By Jay Landers

In May 2020, the Edenville Dam in Michigan failed after heavy rains, unleashing a deluge of water that overtopped and destroyed the downstream Sanford Dam. Although no deaths or injuries occurred, the failures of the two 1920s-era dams forced the evacuation of tens of thousands of central Michigan residents and caused hundreds of millions of dollars’ worth of damage (see “Michigan Dam Failures Prompt Investigations, Lawsuits, and Safety Concerns,” Civil Engineering, July/August 2020, pages 20-22).

Edenville Dam post collapse
In May 2020, a section of the embankment to the left of this spillway on the Edenville Dam gave way, sending a surge of water downstream that overtopped and destroyed the Sanford Dam. (Courtesy of EGLE Water Resources Division)

As it turns out, the physical failure mechanism for the Edenville Dam — static liquefaction — had its genesis in improper construction practices, according to the final report, released on May 4, from the independent forensic team established to investigate the disaster. However, the failure likely could have been avoided if a comprehensive review had been conducted of the earth embankment that eventually gave way in 2020, the report notes.

The absence of such a review is cited by the IFT as one of a host of mistakes, missed opportunities, and unexpected conditions that contributed to the catastrophe. “With respect to the human judgments, decisions, actions, and inactions during the project history leading up to the May 2020 event, the dam failures were foreseeable and preventable,” according to the report.

“The extent to which the numerous contributing factors combined and aligned to result in the failure primarily reflects both deficiencies in the construction of Edenville Dam and deficiencies in subsequent industry practices during the history of the project,” the report notes. “The failure also secondarily involves an unfortunate combination of factors related to the variability of the dam along its length, the variations in the seepage behavior of the dam, the embankment stability analyses that were and were not performed, the hydrologic characteristics of the May 2020 storm event, and the timing of that storm event relative to planned upgrades to the Edenville gate hoist systems and spillways.”

Troubled history

At the time of their failure, the Edenville and Sanford dams were owned by Boyce Hydro Power LLC. Both dams comprised earth fill embankments, gated concrete spillways, and powerhouses. The Edenville Dam was situated at the confluence of the Tobacco and Tittabawassee rivers, while the Sanford Dam was downstream on the Tittabawassee River. The Edenville Dam impounded Wixom Lake, a reservoir that had a gross storage capacity of 40,000 acre ft.

In 2018, the U.S. Federal Energy Regulatory Commission issued an order revoking the company’s license for the Edenville Dam, citing the failure of Boyce Hydro Power to correct various noncompliance issues at the facility. Among the issues were the inability of the dam to accommodate the probable maximum flood because of inadequate spillway capacity. After the failure of the Edenville Dam, Boyce Hydro Power maintained that it had been unable to afford the cost of the modifications needed to increase the structure’s spillway capacity.

Following FERC’s revocation of the company’s license for the Edenville Dam, regulatory authority for the structure reverted to the Michigan agency now known as the Department of Environment, Great Lakes, and Energy. In 2018, the department directed Boyce Hydro Power to conduct a comprehensive analysis of the dam’s structural integrity and ability to meet state safety rules. However, the company had not reported the findings of such an analysis at the time of the dam’s failure.

In August 2020, FERC commissioned the IFT to investigate the physical and human factors that led to the failures of the Edenville and Sanford dams. John W. France, P.E., D.GE, D.WRE, an independent engineering consultant at JWF Consulting LLC, led the team. The other members were Irfan Alvi, P.E., the president and chief engineer of Alvi Associates Inc.; Arthur Miller, Ph.D., P.E, PH, D.WRE, a technical adviser at AECOM; Jennifer L. Williams, P.E., the associate vice president of geotechnical dam safety at AECOM; and Steve Higinbotham, P.E., an independent hydraulic structures engineering consultant.

In September 2021, the IFT released an interim report detailing its determination that the Edenville Dam failed as a result of static liquefaction, rather than internal erosion or overtopping. In support of this unexpected conclusion, the IFT noted the presence of loose, uniform fine sand in the failed embankment as well as in remnant breach material.

Laboratory testing on the loose sand found that the material lost strength in a manner similar to liquefaction. Meanwhile, eyewitness video recorded at the time of the failure showed the failing soil mass to have accelerations and velocities consistent with liquefaction.

The interim report highlighted the possibility that construction practices used during the building of the Edenville Dam may have contributed to its failure. Although construction records indicate that the soil used to build the dam was to be placed in layers and compacted, the IFT found no evidence indicating that this was done.

Geotechnical factors

In its final report, the IFT again identifies static liquefaction as the “most plausible principal mechanism” for the failure. “Edenville Dam was constructed in the 1920s in a manner which significantly deviated from the design plans and construction specifications, and this resulted in the embankments being constructed with sections of very loose to loose sands which created the fundamental physical condition required for static liquefaction,” the report states.

The presence of these sands was the “primary geotechnical factor contributing to the failure” of the Edenville Dam, the report notes. “Had the embankment been constructed in compacted layers in accordance with the original construction specification, the sands in the embankment would not have been loose and the embankment instability failure would almost certainly not have occurred.”

