The I-35W Bridge Collapse, Ten Years On

August 1 marks the tenth anniversary of the tragic and shocking collapse of the I-35W Mississippi River Bridge in Minneapolis.

Around the fifth anniversary, I did an 8-part blog series on the subject, which is worth re-upping:


Motivated by the collapse, we conducted a number of inter-related studies, powered empirically by traffic and GPS data collected before and after the bridge reopening:

The nice thing from a scientific perspective was the ability to use the GPS data collected before and after the bridge reopening for other studies as well, the data comprised part of four of my student’s dissertations, and several from Henry Liu’s students.


Another thing to note, from a career perspective, was that this research agenda was an unanticipated turn. Though I had done some empirical route choice studies before hand, and so was primed to take this direction, I was moving more into the transport and land use and network evolution realm. If you had asked me on July 31, 2007 what I would start working on on August 2, 2007, this was not it.

I-495 bridge closure a ‘national issue’ | Delaware online

The Leaning Bridge of Wilmington (I-495) is to be straightened. This  article by Jeff Montgomery and Melissa Nann Burke , I-495 bridge closure a ‘national issue’, discusses.

I was interviewed as part of the article.

University of Minnesota engineering professor David Levinson, who studied the effects of I-35W bridge collapse on the Twin Cities area, noted that the Minneapolis span was rebuilt in less than a year.

Considering the I-495 bridge is 40 years old, Levinson suggested replacing the whole structure for efficiency’s sake, he said.

“It’s logical to do as much as you can upstream and downstream from the bridge while it’s closed,” Levinson said.

Path dependence

If you don’t know where you’re going, any path will get you there.

Path dependence is the idea that where we are today depends critically on where we were yesterday. Some systems are path independent, those that have a single unique equilibrium. Finding the solutions to some math problems is independent of where you start, as long as you follow a particular algorithm.

However, most systems we deal with on a daily basis have some characteristics of path dependence. Where you live might depend on what job you took, which depends on what your previous job was and where you went to school, and a different decision anywhere along the way would change today’s position.

Nowhere is this more true than transportation. On the one hand, it is obvious that certain locations were destined to be important cities because of significant natural advantages. New York has a deep harbor at the confluence of major navigable river. Chicago is at the pivot point between vast agricultural lands to the Northwest and the shortest land path to the East Coast. It was natural railroads would flow through the point on the map we now call Chicago.

On the other hand, many city sites that were selected for natural advantages in one technological era (The Romans selected London and the Dutch and English chose New York in large part for their capability as ports), remain important even after that technology becomes obsolete. With the logistics revolution and the new dominance of container shipping, London’s shipping has moved northeast to Felixstowe as large container ships cannot easily ply the Thames, while New York’s shipping has migrated to the wide open spaces of New Jersey.

The one-time advantages result in a set of complementary investments and inter-related decisions that take on a life of their own. Because of local trading advantages, commodities markets, banks, insurers, and other related organizations located nearby. A critical mass of those institutions felt no need to migrate just because their initial raison d’être vanished. While a building is under construction, temporary framing will often be used until the more permanent structure is erected. Once the final building can stand on its own, the falsework is dismantled. In a sense, everything is falsework for what comes after.

This kind of mutual complementarity happens repeatedly in transportation. Airplanes are the perfect example of mobile capital. If Amalgamated Airlines no longer wants to serve a particular city pair, the airplane can easily be redeployed elsewhere. Yet 80 years into the commercial aviation industry, airlines today serve mostly the same hubs their predecessors did on the Airmail routes of the 1930s. American Airlines is still in Dallas, United in Chicago, Delta (Northwest) in Minneapolis, and so on.

While very few decisions are completely irreversible, transportation decisions come close. Where we place a right-of-way, or an airport will explain where that facility will be decades, or even centuries from now.

A slight deviation from the efficient path to solve a short term problem today will cost travelers time for years to come. It is important to get the design right for the long term. (Undoubtedly this has social costs, see e.g. I-94 through the Rondo in Saint Paul).

