006-2015, this failure analysis found that failure rates increased significantly around 25-50 years after installation. Among all variables considered, pipe material emerged as the most influential factor affecting service line longevity. Materials such as polybutylene, blue poly, and black poly exhibited the highest failure rates, while copper and iron demonstrated significantly greater durability. The combined application of KM, Weibull, and RSF models revealed critical failure thresholds and material-specific risk profiles. Despite being trained on failure-only data, the RSF model proved capable of identifying meaningful predictive relationships that can inform proactive maintenance strategies. Based on the analysis, the following recommendations are proposed to guide utilities and municipalities in asset management and replacement planning: Prioritize the replacement of high-risk materials, particularly polybutylene and polyethylene pipes, with more durable alternatives such as copper. Leverage machine-learning-based and survival model predictions to refine maintenance schedules, preemptively replace assets before failure, and allocate resources more effectively to optimize infrastructure investment and reduce long-term costs. Integrating survival analysis and machine learning into asset management empowers utility managers and engineers to make informed, data-driven decisions. This approach enhances system reliability, improves long-term operational efficiency, and supports the development of safer, more sustainable water infrastructure systems. References 7th Drinking Water Infrastructure Needs Survey and Assessment . (n.d.). Retrieved June 30, 2025, from https://www.epa.gov/system/files/documents/2023-04/Final_DWINSA%20Public%20Factsheet%204.4.23.pdf AWWA. (2024). M22 Sizing Water Service Lines and Meters, Fourth Edition . https://store.awwa.org/M22-Sizing-Water-Service-Lines-and-Meters-Fourth-Edition Checkbook, J. B. and K. B. | W. C. (2021, February 15). Do you really need insurance for your water and sewer lines? Here’s what to know. The Washington Post . https://www.washingtonpost.com/lifestyle/home/do-you-really-need-insurance-for-your-water-and-sewer-lines-heres-what-to-know/2021/02/15/651f3dd6-6baa-11eb-9f80-3d7646ce1bc0_story.html Ishwaran, H., Kogalur, U. B., Blackstone, E. H., & Lauer, M. S. (2008). Random survival forests. The Annals of Applied Statistics , 2 (3), 841–860. https://doi.org/10.1214/08-AOAS169 Jackson, C. (2016). flexsurv: A Platform for Parametric Survival Modeling in R. Journal of Statistical Software , 70 , 1–33. https://doi.org/10.18637/jss.v070.i08 Keller, J., Burchett, R., & Gallet, D. (2025). Strategies for Effective Lead Service Line Replacement Communication. Journal AWWA , 117 (3), 32–40. https://doi.org/10.1002/awwa.2418 Kleinbaum, D. G., & Klein, M. (2012). Survival Analysis: A Self-Learning Text . Springer New York. https://doi.org/10.1007/978-1-4419-6646-9 Lee, J., Burkhardt, J., Buchberger, S., Grayman, W., Janke, R., Murray, R., & Platten, W. (2023). Premise Plumbing Modeling . ASCE. Lee, J., Kleczyk, E., Bosch, D. J., Dietrich, A. M., Lohani, V. K., & Loganathan, G. (2013). Homeowners’ decision-making in a premise plumbing failure–prone area. Journal-American Water Works Association , 105 (5), E236–E241. Lee, J., & Meehan, M. (2017). Survival analysis of US water service lines utilizing a nationwide failure data set. Journal-American Water Works Association , 109 (9), 13–21. NRC. (2006). Drinking Water Distribution Systems: Assessing and Reducing Risks . National Academies Press. https://doi.org/10.17226/11728 Rhyan, C., Miller, G., Betanzo, E., & Hanna-Attisha, M. (2023). Removing Michigan’s Lead Water Service Lines: Economic Savings, Health Benefits, And Improved Health Equity. Health Affairs , 42 (8), 1162–1172. https://doi.org/10.1377/hlthaff.2022.01594 Therneau, T. (2023). A Package for Survival Analysis in R (Version 3.5–5). 2023 . US EPA, O. (2019, October 9). Strategies to Achieve Full Lead Service Line Replacement [Reports and Assessments]. https://www.epa.gov/ground-water-and-drinking-water/strategies-achieve-full-lead-service-line-replacement Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8187953","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Short Report","associatedPublications":[],"authors":[{"id":549521103,"identity":"3192598f-d18a-4663-8433-aaa694e0d22e","order_by":0,"name":"Sarah Wohlfahrt","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Sarah","middleName":"","lastName":"Wohlfahrt","suffix":""},{"id":549521104,"identity":"f4963128-d5ef-4ea7-8eec-c3a2d914668c","order_by":1,"name":"Myles Meehan","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Myles","middleName":"","lastName":"Meehan","suffix":""},{"id":549521105,"identity":"f59184d3-abc6-48be-989c-4512d4b8b68d","order_by":2,"name":"Juneseok Lee","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0002-7942-4273","institution":"","correspondingAuthor":true,"prefix":"","firstName":"Juneseok","middleName":"","lastName":"Lee","suffix":""}],"badges":[],"createdAt":"2025-11-23 23:34:53","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8187953/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8187953/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96716186,"identity":"d46d9afb-6934-4d70-ac4e-448ea16aa097","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4593956,"visible":true,"origin":"","legend":"","description":"","filename":"EvaluationofWaterServiceLineFailuresrevised.FINAL.docx","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/6c8604311816336045136650.docx"},{"id":96913659,"identity":"7d30bf80-acab-4334-a73c-a98eee53995d","added_by":"auto","created_at":"2025-11-27 