Why a small change had tragic consequences

A change intended to simplify assembly led to the collapse of the Hyatt Regency atrium walkways. The change unintentionally doubled the force at a local connection. The case shows why design changes must be assessed through the entire load path rather than only for assembly convenience.

On 17 July 1981, the second- and fourth-floor walkways in the atrium of the Hyatt Regency Hotel in Kansas City, Missouri, collapsed. 114 people died and more than 200 were injured. The National Bureau of Standards, now NIST, identified inadequate capacity of the hanger-rod-to-box-beam connection as the most probable cause.

The decisive change in the load path

The original design used a continuous hanger rod from the roof through the fourth-floor box beam to the second-floor box beam. During shop detailing, the rod was split. The lower walkway was then suspended from the upper walkway’s box beam, with the two rod axes offset by about 100 mm.

This fundamentally changed the local structural action. The fourth-floor connection now had to receive the load from both walkways and transfer it over the eccentric distance between the rods. The local force was approximately doubled and an additional moment increased bending and shear in the built-up box beam.

What actually failed?

The hanger rods were not the primary failed members. The critical component was the box beam formed from two channels welded together along their flange edges. The concentrated rod force was introduced by a nut and washer into the thin built-up section.

Under the high local force, the channel flanges deformed and the box opened. The washer and nut lost their bearing support and pulled through the distorted section. NIST found no unusual material or weld properties that played a significant role in initiating the collapse. The entire connection detail was inadequately designed.

Almost no reserve remained

NIST determined a mean capacity of about 83 kN for the upper walkway connections, while the highest estimated connection force at the time of collapse was about 95 kN. Dead load and comparatively modest occupancy were already sufficient to exceed capacity.

The original design was also inadequate

The frequently repeated claim that the original continuous-rod arrangement was safe is incomplete. Its connection load would have remained around 90 kN, but the mean capacity of the original detail was only about 91 kN. The applicable requirements called for a minimum capacity of approximately 151 kN.

The original detail would probably have carried the load present at the time, but it did not satisfy the required safety concept either.

Professional consequences

The responsible licensing authority revoked the engineering licences of two engineers for gross negligence and professional misconduct. The case demonstrates that review and approval of changed connection details cannot disappear in an unclear division of responsibility between design, detailing and fabrication.

What the case teaches today

A comparable connection would now be designed under the applicable steel connection rules, including DIN EN 1993-1-8 and its National Annex in a German context. Execution and documented implementation fall within DIN EN 1090-2. This is a present-day classification, not a retroactive application to the historic US case.

Changing a load path requires more than changing a drawing: the changed structural action must be recalculated and formally approved.

Back to bolting technology