But a slight deviation from the path will also change what the long term is. Build a bridge “here” rather than “there”, and then you will adjust all of the roads feeding into the bridge to meet it “here” (instead of “there”). And then land will be developed along the road to “here” to take advantage of the newly created accessibility, properties will be platted, buildings will be built, travel and trade patterns established, and other critical dependencies will come to assume that the bridge is “here”. At some point, say 50 years in the future, the bridge will need to be replaced. Even if “there” was a better location than “here” initially, after five decades of adaptation, it is quite likely that “here” is better now. The whole may have been better were a different initial decision been made, given conditions at the time. Given current realities, that path must now be foregone.

In transportation we say build it right the first time, because there won’t be a second chance. And that is true. But also remember the world will adapt to whatever we do, and we cannot let the perfect be the enemy of the good.

Tunnel Collapses Outside Tokyo Kills Nine –

NY Times: Tunnel Collapses Outside Tokyo Kills Nine :

“The police said they were investigating the cause of the collapse on Sunday at the Sasago Tunnel — a three-mile passage near the city of Otsuki and about 50 miles west of Tokyo — and for evidence of negligence by the company that operates the highway.
News reports said investigators believed that supports in the ceiling of the 35-year-old tunnel might have grown brittle, allowing hundreds of the slabs to fall onto passing vehicles. Each slab weighed 1.2 tons, officials said.”

Getting from here to there: Bridge Collapse Causes Train Wreck

David King on Bridge Collapse Causes Train Wreck:

A train derailed in New Jersey after the bridge it was crossing collapsed. Here is a CNN story. At this point no one knows if the bridge collapse was the cause of the derailment (or was there something with the train that caused the collapse), but will this event serve as a reminder than we tolerate catastrophic failures of our infrastructure far more commonly than most people think? I am not confident that knowledge of potential failure will spur action, nor am I very confident that actual failure will change priorities to fix our infrastructure first. It seems most likely that we will continue to tolerate occasional failure even though everybody knows this is the wrong  way to go about things. Collective action problems are hard.

Timelapse video: New Hastings Bridge is floated into place

Pioneer Press: Timelapse video: New Hastings Bridge is floated into place:

“Watch the entire process of moving the 6.5 million pound bridge down the river and being lifted into place, which took about 60 hours from start to finish, according to the Minnesota Department of Transportation.
The roadway for the bridge won’t be poured until next spring.”

The Fall and Rise of the I-35W Mississippi River Bridge – Part 8: Policy Implications |

Cross-posted from The Fall and Rise of the I-35W Mississippi River Bridge – Part 8: Policy Implications


The Fall and Rise of the I-35W Mississippi River Bridge – Part 8: Policy Implications


Unlike bridges, transportation networks are seldom “fracture critical”. While the Interstate Highway System did what it could to sever local streets and channelize traffic onto fewer, larger, limited access link to achieve economies of scale and higher speeds and throughputs and the expense of redundancy, there was enough remaining redundancy to ensure that this one bridge collapse would not have devastating transportation implications. Had the I-94 bridge collapsed instead, I am sure the consequences would have been much worse, as the alternative paths are not as convenient. But again, the Twin Cities would have muddled through.

Several lessons can be drawn. If there is redundancy, maintaining road operations during construction may not be necessary, and may needlessly delay construction while exploding its cost. There have been projects, in Minnesota notably Mn36, where closing the road entirely to speed construction has been successful. Rebuilding a bridge while keeping it open is like doing brain surgery on one-self. In principle it is possible, but why? The length of bridge construction in other examples (e.g. the Oakland Bay Bridge, admittedly a somewhat more complicated project)
versus the less than one year from design to opening that this bridge took is instructive.

Actual effects were much less than forecast. People are quite adaptable in their travel patterns, especially if there are alternatives routes and destinations.

As a consequence of the Bridge collapse, many other bridges were repaired more quickly than they otherwise would have been. A list of recent Bridge closures in the region below is illustrative.

