bridges with

no bottom

Problem statement

Scour — the erosion of streambed around piers and abutments during floods — is the leading cause of bridge failure in the United States, and evaluating a bridge's scour vulnerability requires knowing how deep its foundation goes. For a large population of older bridges the plans are lost, were never drawn, or omit the as-built pile lengths, so the foundation is "unknown": FHWA defines these as bridges "having insufficient details such that they cannot be evaluated for vulnerability to scour." As of December 2012 the National Bridge Inventory identified 36,076 bridges over waterways with unknown foundations — and the 2025 NBI release still carries 24,250 highway bridges coded "U" in Item 113 (verifier's count from FHWA's 2025 all-states delimited file, 2026-08-18) — and FHWA notes the count for bridges over land "is not known because this qualifier is not a reportable item." State and local bridge owners — the practitioners who must code these structures in NBI Item 113 — are told that non-destructive methods "can" determine the needed characteristics but that "currently available technology has limitations and the cost in some circumstances may prove prohibitive to many bridge owners."

Why this matters

Bridges that remain coded "U" cannot be moved into the regular scour program; since November 2010 each requires a Plan of Action — typically flood monitoring, inspection triggers or closure criteria — "until properly designed countermeasures are installed to protect the bridge foundations or until the bridge is replaced." That converts a knowledge gap into a permanent operational burden for thousands of small owners, and it leaves genuinely vulnerable structures indistinguishable from safe ones. FHWA's stated goal is explicitly "to reduce, not eliminate the inventory of bridges coded U," conceding that "there will be an inventory of bridges that remain coded U because sufficient information could not be obtained." The same characterization problem blocks a second professional need: reusing existing foundations when a bridge is replaced, where "the main issues … are their condition assessment, their load-carrying capacity, their remaining service life."

What’s been tried and why it hasn’t worked

FHWA's risk-based procedure (2009) sorts unknown-foundation bridges into risk categories and allows three routes: positive discovery (field determination of embedment depth), inference (assuming foundation type and depth from era, construction practice, neighboring bridges and pile-driving records), or an accepted-risk Plan of Action. Positive discovery relies on geophysical NDT methods — parallel seismic (which needs a borehole drilled alongside the foundation), sonic/ultrasonic echo, induction-field, ground-penetrating radar, resistivity — catalogued since Geotechnical Guideline No. 16 (1998) and rehearsed again at FHWA's 2013–2014 foundation-characterization workshops. They work in favorable conditions but are limited by access (piers in water, pile caps under fill), by ambiguous signals in stiff soils or with battered or spliced piles, and by cost per bridge that is hard to justify for a low-traffic rural structure — hence FHWA's own caveat about prohibitive cost. Inference is cheap but carries exactly the uncertainty the scour evaluation was meant to remove, and owners differ on how much inference to accept; the workshops found that "States have their own individual procedures; there is no ready means of assessing the present practice." The research agenda FHWA set — better field-evaluation guidelines, load testing of existing foundations, and "smart pile" instrumentation for new ones — addresses future bridges more than the legacy inventory.

What would unlock progress

Progress needs a low-cost, access-tolerant way to estimate embedment depth from the accessible parts of a pier — for example, dispersive-wave or vibration-based methods interpreted with modern signal processing, or fusion of several cheap partial measurements with a Bayesian prior built from era-specific construction practice — packaged so that a county engineer can apply it without a specialist crew. Adjacent fields with transferable methods: pile-integrity testing in new construction, borehole-free seismic characterization in geotechnics, and utility-locating practice that combines weak signals from multiple sensors.

Entry points for student teams

A civil/geotechnical team could build a scaled test bed (driven model piles of known lengths in a soil box or field plot) and quantify how accurately a surface-only method — impact-echo or flexural-wave dispersion with modern spectral analysis — recovers embedment depth, then test on an accessible instrumented pier. A data team could build the inference side: a model that predicts pile type and depth from bridge age, owner, region, and structure type using state records where plans do exist, and validate it against known bridges to give owners a defensible prior. A systems team could analyze one state's inventory of U-coded bridges to show where positive discovery is worth its cost. Skills: structural/geotechnical engineering, signal processing, statistics, GIS.