Do Two Failed Dams Foretell a Dire Future?

by Upmanu Lall and Paulina Concha Larrauri |May 27, 2020
map of us with red and blue dots

U.S. dams listed as high (red) or significant (Blue) hazard, overlaid on major rivers and states. (The list for Texas is not complete.) Source: Columbia Water Center

Two dams down last week, a few thousand more to go? It was a narrow escape for Michigan and luckily no one died. Altogether 10,000 people were evacuated and are presumably homeless at the moment. Further complicating matters is the ongoing disaster of COVID-19 and the challenges of having to evacuate while social distancing. What happened in Michigan shines a light on our aging and failing infrastructure. The question remains, will we do something now to fix it?

The 2018 National Climate Assessment report noted: “Aging and deteriorating dams and levees … represent an increasing hazard when exposed to extreme or, in some cases, even moderate rainfall. Several recent heavy rainfall events have led to dam, levee, or critical infrastructure failures, including the Oroville emergency spillway in California in 2017, Missouri River levees in 2017, 50 dams in South Carolina in October 2015 and 25 more dams in the state in 2016, and New Orleans levees in 2005 and 2015. The national exposure to this risk has not yet been fully assessed.” The failure of the Spencer dam in Nebraska in 2019, the failure of the two Michigan dams and the near failure of the dam near Roanoke, VA last week exemplify the concern. Still, dam failures are relatively rare, and hence despite their potentially catastrophic impacts, they seem to not get the attention that they deserve.

Climate change is increasing the frequency of extreme rainfall events and hence the risk for filling and overtopping dams, which is the predominant mechanism of dam failure. However, using climate change as a bogeyman for aging infrastructure failure is an unfortunate trend, since it takes attention away from an urgent and potentially fixable problem. The 2017 Oroville Dam Spillway failure that led to an evacuation of 200,000 people is a case in point. The wet conditions were anomalous, but not extreme. The flow over the spillway was merely 5 percent of what it was designed to handle. Known structural problems with the spillway were ignored for over a decade. The projected maintenance costs were in the millions of dollars, and maintenance was deferred by state and federal officials, despite a formal study of the situation, warnings, and a Federal Energy Regulatory Commission hearing. The final repair bill was over $1.1 billion —about 10 times more than the cost of the dam’s construction when adjusted to 2018 dollars — and was passed on to the federal government. The city of Oroville filed a legal complaint against the State of California for negligence in maintaining the dam.

The majority of the 90,000+ dams in the U.S. are older their nominal design life, which would not be a big issue if they were well maintained. Their state of maintenance is a concern, and this is the silent hazard that needs to be addressed. The National Inventory of Dams has approximately 25,000 dams in the U.S. that are listed as high (>10,000) or significant hazard! This is a national concern — every state has such dams, and people and critical infrastructure below them in harm’s way.

Flooding from dam failure can be much more catastrophic than what may be expected from an extreme rainfall event. The Midland, MI dam exposed one aspect of this concern. The Michigan dam failures are an example of cascading dam failure — the failure of the “unsatisfactory” Edenville dam led to the failure of the downstream Sanford dam that was rated to be in fair condition. The Buffalo Creek, West Virginia dam failures in 1972 provide another example when three dams failed in sequence failed, mobilizing a large amount of coal mining waste, rendering 4,000 of 5,000 people downstream homeless with accompanying loss of life.

The Columbia Water Center’s recent report (see summary) highlights that much critical infrastructure (dams, electricity generating plants, highways, bridges, water treatment and wastewater treatment plants) lies below dams and would be incapacitated in the event of dam failure, leading to significant and potentially chronic economic impacts. There continues to be no comprehensive evaluation of these risks or a prioritized list of dams for remedial action. Does it make sense to keep doing presidential disaster declarations each time a dam fails, or is it better to prioritize which ones to fix or remove before disaster strikes? Our report provides and exemplifies an approach that could be used to do this “rapidly,” but has not yet been applied across the U.S. due to financial and time constraints.