  1. August 1, 2007: I-35W
  2. January 2008 Hastings Bridge
  3. March 20, 2008 St. Cloud Bridge
  4. March 26, 2008 University of Minnesota Pedestrian Bridge
  5. April 25, 2008 Lowry Avenue
  6. May 6, 2008 Blatnik Bridge (I-535) in Duluth
  7. June 4, 2008 MnDOT barricades Hwy. 43 bridge over Mississippi River at Winona

Three Successes

  • Emergency Response
  • Immediate Traffic Restoration
  • Quick Completion of Bridge with Design/Build

Three Failures

  • Bridge Collapse
  • Removal of Successful, Low Cost Restoration Measures
  • Overpaying for Bridge

Overall, Americans seem good at short-term tactics but poor at longer term strategy. This needs to be rectified. While there were positive outcomes, including the passage of the gas tax, and increased attention to maintenance, there may have been some over-reaction in terms of replacement of bridges.

While there was some accountability, some key personnel remained in place. Firings and resignations need to be used more often in the aftermath of events like this, some people should assume responsibility.

We need to address the question of the appropriate role of politics in infrastructure decisions. We don’t expect politicians to make recommendations about which electric power cables are replaced, why are they so involved in maintaining existing roads. (The construction of new roads is on the other hand, more obviously political in nature). Should Mn/DOT and other similar agencies move toward a public utility model to de-politicize?

We are not good at dealing with low probability, high consequence decisions. We are not good at assessing the value of either.

Unless something is done, failure will be more frequent, the Interstate system is aging and nearing the end of useful life for many components.

In The Transportation Experience, we identify a set of strategies for maturity:

  • Abandonment
  • Cash cow – using resources for something else
  • Maintenance and Rehabilitation
  • Replacement

Abandonment of the Interstate system as a whole seems premature, and while a handful of selected urban links have been closed, these are very much the exception. The Cash cow strategy is in fact what has been employed for the past several decades, as the gains from the built infrastructure exceeded by far the amount that was reinvested in it. This is of course economically correct, infrastructure should be fully exploited to its capacity. But this needs to be coupled with appropriate levels of maintenance and rehabilitation, and ultimately planned replacement, otherwise the goose that lays the golden egg will die.

The “Weak Bridge” sign that one sees all over England is not terribly reassuring, and indicates an unwillingness to recapitalize the network. In the US context, there should be more money for transportation maintenance and rehabilitation, we are eroding our physical capital. If there isn’t … we should spend our money more carefully, taking care of the existing systems and users first, and engaging in graceful abandonments as necessary, but not building new infrastructure which will require long-term maintenance without any means for doing so. The mantra Fix it first has been suggested for this strategy, and I agree. Unless something is done, the problem of decaying infrastructure will only get worse (the physical world being subject to entropy and all), bridges do not repair themselves.

Other Parts in Series: Part 1 – IntroductionPart 2 – StructurePart 3 – Communication,Part 4 – PoliticsPart 5 – EconomicsPart 6 – TrafficPart 7 – Replacement, Part 8 – Policy Implications

The Fall and Rise of the I-35W Mississippi River Bridge – Part 7: Replacement |

Cross-posted from The Fall and Rise of the I-35W Mississippi River Bridge – Part 7: Replacement


The Fall and Rise of the I-35W Mississippi River Bridge – Part 7: Replacement


The replacement bridge cost $251 million, funded almost entirely by federal government. We can debate whether the federal government should have paid for it (it was originally built with 90 percent federal contributions, 10 percent state, but matches recently are much more balanced), since most traffic using the Bridge both originated in and is destined for Minnesota. With Minnesota Congressman James Oberstar then Chair of the House Transportation Committee, there was plenty of political support behind this.

The replacement bridge was hurried, completed by September 2008, several months ahead of the original schedule. This is good, a lack of a bridge costs the economy. (Xie and Levinson estimated between $127,000 and $170,000 per day, MnDOT estimated $400,000 per day). The contractor received $200,000 per day bonus for early completion. So perhaps in an economic sense, too much was paid to complete it a few months early.

