The Collapse of the Kinzua Viaduct

Thomas Leech. American Scientist. Volume 93, Issue 4. Jul/Aug 2005.

With rapid growth in the extraction of mineral resources during the late 19th century, the Commonwealth of Pennsylvania saw the development of a complex railroad-transportation network, one that spanned the many gullies, valleys and gorges of the western and north-central parts of the state. Fossil fuels, especially coal, were needed in Great Lakes cities, and rail was the transportation of choice.

General Thomas L. Kane, owner of the New York, Lake Erie and Western Railroad and Coal Company, wanted to know if a bridge could be built higher than any yet known. The target was Kinzua Creek gorge, which lies within a high plateau of north-central Pennsylvania 17 miles south of the New York border and about two miles from the eastern continental divide, which separates waters draining east to Chesapeake Bay and west to the Gulf of Mexico. The name Kinzua is Native American in origin and means “waters of many and large fishes,” and the creek remains a good trout-fishing stream to this day. A bridge would save miles of track required to go around the valley.

Aldolphius Bozano, chief engineer of the Phoenix Bridge Company, claimed to be up to the task, boasting that “we’ll build you a bridge a thousand feet high, if you provide the money….” It required only 300 vertical feet to span Kinzua gorge, and Bozano accomplished the task within 94 days. The 41-span, 2,053-foot-long structure was constructed principally of iron, with its signature patented wrought-iron Phoenix columns. When it was finished in 1882, the Kinzua Viaduct was the tallest bridge in the world, although it held the record for only two years before being eclipsed by the Garabit Viaduct over the Truyere River, Massif Central, France.

Trains rumbled over Kinzua gorge for more than 100 years, riding rails atop a marvelously slender structure. On the afternoon of July 21, 2003, however, the viaduct’s service came to an abrupt end when it took a direct hit from a tornado, and 23 of its 41 spans collapsed in spectacular fashion. The structure, listed on the National Register of Historic Civil Engineering Landmarks and the jewel of the Pennsylvania State Park System, was ravaged in only 30 seconds.

How did a structure that had stood the passage of trains and the capriciousness of nature for more than a century succumb in such an abrupt and unfitting climax? To unravel the mystery of the collapse, a team of forensic engineers, meteorologists and material engineers studied turn-of-the-century engineering journals, probed the wreckage, pored over weather data, looked at electron micrographs and literally pulled apart pieces of crumpled metal. What emerged was a clear and precise understanding of the collapse, a better understanding of the complex flow of wind within a tornado and a warning for future generations of engineers.

Kinzua Two

As is so often the case in engineering forensics, understanding a failure requires fully investigating not only the original construction but also subsequent reconstruction and remodeling over time. In the case of Kinzua Viaduct, the revamping was almost total. Rail loads increased precipitously soon after the original bridge was finished, leading to its being dismantled and rebuilt in 1900. The new bridge had span and tower arrangements identical to the original, and the old structure was used as an erection platform for the new one. The entire metal superstructure, originally comprising wrought-iron truss spans and towers, was replaced with structural steel. An innovative tower design was employed in the redesign, which from certain vantage points afforded striking studies in perspective and geometry.

The 1900 superstructure towers were mounted on the existing (1882) masonry pedestals. In order to secure the new steel superstructure to the existing masonry, the original 1-1⁄4” anchor bolts, the only metal elements remaining from the original viaduct, were reused-a decision that would prove disastrous a century later. Indeed, perhaps with a certain sense of foreboding, the second viaduct’s designer, C. R. Grimm, understood his structure’s weakest element and in so many words said, “If I could do this over, I would.”

As the 20th century moved on, coal resources became depleted, and rail traffic became less and less frequent. The railroad remained in service until 1959, when it was sold by the railroad company for scrap. The new owner, a scrap dealer, maintained title but didn’t have the heart to dismantle the bridge and sold it to the Commonwealth of Pennsylvania in 1963-for the identical price that he had purchased it from the railroad company. The Commonwealth then established a park that featured the bridge as a single attraction. The bridge remained in rail service for a private excursion railroad through early 2002, when the Commonwealth closed it for repairs, as inspection had uncovered visible deterioration. Repair work began in February of 2003.

Weather Report

On the afternoon of July 21, 2003, a light rain started to fall at Kinzua State Park, and by 3:00 p.m. the crew repairing the viaduct left the work site and returned to the construction compound immediately adjacent to the bridge. The crew was unaware of either hidden deterioration within the structure or of a wide range of severe weather that was moving into western Pennsylvania. An intense weather system, called a mesoscale convective system, formed that afternoon over a wide area including eastern Ohio, western Pennsylvania, western New York and southern Ontario. The system produced a series of spiral-like cloud banks, which moved in a counterclockwise manner as the entire system tracked in a northeasterly direction. At the leading edge of the front, the combination of wind shear and moisture with summertime afternoon atmospheric instability initiated intense thunderstorms.

As the storm matured, tornado activity appeared in many locations along the front line. These tornadoes were classified by the National Oceanic and Atmospheric Administration as F-1 on the Fujita scale, with associated wind speeds greater than 72 and equal to or less than 112 miles per hour. In Kinzua State Park, a tornado touched down at approximately 3:20 p.m. The tornado, with estimated speeds exceeding 90 miles per hour, produced a complex pattern of high-velocity winds that attacked the Kinzua Viaduct, precipitating its collapse.

