{"resourceId":"aalto-ml-sewer-failure-tracer-infiltration-jema-2026","versions":[{"version":"external-9553543e4d00a50e96b75709344175a8067cf7eea82189a9b558f45c26062f6f","resource":{"id":"aalto-ml-sewer-failure-tracer-infiltration-jema-2026","title":"Integrating ML failure probabilities and tracer-based infiltration for sewer asset-management screening","organization":"Aalto University / Journal of Environmental Management","sector":"Wastewater / sewer asset management research","geography":"Finland (academic case); methods transferable with local validation","publishedAt":"2026-11-01","publicationDate":"2026-11-01","eventDate":null,"sourceName":"Aalto research portal (Journal of Environmental Management)","sourceLabel":"Peer-reviewed journal article abstract and record (Vol. 419, article 131130)","sourceUrl":"https://research.aalto.fi/en/publications/integrating-machine-learning-failure-probabilities-and-tracer-bas/","evidenceClass":"academic-research","outcomeClass":"mixed","topics":["infrastructure","operating-model","data-security"],"finding":"A 1 Nov 2026 Journal of Environmental Management article (Vol. 419, 131130) combines IPCW-weighted ML failure probabilities from CCTV with independent dry-weather tracer infiltration estimates (EC and COD) across 21 operational zones. Zone rankings from ML structural failure probability and tracer infiltration severity showed positive but statistically uncertain agreement (Kendall tau-b 0.254-0.305; p 0.108-0.056; concordance index 0.628-0.655). Authors present dual-evidence screening as practical when CCTV coverage is incomplete, not as a validated pipe-level predictor. Full PDF returned 403 during this run; findings are from the opened abstract/record.","sledRelevance":"Methods lesson for U.S. municipal wastewater AI condition models: require independent corroboration and report uncertainty—do not treat ML rankings alone as capital truth.","evidence":"Opened Aalto portal record with abstract, journal, publication date 1 Nov 2026, and MoE A1 refereed type. DOI 10.1016/j.jenvman.2026.131130; PDF fetch blocked (403).","architectureImplications":"Aggregate ML outputs to zones when pipe-level labels are sparse; fuse with tracer or flow indicators; retain human capital prioritisation.","governanceImplications":"Document uncertainty metrics when ML informs rehab budgets; avoid single-model capital decisions.","securityPrivacyImplications":"Tracer and CCTV datasets are operationally sensitive; control sharing of network condition models.","caveats":"Abstract-level inspection; PDF 403. Finnish network; uncertain statistical agreement means mixed. Not a product evaluation.","streamIds":["utility-management"],"roles":{"sales":"Sales framing: this peer-reviewed mixed result is useful when buyers want AI sewer scoring—lead with the need for independent corroboration and honest uncertainty, not perfect accuracy claims. Position discovery around CCTV coverage gaps and whether tracer or flow evidence already exists. Do not sell guaranteed Kendall agreement or pipe-level failure prediction from this abstract alone.","engineering":"Engineering framing: fit is zone-level screening that fuses machine-learning failure probabilities from CCTV with independent tracer infiltration indicators. Prerequisites include partial CCTV defect labels plus a dry-weather sampling plan for EC/COD or similar tracers. Proof of value is recomputed local concordance and documented decision usefulness despite imperfect tau—not a claim of statistical certainty or pipe-level prediction.","delivery":"Delivery framing: pilot dual rankings on a subset of zones, present uncertainty bands to asset managers in writing, and only then consider CIP list changes. Ownership stays with wastewater asset management throughout. Risks include over-trusting weak concordance, sampling bias in tracers, and black-box vendor scores substituted for the transparent dual-evidence method described in the abstract."},"roleDetails":{"sales":{"customerProblem":"Incomplete CCTV coverage forces weak rehab prioritisation.","stakeholders":["Wastewater asset manager","City engineer","Environmental compliance","Finance/CIP"],"discoveryQuestions":["What fraction of sewers have recent PACP-coded CCTV?","Do you collect dry-weather EC/COD or similar tracers?","How are ML scores audited before CIP inclusion?"],"valueHypothesis":"Dual-evidence screening can improve prioritisation when CCTV is incomplete if uncertainty is managed.","potentialEngagement":"Methods review and zone-pilot design with uncertainty dashboard.","unsupportedClaims":["Pipe-level leak prediction","Statistically certain ranking agreement","Universal transfer without local data"]},"engineering":{"fit":"Research method for zone screening—adapt, do not copy weights blindly.","architectureIntegration":"CCTV ML scores plus tracer mixing estimates aggregated to zones; GIS ranking views.","prerequisites":["CCTV defect database","Dry-weather sampling plan","Zone topology"],"constraints":["Uncertain tau/p-values","PDF methods not fully retrieved","Local hydrogeology differs"],"security":"Protect inspection and network schematics used in models.","proofOfValue":["Recomputed concordance on local zones","Engineer acceptance of dual ranks","No sole-ML CIP changes"]},"delivery":{"work":"Data prep, model rerun or vendor score ingest, tracer support, uncertainty briefing.","dependencies":["Lab capacity for tracers","CCTV contractor exports","CIP calendar"],"ownership":"Municipal wastewater asset management owns rankings used in budgets.","skillsAdoption":"Train planners to read concordance/uncertainty, not only red/green scores.","governanceCheckpoints":["Methods/data-use review","Pilot zone selection","CIP committee uncertainty review"],"acceptanceCriteria":["Dual ranks for pilot zones","Uncertainty stated in memo","Human sign-off before budget use"],"risks":["Over-trusting weak concordance","Sampling bias","Vendor black-box scores"]}},"retrievedAt":"2026-11-06T22:04:17.905500-06:00","enrichedAt":"2026-11-06T22:13:30-06:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Aims to reduce exclusive dependence on scarce CCTV coding labour by adding tracer evidence.","procurementImplications":"When buying ML sewer tools, require cross-evidence validation plans and published uncertainty metrics.","operatingModelImplications":"Rehab screening becomes dual-rank (ML + tracer) with explicit uncertainty communication to boards.","peerReview":"peer-reviewed","sourceVerification":{"openedUrl":"https://research.aalto.fi/en/publications/integrating-machine-learning-failure-probabilities-and-tracer-bas/","referenceExcerpt":"The agreement between the two rankings was positive but statistically uncertain","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}