Investigators have published their findings on the Spencer Dam failure that occurred in the US during March 2019. Ignorance of ice runs and underestimating dam-hazard classification were said to be key factors in the incident. The Association of State Dam Safety Officials hopes that industry-wide lessons will now be learned from this failure.
Early on 14 March 2019, the Spencer Dam on the Niobrara River in Nebraska, US, failed during a major flood and ice run on the river. At the request of the state regulator, Nebraska Department of Natural Resources, and dam owner Nebraska Public Power District (NPPD), the Association of State Dam Safety Officials (ASDSO) undertook an independent investigation to examine the failure.
“Unlike the transportation industry after an air crash, the dam industry does not have an automatic way of initiating an investigation following a dam failure,” says leader of the Spencer Dam Independent Investigation Panel, Mark Baker.
“So it is credit to the Nebraska Dam Safety Programme and Spencer Dam owner, NPPD, that this investigation was completed.”
ASDSO selected individuals to work on the investigation panel who had decades of experience in dam safety failure investigations, hydrology, hydraulics, ice and debris flow and hydraulic structures. The panel comprised:
- Mark Baker – Principal of Dam Crest Consulting
- Robert Ettema – Professor, Department of Civil and Environmental Engineering, Colorado State University
- Martin Teal – Senior Vice President, WEST Consultants
- John Trojanowski – President of Trojanowski Dam Engineering
ASDSO also established the Spencer Dam Failure Investigation Oversight Group. Both the state regulator and dam owner co-operated completely with the investigation, which was funded by NPPD.
“All work submitted by the investigation panel, as well as conclusions and opinions presented, are the sole work of the team and completed without input or influence from Nebraska Department of Natural Resources, and dam owner the Nebraska Public Power District,” ASDSO states.
Published in April 2020, the report focuses on the physical causes of the failure, the human and organisational causes, and lessons to be learned.
“As engineers,” Baker adds, “our profession holds paramount the safety of the public, and we believe it is essential to learn from this event to prevent similar events in the future.”
Most likely scenario of the Spencer Dam failure
Spencer Dam consisted of a power house, a 122m-long spillway and a long earth embankment. The panel gathered and viewed data from local, state and federal agencies, as well as conducting face-to-face and telephone interviews with those who could provide more of an insight into events.
However, its efforts are described as being hampered by:
- A lack of first-hand accounts due to the remoteness of the site
- The evening and early-morning timing of the failure
- Severe weather conditions during the failure – the dam operators were only able to provide descriptions of what they saw at specific times and locations during the event as their visibility was limited
Given the lack of first-hand accounts, the report says it “describes the range of what might have happened and details the panel’s opinion of the most-likely scenario for the dam’s failure”.
According to the panel, an adverse convergence of factors related to the physical setting of Spencer Dam led to an overwhelming ice event that led to its failure.
“Though such an event could be deemed infrequent,” the panel says, “the dam’s mode of failure was foreseeable because the dam was on a river with a history of ice runs.”
Based on the accounts of the dam operators, evidence left after the failure, and other observations and data, the panel found that the most likely scenario for the dam’s failure is as follows:
- A wet autumn and colder-than-normal winter produced substantial thickness of river ice cover and snow pack. A winter storm produced flooding and dynamic break-up of the river’s ice cover, as weather conditions became colder and windier
- On the evening of 13 March 2019, the dam operators fully opened all four of the dam’s radial gates and then later released stoplogs from some of the other bays to increase outflow. However, the accumulation of ice prevented them from opening most stoplog bays
- On midnight of 13 March, a major ice run came down the Niobrara River failing two bridges upstream from Spencer Dam. One or more of these upstream ice jams backed up flood waters and burst, sending rubble and flood water towards the dam
- It is likely that ice rubble clogged the opened gates and stoplogs of the dam’s spillway, and the reservoir rose to the dike crest
- Ice pushed through the upstream brick wall of the powerhouse and flow overtopped the dike. Erosion on the downstream side of the dike led to head-cutting and the dam’s embankment breached in two locations discharging water and ice rubble downstream
- The flow of water and ice failed the dam and swept through a house and other buildings immediately downstream. Although the sole resident was warned to evacuate, he was swept downstream with his house and drowned
- The ice run carrying ice and debris continued downstream where several other bridges were destroyed or damaged
- The flood of water and ice greatly exceeded the capacity of the dam and its spillways. In its opinion, the panel says there was nothing the dam operators could have done to have kept the dam from failing given the magnitude of the flood and ice run
“In the past,” the report explains, “failures of engineered structures during large natural events were often described as Acts of God. Today, engineers know that structures need to be designed, constructed and operated to have a low probability of failure for reasonably-foreseen loading conditions and potential failure modes.”