Proposed framework to assess the probability and consequences of dam failures, from a report by the Columbia Water Center.

Michigan, the location of the most recent dam failure, is in the Great Lakes region. There is a profusion of high hazard dams older than 60 years (nominal design life is 50 years) in counties with high population (which are also likely to have thermo-electric power plants, waste facilities and drinking water intakes located on rivers). Governor Whitmer of Michigan characterized the Midland dam failure as a 500-year event, or something that would have a 1/500 chance of occurring in any given year. If we consider dams older than 60 years in counties with a population larger than 500,000 in the Great Lakes region, we have 317 high hazard dams, and the chance of one or more dams experiencing a 500- or 1,000-year event in a year would be 47 percent and 27 percent respectively — pretty high, and while not all would lead to a failure, it is something to think about. The Great Lakes have varied with approximately 10-year cycles for over a century and are currently at or above record high levels. Extreme rainfall events are happening much more frequently than in the last 100 years. In this “wet” cycle, it is possible that more dams will fail. The question to ask is what the state governments in the region are preparing to deal with the potential floods and dam breaks.

great lakes maps

High hazard dams in the Great Lakes region, by age and with county populations shown. Source: Columbia Water Center

Who needs to act, and what needs to be done to address this massive problem? There is an Association of State Dam Safety (ASDS), with state dam safety offices as members. FEMA, the U.S. Army Corps of Engineers and the U.S. Bureau of Reclamation each have a National Dam Safety Program. Despite the ongoing efforts of these groups, and existing legislation and policies for the inspection and re-certification of dams, we lack a cohesive strategy to fix the problem. The Edenville Dam was decertified in 2018, yet it was still operating when it broke last week. Boyce Hydro, its owner, alleges that the failure is due to the State of Michigan requiring it to keep water levels high for environmental reasons. This is disputed. The New York Times and the Associated Press have reported on the state of dams, and have published information as to the inspection frequency and state of the dams using FOIA processes for at least a decade. The National Inventory of Dams provides information on some basic statistics of the dams, but selected information is no longer made available in the interest of “national security.”

A hazard and prioritization analysis, such as the one we prototyped, could be done quite rapidly given data and resources, to at least screen and identify the subset of dams that needs urgent attention and investment by the federal government to avoid a future disaster declaration and the associated cash outlays. Yet, we do not see any of the agencies responsible pursuing such a strategy and/or informing the public of the collective risk the nation faces. In the age of big data, it is imperative that this be done as the first step in a process that eventually fixes the dams or removes them, with stakeholder input.

COVID-19 should at least remind us that an ounce of prevention is worth a ton of post-event anguish. As the federal government puts $3 trillion toward COVID recovery, it puts $10 million a year toward dam repair — a paltry sum compared to the $70 billion the ASDS estimates is needed to repair dams. Put that in the context of Oroville, where the post-failure repair cost $1 billion. If Oroville had actually failed, the damages would have been in the several billions with widespread havoc. What should we spend now to avoid spending trillions later?

Read another version of this story in the New York Times Opinion section. 

Upmanu Lall is the director of the Columbia Water Center. Paulina Concha Larrauri is a researcher at the Columbia Water Center.

Get our newsletter

I'd like to get more stories like this.
Email address
Secure and Spam free...

Leave a Reply

1 Comment threads
0 Thread replies
Most reacted comment
Hottest comment thread
1 Comment authors
Ron Corso Recent comment authors
newest oldest most voted
Notify of
Ron Corso
Ron Corso

The article “Do Two Failed Dams Foretell a Dire Future?” is very well done, but there is one glaring omission. There is no mention of the Federal Energy Regulatory Commission (FERC). The FERC issues licenses for the construction and operation of non-Federal dams that produce hydroelectricity. The FERC has an extensive dam safety program and has more dams under its jurisdiction than the number of dams collectively that the USBR and the Corps are responsible for. The FERC dam safety program for the more 2300 dams under its jurisdiction is a subject you failed to recognize.. For more information on… Read more »