14:03:32","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342,"visible":true,"origin":"","legend":"","description":"","filename":"rs8187953.json","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/8c869536ef3f19bd0dfaf4ba.json"},{"id":96913734,"identity":"732e07d4-cc58-48e0-83f4-8c56ac9319ce","added_by":"auto","created_at":"2025-11-27 14:04:08","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":52756,"visible":true,"origin":"","legend":"","description":"","filename":"rs81879530enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/18abc5b3e47cd29ae9056766.xml"},{"id":96913328,"identity":"9ac59d0a-c96a-4668-a8af-64df6fb6797b","added_by":"auto","created_at":"2025-11-27 13:58:36","extension":"jpeg","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1157270,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/f1d5c4a3decd49f32eb27781.jpeg"},{"id":96716174,"identity":"f6578369-8d55-4a44-84fb-a7a50959a28b","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":905782,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/cc629a36f14f2bad33067513.png"},{"id":96913257,"identity":"57c69e0c-e641-41f6-8e68-cc583c23dd23","added_by":"auto","created_at":"2025-11-27 13:56:21","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":293944,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/6e3908a04cd338d976f10f29.jpeg"},{"id":96716176,"identity":"e5a35796-a948-477f-8e74-7cc85a7a9bfa","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":120991,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/bd526887fc69d4ef0c0c162c.png"},{"id":96716175,"identity":"b8bd6b0b-149d-4daa-a9d1-2a9f768952c4","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":74541,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/81db9f85d3f8632dfce395ac.jpeg"},{"id":96913039,"identity":"95fe2f91-43d9-4291-98a6-0b3a31ea66f8","added_by":"auto","created_at":"2025-11-27 13:50:55","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":172255,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/2d76f6d6052d65689fae0fcf.png"},{"id":96913024,"identity":"5cf36443-13d1-462a-aa78-7eeab69d70c2","added_by":"auto","created_at":"2025-11-27 13:50:07","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":337720,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/5f4f738d2b6a2f32c1503f51.png"},{"id":96716178,"identity":"d53673f7-f8d0-4a4f-ac21-9e43a4b175d0","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":142765,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/eac36ac5f74e1dde3100fad1.png"},{"id":96716180,"identity":"3eceec8c-2e29-4d53-ab64-1e4f522b1b12","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":45095,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/c8e35a8076bce43b98730640.png"},{"id":96913083,"identity":"af32e403-f048-4610-9177-c6620fa29f55","added_by":"auto","created_at":"2025-11-27 13:51:50","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":30210,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/8b63064fcdbb3a9558957bba.png"},{"id":96716184,"identity":"8a0129e5-9985-47cc-aae9-65986e0bc093","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13459,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/318c9d319a8438480a5a5ebe.png"},{"id":96913753,"identity":"3235046a-62a8-45b9-852e-d03062518034","added_by":"auto","created_at":"2025-11-27 14:04:11","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":41144,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/03a21d06ffccab6cfbe131de.png"},{"id":96716185,"identity":"02d9d20b-292f-4421-a0dd-00a9bf14b5ea","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"xml","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":51604,"visible":true,"origin":"","legend":"","description":"","filename":"rs81879530structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/022ae4b5954e6d0fba507eb0.xml"},{"id":96716182,"identity":"24e6f3d3-d0f7-4383-a8f2-c16dfcdd3c6f","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"html","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":57931,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/883043015758170468b0f157.html"},{"id":96716167,"identity":"56d0ee66-8742-4278-aae2-1ca13bec9382","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":706294,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLocation Map of all Datapoints by ZIP Code\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/647d4644479400d2da8dbda0.png"},{"id":96913464,"identity":"0501a604-7371-4f46-af76-dd6ccd7b2993","added_by":"auto","created_at":"2025-11-27 14:01:58","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":512781,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLead Service Line Replacement Locations\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/d4ae17a4a3ff696d85d2ce31.png"},{"id":96912980,"identity":"5a8cb19a-16c6-45f1-ba0a-08c06009a717","added_by":"auto","created_at":"2025-11-27 13:47:56","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":252946,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eKaplan-Meier Survival Curves\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/8a8735718302935cee5476a2.png"},{"id":96716165,"identity":"37fa6f35-1512-46e9-aa92-a66b8d32a375","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":108665,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eWeibull Survival