Rebuilding a collapsed bridge is of course a crisis, but it is also an opportunity to do something interesting. Rushing designs may mean that ideas are missed. What was built is a functional bridge, and there are state-of-the-art real-time structural health monitoring systems installed, so I have no fear driving over it. It was also ensured that the Bridge would be compatible with any future light-rail transit lines, though none are planned for this bridge (and how they would transition from the center of the Bridge to any reasonable destination is extremely unclear). But could more have been done?

The snow removal and icing problem was not considered. Minnesota is famous for its winters. The previous I-35W bridge had installed a de-icing system (in response to earlier crashes), which had been speculated to be responsible for corrosion of the structure. While the NTSB did not find that, de-icing chemicals do have environmental consequences.

A solution not considered was air rights. A bridge over the Mississippi is expensive. But imagine having a 2 or 3 story office building hanging from below, and/or built above the highway. The views from the River are fantastic. It will not impair other views of the river especially much, and would generate a significant amount of revenue to pay for reconstruction. One example would be the historic London Bridge, which had houses and stores along the side, encroaching on travelways. There are better ways to combine transportation arteries with development opportunities, and creative design can show the way.

Why did we not build a habitable structure above the road deck, so that offices, residences, hotels, etc. could share the view? It would have increased the profile of the Bridge by two stories, but created an enormous amount of leasable space, leases which could have paid for themselves and the Bridge. Obviously it would increase the initial construction cost, and perhaps time, but that would be amply repaid over the long-term. That structure would further have shielded the roadway from ice and snow, reducing road snow clearance costs and crashes.

The bridge opened in the early darkness of September 18, 2008. A parade of first responders, and then a bulge of traffic all hoping to be the first (and none succeeding) went across. Soon the Bridge was attracting 120,000 vehicles per day, measurably off the pre-collapse levels.


Other Parts in Series: Part 1 – IntroductionPart 2 – StructurePart 3 – CommunicationPart 4 – PoliticsPart 5 – EconomicsPart 6 – TrafficPart 7 – ReplacementPart 8 – Policy Implications

The Fall and Rise of the I-35W Mississippi River Bridge – Part 6: Traffic |

Cross-posted from The Fall and Rise of the I-35W Mississippi River Bridge – Part 6: Traffic

The Fall and Rise of the I-35W Mississippi River Bridge – Part 6: Traffic


Travel behavior changes after network disruption as well as after the replacement of disrupted links are not well-understood. The first table shows the number of river crossing trips by type of facility before and after the collapse and the reopening.

We discover 46,000 lost trips daily after the collapse (nearly a third of what the I-35W Bridge had carried), and 20,000 found trips after the new bridge opened. Those lost trips may not have been made, or more likely, found different destinations not requiring a river crossing. This provides additional evidence to the phenomenon of induced demand.

Bridge Collapses: 1-Aug-2007 Bridge Reopens: 18-Sep-2008
Bridge Before After Change Before Reopen After Reopen Change
I-35W 140000 0 -100.00% 0 120349
Arterial total 152311 197566 29.70% 169983 95895 -43.60%
Freeway total 572274 481040 -15.90% 488717 583127 19.30%
Total 724585 678606 -6.30% 658700 679022 3.10%


Gains from the Bridge for three peak periods re-estimated with accurate (observed) travel times (but fixed and not-observed OD matrix), shown in the second Table, were on the order of $70,000 per day (somewhat below our initial low all-day estimates ($127,000), far below Mn/DOT’s, and also below the contractor early completion bonus). Much of that gain is lost once the I-94 bridge lane disappears, as consistent with the original results. $42,000 annualizes to about $15 million in benefit, for a $250 million dollar bridge (which pays off in about 23 years at 3% interest). The I-94 lane restriping paid off in a matter of a month.


Other Parts in Series: Part 1 – IntroductionPart 2 – StructurePart 3 – CommunicationPart 4 – PoliticsPart 5 – EconomicsPart 6 – TrafficPart 7 – ReplacementPart 8 – Policy Implications