The Surfside collapse started three weeks early. Sensors that could have flagged it now exist commercially.

Five years ago today, Champlain Towers South killed 98 people when it collapsed at 1:22 AM in Surfside, Florida. On June 22, NIST released its technical findings on why — and the most significant detail isn't the structural failure itself. It's when the failure began.

According to NIST, two slab-column connections beneath the building's pool deck experienced punching-shear failures in early June 2021 — roughly three weeks before the building fell. Those connections cracked. Loads redistributed to adjacent columns. Those columns, already weakened by four decades of corrosion and carrying more weight than the original design intended, couldn't hold. The collapse on June 24 was the end of a three-week progressive failure that nobody detected.

What NIST says caused it

NIST identified four compounding factors, each insufficient on its own but collectively fatal:

• Design deficiencies from original construction. The structural design did not meet the building codes in force when the building was built — the margins against failure were too narrow from the start.
• Deviations during construction. Actual field construction deviated from the design drawings.
• Added loads over 40 years. Pool deck modifications — heavy planters, sand fill, pavers — added weight the structure was never designed to carry.
• Long-term corrosion. Coastal exposure degraded the reinforced concrete over decades, reducing the capacity of the connections that eventually failed.

NIST's final report, expected by end of 2026, will include recommendations for changes to building codes, standards, and industry practice. Engineers and code officials are watching closely.

Why the three-week window matters

Punching shear failure at a slab-column connection doesn't happen in an instant. The progression NIST describes — cracking at the connection, load redistribution across the deck, widening cracks across the pool slab — produces physically measurable signatures over time:

Slab deflection. As a column connection begins to fail, the slab starts to settle, even slightly. Precision tilt sensors and settlement monitoring points can detect millimeter-scale changes in slab position over days or weeks.

Crack propagation. Widening cracks in reinforced concrete produce acoustic emissions — stress waves that piezoelectric sensors attached to the structure can pick up and log continuously. Machine learning models trained on acoustic emission patterns can distinguish normal concrete behavior from crack growth.

Corrosion rate. Electrochemical sensors can track the rate of active corrosion in reinforced concrete, flagging accelerating degradation in sections where the protective concrete cover has been compromised.

None of this is speculative. These are the methods used to monitor bridges, dams, tunnels, and transit infrastructure where failure consequences are severe. The gap at Surfside wasn't that the technology didn't exist. It's that mid-rise residential and commercial buildings typically aren't instrumented this way, even when deterioration is visible and documented.

The building's management knew about concrete deterioration and was planning remediation. They did not have sensors that would have flagged the specific failure mode that actually initiated the collapse. Three weeks of progressive structural failure went undetected because there was nothing installed to detect it.

Where this lands for GCs

This isn't an argument for instrumenting every building you touch. Most scopes don't justify it, and the liability isn't always yours to carry. But there are specific situations where a GC doing work on an existing concrete structure has clear exposure:

You're modifying the load path. Rooftop mechanical additions, pool deck modifications, new equipment pads — any scope that changes dead or live load on an aging concrete structure is doing exactly what Surfside's pool deck modifications did over 40 years. Machine learning-based anomaly detection on continuous sensor streams — the same class of capability now entering prefab and fabrication operations — is increasingly deployable for in-place structural monitoring during and after renovation work as sensor costs fall.

You're working on a building with documented concrete deterioration. If the conditions report shows spalling, exposed rebar, or active corrosion, establishing a baseline with settlement or tilt monitoring before your scope starts — and maintaining it through completion — creates a defensible record. It's documentation that shows the structure was behaving normally when you left it.

Special inspection requirements are likely to expand. NIST's final recommendations will almost certainly drive code updates requiring more rigorous assessment of aging concrete structures before any structural alteration. Knowing what monitoring can and can't tell you now — before those requirements show up in a spec section — puts you ahead of the RFI.

What sensors can't do

Installing monitoring doesn't fix a failing structure — it tells you it's failing sooner. Sensor systems are detection tools, not remediation. They require installation before the degradation they're designed to catch; you can't put in sensors retroactively and get historical data on what happened.

The value of the three-week window NIST identified is not that sensors would have saved the building. It's that three weeks of warning is enough time to evacuate an occupied building before 98 people die in the middle of the night.

NIST's code recommendations will define what the industry is required to do on aging concrete starting in 2027. The GCs who understand the monitoring toolkit — what it detects, what it costs to install, and what it can actually prove in a claims context — will be better positioned to scope the work, price the risk, and answer engineer and owner questions when the requirement shows up in the next set of contract documents.

Friday one chart. Every week, one piece of data that should change a decision on your project. Subscribe at constructionaibrief.com.