the pole that whips once

Problem statement

High-mast lighting towers (HMLTs) — the 100-foot-plus steel poles that light interchanges and stand in medians beside live traffic — are known to occasionally enter first-mode harmonic resonance so violent that the tip swings more than five feet, and owners in Alaska, Kansas, Nebraska, South Dakota, Utah, Wisconsin, and Wyoming have reported collapse and severe fatigue cracking, some within a few years of installation. The problem statement's authors say limited field time-history data suggests a single such event can expend the entire fatigue life of a pole. The unsolved problem is not the physics of fatigue but the observation: the wind/structure condition that triggers these large-amplitude events is so rare and unpredictable that a national 24-month monitoring campaign (NCHRP Report 718, 11 towers at 8 sites) found only two possible instances — neither at the magnitude captured on video — and a dedicated Wyoming DOT study begun in 2017 with four instrumented towers had captured just three events at two sites four years later. Design specifications and mitigation strategies cannot be written for a loading condition nobody has adequately measured.

Why this matters

HMLTs stand directly adjacent to travel lanes, so a fatigue failure is a potential mass-casualty event, and passing motorists have filmed towers whipping in medians between traffic. Removal and replacement of a single tower runs about $80,000, and the current AASHTO specifications for structural supports of signs, luminaires, and traffic signals were calibrated on fatigue loading data (NCHRP 718) that essentially did not contain the large-amplitude events that are now known to occur. The FHWA evaluator's objection to the proposal captures the bottleneck exactly: "In four years, that project has only observed three events, which is likely not enough data on which to make conclusions and establish design criteria and mitigation strategies" — and asked how a 36-month project could gather more.

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

NCHRP Report 718 instrumented 11 towers at eight locations for 24 months to develop fatigue design loading; when the March 2011 Watertown, South Dakota video surfaced near the end of that study, researchers re-examined their triggered time histories and found two candidate events, neither with calculated tip displacements approaching the video's. Wyoming DOT then funded a four-tower monitoring study in 2017 specifically to capture the phenomenon; by the time of the FY2023 problem statement it had recorded three events, the largest with a peak stress range of 78 ksi and a calculated tip displacement range of 216 inches — enough to confirm the severity but not to characterize the loading. A 2024 University of Kansas study using wireless triaxial accelerometers on one 30.47 m pole in Wakeeney, Kansas for three months (1,239 datasets at 50 Hz) found vortex-induced vibration frequent but low-amplitude at 4.47–7.60 m/s in the second mode, and buffeting during a December 2021 derecho as the largest response (about 0.24 m tip displacement); it also reported a battery-depletion gap when repeated wind triggers drained the sensors during cloudy conditions — an illustration of why long-duration, many-site monitoring is hard. The pattern across all three efforts is the same: dense instrumentation on a few towers for a limited time is the wrong sampling strategy for a rare event distributed across thousands of towers, and the proposed $650,000 fix was simply more of the same. The problem statement's own reviewers judged the benefit "may be limited due to the rare occurrence" — which is precisely the constraint.

What would unlock progress

The reframing is from high-fidelity monitoring of a few structures to low-fidelity, always-on detection across the whole inventory: a very cheap, solar/battery-powered event trigger (a threshold accelerometer or a tip-motion detector) on hundreds of towers that records only when large-amplitude motion begins, paired with wind data, would capture in one season what a four-tower study cannot in four years. A complementary route is opportunistic sensing — the events are already being filmed by motorists and traffic cameras, so computer-vision detection of pole motion in existing DOT camera feeds or crowdsourced dashcam video could build an event catalogue with time stamps that can be matched to weather records. Adjacent fields have solved the rare-event capture problem this way: seismology's dense low-cost sensor networks and bridge-cable galloping detection both trade per-sensor precision for coverage.

Entry points for student teams

A team could design and bench-test a sub-$100 event-triggered vibration logger with an LTE/LoRa alert, validate its trigger thresholds against the published Kansas and Wyoming response spectra on a scaled shake rig, and propose a deployment plan for a state's HMLT inventory. A computer-vision team could build a pole-motion detector for fixed traffic-camera video (many DOT feeds show HMLTs in frame) and evaluate false-alarm rates in wind, rain, and camera shake. A structural/aerodynamics team could use the recorded events' wind conditions to test whether buffeting or vortex shedding is the plausible driver and what damper retrofit would suppress first-mode response — the Kansas authors recommend structural damping over aerodynamic fixes. Relevant skills: embedded sensing, structural dynamics, computer vision, and DOT asset-management practice.