Function\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/166516651ffc249abdba780e.png"},{"id":96716170,"identity":"5344b42c-ed3d-455e-9b49-d60f29070e01","added_by":"auto","created_at":"2025-11-25 10:31:57","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":37196,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVariable Importance Plot\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/86bd406bf5e68805e9c6a078.png"},{"id":96913290,"identity":"1d7761e6-5f18-4d7c-af2c-97d54cb47609","added_by":"auto","created_at":"2025-11-27 13:57:06","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":169485,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRSF Survival Curves\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/cfe0a3fe0397d06da170b39a.png"},{"id":96922153,"identity":"99f9389f-725e-4d03-b874-e6c17534e95d","added_by":"auto","created_at":"2025-11-27 14:18:16","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2170731,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8187953/v1/cd79cf71-4128-406b-beb2-494f88f0a330.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eNationwide Patterns of Water Service Line Failures: Insights from 20 Years of U.S. Data\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eWater service lines (WSLs) are the essential link between municipal water mains and building plumbing systems, delivering potable water directly to homes, businesses, and public institutions. The material and condition of these lines directly affect water quality and public health, particularly in systems that contain lead or other high-risk materials (Lee et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Lee \u0026amp; Meehan, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Service lines have some of the highest failure rates within the water distribution system and often exhibit the lowest chlorine residuals, making them a critical focus for both infrastructure reliability and water quality protection (AWWA, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; NRC, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2006\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eAccording to the 7th Drinking Water Infrastructure Needs Survey and Assessment (DWINSA), there are an estimated 100\u0026nbsp;million service lines nationwide, spanning materials such as lead, galvanized steel, copper, and plastic (\u003cem\u003eUSEPA\u003c/em\u003e, 2023). Failures in these lines, through cracking, bursting, leaking, or corrosion, pose growing challenges for homeowners, utilities, and municipalities. As infrastructure continues to age, understanding how long different materials last is essential for developing proactive, cost-effective replacement and maintenance strategies.\u003c/p\u003e\u003cp\u003eThis article evaluates 20 years (2006–2025) of nationwide WSL failure and replacement data, sourced through a partnership with HomeServe USA Corp. (HomeServe). The analysis examines pipe age, material type, replacement trends, and spatial distribution using multiple survival analysis techniques, including Kaplan-Meier (KM) survival curves, Weibull models, and Random Survival Forest (RSF) machine learning.\u003c/p\u003e\u003cp\u003eThis study builds upon earlier research published in the \u003cem\u003eJournal of American Water Works Association\u003c/em\u003e by Lee and Meehan (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2017\u003c/span\u003e), which analyzed WSL failures from 2006 to 2015. That work revealed significant variation in material longevity, for instance, copper pipes generally outperformed poly-based materials. In that study, survival analysis was performed, indicating failure rates increasing significantly after 30–60 years of service. The findings were also featured in \u003cem\u003eThe Washington Post\u003c/em\u003e as part of a broader national discussion on aging infrastructure and consumer decisions (Checkbook, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The current study extends that analysis through 2025, offering a comprehensive 20-year view of material performance trends and failure risks across the United States.\u003c/p\u003e\n\u003ch3\u003eData\u003c/h3\u003e\n\u003cp\u003eThe dataset comprises nationwide WSL records from 2006 to 2025, collected through a collaboration with the home services company HomeServe. The original dataset included approximately 3.8\u0026nbsp;million rows of job site information related to water service line repairs and replacements. Key attributes in the dataset included:\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eExisting Material: The type of pipe material previously in use.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eBuilt Date: The year the house was constructed and assumed pipe installation date.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eJob Date: The date associated with the pipe failure.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eJob Zip Code: Zip code of the failure location.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eReplacement Material: The new material used for pipe replacement.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eDiameter: The pipe diameter (inches).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eTotal Cost: The cost of repairing/ replacing the pipe.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe data were cleaned and organized into a workable format using R Studio. To improve accuracy for this analysis, rows with missing values, conflicting entries (e.g., multiple materials or diameters listed), or logical errors (e.g., built dates later than job dates resulting in negative pipe ages) were removed. After cleaning, the final dataset included approximately 76,000 valid records of service line repairs or replacements between 2006 and 2025.