As people, and the organisations they represent, are involved in every stage of a dam from design through to maintenance, the report panel said that dam “incidents and failures evoke important human and organisational lessons learned beyond the physical processes of the failure itself”.
According to the panel, there are three defined categories of inadequacy due to human errors in risk management: complacency, overconfidence and ignorance.
While the panel did not find complacency or overconfidence to be “major contributing factors to the failure”, general ignorance relating to the nature and risk of ice runs, and ignorance resulting in underestimation of the downstream hazard posed by the dam, were two human factors which did contribute to the failure of Spencer Dam.
Ignorance of ice runs contributed to Spencer Dam failure
Spencer Dam had experienced prior difficulties due to the dynamic break up of ice cover upstream on the Niobrara River in 1935 (the dam failed), 1960 and 1966 (the gates and powerhouse were damaged by ice).
“Ice runs had happened three times previously, and after each one no one said ‘what do we have to do to modify the dam to get it to pass ice runs better?’,” Mark Baker comments. “Instead they just repaired the dam and kept on going.”
Although dam operators had a “vague knowledge from brief discussions with previous operators that ice runs had occurred”, there was no specific provision for handling such risks.
As the report states: “There exists a pervasive ignorance about ice run-related risks in the dam industry generally, and for Spencer Dam specifically.” Indeed, the dam failure database which is maintained by ASDSO lists 380 dam failures across the states, but none of these failures have been attributed to ice runs.
Baker says: “We found that this dam was well maintained but the missing part of the puzzle that contributed to the failure was that maintenance was being done for typical dam activities, such as controlling seepage and erosion on the upstream slope – but no-one was asking the question ‘what about the ice runs?’, or ‘how is this dam performing in the spring when ice runs are coming down the river?’.”
Currently, the panel finds that the dam industry “still lacks adequate technical understanding of ice mechanics, and thus a lack of guidance in the industry’s technical literature on how to evaluate the risk”. It adds that more research is needed in this area.
“It was obvious to the team after many months of collecting evidence that we need a lot more research into this area of ice runs,” Baker says. “This ice run [at Spencer Dam in 2019] was much bigger than any others in the dam’s history.
“We hope that dam owners from here onwards will look at their dams and vulnerability to ice runs and see if those dams need to be modified – or if it is too expensive to do so, consider taking them out if necessary.”
Baker continues: “Our report shows that all through the dam’s history the impact of ice runs on Spencer Dam has been missed out. I’ve been doing dam safety work for 28 years, and seen the industry begin to turn from dam inspections to evaluating dams for something we call Potential Failure Modes.
“The industry is learning from past dam failures and how dams can fail. The Potential Failure Mode for ice runs has not been included into this analysis. This is new, and our report intends to encourage those who do dam-safety evaluations and guidelines to include ice runs in how dams can fail.”
The importance of historical knowledge
Ignorance of historical ice runs and how they affected Spencer Dam was another failing, the report finds. It states: “Past performance is an important indicator of possible future performance, and review of the prior failures and incidents would have yielded valuable insight on the dam’s ice run performance.”
Baker explains: “We have learned from the Spencer Dam failure, and from Oroville Dam spillway failure too, that the history of dams is really important. Engineers get asked to look at and inspect dams, but the dam history tells a remarkable story and highlights the vulnerabilities of the dam.
“My analogy is that if you go to the doctor, they don’t just take a look at your skin and say you’re healthy. They’ll look at your history of surgery, genetics and parental diseases etc.”
The panel report adds: “Reliance on visual inspections to detect dam deficiencies and vulnerabilities is common in the dam industry, but may miss important, non-visible dam vulnerabilities.”
Baker stresses that engineers need to understand the history of their dams but admits that this does require effort, such as going through archives and maintaining effective information management.
Indeed, during the 1966 ice-run incident at Spencer Dam, the panel discovered that dam records were destroyed. Remaining records were disorganised and not shared adequately, and operator experience was not sufficiently passed on to successive new operators.
