Structural health monitoring: sensors that watch bridges over time

A bridge cannot tell you it is healthy, so engineers fit it with sensors and study the readings over time. This is structural health monitoring (SHM). Typical sensors measure vibration (accelerometers), tilt (tiltmeters), whether a known crack is opening (crackmeters) and movement (displacement sensors). Some systems also log the weather. Readings go to a data system, and engineers compare them with what the structure normally does. It matters because inspections happen only every so often, while sensors can show a change between visits or after an event such as an earthquake or a ship strike.

This article separates three kinds of evidence: what is routine, what has been shown in field work, and what is still a plan. Most of the detail concerns bridges, because that is where the sources we checked were strongest.

Routine practice

The routine rule we could source is inspection, not monitoring. The US Federal Highway Administration (FHWA) says it requires a visual inspection of bridges every two years. We did not find a source showing that permanent sensors are a routine requirement on bridges. Owners fit them where they judge the risk high enough. The owner of the West Virginia bridge described below says traditional inspections are insufficient for proactive maintenance. That is the owner's view, not an independent finding.

Shown in field trials

A cable-stayed bridge in West Virginia. The West Virginia Department of Transportation installed a 57-sensor system on the East Huntington/Gunner Gatski Bridge, a cable-stayed bridge over the Ohio River. Installation began on 18 November 2024 and was substantially complete on 22 November. The system included 39 accelerometers (20 on the cables, 19 on the deck and towers), six tiltmeters, four crackmeters, six displacement sensors, a weather station and a vessel collision detection system. The owner's report, submitted in June 2025, makes these claims, all of them the owner's own:

• Cable frequencies: 19 of 20 measured cable frequencies were within 2% of the 2018 and 2022 values. The earlier values came from manually exciting the cables, while the 2025 values came from ambient excitation such as wind and traffic, so the comparison is a consistency check rather than an accuracy test. One cable, W18, fell from 1.54 Hz in 2022 to 1.47 Hz in 2025, which the report says indicates a change in one cable parameter. Inspection reports from 2018 and 2022 had already noted distortion along this cable, and the report suggests it may have deteriorated further since 2022.
• Alerts: in a test where one sensor's alert threshold was manually lowered, emails arrived in an average of five minutes and at most seven, inside a ten-minute target.
• Earthquake: the pylon accelerometer recorded a magnitude 3.3 earthquake on 16 December 2024. The report says the pylon's dynamic behaviour did not change, and it infers the pylon likely did not suffer significant damage. That is an inference from data, not an inspection finding.
• Dampers: the report says vortex-induced vibration on several stay cables probably led to early failure of the cable dampers, many of which were replaced in 2021 and showed problems by 2023. The owner puts each replacement at about $20,000 and says repeated like-for-like replacement can now be reconsidered in favour of a damper type better suited to this vibration.
• Cost: an example calculation suggests that delaying a hypothetical $500 million replacement by ten years is worth about $85 million in present-day terms, after 3% discounting and $5 million a year in maintenance. It is an illustration, not a measured saving.

Raw data for 1 January to 28 February 2025 was published in the US Department of Transportation's ROSA P repository, which the owner says the SMART grant reporting requires.

An earlier bridge foundation study. FHWA's 2014 report, FHWA-HRT-09-040, covers monitoring of the foundations of the replacement I-35W bridge in Minneapolis, built after the original bridge collapsed in August 2007. One chapter says eight weighed, fully loaded dump trucks, about 400 kips (roughly 181 tonnes) in total, were driven across the bridge four days before it opened on 18 September 2008. When temporary data boxes were removed to build a public viewing platform, nearly one month of data was lost. Six months after opening, the chapter says, no live-load, daily or seasonal data were available beyond several days of monitoring that included the truck tests.

Wireless sensors. A 2017 FHWA summary reports that in one field test, wireless strain and displacement gauges gave accuracy comparable to wired sensors, with significantly shorter installation times and lower costs. It gives no figures for accuracy, time or cost. That test was on a multi-span bridge where motorists had reported high levels of vibration. A second test, on a multi-span steel bridge over the Delaware River, used wireless accelerometers to measure vibration properties and calibrate a computer model. A separate project in the same programme is developing sensors that need no batteries and harvest energy from the structure itself.

Aircraft. A UH-60L Black Hawk helicopter has flown with piezoelectric SMART Layer sensors since 2009. A 2015 conference paper reports on 72 months of sensor data from the aircraft. The abstract does not say whether the sensors survived, and the full text is for subscribers only, so this is long exposure data rather than a verified result.

Plans and proposals

FHWA's Long-Term Bridge Performance Program was launched in April 2008 as a planned 20-year effort. It aims to build a database of quantitative information from a representative sample of US bridges, combining nondestructive testing with visual inspection and sensors that track traffic loads, fatigue and corrosion cracks, and environmental conditions. The page we checked is an archived page last modified in 2016, and it describes planned work, not results.

The 2017 FHWA summary also describes portable wireless sensors meant to be installed as needed after floods, crashes or similar incidents. We did not find field results for that rapid-response use, so it remains a proposal.

Limits and open problems

• Sensor life and power. The owner says the battery-powered wireless sensors last up to eight years, depending on how often they collect data. The report also says battery limits stop the sensors capturing long events, so the extent of the cable vibration problem needs more work.
• Placement. The owner's report says that, because of budget, sensors sit only at chosen points, leaving blind spots where damage could go unseen.
• False alarms. The report warns that anomalies in the data do not always mean damage, and that without corroborating evidence the system risks false positives and false negatives.
• Damage or environment? The report says signal anomalies can come from harmless sources or environmental factors, and that limited pre-damage baseline data makes change harder to judge.
• Cost against inspection. The owner's savings figure rests on assumptions. The sources we checked give no verified comparison of monitoring costs against inspection costs.
• Data and ownership. The system vendor holds the data on Microsoft's Azure cloud platform. Only two months of raw data is public so far.

What to watch next

• Stage 2 of the West Virginia project, which the owner plans to extend to six more Ohio River bridges.
• Whether public raw data is extended beyond two months, so outside researchers can test the claims.
• Evidence on sensor life beyond the first years, against the eight-year figure.
• Full results from the six-year helicopter study.
• Results from FHWA's long-term bridge programme; the archived page we checked describes aims only.

For how close bridges are to giving reliable early warning, see our tracker on bridge sensors that flag possible damage.