You might be surprised to learn that the word “tornado” is not rigorously defined from a meteorological standpoint. A common working definition for our purposes, however, describes a tornado as a violently rotating column of air… fed by spiraling inflows of air. Both features-the rotating vortex and inflow-prove crucial in understanding the viaduct’s failure. At Kinzua Bridge State Park, the counterclockwise (or cyclonic) spiraling vortex extended to as much as one-third mile in diameter. At the surface, the rotating vortex produced high leading-edge winds in the vicinity of the viaduct The vortex was fed by concentrated inflows of air channeled along separate and seemingly arbitrary paths shaped by the hilly and varied topography. The leading edge winds attacked the structure initially from the east because of the cyclonic rotation of the winds within the vortex. As the tornado tracked northward, moving past the viaduct, the structure was re-attacked from the south by a strong inflow of air spiraling into the tornado vortex. The inflow winds were capricious, slamming directly into the viaduct but entirely missing the compound where the construction workers and park attendants had gathered.

The Collapse

A tall, slender structure such as the Kinzua Viaduct is a flexible body that quickly begins to vibrate when the wind blows. As the tornado’s winds intensified, all the towers began swaying back and forth at the their natural frequency of approximately one cycle per second. When the wind’s speed quickly increased, the amplitude of vibrations was magnified continually and dramatically. The tallest tower likely experienced at least four cycles prior to collapse, increasing internal forces as a result of inertial effects.

As the wind reached an estimated critical speed of 94 miles per hour, a failure occurred within the anchor-bolt system-the one element of the bridge that the 1900 designer, C. R. Grimm, had reflected on with some concern. East winds from the vortex actually pushed three of the towers off their bases to the west by as much as 20 feet, but the three towers momentarily remained upright. The winds from the vortex, which was moving northward, then attacked the tallest towers. One by one, the towers separated from their masonry bases, and one by one, they leaned and then collapsed westward. As the vortex moved past the structure, the three towers that had initially been pushed off their bases but had remained upright were re-attacked by the spiraling inflow air and collapsed northward, falling on top of the remaining collapsing structure. A progressive failure took place, as the towers collapsed one after the other. Each collapse was accompanied by a loud noise as each tower, moving at nearly 100 miles per hour, struck the ground dissipating its kinetic energy and distorting all structural elements.

Forensic Markers

In the course of a forensic investigation, numerous indicators, or forensic markers, are identified that, when taken collectively, reveal the secrets of the event. At the Kinzua site, there were four particularly apparent forensic markers.

Order markers: The distribution of materials clustered within the debris field reveals the chain of events. In particular, the order of collapse can be deduced from what debris is beneath other debris.

Directional markers: The direction of fallen trees and collapsed towers show wind directions. The one-third-mile-wide debris path contained trees and pieces of structure oriented to the west-owing to leading-edge winds-overlain by trees and pieces of structure oriented to the north-owing to the inflow winds.

Separation markers: The location of “clean”-that is, fresh-breaks show failure points in materials. Consistently clean breaks were observed at the interface of the 1882 and 1900 construction.

Fracture markers: The presence of consistent small-scale fractures within members identifies high stress points. Such small fractures were found in the anchor-bolt couplings connecting the 1882-era construction to the 1900-era construction. During the 1900 rebuilding, the couplings were placed only on the eastern faces of the towers to extend the connection of the anchor bolts to the towers through metal roller-bearing assemblies, which were intended to relieve thermal stresses within the towers. (These metal roller bearing assemblies quickly corroded, induced fracture in the couplings and never served their design intention.) Laboratory analyses of these small fractures revealed long-term fatigue failure with secondary fractures produced by overload during collapse.

Evaluation of these four forensic markers presented the team with a clear picture of the mechanisms of failure at work during the collapse. The fracture markers revealed that the structure had over time developed a specific weaknessfatigue fracture of the coupling assemblies. Furthermore, this weakness would only be critical if winds came from other than the prevailing westerly direction, which was and could only be the case in a tornado with strong easterly winds. The separation markers disclosed the initiating point of collapse-the anchor-bolt couplings. The directional markers quantified the remarkably complex wind patterns during the tornado. And the order markers confirmed the collapse mechanism revealed by the other markers.

These details also allowed the investigative team to describe the evolution of the collapse. As the wind speed exceeded 85 miles per hour, the wind pressure on the windward (eastern) face of the structure was sufficient to shift the structure’s weight beyond the leeward (western) bearings, and towers began pulling on the anchor bolt system on the windward face. As the wind reached the critical threshold of 94 miles per hour, the already weakened anchorbolt system on the windward face experienced a pull of sufficient magnitude to initiate a separation failure. With each progressive anchorbolt failure on the windward side, the towers, one by one, lifted on the windward side, and the leeward anchor bolts quickly sheared off. No longer attached to their bases, the towers became airborne. It took only about 30 seconds for the 23 spans and 11 towers to fall, and the order and direction of collapse was related to the complex and changing wind directions as the tornado crossed the valley floor. C. R. Grimm’s eerie premonition of 1901 had proved correct.


Tornadoes are the most violent expression of nature’s meteorological fury, and it may not be practical to build major structures to resist their most extreme winds. It should not be unreasonable, however, to protect structures against at least the forces of an F-1 tornado. In particular, the collapse of Kinzua Viaduct forces the realization that tall railroad viaducts may be vulnerable to extreme wind events when their anchor-bolt systems have been either inadequately designed or weakened through corrosion or fatigue. We can hope that a new appreciation for the engineering realities of such “wind-susceptible” structures will be an outgrowth of this sad catastrophe.