No consolidated history of Spencer Dam was established and maintained with references to key documents.
Another criticism was that state dam-safety inspections did not specifically address ice-run performance. They are described as being focused on “observable deficiencies and not on latent vulnerabilities such as performance during ice runs”.
Inspections were carried out in warm weather months, and so performance during ice events was not observed.
Underestimation of downstream hazard
The potential of Spencer Dam to cause life-threatening flooding at the downstream house and property was also underestimated. “There was a lack of recognition that the house, Strawbale Saloon and RV campground situated just downstream from the dam would be at risk if the dam failed,” the report states.
“One reason is that the Downstream Hazard Potential Classification (DHC) for the dam was ‘significant’ when, in the panel’s opinion, it should have been ‘high’. Its significant DHC rating resulted in less dam-safety regulation, including no requirement for an Emergency Action Plan (EAP).
“If the dam was designated a high hazard potential dam, there would have been a requirement for an EAP and there might have been a requirement to modify the dam to increase flood handling capacity.”
According to the investigation panel, Nebraska Dam Safety Programme (NebDSP) may have not classified Spencer Dam as high hazard for several reasons:
- The original classification was performed by the US Army Corps of Engineers (USACE) in the 1970s. It may have taken effort and justification to revise as “once a classification is made by a reputable federal agency such as the USACE, it is easy for a state regulator to treat that classification as one that needs no review or update”
- NebDSP relied on inaccurate methods for screening dams and determining whether additional hazard-classification analysis was required
- There is ambiguity in the language used for state and federal classification of hazard
- There is inadequate documentation in NebDSP’s DHC review procedure to ensure that all dams are adequately reviewed
Signs of stress for Nebraska Dam Safety Programme
Nebraska Dam Safety Programme has almost 3,000 dams across the state to regulate, and 600 of these are currently assessed as being in poor condition.
The investigation panel said it believes that “the many responsibilities of the programme and high workload relative to the size of staff had the potential to negatively impact the quality of the programme”.
Bakers says: “There needs to be greater awareness of the excessive workload in state dam safety programmes. Each Nebraska dam-safety person had a higher number of dams they were responsible for than other dam-safety personnel across the country.
“Three thousand dams are a lot of structures to keep track of and inspect. There is a greater number of dams corresponding to a greater workload for staff.
“The panel found that there might have been signs of stress within the programme, as NebDSP was behind with updating downstream hazard classifications and completing some significant hazard potential dam inspections; there were unused training funds every year; and they had difficulty in attracting and retaining engineers to their programme.”
As a result of an inability to attract and retain engineers, NebDSP said it had been using non-engineers in the region to perform all low-hazard-potential dam inspections and about half of significant-hazard dam inspections.
The most recent 2018 inspection of Spencer Dam was performed by two non-engineers.
“I don’t believe that the industry has made any conclusive statement about whether engineers should only be doing dam inspections,” says Baker. “However, I do know that if non-engineers are used to inspect dams then the process needs to be peer reviewed and have a lot of oversight by engineers themselves.”
No major mistakes by operators
When looking into the action of the dam operators on the night that Spencer Dam failed, the investigation panel said that despite stress, fatigue and adverse working conditions, it found “no major mistakes, lapses or errors on the part of the operators”.
Indeed, despite knowledge that the dam was likely to fail soon, the operators drove to the downstream home of the resident to warn him and tell him to evacuate. Sadly, the resident did not evacuate in time and may not have understood or heeded the warning of the dam operators.
“The operators were diligent and did all they could that night,” Baker adds. “We concluded that there wasn’t a lot the operators could have done to save the dam. They were fortunate to escape with their lives.”
As the Spencer Dam Failure Investigation Report concludes: “The failure of Spencer Dam was a tragedy. The dam-safety community has a responsibility to learn from this event and prevent future failure.
“By documenting and stating the lessons learned, the panel hopes that the dam-safety industry will incorporate what the Spencer Dam failure can reveal about the status of dam-engineering practice and where the industry needs to improve.”
Baker adds: “This was a tragedy and must not happen again. Our report has received a lot of attention from the dam industry, and we will do everything we can to help learn from this event and prevent it from happening again.”
This article originally appeared in International Water Power & Dam Construction (IWP&DC) magazine