\u003c/p\u003e\u003cp\u003eThe national distribution of failed WSLs shows clear regional patterns, in part due to the areas served by HomeServe, with the highest concentrations in the Northeast, Midwest, and parts of California (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Dense clusters are visible around legacy urban centers such as New York City, Chicago, and Philadelphia, reflecting the prevalence of older infrastructure. Additional hotspots in the Southeast and Southwest suggest emerging risks in rapidly developing areas, likely driven by material degradation under environmental stress. In contrast, failures are less frequently reported in the Mountain West and Great Plains, which may reflect both lower population density and limited data coverage.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eA descriptive statistical analysis was conducted to characterize the dataset. The majority of replacement costs ranged from \u003cspan\u003e$\u003c/span\u003e0 to \u003cspan\u003e$\u003c/span\u003e5,000, with the mean around \u003cspan\u003e$\u003c/span\u003e1,000. Pipe diameters spanned from 0.25 to 6 inches, with 0.75 inches being the most common size. The most frequently encountered existing pipe material was copper, followed by galvanized pipe, while the most common replacement material was copper, followed by cross-linked polyethylene (PEX).\u003c/p\u003e\u003cp\u003eOf the 76,000 service line repairs analyzed, 2,711 involved the replacement of lead pipes, representing approximately 3.6% of all documented failures. While this may appear to be a relatively small proportion, the presence of lead carries outsized public health implications, as even intact lead lines pose contamination risks. The concentration of these replacements in older urban areas, such as New York City, Boston, Chicago, and Columbus (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), reflects historical installation practices and highlights regional equity and infrastructure aging challenges. These findings underscore the continued urgency of identifying and replacing lead service lines, not solely based on failure likelihood, but also on regulatory compliance and urgent health risk mitigation.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo further understand the factors driving WSL failures, including material performance over time, this study applied a combination of statistical and machine-learning techniques. The following section outlines the analytical methods used to evaluate failure timing, model survival behavior, and identify the most influential variables across the full dataset of 76,000 service line replacements.\u003c/p\u003e"},{"header":"Methods: Survival Analysis and Machine Learning","content":"\u003cp\u003eThis study applied survival analysis to evaluate the longevity of WSLs, where the event of interest was failure, defined as the need for repair or replacement. Survival analysis models estimate the probability that an asset will continue functioning over time. Because all data points in this study represent actual failures (\u003cem\u003ei.e.\u003c/em\u003e, no censored data), traditional survival models may introduce selection bias. However, these methods still provide valuable insights into relative material performance and failure timing (Lee \u0026amp; Meehan, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe following techniques were used to characterize failure behavior across pipe materials:\u003c/p\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cem\u003eKaplan-Meier (KM) Survival Curves\u003c/em\u003e: A nonparametric method to estimate survival probability over time.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cem\u003eWeibull Models\u003c/em\u003e: Parametric models useful for identifying failure patterns and estimating time-to-failure distributions.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cem\u003eRandom Survival Forests (RSF)\u003c/em\u003e: A machine learning approach for modeling survival without assuming specific functional forms.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003cp\u003eKM and Weibull models were applied to estimate survival over time. KM, a nonparametric method, captures the proportion of assets remaining in service at each time point and is useful for comparing materials with different longevity profiles. Weibull modeling, on the other hand, introduces a parametric framework that quantifies how failure risk evolves with age, offering interpretable parameters and the ability to forecast future performance (Jackson, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Kleinbaum \u0026amp; Klein, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Therneau, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Lastly, machine learning was introduced through the Random Survival Forest (RSF) model, which can capture complex, nonlinear relationships and interactions among predictor variables. Unlike traditional models, RSF does not rely on prior assumptions about data distribution and can handle high-dimensional data effectively (Ishwaran et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). RSF provides several analytical advantages over traditional survival models:\u003c/p\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eNo parametric assumptions: RSF is fully nonparametric, improving flexibility and robustness.