Fracture-Critical Bridges

The Golden Gate – a fracture-critical bridge

The story of the I-5 collapse continues to evolve. Currently the focus seems to be on the fact that I-5 (not to mention I-35W) was classified as “fracture-critical”, and the danger it and other fracture critical bridges pose. As there’s a great deal of misinformation being spread about the nature of fracture-critical bridges, an explanation of what it actually means is in order.

AASHTO defines a fracture-critical member (FCM) as “a component in tension whose failure is expected to result in the collapse of the bridge or the inability of the bridge to perform it’s function.” Namely, a fracture-critical element is a non-redundant tensile member. Non-redundant compression members, such as top truss chords or support columns, do not qualify as fracture-critical. Thus, a bridge can have many non-redundant elements and yet not qualify as a fracture-critical bridge.

This definition reflects the very specific risk that fracture-critical members pose. All structural elements are at risk of developing small flaws or cracks in their section. However, if the member is in tension, the force will cause the crack to gradually enlarge over time, until the member fails. Thus, any tensile elements that aren’t redundant receive special attention, and special terminology.

Wreckage of the Silver Bridge

The term fracture-critical first came into use in the late 60’s, following the high-profile collapse of the Silver Bridge in 1967. The collapse was due to the failure of a fracture-critical member – one of the suspension chains supporting the main span. As a result of the disaster, the  National Bridge Inspection program was first created as part of the Federal Aid Highway Act of 1968, requiring inspection of all federally funded bridges. The first inspection standards followed several years later. So a fracture-critical failure is why we have a bridge-inspection program at all.

Also following this collapse, AASHTO initiated a Fracture Control Plan (FCP) to reduce the risk posed by fracture-critical members. This included, among other things, more stringent steel toughness requirements and connection detail limitations. Because of this, modern fracture-critical members are much stronger than those made prior to the mid-70’s, and unlikely to fail due to welds, material defects, fatigue, or other issues.

Another high-profile collapse resulted in further changes in the requirements for fracture-critical members. In 1983, the Mianus River Bridge collapsed due to corrosion on a pin resulting in the development of a fatigue crack. Following this, the National Bridge Inspection Standards were revised in 1988, requiring much stricter inspection procedures for fracture-critical members.

In addition, fracture-critical elements are discouraged by the bridge design code, which requires higher loads for non-redundant members. Because of these material and inspection requirements, fracture-critical bridges are much more expensive to build. As a result, of the 11% of bridges around the US which are classified as “fracture-critical”, 75% were built prior to the mid 1970’s. The onerous requirements mean they’re no longer an attractive option for bridge builders.

Train derailment on a fracture-critical truss bridge
Train derailment on a fracture-critical truss bridge

A bridge is ostensibly classified as fracture-critical if it contains any fracture critical members. However, this designation is seldom based on actual analysis of the bridge superstructure, and is instead based on the structural systems being used. Certain systems will, theoretically, be inherently fracture-critical – truss bridges, suspension bridges, two-girder bridges. But depending on the actual details of construction, alternate load paths may exist which are not accounted for. There have been a large number of failures of supposed fracture-critical members which did not result in the collapse of the bridge. In one case, a train derailment destroyed a number of fracture critical members of a truss as it plunged into the river. However, the bridge remained standing, even with a large portion of the train still on it.

Table of which bridge types are classified as fracture-critical.
Results of a survey of state DOTs asking which bridge types are classified as fracture critical.

Because it’s based on broad classification types, different systems will be classified as “fracture-critical” in different states. While nearly all states classify truss bridges and two girder bridges as fracture-critical, there is substantial disagreement about other structural systems. In Europe, no distinction of fracture-critical members is made, and structures that would be classified as fracture-critical and discouraged in the US are commonly built. The focus there is on using 3D analysis to determine load paths, and on overall system reliability. A fracture critical, non-redundant bridge can be made as reliable as any other bridge type by simply requiring higher load capacity.

Fracture-critical has the interesting property of being both a very precise and very vague term. It’s precise as it’s defined, a non-redundant tensile member. It’s vague as it’s applied, to broad classes of bridges which may or may not actually have any fracture-critical members. The fracture-critical classification is due to specific failure modes which occurred more than 30 years ago, and is often applied to bridges which are in fact not-fracture critical. Despite dire pronouncements, it’s quite common for all types of structures to have single points of failure, and a structure that has them is not inherently less reliable or safe than one that doesn’t.


AASHTO LRFD Bridge Design Specifications, 5th Edition (2010)
NCHRP Synthesis 354: Inspection and Management of Bridges with Fracture-Critical Details
Wikipedia: Federal-Aid Highway Act of 1968
Wikipedia: Mianus River Bridge
Wikipedia: Silver Bridge
TxDOT Bridge Inspection Manual
Bridge Redundancy and Fracture Critical Members

4 thoughts on “Fracture-Critical Bridges

  1. Love your thoughts on engineering failures and failure mechanisms. Will read with some care and comment further.

  2. It’s evident from photos that the Skagit bridge collapsed from compressive buckling at top chord joints, above where the deck folded. Unstoppable buckling and total span failure was inevitable once those top chord beams got pulled slightly out of alignment. Their joint was pulled downwards somehow. Probably when the vertical beam supporting that joint got twisted and buckled by the cargo’s impact. 3 other nearby vertical beams show similar but less-fatal buckling from the same swerving truck:

    I think there was no cracking or fracturing involved until the tortured joints and beams were already hopelessly compromised. This bridge did not fall from critical fractures, ageing, bad maintenance, or tax policies. Its design was vulnerable. Low, curved clearances invited strikes. The simple pre-computer-era truss design could not survive severe shortening of a single vertical beam.

    The truck’s hits were all on the horizontal sway struts that define the clearance. Those struts normally help the bridge keep its top chords aligned, despite wind pressures. But maybe their connection to the vertical beams was too strong. If the sway struts had simply broken away when struck hard, without bending the vertical, the bridge span would still be standing today.

    1. Current reports say the cargo struck 10 beams. The bridge’s 4 spans had 26 low-clearance beams total, 6 on the collapsed span.

      The folding occurred at the first sway strut and its verticals, after the entrance portal. This vertical appears to be particularly vulnerable to top-chord buckling in this Warren+verticals truss design. The vertical posts carry little of the bridge’s dead load or live load forces. See and

      The vertical posts were added to allow the top-chord beams to be half as long as on Warren trusses. This normally helps avoid buckling of top chords between diagonals. But at these points it supports and stabilizes that added top chord joint with just one vertical, rather than with 3 members (2 diagonal tops + vertical) as on the odd-numbered joints. This makes it unstable and critical to span failure if that particular vertical gets buckled by impact. The odd-numbered verticals are countered by diagonals and are less likely to move their top chord joint. I think a plain Warren truss with no verticals and with thicker top chord beams is less vulnerable to this kind of accident.

      The collapse began with the truck clipping sway struts. That bent verticals, which unstabilized a compressive joint in end-to-end top chords. The now-misaligned existing forces expanded the joint buckle to where it folded downward completely. This dragged the west-side top chord into a similar downfolding failure. As the span folded, it shortened and fell off of its piers.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s