Structures · BVK-PRO

Citicorp Center: the skyscraper put at risk by a single changed detail

In 1978, welders worked night after night on a finished, fully occupied 59-storey skyscraper in New York. By day, office life went on as usual; by night, the connections of the load-bearing structure were being strengthened. The reason: the building’s own structural engineer had discovered that, for a certain wind direction, his building was more vulnerable than he had thought. The Citicorp Center story is now taught at engineering schools as a case study in professional ethics. For us as structural engineers, however, what matters most is where the error came from – not from the calculation, but from a change to a detail that nobody re-checked.

The structural system was decided by a church, not by the engineer

St. Peter’s Lutheran Church stood on the corner of the site. The congregation agreed to the sale only if it received a new building on the same spot, structurally separate from the tower, and if at least 63 % of the church had nothing built above it. The columns therefore could not stand at the corners, where an ordinary high-rise has them.

The solution: four massive columns were moved to the middle of each side, and the corners of the building became cantilevers reaching 22 m from the core, with a 34-metre-high open space beneath. From this single condition followed a whole chain of structural consequences.

The wind has only one path

Both the vertical load from the cantilevered corners and the horizontal wind load must reach the four columns. This is done by a system of diagonals in the facade: each wall has six chevrons (inverted Vs), each eight storeys high. The diagonals collect the load from the corners and lead it to the middle of the wall. No shear walls at the corners, no back-up load path. The diagonals and their connections are therefore critical for the entire building.

Quartering wind

LeMessurier originally calculated wind only perpendicular to the walls – exactly as the New York building code of the time required. For typical buildings, wind acting diagonally on a corner is less unfavourable, so this was a standard and legal approach.

In June 1978, following a question from Princeton engineering student Diane Hartley, he recalculated the building for quartering wind as well. The result: such wind significantly increases the load in the diagonal connections. With wind perpendicular to a wall, it is resisted mainly by the two walls parallel to the wind; with wind on the corner, all four walls are loaded at once and some diagonals receive more than in the perpendicular case.

Note: it is often said that for quartering wind the exposed width is √2 times larger, i.e. 41 % more. Geometrically this is true, but the pressure coefficient changes with the angle, so a 41 % higher load does not automatically follow. The sources give no specific percentage increase in forces, so neither do we.

A connection changed without a re-check

This is the core of the case. The original design of the diagonals specified welded connections. In 1974 the steel supplier, Bethlehem Steel, changed them to bolted ones to save money. The change was accepted by LeMessurier’s office, but the engineer of record himself did not know about it.

Under the loads considered at the time, the bolted connections were adequate – which is why the change went through as routine. They stopped being adequate only when the load assumption changed. That was four years later, with people already working in the building. Each of the two things – wind only perpendicular to the wall, and bolts instead of welds – would have been fine on its own. Together they were not.

The damper: part of the design and part of the risk

The building is light and slender, so the original design already placed a tuned mass damper (TMD) on the roof: a 9.1 × 9.1 × 1.8 m concrete block weighing 360 t, movably supported. When the building sways, the block moves with a lag against it and takes energy out of the motion. According to the source, it reduces wind-induced movement by up to 50 % and cost about USD 1.5 million.

The catch: the damper needs electricity. LeMessurier therefore had to consider two states – with the damper running and without it. And strong winds and power cuts are events that often come together.

The risk in numbers

LeMessurier quantified the probability of a wind capable of toppling the building. The source gives annual values. The last column is our conversion to 50 years – and it shows why the repair had to happen at once:

“1 in 55 per year” sounds like something that will not happen. Over fifty years, it is more likely than not.

The repair: at night, in an occupied building

The structural fix was simple: a 51 mm (2-inch) steel plate was welded over each affected bolted connection, bringing the joint back to what had originally been designed. The welding was carried out by Karl Koch Erecting at night from August 1978; the work was completed in October 1978.

The public learned the true extent of the problem only in 1995, from the article “The Fifty-Nine-Story Crisis” in The New Yorker. Accounts differ on who exactly asked the question that started it all – besides Diane Hartley, another student, Lee deCarolis, is also mentioned.

What we take from Citicorp Center into practice

• Load direction is an assumption, not a given. For asymmetric or diagonally braced systems, oblique directions must be checked too. The question is not whether we calculated wind, but whether we calculated the most unfavourable direction for this particular system.
• A change of detail during fabrication is a structural decision, not paperwork. Welds instead of bolts, a different section, a different anchor – the engineer responsible for the structure must know about every change.
• When a load assumption changes, re-check what has already been approved. A detail that worked under the old assumption is not automatically fine under the new one.
• An active measure is not safety. Damper, pump, control system – consider the state in which they do not work, and report the result for both states.
• Annual probability is misleading. When we talk about risk, convert it to the period for which the structure will stand.
• A non-structural condition at the start has structural consequences at the end. An existing building, a site boundary, an open ground floor – write down straight away which decisions follow from them, so they are not forgotten four years later.
• Anyone can ask the question. “It has already been approved” is not an answer. In this case a student’s question led to the re-check that saved the building.

What we verified – and what we did not

All figures on the building, chevrons, damper, probabilities and repair come from the sources listed. Only the diagrams and the 50-year probability conversion are ours. The sources give no specific forces in the diagonals, no number or grade of bolts and no number of strengthened connections – so we do not give them either.

Sources

• Wikipedia EN – Citigroup Center
• Wikipedia EN – William LeMessurier
• Structurae – Citigroup Center
• J. Morgenstern: The Fifty-Nine-Story Crisis, The New Yorker, 29 May 1995 (cited via Wikipedia)
• Photographs: Wikimedia Commons (authors and licences with each image). Diagrams: BVK-PRO. Data retrieved 22 Sep 2026.

Is your structure going through design or fabrication changes? At BVK-PRO we check structures from concept to fabrication documentation – including the changes that come from site. Get in touch.

Ing. Csaba Baji
Authorized Structural Engineer, Reinforced Concrete Structures Specialist
BVK-PRO, s.r.o.