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eNonlinear interaction detection: RSF identifies interactions among variables (e.g., pipe material, diameter, installation year) that KM and Weibull models cannot detect.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eVariable importance ranking: RSF quantifies the relative influence of each input variable on survival outcomes, even if those variables are not statistically significant in parametric models.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003cp\u003eThis combination of statistical and machine-learning approaches enables a more comprehensive assessment of pipe material performance and failure risk over time.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eKaplan-Meier Survival Analysis. \u0026nbsp;\u003c/em\u003e\u003c/strong\u003eThe KM survival curves illustrate the probability that a given pipe material will continue functioning over time since installation (Figure 3). Steeper declines and curves that drop closer to the y-axis indicate shorter service life and earlier failures. Materials such as PEX, polybutylene, and blue polyethylene (poly) show faster declines, indicating more rapid degradation and shorter lifespans. In contrast, lead and galvanized steel exhibit higher survival probabilities for longer durations, suggesting greater longevity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eWeibull Survival Function Analysis\u003c/em\u003e\u003c/strong\u003e. The Weibull survival curves provide insight into how quickly different pipe materials deteriorate over time (Figure 4). Steep declines in the curves closer to the x-axis indicate shorter lifespans and earlier failures, while initially flatter curves with delayed declines suggest greater durability and longer service life. \u0026nbsp;The overall plot shows a sharp decline in survival probability within the 100 years, indicating that most service lines fail within this timeframe. The survival curves are closely grouped, suggesting broadly similar failure patterns across materials, though subtle differences are evident.\u003c/p\u003e\n\u003cp\u003eThe failure dynamics of different pipe materials can be described through the Weibull survival model by estimating the shape and scale of deterioration. The Weibull shape parameter (\u003cem\u003e\u0026beta;)\u003c/em\u003e clarifies how each pipe material\u0026rsquo;s failure risk changes with age. Values below 1 indicate a high incidence of early (infant‑mortality) failures that taper off, whereas values above 1 signal an aging‑related hazard that grows over time. In this dataset, PEX (\u003cem\u003e\u0026beta;\u003c/em\u003e = 0.53), brass (0.68), and PVC (0.75) exhibit the strongest early‑failure tendencies; they are more likely to fail soon after installation but become comparatively stable thereafter. Materials such as copper (0.82), polybutylene (0.80), blue poly (0.89), and black poly (0.84) show milder early‑failure behavior that likewise improves with age. By contrast, lead (1.46), galvanized (1.10), and steel (1.08) display \u003cem\u003e\u0026beta;\u003c/em\u003e \u0026gt; 1, meaning their failure risk increases as they age, consistent with corrosion and long‑term degradation processes, while iron (0.99) sits near the exponential boundary, suggesting a roughly constant hazard throughout its life cycle.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe scale parameter (\u003cem\u003e\u0026lambda;\u003c/em\u003e) of the Weibull survival model provides insight into the typical time frame over which failures are expected to occur, effectively reflecting the material\u0026apos;s average service life. Higher \u003cem\u003e\u0026lambda;\u003c/em\u003e values indicate longer-lasting performance, while lower values suggest that failures are likely to occur earlier in the material\u0026apos;s lifespan. In this dataset, lead (\u003cem\u003e\u0026lambda;\u003c/em\u003e = 4.64) and steel (4.26) exhibit the highest scale values, implying these materials generally persist longer before failure, despite aging-related risk increases. Copper (4.07) and iron (4.28) also show relatively long service lives, consistent with their known durability. In contrast, PEX (3.67), polybutylene (3.68), and blue poly (3.65) have the lowest scale values, indicating shorter average lifespans and a higher concentration of early failures. These scale differences help contextualize the failure timing observed across materials and complement the shape parameter (\u003cem\u003e\u0026beta;\u003c/em\u003e) in characterizing both how and when different pipe types are likely to fail (Table 1).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1. Pipe Material Failure Trends Described by Weibull Model Parameters\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"642\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMaterial\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta; (Shape\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eParameter)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026Lambda; (Scale Parameter)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 342px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eExpected Failure Trend\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCopper\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e4.