The steep downstream slope of the dam’s east embankment, where the failure occurred, also appears to have been a contributing geotechnical factor. Although the original design called for the embankment to have a slope of 2:1 horizontal to vertical, the embankment as constructed had a steeper slope of 1.8:1, France said during a May 5 call with media held by the Association of State Dam Safety Officials. “The steeper the slope, the higher the shear stresses” in the embankment, he noted, increasing the likelihood of static liquefaction.

Although other sections of the approximately 6,000 ft of embankments at the Edenville Dam had been subjects of previous stability analyses, no such analysis was ever conducted for the section that eventually would fail. “Therefore, the stability deficiency of the embankment section that failed was never recognized,” the report states. “If this vulnerability had been recognized, it could have been remediated by slope flattening or buttressing, which the dam owners would have been able to afford and which had already been done at other locations. This remediation would likely have prevented the embankment failure.”

Unusual and unfortunate factors

High water levels in the reservoir behind the Edenville Dam appear to have precipitated the failure of the structure’s east embankment. “The static liquefaction occurred when Wixom Lake reached a level that was about 3 feet higher than the previous high level which occurred in 1929,” according to the final report. At the time of failure, the lake level was about 5.5 ft higher than normal and about 1-1.5 ft below the dam’s crest. The abnormally high water levels in the reservoir resulted from an “unusual and unfortunate combination of factors,” the report states.

Among these factors were the “very concentrated” rainfall that fell in the watershed immediately above the Edenville Dam in the run-up to its failure, France said. “Almost all of the rainfall, something in the neighborhood of 90% of it, fell within 18 hours at an almost constant rate of 0.22 inches per hour, which is a pretty heavy rate of rain,” he said.

Unfortunately, saturated ground conditions, along with frozen conditions in forested parts of the watershed, “resulted in a relatively high percentage of conversion of the rainfall into runoff,” France said. “Of the rainfall, 35% of it was converted into runoff rather than percolating into the ground. That's quite high.” As a result, a 25- to 50-year storm generated a 100- to 200-year flood event.

Meanwhile, the Edenville Dam spillway gates were not opened to a height sufficient to allow the free flow of water over the concrete spillway crests. The original gate hoist system could not open the gates more than 7 ft. The owner added an A-frame system to facilitate a 10 ft opening of the gates, the height needed for free flows over the spillway crests. However, in 2019 the system was found to be potentially unsafe for operators, and so the gates were not opened more than 7 ft during the May 2020 event.

Had the gates been opened the full 10 ft at that time, the peak lake level would have been 1 ft lower at the time of failure. “Because of uncertainty regarding the triggering of static liquefaction, the IFT cannot conclude that this reduction in lake level would have prevented the failure, but it might have,” the final report states.

‘Less than comprehensive’ evaluations

In investigating the failure, the IFT determined that previous hydraulic and hydrologic evaluations of the Edenville Dam “were less than comprehensive,” France said. For example, he noted the failure by project engineers to recognize the potential for unusually high runoff from the upstream watershed.

Meanwhile, prior evaluations of the dam’s spillway discharge capability had determined that the dam could accommodate 50% of the probable maximum flood. However, these evaluations failed to account for the fact that the spillway gates could not be fully opened. “In reality, with this limitation on the gates,” France said, “the flood-handling capability (of the Edenville Dam) was really more in the range of about a 200-year flood,” an event much more likely to occur than the probable maximum flood.

Had the higher runoff rates and an accurate estimate of the spillway discharge capability been accounted for previously, the owner and regulators of the Edenville Dam might have proceeded differently, France said. “There may have been a greater urgency to at least modify the spillways” in an effort to increase their capacity, he said, resulting in a lower lake level and decreasing the likelihood of static liquefaction occurring in May 2020.

Lessons learned

In its report, the IFT cites several lessons to be learned from the disaster. Foremost among them is the need to consider static liquefaction as a potential failure mode “for water storage or flood management dams when saturated or potentially saturated, loose or very loose sands, silty sands, or nonplastic silts are present in the embankment or foundation of the dam,” the report notes. Historically, such a failure mechanism has been rare for water storage dams.

The report continues: “The challenge for water dam engineers now is to develop procedures and protocols to screen and evaluate static liquefaction potential and determine when risk reduction actions to address this (potential failure mode) are appropriate. Developing these procedures and protocols should leverage the work that has been done by tailings dam practitioners.”

The IFT also calls for dam safety evaluations to include “periodic comprehensive reviews of original design and construction, performance, operations, analyses of record, maintenance, and repairs,” the report states. “These evaluations need to be an independent review, unbiased by previous conclusions by others.”

Although most U.S. state dam safety programs rely mainly on periodic physical inspections, comprehensive reviews are needed to discover problems that are not readily detectable, France said. “No one has X-ray eyes,” he noted.

Along with calling for all levels of government to provide greater financial support for owners needing to rehabilitate dams, the IFT recommends efforts to address situations in which those who enjoy benefits from a dam and its reservoir — such as higher tax revenue, increased property values, and recreational opportunities — do not share any of the financial responsibility associated with it. In such cases, “sales of dams to local public entities should be considered along with more creative solutions such as public-private partnerships,” the report states.

On the enforcement side, the IFT recommends that all dam safety regulatory agencies “should have the regulatory authority to order a dam breach if dam safety risks are judged to be unacceptable and an owner does not have the financial resources to reduce the risks or does not comply with a directive to reduce the risks,” the report states.