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eModerate early failure risk, but generally long-lasting performance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBrass\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e3.91\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003ePronounced early failures, shorter average service life\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGalvanized\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e1.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e4.35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eAging-related failures increase over time, relatively long service life\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIron\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e4.28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eConstant failure risk, moderate-to-long expected lifespan\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSteel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e1.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e4.26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eMild aging effect, stable performance over moderate duration\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePEX\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e3.67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eHigh infant-mortality rate, short lifespan unless early failures avoided\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePVC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e3.77\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eEarly failures present, but relatively stable afterward\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePolybutylene\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.80\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e3.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eEarly failure trend, shorter average lifespan\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLead\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e1.46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e4.64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eStrong aging effect, long physical life but increasing failure risk\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBlue Poly\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e3.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eEarly failure is common, but risk declines with age\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 108px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBlack Poly\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e3.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 342px;\"\u003e\n \u003cp\u003eSimilar to blue poly, with early failures tapering over time\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eRandom Survival Forest (RSF) Analysis. \u0026nbsp;\u003c/em\u003e\u003c/strong\u003eThe RSF model was used to evaluate the factors influencing WSL longevity. A variable importance analysis was performed to find the parameters with the strongest influence on failure likelihood. With a higher importance value, the analysis revealed that pipe material was the most significant predictor of survival, followed by pipe diameter as a minor predictor (Figure 5). Replacement cost has a negative importance value, and therefore, has no influence on failure likelihood.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe RSF survival curves illustrate the probability of each material type remaining functional over time, with survival probability on the y-axis and years since installation on the x-axis (Figure 6).\u003c/p\u003e\n\u003cp\u003eGalvanized and lead exhibited the slowest decline in survival probability, indicating superior durability and longer service life. Lead, in particular, maintained survival probabilities above 50% beyond 75 years. In contrast, polybutylene, blue poly, and brass showed the steepest declines. Their survival probabilities dropped rapidly within the first 50 to 75 years, suggesting significantly shorter service lives. Black poly and blue poly approached zero survival probability well before the 100-year mark. The middle-ground materials include copper, steel, and iron showing high probabilities of survival (\u0026gt;85%) through the first 30 years and declining to approach zero around 100 years.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eInterestingly, while KM results suggest that copper consistently outperforms PEX, the Weibull model indicates that PEX may demonstrate longer survival under certain installation or environmental conditions. \u003cem\u003eThese differences underscore the value of using multiple modeling approaches\u003c/em\u003e. The Weibull model\u0026rsquo;s ability to describe the changes in risk trends over time allows predicting pipe longevity behavior while considering additional factors to pipe material. Additionally, the RSF model\u0026rsquo;s ability to account for complex, nonlinear interactions make it particularly well-suited for context-specific asset management, rather than relying on uniform material replacement strategies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImplications \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe findings from this 20-year analysis offer critical guidance for infrastructure planning, material selection, maintenance scheduling, and public health policy. Materials with longer survival times, such as copper, should be prioritized for long-term installations, while poly-based materials, which exhibit earlier and more frequent failures, may require more frequent monitoring and earlier replacement.\u003c/p\u003e\n\u003cp\u003eHomeowners should be aware of the material composition of their service lines and the associated risks. Proactive replacement or having measures in place to help manage replacement of high-risk materials can reduce long-term costs, mitigate failure rate, and prevent property damage. However, material selection is rarely based on performance data alone. Factors such as health concerns, taste and odor, corrosion resistance, installer recommendations, insurance requirements, and even marketing play significant roles in decision-making. Prior research has shown that homeowners value health and water quality above cost or installation convenience (Lee et al., 2013).\u003c/p\u003e\n\u003cp\u003eUtilities and engineers should adopt data-driven, predictive maintenance strategies, leveraging survival models and RSF techniques to anticipate failures before they occur. This shift from reactive to proactive asset management enables more strategic resource allocation, minimizes emergency repair costs, and improves service reliability.\u003c/p\u003e\n\u003cp\u003eFinally, the implications extend beyond engineering into public health. While materials like lead may exhibit structural longevity, their continued use poses significant health risks. As of January 2025, an estimated 9.2 million lead service lines remain in use across the U.S. (Keller et al., 2025). In response, the EPA\u0026rsquo;s Lead and Copper Rule Improvements mandate the full replacement of lead pipes by 2035. Although the average replacement cost is approximately $6,000 per line, federal, state, and local programs offer funding support and incentives to assist homeowners and utilities with this critical transition (Keller et al., 2025; Rhyan et al., 2023; US EPA, 2019).\u003c/p\u003e\n\u003cp\u003eLimitations\u003c/p\u003e\n\u003cp\u003eThis study relies on a dataset composed exclusively of failed water service lines. While the absence of censored data (\u003cem\u003ei.e.,\u0026nbsp;\u003c/em\u003eWSL still in service\u003cem\u003e)\u0026nbsp;\u003c/em\u003elimits the application of traditional survival models, the large volume of failure records, 76,000 cases over 20 years, still enables meaningful insights into material performance, temporal patterns, and regional variation in failure behavior.\u003c/p\u003e\n\u003cp\u003eApparent differences in the results from KM, Weibull, and RSF models are not contradictions but reflections of each method\u0026rsquo;s underlying assumptions and analytical perspective. KM and Weibull provide aggregate survival insights, with KM capturing nonparametric survival patterns and Weibull enabling parametric forecasting. In contrast, RSF captures complex, nonlinear interactions among multiple variables, such as geography, pipe diameter, and installation era. These differences underscore the importance of using multiple modeling approaches to capture diverse failure dynamics and highlight the need for utility-specific modeling when planning replacement and maintenance strategies.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study is based on a failure-only dataset, where all observations represent WSLs that were replaced due to failure. While this limits the ability to estimate unbiased survival probabilities, since censored data are not available, it still provides valuable insights for comparative analysis. With more than 76,000 real-world failure records collected over 20 years across a wide range of geographic and climatic conditions, the dataset enables a robust evaluation of material-specific degradation trends, failure timing, and regional risk patterns.\u003c/p\u003e\n\u003cp\u003eSimilar to the Lee \u0026amp; Meehan 2017 study based on data from 2006-2015, this failure analysis found that failure rates increased significantly around 25-50 years after installation. Among all variables considered, pipe material emerged as the most influential factor affecting service line longevity. Materials such as polybutylene, blue poly, and black poly exhibited the highest failure rates, while copper and iron demonstrated significantly greater durability. The combined application of KM, Weibull, and RSF models revealed critical failure thresholds and material-specific risk profiles. Despite being trained on failure-only data, the RSF model proved capable of identifying meaningful predictive relationships that can inform proactive maintenance strategies.\u003c/p\u003e\n\u003cp\u003eBased on the analysis, the following recommendations are proposed to guide utilities and municipalities in asset management and replacement planning:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003ePrioritize the replacement of high-risk materials, particularly polybutylene and polyethylene pipes, with more durable alternatives such as copper.\u003c/li\u003e\n \u003cli\u003eLeverage machine-learning-based and survival model predictions to refine maintenance schedules, preemptively replace assets before failure, and allocate resources more effectively to optimize infrastructure investment and reduce long-term costs.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eIntegrating survival analysis and machine learning into asset management empowers utility managers and engineers to make informed, data-driven decisions. This approach enhances system reliability, improves long-term operational efficiency, and supports the development of safer, more sustainable water infrastructure systems.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003e\u003cem\u003e7th Drinking Water Infrastructure Needs Survey and Assessment\u003c/em\u003e. (n.d.). Retrieved June 30, 2025, from https://www.epa.gov/system/files/documents/2023-04/Final_DWINSA%20Public%20Factsheet%204.4.23.pdf\u003c/li\u003e\n\u003cli\u003eAWWA. (2024). \u003cem\u003eM22 Sizing Water Service Lines and Meters, Fourth Edition\u003c/em\u003e. https://store.awwa.org/M22-Sizing-Water-Service-Lines-and-Meters-Fourth-Edition\u003c/li\u003e\n\u003cli\u003eCheckbook, J. B. and K. B. | W. C. (2021, February 15). Do you really need insurance for your water and sewer lines? Here\u0026rsquo;s what to know. \u003cem\u003eThe Washington Post\u003c/em\u003e. https://www.washingtonpost.com/lifestyle/home/do-you-really-need-insurance-for-your-water-and-sewer-lines-heres-what-to-know/2021/02/15/651f3dd6-6baa-11eb-9f80-3d7646ce1bc0_story.html\u003c/li\u003e\n\u003cli\u003eIshwaran, H., Kogalur, U. B., Blackstone, E. H., \u0026amp; Lauer, M. S. (2008). Random survival forests. \u003cem\u003eThe Annals of Applied Statistics\u003c/em\u003e, \u003cem\u003e2\u003c/em\u003e(3), 841\u0026ndash;860. https://doi.org/10.1214/08-AOAS169\u003c/li\u003e\n\u003cli\u003eJackson, C. (2016). flexsurv: A Platform for Parametric Survival Modeling in R. \u003cem\u003eJournal of Statistical Software\u003c/em\u003e, \u003cem\u003e70\u003c/em\u003e, 1\u0026ndash;33. https://doi.org/10.18637/jss.v070.i08\u003c/li\u003e\n\u003cli\u003eKeller, J., Burchett, R., \u0026amp; Gallet, D. (2025). Strategies for Effective Lead Service Line Replacement Communication. \u003cem\u003eJournal AWWA\u003c/em\u003e, \u003cem\u003e117\u003c/em\u003e(3), 32\u0026ndash;40. https://doi.org/10.1002/awwa.2418\u003c/li\u003e\n\u003cli\u003eKleinbaum, D. G., \u0026amp; Klein, M. (2012). \u003cem\u003eSurvival Analysis: A Self-Learning Text\u003c/em\u003e. Springer New York. https://doi.org/10.1007/978-1-4419-6646-9\u003c/li\u003e\n\u003cli\u003eLee, J., Burkhardt, J., Buchberger, S., Grayman, W., Janke, R., Murray, R., \u0026amp; Platten, W. (2023). \u003cem\u003ePremise Plumbing Modeling\u003c/em\u003e. ASCE.\u003c/li\u003e\n\u003cli\u003eLee, J., Kleczyk, E., Bosch, D. J., Dietrich, A. M., Lohani, V. K., \u0026amp; Loganathan, G. (2013). Homeowners\u0026rsquo; decision-making in a premise plumbing failure\u0026ndash;prone area. \u003cem\u003eJournal-American Water Works Association\u003c/em\u003e, \u003cem\u003e105\u003c/em\u003e(5), E236\u0026ndash;E241.\u003c/li\u003e\n\u003cli\u003eLee, J., \u0026amp; Meehan, M. (2017). Survival analysis of US water service lines utilizing a nationwide failure data set. \u003cem\u003eJournal-American Water Works Association\u003c/em\u003e, \u003cem\u003e109\u003c/em\u003e(9), 13\u0026ndash;21.\u003c/li\u003e\n\u003cli\u003eNRC. (2006). \u003cem\u003eDrinking Water Distribution Systems: Assessing and Reducing Risks\u003c/em\u003e. National Academies Press. https://doi.org/10.17226/11728\u003c/li\u003e\n\u003cli\u003eRhyan, C., Miller, G., Betanzo, E., \u0026amp; Hanna-Attisha, M. (2023). Removing Michigan\u0026rsquo;s Lead Water Service Lines: Economic Savings, Health Benefits, And Improved Health Equity. \u003cem\u003eHealth Affairs\u003c/em\u003e, \u003cem\u003e42\u003c/em\u003e(8), 1162\u0026ndash;1172. https://doi.org/10.1377/hlthaff.2022.01594\u003c/li\u003e\n\u003cli\u003eTherneau, T. (2023). \u003cem\u003eA Package for Survival Analysis in R (Version 3.5\u0026ndash;5). 2023\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eUS EPA, O. (2019, October 9). \u003cem\u003eStrategies to Achieve Full Lead Service Line Replacement\u003c/em\u003e [Reports and Assessments]. https://www.epa.gov/ground-water-and-drinking-water/strategies-achieve-full-lead-service-line-replacement\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8187953/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8187953/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eKey Takeaways\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e- This study analyzes 20 years (2006-2025) of nationwide water service line failure and replacement data, collected through a unique collaboration with a home services company, covering over 76,000 real-world incidents.\u003c/p\u003e\n\u003cp\u003e- Multiple survival analysis methods, including Kaplan-Meier (KM) and Weibull models and Random Survival Forest (RSF) machine learning, were used to assess failure risk across different pipe materials.\u003c/p\u003e\n\u003cp\u003e- Municipalities and homeowners are encouraged to inventory existing service line materials and put measures in place to assist in managing high-risk types (e.g., polybutylene, polyethylene) to reduce failure rates, improve reliability, and lower long-term maintenance costs.\u003c/p\u003e","manuscriptTitle":"Nationwide Patterns of Water Service Line Failures: Insights from 20 Years of U.S. Data","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-25 10:31:52","doi":"10.21203/rs.3.rs-8187953/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"25bf19b8-112c-40be-ab19-e6e6d91334d5","owner":[],"postedDate":"November 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":58466720,"name":"Civil Engineering"},{"id":58466721,"name":"Environmental Engineering"}],"tags":[],"updatedAt":"2025-11-25T10:31:53+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-25 10:31:52","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8187953","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8187953","identity":"rs-8187953","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}